Tracking the Metabolic Pulse of Plant Lipid Production with Isotopic Labeling and Flux Analyses: Past, Present and Future

Abstract

Metabolism is comprised of networks of chemical transformations, organized into integrated biochemical pathways that are the basis of cellular operation, and function to sustain life. Metabolism, and thus life, is not static. The rate of metabolites transitioning through biochemical pathways (i.e. flux) determines cellular phenotypes, and is constantly changing in response to genetic or environmental perturbations. Each change evokes a response in metabolic pathway flow, and the quantification of fluxes under varied conditions helps to elucidate major and minor routes, and regulatory aspects of metabolism. To measure fluxes requires experimental methods that assess the movements and transformations of metabolites without creating artifacts. Isotopic labeling fills this role and is a long-standing experimental approach to identify pathways and quantify their metabolic relevance in different tissues or under different conditions. The application of labeling techniques to plant science is however far from reaching it potential. In light of advances in genetics and molecular biology that provide a means to alter metabolism, and given recent improvements in instrumentation, computational tools and available isotopes, the use of isotopic labeling to probe metabolism is becoming more and more powerful. We review the principal analytical methods for isotopic labeling with a focus on seminal studies of pathways and fluxes in lipid metabolism and carbon partitioning through central metabolism. Central carbon metabolic steps are directly linked to lipid production by serving to generate the precursors for fatty acid biosynthesis and lipid assembly. Additionally some of the ideas for labeling techniques that may be most applicable for lipid metabolism in the future were originally developed to investigate other aspects of central metabolism. We conclude by describing recent advances that will play an important future role in quantifying flux and metabolic operation in plant tissues.
Copyright © 2015. Published by Elsevier Ltd.

Full text

1
2Review
4Tracking the metabolic pulse of plant lipid production with isotopic
5labeling and flux analyses: Past, present and future
6
7
8Doug K. Allen
a,b,
⇑
, Philip D. Bates
c
, Henrik Tjellström
d,e
9
a
United States Department of Agriculture, Agricultural Research Service, 975 North Warson Road, St. Louis, MO 63132, United States
10
b
Donald Danforth Plant Science Center, 975 North Warson Road, St. Louis, MO 63132, United States
11
c
Department of Chemistry and Biochemistry, University of Southern Mississippi, Hattiesburg, MS 39406, United States
12
d
Department of Plant Biology, Michigan State University, East Lansing, MI 48824, United States
13
e
Great Lakes Bioenergy Research Center, Michigan State University, East Lansing, MI 48824, United States
14
16
article info
17 Article history:
18 Received 15 October 2014
19 Received in revised form 30 January 2015
20 Accepted 11 February 2015
21 Available online xxxx
22 Keywords:
23 Metabolic flux analysis
24 Isotopic labeling
25 Acyl editing
26 Central metabolism
27 Mass spectrometry
28
29
abstract
30
Metabolism is comprised of networks of chemical transformations, organized into integrated biochemical
31
pathways that are the basis of cellular operation, and function to sustain life. Metabolism, and thus life, is
32
not static. The rate of metabolites transitioning through biochemical pathways (i.e., flux) determines
33
cellular phenotypes, and is constantly changing in response to genetic or environmental perturbations.
34
Each change evokes a response in metabolic pathway flow, and the quantification of fluxes under varied
35
conditions helps to elucidate major and minor routes, and regulatory aspects of metabolism. To measure
36
fluxes requires experimental methods that assess the movements and transformations of metabolites
37
without creating artifacts. Isotopic labeling fills this role and is a long-standing experimental approach
38
to identify pathways and quantify their metabolic relevance in different tissues or under different condi-
39
tions. The application of labeling techniques to plant science is however far from reaching it potential. In
40
light of advances in genetics and molecular biology that provide a means to alter metabolism, and given
41
recent improvements in instrumentation, computational tools and available isotopes, the use of isotopic
42
labeling to probe metabolism is becoming more and more powerful. We review the principal analytical
43
methods for isotopic labeling with a focus on seminal studies of pathways and fluxes in lipid metabolism
44
and carbon partitioning through central metabolism. Central carbon metabolic steps are directly linked to
45
lipid production by serving to generate the precursors for fatty acid biosynthesis and lipid assembly.
46
Additionally some of the ideas for labeling techniques that may be most applicable for lipid metabolism
47
in the future were originally developed to investigate other aspects of central metabolism. We conclude
48
by describing recent advances that will play an important future role in quantifying flux and metabolic
49
operation in plant tissues.
50
Ó2015 Published by Elsevier Ltd.
51
52
53
54
Contents
55
1. Introduction . . . ....................................................................................................... 00
56
1.1. Network fluxes define cellular phenotype and are measured with isotope labeling . . . . . . . .................................... 00
57
1.2. Cellular roles for lipids in metabolism are diverse and complex . . . . ....................................................... 00
58
1.3. The scope of opportunities for lipid production and scientific discovery . . . . . . . . . . . . . . . . .................................... 00
59
2. General considerations. . . . . . . . . . . . . . .................................................................................... 00
60
2.1. Elemental isotopes used in biochemical studies . . . . . . . . . . . . . . . . . ....................................................... 00
61
2.2. Measuring isotopic enrichment with MS and NMR . . . . . . . . . . . . . . ....................................................... 00
62
2.3. Steady state and transient labeling . . . . . . . . .......................................................................... 00
63
3. Pioneering investigations in lipid labeling to modern day . . . . ................................................................. 00
64
3.1. Enzymatic steps elucidated by tracking isotopes . . . . . . . . . . . . . . . . ....................................................... 00
http://dx.doi.org/10.1016/j.plipres.2015.02.002
0163-7827/Ó2015 Published by Elsevier Ltd.
⇑Corresponding author at: United States Department of Agriculture, Agricultural Research Service, 975 North Warson Road, St. Louis, MO 63132, United States.
E-mail address: doug.allen@ars.usda.gov (D.K. Allen).
Progress in Lipid Research xxx (2015) xxx–xxx
Contents lists available at ScienceDirect
Progress in Lipid Research
journal homepage: www.elsevier.com/locate/plipres
JPLR 873 No. of Pages 24, Model 5G
16 March 2015
Please cite this article in press as: Allen DK et al. Tracking the metabolic pulse of plant lipid production with isotopic labeling and flux analyses: Past, pre-
sent and future. Prog Lipid Res (2015), http://dx.doi.org/10.1016/j.plipres.2015.02.002

65
3.1.1. Elucidation of lipid assembly and the Kennedy pathway . . . . . . . . . . ............................................... 00
66
3.1.2. Desaturation substrates and pathways . . . . . . .................................................................. 00
67
3.1.3. Acyl chain formation and trafficking between the chloroplast and ER in leaves . . . . . . . . . . . ............................ 00
68
3.1.4. Metabolic labeling reveals alternative fluxes of acyl groups into membrane lipids and TAG . ............................ 00
69
4. Extending isotopic labeling interpretation . . . . . . . . . . . ....................................................................... 00
70
4.1. Computational metabolic flux analysis using stable isotopes ............................................................. 00
71
4.1.1. Central carbon metabolism supplies carbon, energy and reducing equivalents for FA biosynthesis . . . . . . . . . . . . . . ......... 00
72
4.2. Assessing resource partitioning through perturbation-response experiments . . . . . . .......................................... 00
73
4.2.1. Alterations in carbon and nitrogen supply to fatty acids . . . . . . . . . . . ............................................... 00
74
4.2.2. Altered flux phenotypes in lipid mutants . . . . .................................................................. 00
75
4.2.3. Oil with modified FA composition. . . . . . . . . . .................................................................. 00
76
4.2.4. Assessing regulation with metabolic control analysis . . . . . . . . . . . . . ............................................... 00
77
4.3. Future techniques and technology to address longstanding challenges . . . . . . . . . . . .......................................... 00
78
4.3.1. Developments in MS technology to enable labeling and flux studies . ............................................... 00
79
4.3.2. Addressing the challenges of multicellular eukaryotic metabolism . . ............................................... 00
80
4.3.3. From CO
2
to lipid: temporal labeling-based MFA approaches . . . . . . . ............................................... 00
81
5. Conclusions and perspective . . . .......................................................................................... 00
82
Conflicts of interest . . . . . . . . . . . . .......................................................................................... 00
83
Acknowledgements . . . . . . . . . . .......................................................................................... 00
84
References . .......................................................................................................... 00
85
86
87
1. Introduction
88
1.1. Network fluxes define cellular phenotype and are measured with
89
isotope labeling
90
Plant cellular function is defined by networks of enzymatic
91
reactions with substrates and products that are linked by mass
92
and energy balances. Through developmental or environmental
93
cues the expression of genes change the operating network and
94
underscore the dynamic nature of metabolism. Fluxes establish
95
the products of metabolism and can be measured through the rate
96
of accumulation for end products (i.e., compounds that are not
97
turned over such as storage protein or oil in a seed). However at
98
metabolic steady state cellular intermediates do not accumulate
99
and are both produced and consumed at rates that cannot be read-
100
ily inferred from metabolite concentrations alone (i.e., metabolo-
101
mics). Thus as a metabolic attribute, fluxes must be assessed
102
through other means. Isotopes can serve as ‘‘tracers’’ that describe
103
the rate of conversion between pools over time, by providing a
104
change in metabolite molecular mass that occurs without
105
perturbing metabolism. Therefore the isotopic ‘‘labeling’’ indicates
106
formation and turnover through movement of an isotope from one
107
metabolite to the next, leading to flux descriptions that are impor-
108
tant to studies on metabolic operation, regulation and control [1].
109
1.2. Cellular roles for lipids in metabolism are diverse and complex
110
In biology few types of molecules serve as many diverse roles or
111
are as poorly characterized as lipids. The Arabidopsis genome con-
112
tains over 600 genes annotated with functions putatively tied to
113
lipid and/or fatty acid catabolism and anabolism, however less
114
than half of the genes have been characterized with in vivo studies
115
of mutants that demonstrate a clear role in lipid metabolism [2,3].
116
In the most rigorously studied pathways many enzymes have
117
multiple isoforms that conduct identical or very similar biochemi-
118
cal reactions, but reflect specialized cellular, subcellular or
119
developmental activities for different aspects of lipid metabolism.
120
Whereas some mechanisms of lipid metabolism are conserved
121
between organisms, the subcellular descriptions of biosynthesis
122
and degradation in plants are distinct from other species [4].
123
Thus textbook descriptions of lipid metabolism are not universal,
124
and the operational network of metabolic reactions can vary
125
between species, tissues, and cells; as well as across
126
developmental and environmental conditions. Fluxes change to
127
accommodate the different cellular demands for lipid production
128
(e.g., membrane, surface, or storage lipids), as well as turnover
129
(e.g., storage oil breakdown during germination) necessary to
130
produce other metabolic precursors, energy, and to maintain
131
homeostasis [5,6].
132
Lipids comprise membranes that are a defining feature of cell
133
biology. Membrane lipids separate cells from their environment
134
(i.e., plasma membranes) and establish subcellular organelles that
135
compartmentalize metabolism in eukaryotes. Layered on surfaces,
136
cutin and suberin provide a resistive, protective barrier to natural
137
elements, while other lipids perform signaling functions or serve
138
in an energy storage capacity. Triacylglycerols (TAG) in the seeds
139
(e.g., soybean, rapeseed, sunflower) or fruit (e.g., olive, avocado,
140
palm) of many plants are an energy dense storage form of biomass.
141
Apart from pericarp tissues, stored oils are remobilized at ger-
142
mination providing carbon and ATP for plant growth until auto-
143
trophic metabolism can be sustained [7]. The biochemistry and
144
genetics of storage lipid accumulation have received extensive
145
attention, and this area of research remains an important focus
146
in part because acyl chains are one of the most highly reduced
147
forms of carbon (i.e., approximately twice the energy content per
148
gram of dry weight than carbohydrates or storage protein) and
149
can be used to supplant non-renewable petroleum in many appli-
150
cations. Our dependence on TAGs for food, fuel and chemicals con-
151
tributes to an industry currently estimated at over $120 billion per
152
year (http://lipidlibrary.aocs.org/market/prices.htm). Despite the
153
many roles for lipids and intense research in plant lipid biochem-
154
istry and genetics over the past half century, significant gaps in
155
our understanding remain, foremost among these are the identity
156
and in vivo function of genes and enzymatic reaction networks
157
involved in lipid metabolism [2,3].
158
1.3. The scope of opportunities for lipid production and scientific
159
discovery
160
Given that oil content varies in plants from less than 1% of dry
161
weight (e.g., lentils, potatoes) to approximately 70% (pecans, wal-
162
nuts) and can exceed 88% in mesocarps (e.g., palm); there exists
163
a large difference in metabolic operation among various cells that
164
is an inviting prospect for engineering increased oil accumulation
165
in plants. Tissues such as leaves are usually less than 5% lipid, how-
166
ever the abundance of leafy biomass vs. seed/mesocarp tissue in
167
most plants has led to significant recent efforts to engineer lipid
168
accumulation into vegetative tissues for biofuels [8–14]. To date
2D.K. Allen et al. / Progress in Lipid Research xxx (2015) xxx–xxx
JPLR 873 No. of Pages 24, Model 5G
16 March 2015
Please cite this article in press as: Allen DK et al. Tracking the metabolic pulse of plant lipid production with isotopic labeling and flux analyses: Past, pre-
sent and future. Prog Lipid Res (2015), http://dx.doi.org/10.1016/j.plipres.2015.02.002

169
there have been a few examples of significantly altered oil levels in
170
crop plants [15–17], and other engineering efforts that demon-
171
strate the potential to change fatty acid composition for health or
172
industrial applications, though mechanisms that control FA syn-
173
thesis, FA modification, and accumulation in plant tissues are still
174
not well understood [18–20]. Future progress will require a more
175
complete understanding of central carbon metabolism including
176
further annotation of genes, networks and enzyme function, reg-
177
ulation of central carbon flux with FA biosynthesis, and the
178
coordination of FA modification with lipid assembly and turnover
179
to produce more oil with specific FA compositions. Advances in
180
genetics and omic-level analyses could be complemented by iso-
181
topic labeling of targeted pathways as well as larger network flux
182
analyses; providing the rationale to revisit what has been learned
183
through prior isotope investigations.
184
In this review we first provide a description of the analytical
185
methods including a description of isotopes, instrumentation,
186
and different types of experiments. Next some of the seminal dis-
187
coveries in metabolism that were established through labeling
188
analyses with substrates commonly used to explore lipids are
189
reviewed. After some general considerations on computational
190
metabolic flux analysis, we describe what insights about regulation
191
and control have been learned from experiments that perturbed
192
metabolism. Finally, the most recent developments in the applica-
193
tion of isotope labeling, novel instrumentation and flux analysis
194
strategies are described that represent a significant opportunity
195
for future explorations in lipid metabolism and that aim to define
196
plant function at the systems level (e.g., the coordinated auto-,
197
mixo- and heterotrophic carbon metabolism and regulation across
198
cells and tissues, Fig. 1).
Fig. 1. Plants operate at a systems level with different cellular metabolic activities and biomass compositions in different tissues. Simplified networks of metabolism are
shown. Leaves that are primarily autotrophic assimilate CO
2
through the Calvin cycle (green arrows) and glycolytic enzymes (red arrows) to make organic compounds, mostly
sucrose, with energy from sunlight. Seeds can operate hetero- or mixotrophically converting sugars and amino acids into storage reserves that can include significant
amounts of storage lipid. Seeds may have duplicated pentose phosphate (blue arrows) and glycolytic pathway activities along with TCA cycle (purple arrows) and significant
amino acid biosynthetic flux necessary to make storage reserves. In some instances green seeds also possess the capacity to reassimilate respired CO
2
. Roots receive carbon,
predominantly as sucrose from other plant tissues and function heterotrophically with significant TCA cycle activities and oxidative pentose phosphate metabolism to
assimilate nitrogen and meet cellular demands. Together the cellular activities characterize the plant as a system. (Abbreviations: ADPG, adenosine diphosphoglucose; AKG,
alpha-ketoglutarate; AA, amino acid; E4P, erythrose-4-phosphate; F6P, fructose-6-phosphate; G6P, glucose-6-phosphate; HP, hexose phosphate; N, nitrogen; OAA,
oxaloacetate; PYR, pyruvate; R5P, ribose-5-phosphate; RuBP, ribulose 1,5-bisphosphate; S7P, sedoheptulose-7-phosphate; TCA, tricarboxylic acid cycle; TP, triose phosphate;
UDPG, uridine diphosphoglucose; 3-PGA, 3-phosphoglyceric acid).
D.K. Allen et al. / Progress in Lipid Research xxx (2015) xxx–xxx 3
JPLR 873 No. of Pages 24, Model 5G
16 March 2015
Please cite this article in press as: Allen DK et al. Tracking the metabolic pulse of plant lipid production with isotopic labeling and flux analyses: Past, pre-
sent and future. Prog Lipid Res (2015), http://dx.doi.org/10.1016/j.plipres.2015.02.002

199
2. General considerations
200
2.1. Elemental isotopes used in biochemical studies
201
Isotopes of low natural abundance can serve as tracers and are
202
used to describe metabolic pathways, enzymatic reaction mecha-
203
nisms and movement of atoms within plant cells. The isotopes
204
for an element contain differing numbers of neutrons resulting in
205
unique masses. Isotopes can be stable or radioactive based upon
206
their half-lives, requiring different detection techniques and con-
207
tributing to their use in scientific explorations. In most labeling
208
experiments, a portion of the most abundant isotope of an element
209
(e.g.,
12
C,
16
O,
1
H,
14
N) is substituted with either a different stable
210
(
13
C,
2
H,
15
N,
17
Oor
18
O), or long-lived radioactive (
14
C,
3
H,
35
S,
211
32
P,
33
P) form. Other radioactive tracers with very short half-lives
212
(e.g.,
11
C, half-life 20 min relative to
14
C, half-life 5730 years)
213
can be synthesized just prior to use and assessed by positron emis-
214
sion tomography (PET) analysis [21–23]. Such short-lived radioiso-
215
tope investigations are more specialized and will not be further
216
considered though their application in photosynthesis, long
217
distance transport and other plant processes are becoming more
218
common [22,24–26].
219
The type of information obtained from stable or radioactive iso-
220
topic enrichment is distinct. Stable isotopic enrichment can be
221
assessed within individual atoms of the molecule (e.g., using
222
NMR), whereas radioactivity generally describes the overall label-
223
ing of the molecule unless additional techniques (e.g., a lipolytic
224
cleavage) are employed. Assaying the incorporation of radioisotope
225
into biomass does not require prior separation of metabolites per
226
se and sensitive detection of radioactivity through liquid scin-
227
tillation counting or autoradiography is not influenced by the pres-
228
ence of unlabeled metabolites. This can be of benefit because the
229
kinetics of labeling can be established using very low ‘tracer’
230
amounts of radioisotope for short labeling durations (a few sec-
231
onds to a few hours) presuming that the endogenous substrate
232
pool labels quickly (i.e., little to no lag time). By definition isotopes
233
contain the same number of protons per atom and maintain very
234
similar biophysical properties; allowing investigations to probe
235
metabolism with minimal perturbation from label introduction.
236
Nonetheless, care is necessary to avoid enzymatic discrimination
237
between labeled and unlabeled elements, (i.e., isotope discrim-
238
ination or kinetic isotope effects) which is well-known and can
239
lead to artifacts [27–30]. In some cases these same distinctions
240
have been used to constructively characterize physiology as
241
described elsewhere [31,32].
242
With stable isotopes there is less of an enzymatic preference
243
[33], and in addition stable isotopes do not require special handling
244
precautions. However, possibly the greatest benefit is that the
245
analysis of stable isotopes by mass spectrometry and NMR can pro-
246
vide additional positional and compositional information at the
247
atom and bond levels of metabolites without laborious prepara-
248
tions. The description of isotopic composition for a compound is
249
referred to as the set of isotopologues whereas isomers with
250
equivalent numbers of each isotopic atom but at different positions
251
in the molecule are referred to as isotopomers (i.e., isotopic iso-
252
mers; [34]). The terms ‘‘mass’’ or ‘‘positional isotopomers’’ are often
253
used to describe molecule composition that varies by the number of
Fig. 2. NMR and mass spectra. A hypothetical set of NMR and mass spectra from
analysis of a three carbon molecule that contains unlabeled (12-carbon; open
circles) and isotopically labeled (13-carbon; closed circles) in various combinations
known as isotopomers. For a 3 carbon molecule there can be up to 2
3
=8
isotopomer labeling descriptions because each of the 3 carbons can be either
labeled or unlabeled. The bar graph is a typical description provided by mass
spectrometry, where the intensity of the [M]
+
to [M+3]
+
peaks represent the relative
abundance of labeled species within the population for the compound. The isotopic
composition is measured based upon mass and does not resolve the location of the
isotope within the metabolite, rather the amount of labeling is distinguished only
by the differences in weight detected through mass spectrometry, thus ‘‘mass
isotopomers’’. For the three carbon molecule there are four mass isotopomers. The
interpretation of NMR spectra is often more complex and indicates the amount of
labeling within a particular position in the molecule as well as its relation to other
labeled molecules within the carbon backbone. In carbon-13 spectra the higher
degree of labeled carbons bonded to each other creates more complex peak splitting
patterns, thus the pattern of peaks for carbon 2 (C2) is more complicated because it
can be connected to zero, one or two
13
C atoms. Each connected
13
C atom resulting
in splitting of the peaks. NMR spectra provide ‘‘positional isotopomer’’ information
indicating the enrichment of a particular atom and also describe bond-con-
nectedness to other
13
C atoms.
Fig. 3. Transient and steady state isotopic (e.g.,
13
C) incorporation from pulse or
pulse-chase labeling experiments. Isotopic labeling experiments display the
incorporation into or dilution of isotope from metabolites that accompanies the
metabolic operation of the biological system. (A) Providing a continuous source or
‘‘pulse’’ of
13
C to metabolism results in an increasing amount of isotopic labeling in
metabolites that eventually approaches a steady state value. If the source of carbon
is completely
13
C, the metabolite will eventually approach 100% labeled, presuming
it is an intermediate that is completely turned over. (B) In a pulse-chase experiment
the isotope is provided for only a limited duration resulting in labeling for a period
of time followed by dilution when the source of carbon is switched from
13
Cto
12
C.
4D.K. Allen et al. / Progress in Lipid Research xxx (2015) xxx–xxx
JPLR 873 No. of Pages 24, Model 5G
16 March 2015
Please cite this article in press as: Allen DK et al. Tracking the metabolic pulse of plant lipid production with isotopic labeling and flux analyses: Past, pre-
sent and future. Prog Lipid Res (2015), http://dx.doi.org/10.1016/j.plipres.2015.02.002

254
each isotope (i.e., causing changes in mass) or that varies by the
255
location within the molecular structure (i.e., position). The analysis
256
of relative labeling with stable isotopes within a molecule was
257
coined Mass Isotopomer Distribution Analysis (MIDA) by
258
Hellerstein and colleagues [35]. Along with Kelleher and
259
Masterson [36], these authors established frameworks to describe
260
FA biosynthesis and applied the concepts with [
13
C]glucose,
261
[
13
C]acetate, and
2
H
2
O experiments to inspect aspects of metabo-
262
lism [37,38]. As methods to correct for natural abundance levels
263
of isotopes were developed [39,40], the labeling description
264
became highly quantitative and well-positioned for computational
265
modeling developments such as metabolic flux analysis [MFA, see
266
Section 4.1;[41–43]] that is frequently used to study aspects of
267
central metabolism.
268
2.2. Measuring isotopic enrichment with MS and NMR
269
Mass spectrometry (MS) and nuclear magnetic resonance
270
(NMR) are complementary techniques to assess stable isotope
271
incorporation (Fig. 2). Mass spectrometry ‘‘weighs’’ the composi-
272
tion of isotopes that result in altered mass. Through interpretation
273
of fragmentation products, MS can assess the degree of labeling in
274
molecular substituents and in theory the positional location of the
275
isotope within a molecule. NMR is an atomic property that distin-
276
guishes isotopes with non-zero nuclear magnetic moments and is
277
well-suited for positional labeling analysis. General principles of
278
NMR spectroscopy emphasize its versatility to assess metabolism
279
in entire tissues or through the analysis of extracted metabolites
280
[44–50]. NMR requires more material, additional transient scans,
281
reduced temperature or special tubes to accommodate the reduced
282
sensitivity of instruments. Multi-dimensional NMR experiments or
283
modification of the solution or metabolites chemically are also
284
common techniques used to improve resolution and sensitivity
285
[51–56]. Though the extra effort and lower throughput of NMR
286
has limited its use to more specialized investigations, quan-
287
tification of positional enrichment with NMR can lead to more
288
intuitive assessments of metabolic pathways (relative to mass iso-
289
topomer descriptions) that may be especially important when the
290
network is not well-characterized a priori. Recent advances in NMR
291
are discussed further elsewhere [49,53].
292
2.3. Steady state and transient labeling
293
Isotope studies take one of several forms based on whether
294
steady state, pulse, or pulse-chase experiments will be most infor-
295
mative. Steady state labeling analyses imply that the isotope is
296
provided in excess for sufficient time (usually minutes to hours)
297
to achieve an unchanging description of labeling in metabolic
298
intermediates (Fig. 3). In developing seeds or cell-culture suspen-
299
sions metabolism operates at a pseudo steady state during periods
300
of development, therefore isotopically labeling these tissues with a
301
mixture of labeled and unlabeled substrates will result in redis-
302
tribution of the isotope that reflects pathway use and flux. Such
303
studies have been regularly performed with metabolic flux analysis
304
to document the partitioning of carbon, energy and reducing
305
equivalents to oil, protein and carbohydrate in seeds or cell cul-
306
tures [57–64]. However steady state labeling analyses are limited
307
to assessments of pathways that enzymatically rearrange the dis-
308
tribution of isotope. For example, in central metabolism of devel-
309
oping seeds the provision of [
13
C]labeled glucose will result in
310
fructose-6-phosphate (F6P) that is labeled at different positions
311
depending on the degree of reversibility of glycolysis and the use
312
of the oxidative pentose pathway (Fig. 4A). The labeling in F6P
313
can be measured directly with LC-MS/MS or inferred from mea-
314
surements of plant storage products such as starch and cell wall.
315
More challenging is the study of autotrophic metabolism which
316
depends upon CO
2
as a source of carbon. CO
2
does not contain car-
317
bon–carbon bonds that can be rearranged and the incorporation of
318
[
13
C]O
2
will lead to complete enrichment of metabolites at steady
319
state which are no more informative for pathway flux than the ini-
320
tial unlabeled description (Fig. 4B, [65]). Thus the study of leaves
321
and other autotrophic cells may be best suited to transient or ‘‘non-
322
stationary’’ isotopic analysis [66–70]. Pathways that have limited
323
carbon bond rearrangements or network branching such as spe-
324
cialized metabolite production or pathways based on a single pre-
325
cursor like fatty acid biosynthesis that utilizes acetyl-CoA units
326
have typically been investigated utilizing the kinetic incorporation
327
of radioisotopes over time (i.e., pulse labeling at metabolic steady
Fig. 4. Isotope distribution in heterotrophic and autotrophic pathways. Bond
breaking and reforming reactions of metabolism alter the metabolic labeling of
metabolites. (A) Provision of both unlabeled (open circles) and isotopically labeled
(closed circles) glucose to cells utilizing the oxidative pentose phosphate pathway
(OPPP) and glycolytic steps will result in different labeling descriptions in
intermediates such as fructose-6-phosphate as well as derived end products such
as starch or cell wall. (B) In autotrophic metabolism the provision of isotopically
labeled carbon as [
13
C]O
2
results in complete labeling of metabolites over time,
therefore information on the fluxes through metabolism must be obtained by
measuring the transient incorporation of isotopes.
ABP
Time
Fig. 5. Labeling kinetics in a linear metabolic pathway. Labeling of network
intermediates in series from a continuous pulse of isotope resulting in intermedi-
ates that approach a steady state labeling value and a product pool that
accumulates label over time throughout the experimental duration.
D.K. Allen et al. / Progress in Lipid Research xxx (2015) xxx–xxx 5
JPLR 873 No. of Pages 24, Model 5G
16 March 2015
Please cite this article in press as: Allen DK et al. Tracking the metabolic pulse of plant lipid production with isotopic labeling and flux analyses: Past, pre-
sent and future. Prog Lipid Res (2015), http://dx.doi.org/10.1016/j.plipres.2015.02.002

328
state as described below). However these pathways may also be
329
prime candidates for nonstationary metabolic flux analysis with
330
stable isotopes in the future (see Section 4).
331
The incorporation of isotopes over time provides a transient,
332
dynamic description of pathway activity and is referred to as a
333
pulse labeling experiment (Fig. 3A). If the experiment is extended
334
after the labeled substrate has been removed and the change in
335
label accumulation or depletion in metabolites is measured during
336
this ‘‘chase’’ period, then the experiment is referred to as a ‘‘pulse-
337
chase’’ experiment (Fig. 3B). Pulse and pulse-chase experiments
338
utilize the kinetics of isotopic labeling over time to determine
339
the precursor–product relationships of intermediates within a
340
metabolic pathway or network [71]. In a pulse experiment the
341
individual pools become labeled in the order which they are gener-
342
ated by metabolism (Fig. 5). Assuming there are no diffusion or
343
uptake issues for the substrate, the lag time for labeling in the first
344
metabolite will be very small and the incorporation of label over
345
time into the first metabolite pool will reflect a hyperbolic pattern.
346
If diffusion is an issue the curve will exhibit a lag period and
347
become sigmoidal. The first metabolite in a sequence incorporates
348
label at a linear rate but then starts to level off as steady state
349
enrichment is achieved. The metabolic intermediates downstream
350
will exhibit sigmoidal curves with an initial lag period followed by
351
more rapid linear labeling then approach a maximal value. The lag
352
reflects the time needed to fill precursor pools. Products of meta-
353
bolism can lag significantly because in the network they are often
354
far from the source of label. As products are not turned over into
355
other pools they will continue to accumulate label during the
356
course of the experiment (Fig. 5). Thus, a pulse labeling experiment
357
with short time points (typically seconds, minutes, to only a few
358
hours) can be used to assess the labeling of each intermediate in
359
the pathway and establish network flow. Following the labeling
360
with a chase period (i.e., pulse-chase experiment) enables the
361
tracking of isotopic enrichment as well as dilution and is
362
potentially more informative because reactions that create as well
363
as consume a pool can both be elucidated. Though conceptually
364
straightforward, the movement through intermediates can be
365
challenging to establish because many metabolic steps operate
366
reversibly with forward and backward reaction rates near-
367
equilibrium. As a consequence isotope flow between successive
368
intermediates can exceed the net flux of molecules transferred
369
from one pool to the next. Thus as with any experiment the
370
information obtained will reflect the biological system and its
371
measureable attributes.
372
Together these considerations contribute to the design of the
373
experimental labeling strategy that minimally addresses: (a) the
374
time scales of metabolism (i.e., both rate of metabolism, and
375
the practical experimental duration including the viability of
376
excised tissue during incubation), (b) the pathway reactions
377
including atoms and bonds that are altered, (c) the sufficient
378
uptake of the labeled substrate and its impact on metabolism,
379
and (d) the capacity to measure label incorporation in relevant
380
metabolites. Given the number of considerations and the time
381
and resources necessary to do labeling studies, in silico methods
Fig. 6. Label introduction into central carbon and fatty acid biosynthetic pathways for lipid analysis. Isotopes can be supplied through substrates that are involved in different
bond breaking and reforming reactions and therefore provide complementary information about network operation and pathway flux. A number of commonly used
exogenous isotopically labeled sources are indicated by green boxes. Sugars and CO
2
are incorporated through production of sugar phosphates by the Calvin cycle, pentose
phosphate pathways, and glycolysis and are therefore augmented significantly in the production of acetyl-CoA. Labeled acetate is frequently used to assess fatty acid and lipid
metabolism because it is directly converted to acetyl-CoA and is cheap relative to other labeled substrates. Fatty acid biosynthesis repeatedly adds an acetyl group to the acyl-
ACP chain until fatty acids are 16–18 carbons long (i.e., for long chain fatty acids). The acyl chains are then released from ACP and used for lipid production that takes place in
the plastid or the endoplasmic reticulum. Closed circles with numbers refer to specific enzymatic steps including: plastidic pyruvate kinase and medium chain acyl-ACP
thioesterase that are specifically referenced in text.
6D.K. Allen et al. / Progress in Lipid Research xxx (2015) xxx–xxx
JPLR 873 No. of Pages 24, Model 5G
16 March 2015
Please cite this article in press as: Allen DK et al. Tracking the metabolic pulse of plant lipid production with isotopic labeling and flux analyses: Past, pre-
sent and future. Prog Lipid Res (2015), http://dx.doi.org/10.1016/j.plipres.2015.02.002

382
that can optimize the experimental design have received con-
383
siderable attention [72–79].
384
3. Pioneering investigations in lipid labeling to modern day
385
3.1. Enzymatic steps elucidated by tracking isotopes
386
Before the advent of modern day molecular biology techniques,
387
isotopic labeling studies were used to elucidate biochemical path-
388
ways including substrates and products of individual reactions. In
389
the 1940’s Ruben and Kamen established that water not carbon
390
dioxide is the source of evolved oxygen using H
2
[
18
O] [80]. The dis-
391
covery and implementation of carbon-14 by the same investigators
392
revealed that the ‘dark’ reactions of carbon fixation need not take
393
place in light [81] and provided inspiration for the work of
394
Benson and Calvin that eventually led to descriptions of the
395
Calvin-Benson cycle [82] as well as C4 metabolism [83]. The early
396
examples made clear the value offered by tracers, but method-
397
ological complexities and the laborious nature of the analytical
398
biochemistry, along with the limited access to tracers suppressed
399
a more widespread application. Cholesterol and fatty acid mea-
400
surements based on
2
H incorporation, for example [84,85],
401
involved tedious measurements of the density of heavy water after
402
combustion of organic compounds. Therefore radioactivity-based
403
approaches gained favor because they could be used with routine
404
lipid extraction methods [86] and could be sensitively measured.
405
Fatty acid biosynthesis and lipid metabolism have been studied
406
extensively using labeled acetate [reviewed in [87,88]]. While not
407
the endogenous substrate for fatty acid metabolism [89], acetate
408
is readily taken up and incorporated into both the cytosolic and
409
plastidic acetyl-CoA pools and subsequently utilized to make
410
newly synthesized fatty acids [90,91] and also track the incorpora-
411
tion of newly synthesized FAs into glycerolipids [e.g., [92,93]].
412
Inorganic carbon [e.g., [89,91,93,94]] isotopically enriched water
413
[e.g., [95]], glycerol [e.g., [96,97]], fatty acids [e.g., [98–101]]as
414
well as other substrates [e.g., [27,97,102–104]] have all been used
415
to probe lipid metabolism depending upon the tissue or reaction(s)
416
of interest (Fig. 6). The selection of substrates remains a critical
417
aspect to experimental design but was particularly important in
418
early studies that required biosynthesis of the radiolabeled sub-
419
strate from inorganic compounds (e.g., [
32
P]inorganic phosphate
420
or [
14
C]bicarbonate) followed by extensive purification steps prior
421
to the labeling experiment.
422
3.1.1. Elucidation of lipid assembly and the Kennedy pathway
423
The use of radioactivity with microsomal preparations and cell
424
free extracts has provided descriptions of the in vitro enzymatic
425
esterification of FA to make membrane and storage glycerolipids,
426
and was used to elucidate the Kennedy pathway (Fig. 7A;
427
[105,106]) comprising a series of four enzymatic steps. The process
428
involves two consecutive acyl-CoA dependent acylations of G3P by
429
sn-1 glycerol-3-phosphate acyltransferase (GPAT) and lysophos-
430
phatidic acid acyltransferase (LPAAT), producing LPA and PA,
431
respectively. Subsequently, the phosphate is removed by phospha-
432
tidic acid phosphatase (PAP) producing de novo assembled DAG
433
that can be further acylated by the acyl-CoA dependent diacylglyc-
434
erol acyltransferase (DGAT) to produce TAG. Phospholipid and TAG
435
biosynthesis from glycerol-3-phosphate (Fig. 7) were demon-
436
strated utilizing [
32
P]glycerol-3-phosphate and [
14
C]palmitate or
437
oleoyl-CoA with microsomes and cell free extracts from animal liv-
438
ers [107–109] as well as plant tissues [106,110]. Pertaining to the
Fig. 7. Network descriptions and precursor–product relationships for lipid assembly. (A) The Kennedy pathway with linear incorporation of acyl chains into glycerol-
backbone to generate TAG. (B) Additional mechanism of acyl editing that exchanges FA between the acyl-CoA pool and PC prior to TAG biosynthesis by the Kennedy pathway.
(C) Acyl editing with multiple DAG pools that further emphasize PC involvement in TAG production that may not involve the Kennedy pathway. The precursor–product
panels to the right of each figure indicate the different qualitative labeling patterns for each of three lipid groups. The patterns are dependent on the network relationships
including the pathway sequence to TAG production. DAG labeling from [
14
C]acetate prior to PC or TAG is indicative of de novo glycerolipid biosynthesis, whereas acyl editing
results in faster labeling of PC relative to DAG and TAG when [
14
C]acetate is provided. If the production of TAG requires PC as an intermediate between two DAG pools (shown
in red) then the [
14
C]glycerol labeling will result in more complicated labeling trajectories. The metabolic labeling patterns are subject to the number and size of individual
pools, net and reversible fluxes and the network that splits and joins reactions for metabolites. Therefore the panels can vary extensively from the most basic description of a
precursor–product relationship presented in Fig. 5. (Abbreviations: G3P, glycerol-3-phosphate; LPA, lyso-phosphatidic acid; PA, phosphatidic acid; DAG, diacylglycerol; TAG,
triacylglycerol; PC, phosphatidylcholine; FAS, fatty acid biosynthesis; DGAT, acyl-CoA:diacylglycerol acyltransferase; PDAT, phospholipid:diacylglycerol acyltransferase; CPT,
CDP-choline:diacylglycerol cholinephosphotransferase; PDCT, phosphatydlcholine:diacylglycerol cholinephosphotransferase; rCPT, reverse-CPT; PLD, phospholipase D; PAP,
phosphatidic acid phosphatase; PLC, phospholipase C; dpm, disintegrations per minute).
D.K. Allen et al. / Progress in Lipid Research xxx (2015) xxx–xxx 7
JPLR 873 No. of Pages 24, Model 5G
16 March 2015
Please cite this article in press as: Allen DK et al. Tracking the metabolic pulse of plant lipid production with isotopic labeling and flux analyses: Past, pre-
sent and future. Prog Lipid Res (2015), http://dx.doi.org/10.1016/j.plipres.2015.02.002

439
preferred acyl substrate, palmitoyl-CoA, synthesized chemically,
440
could be used in place of ATP, CoA and fatty acids to esterify G3P
441
providing direct proof that acyl-CoA is used in lipid assembly. PC
442
biosynthesis from phosphocholine was similarly outlined using
443
[
14
C]- and [
32
P]phosphocholine [109], however repeating some of
444
the basic isotope labeling and dilution studies in plants proved
445
challenging [111,112] in part because little was known at the time
446
about the organelle location of lipid production. Eventually a
447
number of aspects of the Kennedy pathway that were analogous
448
to mammalian tissues were confirmed [97,106,113,114]. More
449
recent studies have leveraged commercially available labeled
450
substrates with or without [115,116], the aid of modern genetics,
451
genetic engineering, and homology-based gene identification to
452
demonstrate enzyme function and activity in cell extracts,
453
microsomes and transgenic plants [116–121]. Regular discovery
454
of new enzymatic reactions [122,123] indicates that not all the
455
enzymatic steps (and their associated genes) within lipid
456
metabolism have been elucidated. Differences among species
457
emphasize the role of in vitro kinetic activity measurements in
458
isolated microsomes to provide considerable insight for network
459
elucidation and operation [124]. Thus efforts to identify and
460
describe the ‘‘parts list’’ and the range of operation of enzymes in
461
biochemical networks are critical to models that link enzymatic
462
steps by precursor–product labeling relationships and describe
463
coordinated network operation.
464
3.1.2. Desaturation substrates and pathways
465
Plant lipid metabolism differs significantly from mammalian
466
systems in several aspects including distinct substrates for desat-
467
uration and multiple subcellular locations for lipid assembly that
468
were elucidated by a combination of in vitro and in vivo labeling
469
experiments. Seminal in vitro studies with [
14
C]palmitate [125]
470
and in vivo pulse chase studies with [
14
C]sucrose [126] or [
14
C]O
2
471
[127] first described the mechanisms for desaturation in plants.
472
Oleic acid was labeled rapidly followed by linoleic and linolenic
473
fatty acids; however during the chase the linoleic and linolenic
474
acids continued to increase at a rate that was elevated relative to
475
the oleic acid indicating that oleate was a precursor for desat-
476
uration to linoleic and linolenic acid [126,128]. Unlike animal sys-
477
tems that utilize acyl-CoA substrates for desaturation, in plants
478
most FAs are bound to plastidic or ER membrane lipids for this pro-
479
cess [129–134] with the exception of stearoyl-ACP desaturation.
480
Early work with safflower labeling showed [
14
C]oleoyl-CoA was
481
esterified to PC leading to [
14
C]linoleoyl-PC but without the forma-
482
tion of detectable [
14
C]linoleoyl-CoA [129] supporting PC-based
483
modification. Desaturation of acyl-CoA’s directly in plants was
Fig. 8. Relationships between plastid- and er-based fatty acid biosynthesis and lipid assembly. Acyl chains are produced on an ACP backbone through fatty acid biosynthesis
(FAS) in the plastid where the acyl-ACP can be utilized for prokaryotic membrane lipid assembly in the plastid, or hydrolyzed to fatty acids before attachment to coenzyme A
(CoA) by long chain acyl-CoA synthetase (LACS) outside of the plastid. The transport mechanisms to and from the ER are unknown though studies suggest a lipid such as PC,
LPC PA, or DAG may be involved, possibly through a channeling mechanism. LPCAT association with the plastid envelope [152] supports the involvement of PC and possibly is
directly tied to acyl editing. Acyl groups are processed with acyl editing and modified on PC for the production of lipids in the ER. Others have suggested that LPCAT in the
chloroplast envelope supports a mechanism for LPC transport in both directions that need not be protein mediated [163,335]. LPC can partition into membranes as well as the
aqueous phase, however multiple studies focused on the return of lipids to the plastid for MGDG and DGDG production implicate trigalactosyldiacylglycerol (TGD) protein
mutants in the transport process [145,156]. At least PA, PC, DAG, and LPC have been implicated as acyl transport lipids; however the movement of fatty acids between the
plastid and ER remains an active area of research thus the figure shows a gradient between blue and green organelles in this region. The figure was inspiredby[136] with
modifications to address the text. Enzymes specifically discussed are presented in yellow ovals. The pathways for phosphatidylglycerol, other phospholipids, and sulfolipid
production are not shown. (Additional Abbreviations: ACCase, acetyl-CoA carboxylase; Mal-CoA, malonyl-CoA; Mal-ACP, malonyl-ACP; KASII, ketoacyl synthase II; SAD,
stearic acid desaturase; FAT A/B, fatty acid thioesterase A/B; NEFA, non-esterified fatty acid; ACT1, genetic mutant from act1 locus of Arabidopsis; LPCAT, lyso-
phosphatidylcholine acyltransferase; LPC, lyso-phosphatidylcholine; TGD, trigalactosyldiacylglycerol mutant; MGDG, monogalactosyldiacylglycerol; DGDG,
digalactosyldiacylglycerol).
8D.K. Allen et al. / Progress in Lipid Research xxx (2015) xxx–xxx
JPLR 873 No. of Pages 24, Model 5G
16 March 2015
Please cite this article in press as: Allen DK et al. Tracking the metabolic pulse of plant lipid production with isotopic labeling and flux analyses: Past, pre-
sent and future. Prog Lipid Res (2015), http://dx.doi.org/10.1016/j.plipres.2015.02.002

484
more definitively ruled out through the use of [
14
C]non-
485
hydrolysable ether analogs of PC [135].
486
At nearly the same time as initial descriptions of desaturation,
487
the assembly of glycerolipids and the production of polyunsatu-
488
rated fatty acids were ascribed to two separate biochemical path-
489
ways present in different organelles. The prokaryotic pathway
490
that takes place in the plastid and the eukaryotic pathway in the
491
ER [87,136] (Fig. 8) are distinguishable because of the attachment
492
of hexadecanoic groups (16 carbon fatty acids) at the sn-2 position
493
of glycerolipids – a result of the acyl-specificity of LPAAT of the
494
prokaryotic pathway [137]. Utilization of prokaryotic pathway dia-
495
cylglycerol (DAG) for phosphatidylglycerol (PG), monogalactosyl-
496
diacylglycerol (MGDG) and digalactosyldiacylglycerol (DGDG)
497
synthesis and the subsequent repeated desaturation of the sn-2
498
hexadecanoic to hexadecatrienoic group in only certain plants
499
has led to the name ‘‘16:3 plants’’ [138]. Plants which assemble
500
galactolipids primarily from a DAG-backbone imported from the
501
ER and containing a sn-2 octadecatrienoic acyl chain are termed
502
‘‘18:3 plants’’ [139]. Thus, the fatty acid composition of galac-
503
tolipids is commonly used to determine if the prokaryotic or
504
eukaryotic pathway is the dominant path for galactolipid synthesis
505
in plants [140].
506
Galactolipid assembly through both routes was described using
507
a combination of continuous pulse, and pulse-chase metabolic
508
labeling experiments with isotopically labeled acetate, glycerol,
509
fatty acids and CO
2
supplied to leaf tissue, isolated chloroplasts
510
and seedlings [99,133,137,139,141]. The transit of molecules
511
between the ER and plastid, as required in the eukaryotic pathway,
512
was observed when label that accumulated in PC during a pulse,
513
was traced to subsequent increases in galactolipid labeling during
514
the chase period. The pattern indicated a precursor–product
515
relationship between ER localized PC and plastid localized MGDG
516
and DGDG [132,141]. Further dual labeling with acetate and glyc-
517
erol demonstrated the same ratio of fatty acid to backbone enrich-
518
ment in PC as in MGDG of the chase experiment. This led Slack and
519
colleagues [96] to conclude that the entire DAG moiety was trans-
520
ferred from the ER to the plastid; thus revealing the modern day
521
description of the extra-plastidial or eukaryotic pathway of galac-
522
tolipid synthesis. The identification of Arabidopsis mutants with
523
changes in FA desaturation or membrane lipid composition by
524
Browse, Somerville, Benning and co-workers combined with
525
altered metabolic labeling of the two pathway system has subse-
526
quently allowed specific genes in lipid metabolism (i.e., both lipid
527
assembly and FA desaturation) to be correlated with these path-
528
ways (Fig. 8)[142–147].
529
3.1.3. Acyl chain formation and trafficking between the chloroplast
530
and ER in leaves
531
Though the basic paths and substrates for acyl chain and lipid
532
formation are known, a number of details about their transport
533
remain unclear. Fatty acid chain lengths of up to 18 carbons are
534
made through a series of condensing enzymes that pass an acyl-
535
ACP group and elongate it with the addition of two carbon acetyl
536
groups repeatedly (Fig. 6). The central role of ACP was established
537
through labeling experiments with [
14
C]malonyl-CoA, purified ACP
538
and ACP antibodies [104]. Ohlrogge, Kuhn and Stump varied the
539
ACP levels to establish dependency of de novo fatty acid biosynthe-
540
sis with this protein and also the plastidic location of fatty acid
541
production. Export of newly synthesized FA into the eukaryotic
542
pathway starts with release of the FA from ACP by a thioesterase
543
[2]. Mechanisms for acyl chain export from the plastid have
544
remained enigmatic, and may not require dedicated fatty acid
545
transporters because non-esterified fatty acids (NEFA) that flip-flop
546
diffuse within phospholipid membranes are immediately con-
547
verted to CoA esters after export by long chain CoA synthetases
548
[LACS; [148]](Fig. 8). Acyl-CoAs cannot diffuse across membranes,
549
thus their re-entry into the plastid is prevented and the overall
550
process results in uni-directional transport. However, whereas lac-
551
s9 mutants resulted in approximately 10-fold less acyl-CoA pro-
552
duced from exogenously supplied [
14
C]NEFA [149] the mutants
553
did not give a distinct phenotype possibly indicating redundancy
554
with other LACS genes that are involved in the process [150].
555
The underlying mechanisms of fatty acid export have been dif-
556
ficult to fully describe in part because the activation and move-
557
ment of fatty acids may be a channeled process as suggested
558
from [
14
C]acetate labeling experiments using spinach,
559
Arabidopsis leaves and T87 Arabidopsis cells [27,151,152].
560
Arabidopsis suspension cells labeled with [
14
C]acetate produced
561
[
14
C]PC very quickly with an estimated lag time of approximately
562
5.4 s. The short lag is significantly less than the time needed for
563
bulk acyl group exchange and may implicate the involvement of
564
PC in shuttling acyl groups to the ER [152]. A form of channeled
565
transport could also be supported by the known physical interac-
566
tions between the endoplasmic reticulum and the mitochondria,
567
plasma membrane, trans-golgi and chloroplasts [153,154].To
568
assess the genes involved the PC-MGDG precursor–product
569
relationship, FA and glycerol backbone labeling of leaf lipids was
570
characterized using mutants (Fig. 8). Arabidopsis act1 mutants
571
(from the act1 locus in Arabidopsis and also referred to as ats1,
572
Fig. 8) pulse-chase labeled with [
14
C]acetate demonstrated the role
573
of plastidial acyl-ACP dependent glycerol-3-phosphate acyltrans-
574
ferase (GPAT) activity in partitioning FA between prokaryotic and
575
eukaryotic pathways. Specifically, the lack of plastidial GPAT
576
activity in the act1 mutant resulted in diverting the flux of acyl
577
groups from the prokaryotic pathway into the eukaryotic pathway,
578
enhanced production of MGDG from PC, and effectively converted
579
a 16:3 plant into an 18:3 plant [146,147]. A second set of lines
580
with altered trigalactosyldiacylglycerol (TGD) proteins that facili-
581
tate the transfer of lipids, presumably either PA or DAG derived
582
from PC, into chloroplasts [155–161] were characterized using
583
[
14
C]acetate pulse-chase experiments [145] (Fig. 8). Whereas
584
wild-type Arabidopsis leaves showed the precursor–product
585
relationship between MGDG and PC that is characteristic of
586
eukaryotic galactolipid production, tgd-1 lines did not exhibit a
587
similar precursor–product pattern. Also the tgd phenotype is unlike
588
act1 lines that have reduced plastidic pathway biosynthesis
589
[146,147]. Instead the tgd-based MGDG pool exhibited strong ini-
590
tial labeling presumably from the plastidic pathway, but without
591
a second labeling peak in MGDG labeling that would signify the
592
conversion of labeled PC into MGDG at later time points. The
593
authors concluded that tgd-1 lines had an impaired ability to trans-
594
fer acyl groups from PC to MGDG and that the TDG1 protein was
595
involved in this process [145]. Other investigations specifically
596
interested in the form of lipid exported, suggest that partial
597
hydrolysis of PC to LPC is a prominent step in trafficking acyl
598
groups back to the chloroplast [162–164] based on acyl labeling
599
assessments of sn-1 and sn-2 positions. Together, these studies
600
highlight a recurring theme – the central role of PC in donating
601
and accepting acyl groups, and emphasize the need for further
602
mechanistic investigations of the lipid carrier(s) that import and
603
export acyl groups between chloroplast and ER. In particular,
604
additional labeling studies with mutants that fail to interconvert
605
phospholipids could help resolve the pathway attributes.
606
3.1.4. Metabolic labeling reveals alternative fluxes of acyl groups into
607
membrane lipids and TAG
608
The flux of FA from the plastid through the Kennedy pathway is
609
minimally required to produce TAG composed of newly synthe-
610
sized fatty acids exported from the plastid (e.g., 16:0, 18:0, 18:1);
611
however, many plants accumulate TAG containing fatty acids that
612
have been further modified (e.g., desaturation, hydroxylation).
613
Because PC is the substrate for most of these modifications
D.K. Allen et al. / Progress in Lipid Research xxx (2015) xxx–xxx 9
JPLR 873 No. of Pages 24, Model 5G
16 March 2015
Please cite this article in press as: Allen DK et al. Tracking the metabolic pulse of plant lipid production with isotopic labeling and flux analyses: Past, pre-
sent and future. Prog Lipid Res (2015), http://dx.doi.org/10.1016/j.plipres.2015.02.002

614
[165,166], acyl fluxes into and out of PC are essential for the pro-
615
duction of TAG, and the rate of PC labeling (Fig. 7) will reflect its
616
participation. A combination of at least three mechanisms allow
617
the flux of fatty acids into/out of PC for modification and eventual
618
TAG synthesis (Fig. 7B–C): (1) TAG synthesis from phospho-
619
lipid:diacylglycerol acyltransferase [PDAT; [123]]; (2) derivation
620
of the DAG substrate for TAG synthesis from PC; and (3) the
621
exchange of acyl groups between PC and the acyl-CoA pool by a
622
PC deacylation and lyso-PC reacylation cycle coined ‘‘acyl editing’’
623
[93,94]. For each of the three mechanisms no net synthesis of
624
PC is required for TAG biosynthesis. Acyl editing proceeds
625
by the forward (and probably reverse) action of LPCAT
626
[116,119,120,167,168], and incorporates nascent 18:1-CoA into
627
PC for modification and PC-modified fatty acids can re-enter the
628
acyl-CoA pool for incorporation into the sn-1, -2, -3 positions of
629
TAG by Kennedy pathway enzymes [Fig. 7B; [169]]. Alternatively
630
the use of PDAT results in the transfer of fatty acids from the sn-
631
2 position of PC to the sn-3 hydroxyl group of DAG resulting in
632
the production of TAG and lyso-PC. The lyso-PC is subsequently
633
reacylated by LPCAT as part of the acyl editing cycle (Fig. 7B). If
634
DAG used for TAG biosynthesis is derived from PC, then DAG is
635
in essence synthesized twice because it is also the precursor for
636
PC production (Fig. 7C). First, the production of de novo DAG occurs
637
through the Kennedy pathway for PC synthesis and second, the
638
removal of phosphocholine from PC produces a PC-derived DAG
639
that can be used to make TAG. Mechanisms that are potentially
640
involved in the de novo DAG ?PC ?PC-derived DAG pathway of
641
TAG production include: (a) both the forward and reverse reactions
642
of CDP-choline:diacylglycerol cholinephosphotransferase (CPT)
643
[170,171], (b) CPT to produce PC and phospholipase C (or phos-
644
pholipase D and PAP) for the subsequent production of PC-derived
645
DAG, and (c) phosphatidylcholine:diacylglycerol cholinephospho-
646
transferase (PDCT) [121] which transfers the phosphocholine head
647
group from PC to DAG, creating a new PC molecular species and a
648
PC-derived DAG (Fig. 7C). Of the possibilities, only PDCT has been
649
experimentally shown to deliver PC-derived DAG for TAG biosyn-
650
thesis based on evaluation of Arabidopsis mutants with in vivo
651
metabolic labeling [121]. Proposed further modifications to the
652
acyl editing scheme have been suggested by Lager and coworkers
653
[122] with the production of PC from two molecules of
654
lysophosphatidylcholine (LPC) using a lysophosphatidylcholine
655
transacylase (LPCT). The other product of this reaction, glyc-
656
erophosphocholine (GPC) could be converted back to LPC using a
657
recently discovered acyl-CoA:glycerophosphocholine acyltrans-
658
ferase (GPCAT). This previously undescribed set of reactions does
659
not require regeneration of CDP-choline and will likely impact
660
future interpretations of acyl editing; however the enzymes
661
catalyzing these steps have not been identified.
662
The use of isotopic labels has been essential in the character-
663
ization of the flux of acyl groups through acyl editing in plants.
664
The exchange of labeled fatty acids from acyl-CoAs into and out
665
of PC by LPCAT enzymes was initially characterized in microsomes
666
from oilseeds [167,168]. However, the involvement of acyl editing
667
as a major component of lipid metabolism was indicated through
668
in vivo metabolic labeling of leaves tracing the precursor–product
669
relationships of nascent fatty acid incorporation into membrane
670
lipids. Traditional descriptions of glycerolipid assembly hypothe-
671
sized that newly synthesized fatty acids were initially esterified
672
to glycerol-3-phosphate to produce mostly 18:1/16:0 and 18:1/
673
18:1 (sn-1/sn-2) molecular species of PA in the plastid and ER,
674
respectively. Then, the molecular species were desaturated while
675
linked to membrane lipids in the plastid (e.g., MGDG) or ER (e.g.,
676
PC). These concepts were brought into question when five minute
677
[
14
C]O
2
pulse labeling of Brassica napus leaves followed by one
678
hour and twenty-four hour chases demonstrated labeled lipid
679
molecular species that fit the hypothesis for prokaryotic MGDG,
680
but not eukaryotic pathway derived PC [94]. Subsequent pulse
681
labeling experiments with [
14
C]glycerol for less than ten minutes
682
revealed the traditional PA ?DAG ?PC precursor–product
683
relationship of the eukaryotic pathway in Pea leaves; however
684
[
14
C]acetate labeling indicated that PA and DAG were not the pri-
685
mary precursors for incorporation of nascent fatty acids into PC
686
[93]. Further stereochemical and molecular species analysis of
687
the labeled products indicated newly synthesized fatty acids were
688
first incorporated into eukaryotic pathway lipids by esterification
689
to mostly the sn-2 hydroxyl of lyso-PC (but also sn-1 acylation of
690
1-lyso-2-acyl-PC) creating a new PC molecular species containing
691
one nascent fatty acid and one previously synthesized fatty acid.
692
The relative amount of sn-1 vs. sn-2 acylation depends on species
693
and tissue [93,172]. Additionally, the pool of acyl-CoA utilized for
694
Kennedy pathway reactions included a significant amount of unla-
695
beled fatty acids that were released from PC during the recycling to
696
lyso-PC. This acyl editing cycle allowed for direct incorporation of
697
newly synthesized 18:1 into PC for desaturation and subsequent
698
production of an acyl-CoA pool containing a mixture of nascent
699
(16:0, 18:1) and further desaturated (18:2 and 18:3) fatty acids
700
for de novo membrane lipid production. The rate of acyl editing
701
was estimated to be up to 20 times that of fatty acid synthesis in
702
rapidly expanding pea leaves, indicating that even if de novo glyc-
703
erolipid assembly in the ER utilizes the same acyl-CoA pool as acyl
704
editing, the much higher rate of the acyl editing cycle ensures that
705
most nascent fatty acids exported from the plastid first enter PC.
706
Apart from leaves, the acyl editing mechanism has also been
707
observed in developing seeds [120,172,173] and Arabidopsis cell
708
cultures [152], and may be a more general mechanism for the
709
incorporation of newly synthesized fatty acids into the extra-
710
plastidial lipids.
711
Similar to the characterization of the acyl editing cycle, in vivo
712
pulse labeling of tissues accumulating TAG has helped decipher
713
when TAG is synthesized from de novo DAG or alternatively from
714
PC-derived DAG. When TAG is synthesized directly from nascent
715
fatty acids using the Kennedy pathway, there is little PC labeling
716
from [
14
C]acetate or [
14
C]glycerol (Fig. 7A, graphs). The addition
717
of acyl editing activity results in more significant labeling of PC
718
from [
14
C]acetate but not [
14
C]glycerol (Fig. 7B; graphs); however
719
the glycerol labeling is useful in helping distinguish the flux of
720
TAG synthesis from de novo- versus PC-derived DAG (Fig. 7A, C;
721
[120,133,172,174,175]). During oil accumulation in developing
722
seeds the vast majority of lipid metabolism is for TAG biosynthesis,
723
therefore significant labeling in non-Kennedy pathway intermedi-
724
ates describes a more complicated path of acyl flux that produces
725
TAG. When TAG is produced from PC-derived DAG (Fig. 7C); label-
726
ing from [
14
C]glycerol results in a significant incorporation into
727
DAG followed by PC and finally TAG after a significant lag. Fig. 7C
728
indicates this through a precursor–product labeling description.
729
The amount of acyl flux through PC by the above mechanisms
730
varies with plant species and tissues accumulating TAG. It is
731
tempting to assume that the amount of PC-modified fatty acids that
732
accumulate in TAG correlates with acyl flux through PC, however
733
this is an underestimate because fatty acids move through
734
PC regardless of whether they are further modified or not
735
[5,172–174]. Thus the pathway of TAG synthesis must be elucidated
736
by in vivo [
14
C]acetate and [
14
C]- or [
3
H]glycerol labeling and by
737
measuring the precursor–product relationships between DAG, PC
738
and TAG. For example, tissue slices of developing cocoa
739
(Theobroma cacao) cotyledons cultured with [
14
C]acetate or
740
[
14
C]glycerol produced labeled TAG but had limited incorporation
741
into PC from either substrate during the stage of rapid TAG accumu-
742
lation, suggesting the classical Kennedy pathway is the major route
743
of TAG synthesis (Fig. 7A; [176]). Cocoa TAG contains less than 2%
744
PC-modified fatty acids [176], and thus did not require extensive
745
flux through PC (Fig. 7A). In coriander (Coriandrum sativum)a
10 D.K. Allen et al. / Progress in Lipid Research xxx (2015) xxx–xxx
JPLR 873 No. of Pages 24, Model 5G
16 March 2015
Please cite this article in press as: Allen DK et al. Tracking the metabolic pulse of plant lipid production with isotopic labeling and flux analyses: Past, pre-
sent and future. Prog Lipid Res (2015), http://dx.doi.org/10.1016/j.plipres.2015.02.002

746
precursor–product relationship was observed between PC and TAG
747
for [
14
C]acetate but not glycerol labeling. Thus coriander lipid meta-
748
bolism is more analogous to Fig. 7B where fatty acids participate in
749
acyl editing cycle, but TAG is synthesized from de novo DAG with the
750
Kennedy pathway [177]. As coriander mostly accumulates 18:1
D
6
751
fatty acids produced in the plastid, the substantial flux of nascent
752
fatty acids through PC may in fact be part of the mechanism of
753
fatty acid export to the ER [152] and not for fatty acid modification.
754
Soybean (Glycine max), safflower (Carthamus tinctorius), flax (Linum
755
usitatissimum), and Arabidopsis thaliana;[120,133,172,174,175]
756
each demonstrate the PC to TAG relationship from both
757
[
14
C]acetate and [
14
C]glycerol labeling. In these species acyl editing
758
and the use of PC-derived DAG for TAG synthesis both contribute to
759
fatty acid flux through PC and result in high levels of polyunsatu-
760
rated-fatty acids in TAG (Fig. 7C). Though these examples indicate
761
preferences for TAG synthesis, the paths are likely in simultaneous
762
operation to varied extents, (e.g., the relative use of PDCT or CPT/li-
763
pase for PC-derived DAG production, and the use of DGAT or PDAT
764
for the final acyl acylation of TAG [115]). In the case of DGAT and
765
PDAT, DGAT may be used in any of the situations described in
766
Fig. 7A–C, however, PDAT activity requires the acyl editing cycle
767
to reacylate the lyso-PC co-product of TAG synthesis, and thus can-
768
not be a part of a classical Kennedy pathway (Fig. 7A).
769
4. Extending isotopic labeling interpretation
770
4.1. Computational metabolic flux analysis using stable isotopes
771
The interpretation of labeling is often not straightforward and
772
has been significantly aided by computational assessments of mul-
773
tiple pathways referred to as Metabolic Flux Analysis (MFA). These
774
methods were initially developed to study steady state metabolism
775
in other systems [e.g., [178–183]] but extensive resources for
776
application in plants have now been described in books
777
[184–186] and special issues of Phytochemistry [187] and the
778
Journal of Experimental Botany edited by Kruger and Ratcliffe
779
[188]. In each case the isotopic labeling experiment is assessed
780
with a computer model that describes mass-balanced atom transi-
781
tions of the biochemical reactions [43,189–191]. The model is a
782
mathematical description of network branching, bond-breaking
783
and reforming reactions, substrates and flux estimates that are
784
used to simulate the label incorporation which can then be com-
785
pared to the experimentally measured values through least
786
squares regression. The process is repeated with different flux val-
787
ues until an optimized fit is achieved resulting in ‘‘best estimates’’
788
of flux and confidence intervals and statistics can then be assessed
789
using Monte Carlo or techniques described elsewhere [192–194].
790
Progress in the quantification of label [195,196] has increased the
791
number and precision of measurements resulting in more informa-
792
tion than is minimally required to establish the flux values com-
793
putationally (known as an ‘overdetermined’ system). The least
794
squares comparison results in the best fit of all measurements
795
and produces the lowest overall residuum (i.e., lowest residual
796
error between in silico and experimental measurements). A more
797
in depth survey of computational considerations including elegant
798
mathematical representations are beyond the scope of this review
799
and are described elsewhere [197–201].
800
4.1.1. Central carbon metabolism supplies carbon, energy and reducing
801
equivalents for FA biosynthesis
802
In developing seeds, the carbon, energy and reducing equiva-
803
lents necessary for fatty acid biosynthesis are derived from multi-
804
ple sources requiring coordination between catabolic and anabolic
805
metabolic pathways. Developing seeds receive organic carbon and
806
nitrogen in the form of sugars and amino acids apoplastically from
807
the maternal plant [202,203]. The supply of carbon and nitrogen
808
constrains the potential metabolic operation; but the final seed
809
composition is a consequence of the fluxes through biochemical
810
pathways within the tissue. Branch points in the metabolic
811
network establish the distribution of carbon, energy and reduced
812
cofactors, thus the diversity in lipid and protein content in seeds
813
is a consequence of differences in flux through these steps.
814
Metabolic flux analysis has enabled the identification and quan-
815
tification of the endogenous carbon sources for fatty acid biosyn-
816
thesis and aided descriptions of the balance of ATP and
817
nucleotide cofactors.
818
Some seeds like pea and soybean produce significant amounts
819
of protein and therefore require a large amount of amino acid
820
nitrogen. Most of the nitrogen supplied to the seed is in the form
821
of glutamine and to a lesser extent asparagine or alanine [202].
822
The nitrogen is redistributed by aminotransferases to other carbon
Fig. 9. Glutamine labeling studies indicate flow through organic acids to acetyl-
CoA. Studies with [
13
C]glutamine in several crops indicate that carbon from amino
acids is used in the biosynthesis or elongation of fatty acids in oilseeds (figure
inspired by [57,207]). Labeling in metabolites was measured by mass spectrometry
resulting in mass isotopomers (i.e., M1, M2, M3 etc) that reflect the increase in
molecular weight by 1, 2, 3 etc with incorporation of
13
C in place of
12
C. (A)
Glutamine label in soybeans is transferred to amino acids derived from pyruvate
and acetyl-CoA most likely through a combination of malic enzyme (ME) and ATP
citrate lyase (ACL). The high labeling in glutamine (i.e., M5) is transferred to four
carbon organic and amino acids such as threonine. Leucine that is synthesized
through a combination of acetyl groups is more highly labeled in even mass
isotopomers (i.e., M2, M4) indicating that a significant amount of acetyl-CoA is fully
labeled. (B) The reversibility of isocitrate dehydrogenase (ICITDH) in brassica results
in a high M5 labeling in citrate which would not otherwise occur from respiratory
steps of TCA. Uniformly labeled glutamine (i.e., M5) is decarboxylated by aerobic
TCA metabolism to produce M4 labeled succinate, fumarate, malate and oxaloac-
etate. When combined with unlabeled acetyl-CoA, the resulting citrate and
isocitrate would contain four or fewer
13
C atoms (i.e., <M5). Thus M5 in citrate
and isocitrate signifies a reversible isocitrate dehydrogenase activity and with ACL
could provide additional acetyl groups for fatty acid elongation.
D.K. Allen et al. / Progress in Lipid Research xxx (2015) xxx–xxx 11
JPLR 873 No. of Pages 24, Model 5G
16 March 2015
Please cite this article in press as: Allen DK et al. Tracking the metabolic pulse of plant lipid production with isotopic labeling and flux analyses: Past, pre-
sent and future. Prog Lipid Res (2015), http://dx.doi.org/10.1016/j.plipres.2015.02.002

823
skeletons for the production of all twenty amino acids and also
824
results in the generation of organic acids. Radioisotopic studies
825
[204,205] and stable isotope investigations with metabolic flux
826
analyses in soybeans [57,206] indicate that organic and amino acid
827
carbon is partially repartitioned into oil (approximately 10–20% of
828
carbon in oil originates from glutamine) through conversion of
829
malate by malic enzyme (ME) or in combination with other steps
830
including ATP-citrate lyase (ACL; Fig. 9A). First, glutamine serves
831
as a donor of nitrogen in transamidation reactions to produce glu-
832
tamate. Glutamate aminotransferase reactions further reallocate
833
nitrogen and result in alpha-ketoglutarate which can be converted
834
to other organic acids (e.g., malate, citrate) using reactions com-
835
monly associated with, though not limited to the TCA cycle. Then
836
malic enzyme can convert malate to pyruvate for acetyl-CoA
837
production or citrate can be cleaved using ATP-citrate lyase result-
838
ing in acetyl-CoA (Fig. 9). Organic acids also serve as precursors for
839
acetyl-CoA production in Brassica ([207];Fig. 9B) and maize [208]
840
embryos.
841
Acetyl-CoA can be derived from multiple sources in different
842
subcellular locations though it cannot readily diffuse through
843
membranes causing speculation about its origin and movement
844
in plant cells (e.g., [209]). As a result the enzymatic paths of
845
acetyl-CoA biosynthesis from hexose catabolism or amino acid
846
metabolism have differing consequences on carbon and energy uti-
847
lization. For example the production of two acetyl-CoA molecules
848
from glucose-6-phosphate using glycolysis and pyruvate dehydro-
849
genase produces: two CO
2
, four NAD(P)H, and two ATP. Since fatty
850
acid biosynthesis can be sustained with two NAD(P)H and one ATP
851
per acetyl group, the supply and utilization of carbon, energy and
852
reducing equivalents are stoichiometrically balanced by glycolytic
853
reactions. Other pathways such as the oxidative pentose phosphate
854
pathway (OPPP) can also play an important role in generating
855
reducing equivalents for fatty acid biosynthesis however there is
856
no strict requirement for a particular catabolic pathway’s involve-
857
ment. Estimates of the amount of OPPP in plant tissues such as
858
seeds are particularly difficult to establish because isoforms for
859
these enzymatic steps are found in both the plastid and cytosol
860
[210] (Fig. 10) and can be further complicated by CO
2
reassim-
861
ilatory mechanisms that improve the seed carbon economy [211].
862
Production of acetyl-CoA from pyruvate involves decar-
863
boxylation and creates one mole of CO
2
for every two moles of car-
864
bon dedicated to FA biosynthesis. RuBisCO in developing rapeseed
865
and soybeans [212] and high internal concentrations of CO
2
[213]
866
led to a description of RuBisCO-aided assimilation that was con-
867
firmed by [
13
C]labeling studies [214]. [1-
13
C]alanine taken up by
868
rapeseed embryos was converted to pyruvate then acetyl-CoA with
869
decarboxylation by pyruvate dehydrogenase producing [
13
C]O
2
.
870
Reassimilation of the [
13
C]O
2
resulted in labeling in glycolytic
871
products that were upstream of pyruvate supporting the role for
872
RuBisCO. It was further shown through
14
C investigations that
873
the carbon reassimilation was light-dependent in green seeds
874
[215]. Rapeseed and other green oilseeds such as soybean appear
875
to have developed mechanisms to utilize sunlight for metabolic
876
efficiency though the extent and descriptions remain unclear and
877
may involve subcellular, cellular or tissue-level coordination
878
[206,216–221]. The biological diversity and challenges associated
879
with quantification of CO
2
respiring and assimilating steps in mix-
880
otrophic metabolism (Fig. 1) has led to wide-ranging estimates for
881
flux through central metabolic pathways. Recent descriptions of
882
plant respiration including an in silico study with label design con-
883
siderations and modeling have summarized some of these chal-
884
lenges and opportunities [222,223].
885
Conversion of glutamine carbon to acetyl-CoA entails a different
886
set of reactions that start with the donation of nitrogen from glu-
887
tamine. Alpha ketoglutarate can be converted to acetyl-CoA by
888
way of sequential production of succinate, fumarate, malate and
889
pyruvate through enzymatic steps commonly ascribed to the TCA
890
cycle along with malic enzyme and pyruvate dehydrogenase
891
(Fig. 9). This series of steps will result in 4 NAD(P)H, an FADH
2
892
and one ATP equivalent, though respiring multiple carbons in the
893
process. The additional reducing equivalents may supply electron
894
transport for oxidative phosphorylation needed for protein biosyn-
895
thesis as approximately 4.3 mol of ATP are required for each addi-
896
tional amino acid added to an elongating peptide [41]. Other seeds
897
such as rapeseed utilize acetyl-CoA originating from the organic
898
acid citrate as a carbon source for fatty acid elongation (i.e., cytoso-
899
lic addition of acetyl groups for acyl lengths of twenty carbons or
900
more). [
13
C]glutamine labeling experiments allowed Schwender
901
and colleagues [207] to show that fully labeled alpha-ketoglutarate
902
resulted in citrate with a significant fraction of highly labeled
Fig. 10. Oxidative and non-oxidative pentose phosphate pathways have multiple
subcellular locations. The elucidation of active pathways in central metabolism has
remained difficult to assess because oilseeds have pentose phosphate pathway
enzymes targeted to multiple locations which may be involved in (A) oxidative, (B)
reductive or (C) both types of metabolism. Carbon, reducing equivalents and energy
necessary for fatty acid biosynthesis can come from metabolic networks operating
in different ways dependent upon the conditions and cell or tissue. For emphasis
the plastids in (B) and (C) are green symbolizing that they make use of light and
operate differently than (A).
12 D.K. Allen et al. / Progress in Lipid Research xxx (2015) xxx–xxx
JPLR 873 No. of Pages 24, Model 5G
16 March 2015
Please cite this article in press as: Allen DK et al. Tracking the metabolic pulse of plant lipid production with isotopic labeling and flux analyses: Past, pre-
sent and future. Prog Lipid Res (2015), http://dx.doi.org/10.1016/j.plipres.2015.02.002

903
carbons (i.e., M5) – a result that would not be anticipated with tra-
904
ditional TCA activity including carbon respiration (Fig. 9B). The
905
authors reasoned that isocitrate dehydrogenase (ICITDH) may act
906
reversibly to more efficiently conserve carbon and produce citrate
907
that can be exported and cleaved by ATP citrate lyase to produce
908
acetyl-CoA for fatty acid elongation. Labeling in the elongated fatty
909
acids at the terminal acetate as detected by mass spectrometry
910
helped confirm the result [207]. These divergent descriptions in
911
carbon handling emphasize that seeds are not merely passive col-
912
lection receptacles [224] but function according to their opera-
913
tional metabolic networks and its physiological context [225].
914
4.2. Assessing resource partitioning through perturbation-response
915
experiments
916
Given that plant metabolic operation is not static, the use of
917
labeling with or without computational analyses can help describe
918
cellular phenotypes and lead to hypotheses about regulation
919
through perturbation experiments. At the biochemical level these
920
studies have provided insight to wild-type metabolic operation
921
as well as the response to genetic or environmental conditions.
922
By changing the temperature [226], oxygen [62,227] nutrient sta-
923
tus [57,61,63,64,228] or through genetic manipulation of central
924
metabolism [59,229,230], or lipid metabolism [120] plant cells
925
are forced to rebalance resource utilization resulting in altered
926
phenotypes.
927
4.2.1. Alterations in carbon and nitrogen supply to fatty acids
928
The supply of carbon for fatty acid biosynthesis was examined
929
with [
13
C]labeling in genetically altered Arabidopsis lines [59] con-
930
taining mutations in the WRINKLED1 transcription factor [231–
931
233] or plastidic pyruvate kinase genes [234]. In the wri1-1 mutant
932
allele seed oil content was reduced by 80% and resulted in an
933
observable wrinkling appearance. Double mutants in plastidic
934
pyruvate kinase pkpb
1
pkp
a
(Fig. 6 ) also had significant reduction
935
in oil presumably because pyruvate generation was limiting for
936
fatty acid biosynthesis. The decrease in oil in wri1 was attributed
937
to reductions in flux from sucrose that is consistent with transcrip-
938
tionally regulated changes in glycolysis [235] and fatty acid meta-
939
bolism [236]. As a result flux through cytosolic pyruvate kinase and
940
malic enzyme was altered in part to compensate for the reduced
941
supply of carbon to fatty acid metabolism by plastid-localized
942
steps [59]. The role of these two enzymes have also been consid-
943
ered by environmental changes to temperature [226] and supplied
944
nitrogen [47,57,61]. When inorganic nitrogen (i.e., instead of
945
amino acids) was supplied to rapeseed the flux through malic
946
enzyme in the mitochondria decreased 50% as a consequence of
947
increased use of TCA intermediates for amino acid biosynthesis
948
[61]. Rapeseed growth was depressed by 50% possibly due in part
949
to the form of nitrogen supplied for metabolism. In soybeans,
950
changes in the amount of organic nitrogen supplied did not signifi-
951
cantly impact growth but did lead to changes in final composition
952
that included more protein [47,57]. Thus the response of metabo-
953
lism to changes in nutrient or environment is variable and provides
954
further rationale for flux analysis in many species [63].
955
4.2.2. Altered flux phenotypes in lipid mutants
956
Genetic mutants have also played a role in elucidating the most
957
active pathways that contribute to fatty acid and lipid production.
958
In the examples of act1(ats1) and tgd1 lines [145,146] each mutant
959
was identified from a forward genetic screen by an easily detected
960
lipid phenotype (e.g., fatty acid or lipid class composition changes),
961
therefore changes in acyl fluxes were expected and confirmed by
962
labeling analysis. However reverse genetics-derived mutants that
963
do not display significant growth or lipid composition phenotypes
964
have also benefited from tracer studies. In particular the acyl
965
editing mechanism has been supported by lyso-phosphatidyl-
966
choline acyltransferase (LPCAT) mutants. Since acyl editing can
967
result in the removal of PUFA from PC for TAG synthesis in seed tis-
968
sue, defects in the LPCAT enzymes required for acyl editing cycle
969
would be expected to reduce the PUFA composition of TAG. Of
970
the four putative Arabidopsis gene products with in vitro lyso-
971
phosphatidylcholine activity [119,237] a double knockout com-
972
prised of the two enzymes with highest LPCAT activity gave very
973
little change to Arabidopsis seed FA composition [120,238]. The
974
TAG composition suggested the two LPCATs may not have been
975
greatly involved in acyl editing, or that other enzymes with
976
lysophospholipid acyltransferase activity can compensate for their
977
loss in seed tissue. However, [
14
C]acetate labeling of the double
978
mutant revealed a different conclusion. Wild-type seeds incorpo-
979
rated nascent fatty acids into sn-2 PC prior to sn-1/sn-2 DAG, char-
980
acteristic of the seed acyl editing mechanism [172]. However, the
981
lpcat1/lpcat2 double mutant had a DAG ?PC precursor–product
982
relationship more similar to the flux of de novo DAG synthesis by
983
the Kennedy pathway, and the stereochemistry of fatty acid
984
incorporation was essentially the same in both lipids. The results
985
suggested that acyl editing was impaired within the lpcat1/lpcat2
986
mutant [5,120,174,238]. It appears that a compensatory increase
987
in the PC-derived DAG pathway of TAG synthesis provided a way
988
for nascent fatty acids to flux through PC for desaturation prior
989
to TAG synthesis. Thus, the true phenotype of a mutant was not
990
revealed by static pool measurements of lipid composition.
991
Analysis of metabolic fluxes with isotopic labeling identified the
992
‘‘hidden’’ mutant phenotype and also underlined the compensatory
993
metabolism needed to maintain a seed composition similar to wild
994
type.
995
4.2.3. Oil with modified FA composition
996
The majority of fatty acids found in plants are 16 or 18 carbons
997
long with 0–3 methylene interrupted double bonds. These acyl
998
groups are valuable sources of nutrition, but have a select range
999
of uses as biofuels and industrial feed stocks. However, within the
1000
Plant Kingdom there are greater than 300 different types of
1001
‘‘unusual fatty acids’’ that have desirable physical properties and
1002
functional groups (e.g., short chain, medium chain, hydroxy, epoxy,
1003
cyclopropane, and conjugated double bonds). Capitalizing on natu-
1004
ral diversity is an attractive proposition for biotechnology and
1005
could result in value-added fuels, lubricants, polymers, coatings,
1006
adhesives, surfactants, resins, and other products that reduce
1007
dependency on petroleum [19,239,240]. However, many plants that
1008
produce these unusual fatty acids have agronomic features which
1009
make them less suitable as major crops. Over the past two decades
1010
attempts to genetically engineer unusual fatty acids into common
1011
oilseed crops or model species with few exceptions have produced
1012
modest proportions of desirable fatty acids [18,19,240–243].To
1013
understand why, isotopic labeling has been used to identify limiting
1014
steps. California bay (Umbellularia californica) plants contain a
1015
medium-chain acyl-acyl carrier protein thioesterase (MCTE) that
1016
terminates FA synthesis at 12 carbons [244] (Fig. 6 ). When the
1017
MCTE was expressed in B. napus, lauric acid accumulation was posi-
1018
tively correlated with MCTE activity up to approximately 30% 12:0
1019
in TAG. A further 30-fold increase in MCTE enzyme activity
1020
increased the concentration of 12:0 to near 60% and indicated that
1021
this enzyme was no longer limiting lauric acid accumulation in TAG
1022
[118,245–247]. Though [
14
C]acetate was predominantly converted
1023
to lipid in wild-type B. napus seeds, in the high lauric transgenic B.
1024
napus line only one-half of the [
14
C]acetate went to lipids with the
1025
rest contributing to water soluble products including sucrose and
1026
malate [246]. Along with high enzyme activities for 12:0-CoA oxi-
1027
dase, isocitrate lyase, and malate synthesis the results suggested a
1028
high rate of beta-oxidation and gluconeogenesis that likely breaks
1029
down newly synthesized 12:0 and limits its accumulation in TAG.
D.K. Allen et al. / Progress in Lipid Research xxx (2015) xxx–xxx 13
JPLR 873 No. of Pages 24, Model 5G
16 March 2015
Please cite this article in press as: Allen DK et al. Tracking the metabolic pulse of plant lipid production with isotopic labeling and flux analyses: Past, pre-
sent and future. Prog Lipid Res (2015), http://dx.doi.org/10.1016/j.plipres.2015.02.002

1030
Even though FA beta-oxidation was increased in these seeds total
1031
oil accumulation was unaffected because of a concomitant increase
1032
in fatty acid synthesis (also elucidated by metabolic labeling)
1033
indicating plants can respond to futile cycles of FA synthesis and
1034
degradation to maintain seed lipid levels.
1035
Significant production of unusual oils does occur in a few
1036
crops including castor. Castor (Ricinus communis) produces a
1037
significant amount of the hydroxylated fatty acid ricinoleic acid
1038
(12-hydroxy-9-cis-octadecenoic acid) through a FA hydroxylase
1039
that is a variant of the FAD2 enzyme and utilizes 18:1 esterified
1040
to PC as a substrate [166,239,248]. The seed endosperm contains
1041
90% ricinoleic acid in castor TAG; though only 5% of the PC fatty
1042
acid profile is hydroxylated [249]. Thus at least 90% of seed 18:1
1043
fluxed through the sn-2 position of PC. An in vitro analysis of TAG
1044
synthesis within castor endosperm microsomes suggested that
1045
supplied [
14
C]18:1 was hydroxylated on PC, released from PC
1046
through an acyl editing mechanism, and the hydroxy-FA-CoA
1047
utilized to synthesize de novo DAG and TAG through a classical
1048
Kennedy pathway (Fig. 7B; [250]).
1049
Attempts to engineer other plants to produce castor type oils
1050
have resulted in significant, but far less, hydroxylated fatty acid
1051
content. Heterologous expression of the castor fatty acid hydroxy-
1052
lase produced only 17% modified fatty acids in Arabidopsis,
1053
suggesting other factors may limit their accumulation in TAG
1054
[251–255]. Based on short time point [
14
C]glycerol pulse labeling
1055
approximately 50% of newly synthesized de novo DAG produced
1056
by the Kennedy pathway contained a hydroxylated fatty acid, but
1057
this de novo DAG was turned over and not utilized for synthesis
1058
of PC or TAG [174]. Since Arabidopsis utilizes the PC-derived
1059
DAG pathway of TAG synthesis, the inefficient utilization of unu-
1060
sual-fatty-acid-containing de novo DAG for PC synthesis represents
1061
a limitation for synthesis of TAG containing unusual fatty acids at
1062
the sn-1 position. Additionally, hydroxy-fatty acid production
1063
within transgenic Arabidopsis seeds reduced total seed oil levels
1064
by 30–50% [251,256] and revealed a concomitant decrease in the
1065
rate of fatty acid synthesis based on [
14
C]acetate and [
3
H]
2
O label-
1066
ing studies [257]. Comparative [
14
C]acetate and [
14
C]malonate
1067
labeling suggested acetyl-CoA carboxylase activity was reduced
1068
by one half in seeds producing hydroxy-fatty acids relative to the
1069
wild-type. Together these results indicated that the lower oil con-
1070
tent of the hydroxy-fatty acid producing lines was a consequence
1071
of reduced fatty acid biosynthesis rather than fatty acid
1072
beta-oxidation [257]. Subsequent engineering attempts to increase
1073
the proportion of hydroxy-fatty acids in seed oils through
1074
co-expression of the castor fatty acid hydroxylase and selective
1075
TAG synthesis enzymes from castor [251,254] have resulted in sig-
1076
nificant increases in the proportion of modified fatty acids in TAG
1077
(i.e., to over 25%), relieved the acetyl-CoA carboxylase inhibition,
1078
and restored near wild-type rates of fatty acid synthesis and oil
1079
production [257]. One implication from this work is that the
1080
inefficient utilization of unusual fatty acids within ER glycerolipid
1081
synthesis can inhibit de novo fatty acid synthesis in the
1082
plastid, potentially indicating an uncharacterized endogenous
1083
mechanism that coordinates ER lipid assembly with plastid FA
1084
synthesis [257]. Similar reductions in TAG in other transgenic
1085
oilseeds [258,259] may suggest a common mechanism that
1086
coordinates fatty acid biosynthesis with TAG containing unusual
1087
fatty acids.
1088
4.2.4. Assessing regulation with metabolic control analysis
1089
Though perturbation experiments can provide insights into car-
1090
bon partitioning, attempts to unify this data with other ‘‘omics’’
1091
technologies have not led to straightforward interpretations. For
1092
example Junker and colleagues measured the protein activities of
1093
22 enzymes of central metabolism at saturating substrate concen-
1094
trations in Brassica [61]. Their data, obtained from cultured
1095
embryos, was reasonably consistent with tissues measured directly
1096
after harvest [260] but activities did not agree closely with meta-
1097
bolic fluxes. This cautionary tale supports the shared control of flux
1098
through multiple pathway enzymes [261] and demonstrated that
1099
metabolism can be poised to adjust to environmental perturbation
1100
without significant regulatory reprogramming at the proteome
1101
level. Therefore catalytic activities may not provide meaningful
1102
estimates of in vivo flux. Analogously, changes in metabolite levels
1103
from perturbed oxygen supply [62] or through in silico explorations
1104
[262] do not require obvious changes in fluxes either; indicating
1105
our understanding of the control of oil biosynthesis and regulation
1106
of its accumulation remain primitive [263].
1107
As a complementary approach the relationship between activi-
1108
ties and fluxes has been considered by metabolic control analysis
1109
[264–266]. MCA first gained prominence because early analyses
1110
described the shared control of pathway flux that was inconsistent
1111
with notions of enzymatic bottlenecking that still pervades the
1112
literature and common thought today. Assessment of control can
1113
entail either an enzyme level exploration that is a ‘‘bottom up’’
1114
approach or alternatively divide metabolism into groups of path-
1115
way steps referred to as ‘‘blocks’’ with top down control analysis
1116
(TDCA) [267]. Through a series of investigations Harwood and col-
1117
leagues have applied the latter approach to assess the control
1118
structure of oil metabolism in crops. Fatty acid biosynthesis and
1119
lipid assembly were considered as separate blocks in soybean,
1120
palm, olive and brassica species [268–274]. The flux through both
1121
blocks was measured using
14
C labeling with acetate and glycerol
1122
and then again after perturbing metabolism. By adding exogenous
1123
fatty acids (i.e., oleate) or enzyme inhibitors, altering temperature,
1124
or through changes in gene expression the studies monitored the
1125
responses of fatty acid biosynthesis and lipid assembly that were
1126
the metabolic blocks. In this case, the blocks were separated by
1127
the cytosolic acyl-CoA pool which was manipulated using oleate.
1128
This approach is referred to as single manipulation TDCA. Then
1129
inhibitors can be used to further probe control, known as double
1130
manipulation TDCA. In soybean and palm, fatty acid biosynthesis
1131
contributed approximately two-thirds of the flux control, slightly
1132
more than in olive [268,270,274]. In oilseed rape, the flux control
1133
coefficient for lipid assembly was predominant [273]. Double
1134
manipulation studies with combinations of inhibitors including
1135
2-bromooctanoate or diazepam that inhibit lipid assembly
1136
[272–274] and diflufenican and triclosan that inhibit fatty acid
1137
synthesis [271,273] provided further confirmation of the findings.
1138
One unique observation came from the specific comparison of
1139
Kennedy pathways between olive and palm callus cultures by
1140
Ramli and coworkers. They observed that radioactive DAG in addi-
1141
tion to TAG accumulated at increased levels in olive relative to
1142
palm [270]. They attributed the change in labeling to enhanced flux
1143
control specific to DGAT in olive and used the DGAT specific inhi-
1144
bitor 2-bromooctanoate to compare olive and palm. Results from
1145
the studies indicated the flux control coefficient for DGAT was
1146
74% in olive but only 12% in palm and therefore DGAT played a
1147
more significant role in controlling olive oil production.
1148
Subsequent studies overexpressing DGAT indicated a shift in the
1149
distribution of flux control, consistent with TDCA theory, but also
1150
demonstrated a related increase in seed oil content [e.g., 14%
1151
increased oil in Brassica [272]]. These studies have since been fur-
1152
ther validated with results from field trials [16]. Together they
1153
describe promising enzymatic targets for enhanced oil biosynthe-
1154
sis on the basis of their contribution of flux control through lipid
1155
biosynthetic pathways. Control analysis is equally well-suited to
1156
other tissues such as leaves [275] and may provide insights to
1157
overcome current challenges that limit oil production in vegetative
1158
tissues. Altogether these investigations establish that even with
1159
shared flux control, the modification of steps exerting greatest con-
1160
trol can result in altered overall flux, however with the additional
14 D.K. Allen et al. / Progress in Lipid Research xxx (2015) xxx–xxx
JPLR 873 No. of Pages 24, Model 5G
16 March 2015
Please cite this article in press as: Allen DK et al. Tracking the metabolic pulse of plant lipid production with isotopic labeling and flux analyses: Past, pre-
sent and future. Prog Lipid Res (2015), http://dx.doi.org/10.1016/j.plipres.2015.02.002

1161
consequence that control of flux is likely shifted to other steps in
1162
the process.
1163
4.3. Future techniques and technology to address longstanding
1164
challenges
1165
4.3.1. Developments in MS technology to enable labeling and flux
1166
studies
1167
Deciphering unknown plant function and the operation of lipid
1168
metabolic networks in the future will greatly benefit from
1169
technological advances. Historically separation techniques such
1170
as thin layer and gas chromatography that are highly reproducible
1171
have been invaluable in segregating lipid species; however trans-
1172
formative progress in lipid analysis was more recently aided with
1173
the development of electrospray ionization (ESI) technology
1174
[276]. The ionization of intact molecular species of lipids through
1175
‘‘soft’’ techniques that minimize fragmentation has helped avoid
1176
artifacts from preparation and allowed direct verification of acyl
1177
composition in different lipids [277–279]. ESI tandem MS can reli-
1178
ably quantify picomole levels of known compounds and identify
1179
hundreds of lipids species [280] leading to estimates that approxi-
1180
mately 90% of all lipid pools can be measured [281,282]. There is
1181
hope that fragmentation patterns may even be capable of discern-
1182
ing the regiospecific attachment at sn-1 and sn-2 positions
1183
[277,279,283], though acyl or double bond migration artifacts
1184
remain a concern [284]. As ESI-related technology is several
1185
orders of magnitude more sensitive than other technologies
1186
[283], it has found application in quantifying isotopic labeling
1187
[67,285] leading to assessments of metabolic flux in central meta-
1188
bolism [e.g., [66,70,286]] as well as recent mammalian studies
1189
examining flux of cellular lipids [278]. By tracking deuterium or
1190
phospholipid head group labeling, newly synthesized PC and
1191
mechanisms of PLA and PLB deacylation-reacylation mechanism
1192
have been described [277] and the status of progress in this area
1193
was recently reviewed by Ecker and Liebisch within this journal
1194
[287]. Presumably similar approaches could be applied to plant tis-
1195
sues if commensurate amounts of labeled substrate were metabo-
1196
lized by plants. Given the number of different labeling experiments
1197
described for plants in the literature that range from in planta,to
1198
leaf disks, embryos, cell or root tip cultures, tissue slices, or
1199
homogenates and organelle preparations, such approaches seem
1200
promising.
1201
4.3.2. Addressing the challenges of multicellular eukaryotic
1202
metabolism
1203
One of the practical limitations to interpretation of labeling
1204
experiments in eukaryotes is the ability to resolve the spatial dis-
1205
tribution of compounds at cellular and subcellular levels. Plant
Fig. 11. Cellular heterogeneity contributes to spatial labeling differences in
metabolites. The complexity of plant cells includes subcellular pools of metabolites
separated at cellular and subcellular levels, however most experimental methods
are incapable of examining pools specific to a compartment. Portions of three
individual cells are shown in orange along with the intercellular space between
them. Individual metabolite pools shown in blue contain both unlabeled and
labeled atoms indicated by white and dark blue filled circles. In this example, triose
phosphate that is labeled autotrophically in the chloroplast (TP
p
) can be exported to
the cytosol (TP
c1
) and converted to sucrose by combining with other metabolites
that are unlabeled. The resulting sucrose (S
c1
) is less labeled and may be further
diluted when exported to the exterior of the cell (S
ext
) and into other cells (S
c2
).
Conversion of sucrose to triose (TP
c2
) results in at least three triose pools that are
mixed when metabolites are extracted from biomass. The heterogeneity in cells and
the distinct metabolism at the subcellular level can limit the conclusions that can be
drawn.
Fig. 12. Analysis of
13
C peptide labeling in developing soybeans for spatial and
temporal metabolic information. Developing soybeans were taken from pods and
cultured with
13
C resulting in partially labeled storage proteins (inspired by [319]).
(A) Initially seed metabolism on the vine results in labeling only at levels consistent
with natural abundance of isotopes. (B) As the labeling experiment progresses new
proteins are made making use of amino acids that contain
13
C. (C) Examination of
the peptides indicated the presence of a small fraction of storage protein that was
labeled at levels consistent with presence of natural abundance (i.e., very little
labeling), as well as a second fraction that was more labeled consistent with
proteins made during the culturing process. Thus the peptide mass isotopomers
formed a bimodal distribution that reflected the temporal labeling process. Some
proteins were the result of unlabeled amino acids present in planta whereas others
were the result of amino acids made by labeled amino acids generated during
culturing. For comparison purposes the same protein harvested after culturing was
hydrolyzed and amino acids mass isotopomers were measured. The amino acid
descriptions were mathematically convolved to generate a labeling description of
the same peptides for comparison. However when hydrolyzed amino acids are
convolved the loss connectivity of labeled or unlabeled amino acids specific to a
time on the vine or in culture is lost. Therefore GC–MS analysis of amino acids and
mathematical convolution cannot account for the specific attachment of labeled
amino acids next to each other which was a result of the temporal labeling process
in seed development and culturing. Thus the direct measurement of peptides
provides enhanced and more accurate labeling information.
D.K. Allen et al. / Progress in Lipid Research xxx (2015) xxx–xxx 15
JPLR 873 No. of Pages 24, Model 5G
16 March 2015
Please cite this article in press as: Allen DK et al. Tracking the metabolic pulse of plant lipid production with isotopic labeling and flux analyses: Past, pre-
sent and future. Prog Lipid Res (2015), http://dx.doi.org/10.1016/j.plipres.2015.02.002

1206
cells come in dozens of different types that perform different
1207
functions (e.g., leaf mesophyll and epidermal cells) and that are
1208
themselves comprised of heterogeneous populations. The
1209
metabolite heterogeneity including lipid compositions at the cellu-
1210
lar level examined through MRI [288,289] and imaging mass spec
1211
[290–293] is visually striking [see figures and further description
1212
in [294]] and reminds us that plants operate at a systems level
1213
(Fig. 1) with individual cells predisposed to differing metabolic
1214
objectives. Lessons learned from a specific tissue such as oil rich
1215
cells within seeds could be used to characterize mechanisms that
1216
limit oil production elsewhere, including leaves and other veg-
1217
etative tissue that are less than 5% lipid [295] but that could be
1218
engineered for increased energy content [8–14,296–299].
1219
Eukaryotic and prokaryotic pathways and the transport
1220
between them in lipid metabolism emphasize the complicated
1221
subcellular compartmentation that convolutes the interpretation
1222
of experiments. Relative to other species, plants have an increased
1223
number of duplicated metabolic pathways within different orga-
1224
nelles [300] including both lipid and central metabolism.
1225
Glycolysis has been reported in at least three locations of a plant
1226
cell [301] including cytosolic, plastidic, and a possible role on the
1227
surface of mitochondria [302]. Likewise both cytosolic and
1228
plastidic forms of enzymes functioning with pentose phosphate
1229
metabolism exist [210] (Fig. 10). These intricate network descrip-
1230
tions are a well-recognized challenge [6,303–308] (Fig. 11) and
1231
the implications for modeling in multiple locations have been
1232
described [222] including labeling experiments to maximize infor-
1233
mation content from the labeling experiment [75].
1234
Experimental techniques that provide information specific to
1235
cellular or subcellular locations represent one approach to more
1236
accurately assess metabolism. Organelle fractionation using non-
1237
aqueous buffers at reduced temperatures to rapidly quench meta-
1238
bolism produces minimal artefacts [309–311] and has allowed
1239
metabolite profiling at the subcellular level [312]. Extraction of sin-
1240
gle cells [313] or subcellular treatments [314] have also been devel-
1241
oped to assess the metabolome and the status of profiling methods
1242
was recently reviewed [315]. Thus the methods to combine very
1243
specific spatial profiling approaches with isotopic labeling are
1244
becoming more refined. A different strategy is to make use of the
1245
known biosynthetic locations of metabolite biosynthesis and ana-
1246
lyze the differences in enrichment from a labeling experiment.
1247
Experiments have capitalized on the distinct biosynthetic locations
1248
of fatty acids [52,59], carbohydrates [51,52], or protein-derived
1249
amino acids [316] to inform isotopic labeling on subcellular meta-
1250
bolism. The production of fatty acids includes a fatty acid synthase
1251
complex located in the chloroplast stroma that synthesizes acyl
1252
chains to lengths of 16 or 18 depending on the species. Further
1253
elongation or assembly to make lipids occurs after export to the
1254
ER. Thus the distinct locations of fatty acid biosynthesis and elonga-
1255
tion result in the incorporation of acetate groups from distinct plas-
1256
tidic or extra-plastidial sources. The inspection of the terminal
1257
acetate on labeled fatty acids of different lengths provides a means
1258
of comparing the spatially distinct acetyl-CoA pools [52,317].
1259
Similarly, methods for carbohydrates [51,52] and amino acids
1260
[316] produced in distinct biosynthetic locations of plants can
1261
provide analogous information about sugars and amino acids from
1262
different locations and have been recently extended with high
1263
resolution MS to indirectly assess amino acids through peptide
1264
labeling descriptions [318,319] assisting flux analysis [320].
1265
Developing soybean embryos were cultured with
13
C to gener-
1266
ate significant amounts of biomass labeled through metabolism
1267
(Fig. 12A). The isotopic labeling was measured in peptides obtained
1268
from proteolysis of storage proteins. A fraction of the storage pro-
1269
tein was nearly unlabeled because it was produced ‘‘on the vine’’
1270
prior to embryo culturing (Fig. 12B). This resulted in a subset of
1271
mass isotopomers with little
13
C incorporation (Fig. 12C inspired
1272
by [319];m/z < 5 has low histogram values). The subsequent meta-
1273
bolic labeling and growth of the embryos in culture resulted in a
1274
second distribution of mass isotopomers in newly synthesized
1275
storage proteins that correspond to significant
13
C in the amino
1276
acids used in protein biosynthesis (i.e., m/z >5 in Fig. 12C). Thus,
1277
the MS measurements produced a bimodal distribution within
1278
the peptides that reflected two distinct metabolic events for the
1279
embryos: growth in planta initially, without isotopic labeling, fol-
1280
lowed by growth in culture with
13
C substrates. Thus the final
1281
labeling in protein observed through the mass isotopomer descrip-
1282
tion, served as a record of the growth over time with and without
1283
provision of isotope. The labeling distributions in the individual
1284
amino acids that were necessary to generate the mass spectral dis-
1285
tribution can be determined through a computational fitting pro-
1286
cess where many peptides of different amino acid compositions
1287
and labeling descriptions are considered. Additionally the unla-
1288
beled fraction can be accurately established and accounted for
1289
based upon the composition and spectral distribution. The study
1290
indicated that protein made with isotopes during metabolism
1291
can be used to track differences in metabolism that occur
1292
temporally [319]. For comparison purposes, the same protein
1293
was hydrolyzed and labeling in individual amino acids measured
1294
using GC–MS. The amino acid mass isotopomer descriptions were
1295
mathematically convolved to regenerate a labeling description in
1296
peptides; however as presented in Fig. 12C, the temporally aver-
1297
aged amino acid labeling resulted in a mass isotopomer profile that
1298
did not exhibit a bimodal labeling pattern. The information about
1299
[
13
C]-amino acids connected to each other in peptides that
1300
occurred when made in culture and the connectivity information
1301
of [
12
C]-amino acids in peptides that were produced on the vine
1302
prior to culturing was lost when all peptides were hydrolyzed
1303
and measured by GC–MS. The GC–MS-based measurement of
1304
hydrolyzed peptides resulted in a labeling description for each
1305
amino acid that was the combination of all growth including the
1306
initial growth in planta without isotopes, as well as in culture
1307
labeling. Thus peptide measurements present a strategy for tem-
1308
poral metabolism (and analogously for spatial metabolism, see
1309
[316]) and compartmentalized flux analyses [320].
1310
4.3.3. From CO
2
to lipid: temporal labeling-based MFA approaches
1311
Steady state MFA descriptions are limited to tissues that exhibit
1312
unchanging metabolism for long durations and to networks con-
1313
taining branch points with enzymatic bond-breaking and reform-
1314
ing reactions. The steady state enrichments of different atoms
Fig. 13. S7P isotopic labeling during autotrophic metabolism in leaves. Isotopic
labeling of sedoheptulose-7-phosphate (S7P), an intermediate of the Calvin cycle.
The mass isotopomers indicate the labeling trajectories that reflect the incorpora-
tion of [
13
C]O
2
sequentially into S7P. Initially the metabolite is unlabeled as
indicated by 100% M0 composition at time zero, then as the time exposed to [
13
C]O
2
increases the M1 pool increases but at some point is replaced by M2. The pattern
continues until S7P becomes highly labeled. The inset graph indicates the average
labeling per carbon. The average labeling is approaching a value of 80% with time,
indicating the presence of pools that were inactive within the labeling duration
(20% of the total S7P pool size).
16 D.K. Allen et al. / Progress in Lipid Research xxx (2015) xxx–xxx
JPLR 873 No. of Pages 24, Model 5G
16 March 2015
Please cite this article in press as: Allen DK et al. Tracking the metabolic pulse of plant lipid production with isotopic labeling and flux analyses: Past, pre-
sent and future. Prog Lipid Res (2015), http://dx.doi.org/10.1016/j.plipres.2015.02.002

1315
within a metabolite or end product can indicate the relative use of
1316
different pathways; however many tissues exhibit only brief peri-
1317
ods of constant metabolism (e.g., leaf metabolism is diurnal) or in
1318
some cases all metabolites become fully labeled with time due to
1319
an exclusive source of carbon (e.g., autotrophic metabolism with
1320
[
13
C]O
2
provision). Still other paths lack branch points and do not
1321
exhibit enzymatic isotope rearrangements within a molecule. The
1322
latter is often a feature of secondary metabolic pathways that are
1323
linear or fatty acid biosynthesis where the repeated addition of
1324
labeled acetyl groups results in nearly completely labeled
1325
intermediates at isotopic steady state. Thus additional information
1326
such as the enrichment over time is needed. For this reason con-
1327
tinuous pulse or pulse-chase metabolic labeling of lipid assembly
1328
are preferred.
1329
The temporal measurement of stable isotopic labeling is analo-
1330
gous to the dynamic radiolabeling approaches (but with stable iso-
1331
topes) and is an appealing development because each time point
1332
contributes information that complements steady state labeling
1333
analysis. Therefore the additional measurements of labeling over
1334
time in a temporal analysis (Fig. 3) provide a richer set of data
1335
for modeling purposes [321] over steady state analysis alone.
1336
Additionally, because the earliest time points are the most sensi-
1337
tive to label provision, they provide a significant amount of infor-
1338
mation and the experimental duration can be shortened
1339
[322,323]. Recent studies have described the incorporation of
13
C
1340
into whole plants [324–326] and specific photosynthetic tissues,
1341
cells or unicellular organisms [66,67,70,327,328]. When [
13
C]O
2
is
1342
provided to plants, the experiment is non-invasive and metabolic
1343
data reflects in vivo operation. Fluxes in Arabidopsis leaves were
1344
recently assessed by kinetic flux profiling that models unlabeled
1345
mass isotopomers levels in metabolites [66]. Estimates for active
1346
and inactive pools were used to overcome the spatial challenges
1347
of a plant system and describe photosynthetic metabolism.
1348
Autotrophic metabolism has also been described mathematically
1349
using non-stationary state MFA applied to unicellular systems
1350
[65,70] and the same approach has now been leveraged to investi-
1351
gate higher plants [69]. The application of non-stationary MFA uti-
1352
lizes all isotopomer data, which is fitted to accommodate the
1353
labeling trajectories between time points (Fig. 13). Thus multiple
1354
evaluations in time provide an enriched data set relative to steady
1355
state investigations (Fig. 3A) and can be used to distinguish path-
1356
ways. The modeling of isotopomer pools allows estimation of the
1357
inactive pools through pool dilution fluxes and therefore presents
1358
an alternative way to obtain information on spatial complexity,
1359
possibly without some of the pitfalls associated with direct experi-
1360
mental measurements. For example, the flux through photores-
1361
piration, a pathway with up to 16 spatially resolved pools was
1362
examined through this strategy and led to quantification without
1363
assumptions about the ratio of carboxylation to oxygenation [69].
1364
A number of other reports, mostly in non-plant systems are start-
1365
ing to surface on related use and applications [199,329–333].
1366
The power of computational nonstationary MFA and other
1367
methods lies in the abundance of measurements that contribute
1368
to establish an overdetermined set of network differential mass
1369
balance equations that describe metabolism. Though the networks
1370
are generally simplifications, future analyses can be expected to
1371
become more complicated and benefit from advances in labeling
1372
and mass spectrometry. In particular mass spectrometry collision
1373
cell technologies are available but largely untapped such as elec-
1374
tron transfer dissociation (ETD), and higher energy collision
1375
dissociation (HCD). Together with CID methods, fragmentation
1376
can be optimized for lipids or other macromolecules depending
1377
on the biological question of interest. Regiospecific information,
1378
higher resolution techniques to distinguish similar compounds,
1379
positive and negative ionization for different lipid classes,
1380
increased MS
n
afforded in linear and orbital trap MS with small
1381
quantities, and separation technologies such as ion mobility that
1382
capitalize on other physical properties all represent technological
1383
opportunities that remain largely unexplored. As these capacities
1384
are further developed, compounds will be identified less ambigu-
1385
ously and positional (not just mass) isotopomer descriptions will
1386
become available. Fig. 14 illustrates the additional measurement
1387
information that can be obtained from tandem MS that is a step
1388
in this direction. Whereas a single quadrupole has the capacity to
1389
measure intact as well as fragments of a metabolite, tandem MS
1390
can link fragmented products back to labeling in their precursors.
1391
Thus the connection between precursor and product ions results
1392
in additional information relative to independent monitoring of
1393
each. As indicated in the figure a four carbon compound (i.e., 4 cir-
1394
cles) results in five mass isotopomers of which four are indepen-
1395
dent (i.e., one is redundant if they must sum to 100% to account
1396
for total labeling description), and the measurement of a second,
1397
two carbon fragment adds an additional three measurements of
1398
which two are independent. Together the MS of these two frag-
1399
ments provides six independent mass isotopomer measurements;
1400
however, by monitoring the transition from precursor to product
1401
ions, nine mass isotopomers groups can be measured of which
1402
eight are independent from one another. Thus the tandem MS of
1403
the same two fragments results in 25% (i.e., 6 vs. 8 measurements)
1404
more information. Further details on the number of independent
1405
measurements from tandem MS is presented elsewhere [334].It
1406
is reasonable to expect that with higher power MS
n
techniques
1407
or different fragment evaluations a complete isotopomer descrip-
1408
tion may be achieved; reducing further the guesswork in model
1409
descriptions and biological interpretation.
1410
5. Conclusions and perspective
1411
The engineering of primary metabolism including the accumu-
1412
lation of lipids in plant tissues remains a challenging endeavor
Fig. 14. Comparison of mass isotopomer measurements for isotopically labeled
fragments using single and tandem mass spectrometers. Assessment of mass
isotopomers with mass spectrometry can benefit from linking fragments (product
ions) to their precursor molecules. In the example a four carbon compound is
fragmented in the mass spectrometer resulting in a 2 carbon product that can be
measured. (A) In a single quadrupole instrument the fragment may be detectable
along with some of the remaining intact four carbon molecule resulting in 8 mass
spectral measurements, 5 from the four carbon product and 3 from the two carbon
fragment. In both cases the mass isotopomers must account for 100% of the
fractional labeling, thus the number of independent measurements are 4 and 2,
respectively. (B) When the measurements for the same four and two carbon
compound can be directly linked through the use of a tandem mass spectrometer,
the number of independent measurements increases to 8 (i.e., 9 measurements, 8
that are independent).
D.K. Allen et al. / Progress in Lipid Research xxx (2015) xxx–xxx 17
JPLR 873 No. of Pages 24, Model 5G
16 March 2015
Please cite this article in press as: Allen DK et al. Tracking the metabolic pulse of plant lipid production with isotopic labeling and flux analyses: Past, pre-
sent and future. Prog Lipid Res (2015), http://dx.doi.org/10.1016/j.plipres.2015.02.002

1413
despite intensive research efforts. Fundamental to our understand-
1414
ing of lipid pathways, isotopic labeling methods provide a dynamic
1415
description of metabolic operation including network fluxes. The
1416
most basic aspects of lipid metabolism in plants including the
1417
eukaryotic and prokaryotic pathways of membrane and storage
1418
lipid assembly were largely defined by isotopic labeling studies.
1419
Future investigations can be anticipated to be equally trans-
1420
formative for our understanding of lipid metabolism and have a
1421
significant impact on modern day problems in food and energy,
1422
in part because of the availability of high purity commercial iso-
1423
topes and technologies such as MS that are sensitive with high res-
1424
olution. Coupled with MS imaging, or more specific metabolic
1425
readouts, isotopic labels can account for heterogeneity, tissue-level
1426
and subcellular differences in metabolism. Other dynamic
1427
lipidomic studies involving isotopic labeling with or without high
1428
resolution MS, and nonstationary MFA will define fluxes quan-
1429
titatively and establish emergent biological properties. Finally,
1430
the use of isotopic labeling methods with environmental or geneti-
1431
cally altered plants will be essential in the further elucidation of
1432
unknown reactions and/or pathways within lipid metabolism,
1433
and is starting to allow metabolic assessment of regulation and
1434
control mechanisms that remain the current frontier of metabolic
1435
research in many species. Elucidation of lipid metabolism is no
1436
longer technologically limited, but possibly our imagination and
1437
ability to leverage available tools in new clever experiments
1438
present the greatest hurdle to Progress in Lipid Research.
1439
Conflicts of interest
1440
The authors declare that there are no conflicts of interest.
1441
Acknowledgements
1442
We gratefully acknowledge conversations with Drs. John
1443
Ohlrogge and Jan Jaworski on topics related to the review and its
1444
revision. Work in the authors’ labs was supported by a
1445
Department of Energy grant (DE-AR0000202; D.K.A.) and the
1446
Great Lakes Bioenergy Research Center Cooperative Agreement
1447
(DE-FC02-07ER64494; H.T.), the National Science Foundation
1448
(EF-1105249; D.K.A.), and the USDA-ARS. Any product or trade-
1449
mark mentioned here does not imply a warranty, guarantee, or
1450
endorsement by the authors or their affiliations over other suitable
1451
products.
1452
References
1453 [1] Gerosa L, Sauer U. Regulation and control of metabolic fluxes in microbes.
1454 Curr Opin Biotechnol 2011;22:566–75.
1455 [2] Li-Beisson Y, Shorrosh B, Beisson F, Andersson MX, Arondel V, Bates PD, et al.
1456 Acyl-lipid metabolism. Arabidopsis Book 2013;11:e0161.
1457 [3] McGlew K, Shaw V, Zhang M, Kim RJ, Yang W, Shorrosh B, et al. An annotated
1458 database of Arabidopsis mutants of acyl lipid metabolism. Plant Cell Rep
1459 2014.
1460 [4] Graham IA, Eastmond PJ. Pathways of straight and branched chain fatty acid
1461 catabolism in higher plants. Prog Lipid Res 2002;41:156–81.
1462 [5] Bates PD, Browse J. The significance of different diacylgycerol synthesis
1463 pathways on plant oil composition and bioengineering. Front Plant Sci
1464 2012;3:147.
1465 [6] Allen DK, Libourel IGL, Shachar-Hill Y. Metabolic flux analysis in plants:
1466 coping with complexity. Plant Cell Environ 2009;32:1241–57.
1467 [7] Theodoulou FL, Eastmond PJ. Seed storage oil catabolism: a story of give and
1468 take. Curr Opin Plant Biol 2012;15:322–8.
1469 [8] Vanhercke T, El Tahchy A, Liu Q, Zhou XR, Shrestha P, Divi UK, et al. Metabolic
1470 engineering of biomass for high energy density: oilseed-like triacylglycerol
1471 yields from plant leaves. Plant Biotechnol J 2014;12:231–9.
1472 [9] Slocombe SP, Cornah J, Pinfield-Wells H, Soady K, Zhang Q, Gilday A, et al. Oil
1473 accumulation in leaves directed by modification of fatty acid breakdown and
1474 lipid synthesis pathways. Plant Biotechnol J 2009;7:694–703.
1475 [10] Durrett TP, Benning C, Ohlrogge J. Plant triacylglycerols as feedstocks for the
1476 production of biofuels. Plant J 2008;54:593–607.
1477 [11] Ohlrogge J, Allen D, Berguson B, DellaPenna D, Shachar-Hill Y, Stymne S.
1478 Driving on biomass. Science 2009;324:1019–20.
1479
[12] Winichayakul S, Scott RW, Roldan M, Hatier JH, Livingston S, Cookson R, et al.
1480
In vivo packaging of triacylglycerols enhances Arabidopsis leaf biomass and
1481
energy density. Plant Physiol 2013;162:626–39.
1482
[13] Chapman KD, Dyer JM, Mullen RT. Commentary: why don’t plant leaves get
1483
fat? Plant Sci 2013;207:128–34.
1484
[14] Fan J, Yan C, Zhang X, Xu C. Dual role for phospholipid:diacylglycerol
1485
acyltransferase: enhancing fatty acid synthesis and diverting fatty acids from
1486
membrane lipids to triacylglycerol in Arabidopsis leaves. Plant Cell
1487
2013;25:3506–18.
1488
[15] Lardizabal K, Effertz R, Levering C, Mai J, Pedroso MC, Jury T, et al. Expression
1489
of Umbelopsis ramanniana DGAT2A in seed increases oil in soybean. Plant
1490
Physiol 2008;148:89–96.
1491
[16] Taylor DC, Zhang Y, Kumar A, Francis T, Giblin EM, Barton DL, et al. Molecular
1492
modification of triacylglycerol accumulation by over-expression of DGAT1 to
1493
produce canola with increased seed oil content under field conditions. Botany
1494
2009;87:533–43.
1495
[17] Shen B, Allen WB, Zheng P, Li C, Glassman K, Ranch J, et al. Expression of
1496
ZmLEC1 and ZmWRI1 increases seed oil production in maize. Plant Physiol
1497
2010;153:980–7.
1498
[18] Haslam RP, Ruiz-Lopez N, Eastmond P, Moloney M, Sayanova O, Napier JA.
1499
The modification of plant oil composition via metabolic engineering – better
1500
nutrition by design. Plant Biotechnol J 2013;11:157–68.
1501
[19] Vanhercke T, Wood CC, Stymne S, Singh SP, Green AG. Metabolic engineering
1502
of plant oils and waxes for use as industrial feedstocks. Plant Biotechnol J
1503
2013;11:197–210.
1504
[20] Carlsson AS, Yilmaz JL, Green AG, Stymne S, Hofvander P. Replacing fossil oil
1505
with fresh oil – with what and for what? Eur J Lipid Sci Technol
1506
2011;113:812–31.
1507
[21] Wu H, Tai YC. A novel phoswich imaging detector for simultaneous beta and
1508
coincidence-gamma imaging of plant leaves. Phys Med Biol
1509
2011;56:5583–98.
1510
[22] Dirks RC, Singh M, Potter GS, Sobotka LG, Schaefer J. Carbon partitioning in
1511
soybean (Glycine max) leaves by combined
11
C and
13
C labeling. New Phytol
1512
2012;196:1109–21.
1513
[23] Jahnke S, Menzel MI, van Dusschoten D, Roeb GW, Buhler J, Minwuyelet S,
1514
et al. Combined MRI-PET dissects dynamic changes in plant structures and
1515
functions. Plant J 2009;59:634–44.
1516
[24] Babst BA, Karve AA, Judt T. Radio-metabolite analysis of carbon-11
1517
biochemical partitioning to non-structural carbohydrates for integrated
1518
metabolism and transport studies. Plant Cell Physiol 2013;54:1016–25.
1519
[25] Minchin PE, Thorpe MR, Farrar JF, Koroleva OA. Source-sink coupling in young
1520
barley plants and control of phloem loading. J Exp Bot 2002;53:1671–6.
1521
[26] Gould N, Thorpe MR, Pritchard J, Christeller JT, Williams LE, Roeb G, et al.
1522
AtSUC2 has a role for sucrose retrieval along the phloem pathway: evidence
1523
from carbon-11 tracer studies. Plant Sci 2012;188–189:97–101.
1524
[27] Pollard M, Ohlrogge J. Testing models of fatty acid transfer and lipid synthesis
1525
in spinach leaf using in vivo oxygen-18 labeling. Plant Physiol
1526
1999;121:1217–26.
1527
[28] Feng X, Tang YJ. Evaluation of isotope discrimination in
13
C-based metabolic
1528
flux analysis. Anal Biochem 2011;417:295–7.
1529
[29] Wasylenko TM, Stephanopoulos G. Kinetic isotope effects significantly
1530
influence intracellular metabolite
13
C labeling patterns and flux
1531
determination. Biotechnol J 2013;8:1080–9.
1532
[30] Kruger NJ, Huddleston JE, Le Lay P, Brown ND, Ratcliffe RG. Network flux
1533
analysis: impact of
13
C-substrates on metabolism in Arabidopsis thaliana cell
1534
suspension cultures. Phytochemistry 2007;68:2176–88.
1535
[31] McNevin DB, Badger MR, Whitney SM, von Caemmerer S, Tcherkez GG,
1536
Farquhar GD. Differences in carbon isotope discrimination of three variants of
1537
D
-ribulose-1,5-bisphosphate carboxylase/oxygenase reflect differences in
1538
their catalytic mechanisms. J Biol Chem 2007;282:36068–76.
1539
[32] Crayn DM, Winter K, Smith JA. Multiple origins of crassulacean acid
1540
metabolism and the epiphytic habit in the Neotropical family Bromeliaceae.
1541
Proc Natl Acad Sci U S A 2004;101:3703–8.
1542
[33] Lin YH, Pawlosky RJ, Salem Jr N. Simultaneous quantitative determination of
1543
deuterium- and carbon-13-labeled essential fatty acids in rat plasma. J Lipid
1544
Res 2005;46:1974–82.
1545
[34] McNaught AD, Wilkinson A. IUPAC. Compendium of chemical terminology.
1546
2nd ed. (the ‘‘Gold Book’’). 2nd Revised edition ed. WileyBlackwell.
1547
[35] Hellerstein MK, Christiansen M, Kaempfer S, Kletke C, Wu K, Reid JS, et al.
1548
Measurement of de novo hepatic lipogenesis in humans using stable isotopes.
1549
J Clin Invest 1991;87:1841–52.
1550
[36] Kelleher JK, Masterson TM. Model equations for condensation biosynthesis
1551
using stable isotopes and radioisotopes. Am J Physiol 1992;262:E118–25.
1552
[37] Turner SM, Murphy EJ, Neese RA, Antelo F, Thomas T, Agarwal A, et al.
1553
Measurement of TG synthesis and turnover in vivo by
2
H
2
O incorporation into
1554
the glycerol moiety and application of MIDA. Am J Physiol Endocrinol Metab
1555
2003;285:E790–803.
1556
[38] Kharroubi AT, Masterson TM, Aldaghlas TA, Kennedy KA, Kelleher JK.
1557
Isotopomer spectral analysis of triglyceride fatty acid synthesis in 3T3-L1
1558
cells. Am J Physiol 1992;263:E667–75.
1559
[39] Lee WNP, Byerley LO, Bergner EA, Edmond J. Mass isotopomer analysis:
1560
theoretical and practical considerations. Biol Mass Spectrom 1991;20:
1561
451–8.
1562
[40] Fernandez CA, Des Rosiers C, Previs SF, David F, Brunengraber H. Correction of
1563
13
C mass isotopomer distributions for natural stable isotope abundance. J
1564
Mass Spectrom 1996;31:255–62.
18 D.K. Allen et al. / Progress in Lipid Research xxx (2015) xxx–xxx
JPLR 873 No. of Pages 24, Model 5G
16 March 2015
Please cite this article in press as: Allen DK et al. Tracking the metabolic pulse of plant lipid production with isotopic labeling and flux analyses: Past, pre-
sent and future. Prog Lipid Res (2015), http://dx.doi.org/10.1016/j.plipres.2015.02.002

1565 [41] Stephanopoulos G, Aristidou A, Nielsen J. Metabolic engineering: principles
1566 and methodologies. San Diego: Academic Press; 1998.
1567 [42] Zamboni N, Fendt SM, Ruhl M, Sauer U.
13
C-based metabolic flux analysis. Nat
1568 Protoc 2009;4:878–92.
1569 [43] Wiechert W, Mollney M, Petersen S, de Graaf AA. A universal framework for
1570
13
C metabolic flux analysis. Metab Eng 2001;3:265–83.
1571 [44] Ratcliffe RG. In-vivo NMR-studies of higher-plants and algae. Adv Bot Res
1572 1994;20:43–123.
1573 [45] Ratcliffe RG, Shachar-Hill Y. Probing plant metabolism with NMR. Annu Rev
1574 Plant Physiol Plant Mol Biol 2001;52:499–526.
1575 [46] Allen DK, Ratcliffe RG. Quantification of isotope label. In: Schwender J, editor.
1576 Plant metabolic networks. New York: Springer; 2009. p. 105–49.
1577 [47] Truong Q, Shanks JV. Analysis of proteinogenic amino acid and starch labeling
1578 by 2D NMR. Methods Mol Biol 2014;1090:87–105.
1579 [48] Truong QX, Yoon JM, Shanks JV. Isotopomer measurement techniques in
1580 metabolic flux analysis I: nuclear magnetic resonance. Methods Mol Biol
1581 2014;1083:65–83.
1582 [49] Boisseau R, Charrier B, Massou S, Portais JC, Akoka S, Giraudeau P. Fast
1583 spatially encoded 3D NMR strategies for
13
C-based metabolic flux analysis.
1584 Anal Chem 2013;85:9751–7.
1585 [50] Masakapalli SK, Ratcliffe RG, Williams TC. Quantification of
13
C enrichments
1586 and isotopomer abundances for metabolic flux analysis using 1D NMR
1587 spectroscopy. Methods Mol Biol 2014;1090:73–86.
1588 [51] Sriram G, Iyer VV, Bruce Fulton D, Shanks JV. Identification of hexose
1589 hydrolysis products in metabolic flux analytes: a case study of levulinic acid
1590 in plant protein hydrolysate. Metab Eng 2007;9:442–51.
1591 [52] Allen DK, Shachar-Hill Y, Ohlrogge JB. Compartment-specific labeling
1592 information in
13
C metabolic flux analysis of plants. Phytochemistry
1593 2007;68:2197–210.
1594 [53] Giraudeau P, Massou S, Robin Y, Cahoreau E, Portais JC, Akoka S. Ultrafast
1595 quantitative 2D NMR: an efficient tool for the measurement of specific
1596 isotopic enrichments in complex biological mixtures. Anal Chem
1597 2011;83:3112–9.
1598 [54] Giraudeau P, Cahoreau E, Massou S, Pathan M, Portais JC, Akoka S. UFJCOSY: a
1599 fast 3D NMR method for measuring isotopic enrichments in complex
1600 samples. ChemPhysChem 2012;13:3098–101.
1601 [55] Jin ES, Jones JG, Merritt M, Burgess SC, Malloy CR, Sherry AD. Glucose
1602 production, gluconeogenesis, and hepatic tricarboxylic acid cycle fluxes
1603 measured by nuclear magnetic resonance analysis of a single glucose
1604 derivative. Anal Biochem 2004;327:149–55.
1605 [56] Shachar-Hill Y, Pfeffer PE, Ratcliffe RG. Measuring nitrate in plant cells by
1606 in vivo NMR using Gd
3+
as a shift reagent. J Magn Reson B 1996;111:9–14.
1607 [57] Allen DK, Young JD. Carbon and nitrogen provisions alter the metabolic flux in
1608 developing soybean embryos. Plant Physiol 2013;161:1458–75.
1609 [58] Alonso AP, Val DL, Shachar-Hill Y. Central metabolic fluxes in the endosperm
1610 of developing maize seeds and their implications for metabolic engineering.
1611 Metab Eng 2011;13:96–107.
1612 [59] Lonien J, Schwender J. Analysis of metabolic flux phenotypes for two
1613 arabidopsis mutants with severe impairment in seed storage lipid
1614 synthesis. Plant Physiol 2009;151:1617–34.
1615 [60] Troufflard S, Roscher A, Thomasset B, Barbotin JN, Rawsthorne S, Portais JC. In
1616 vivo C-13 NMR determines metabolic fluxes and steady state in, linseed
1617 embryos. Phytochemistry 2007;68:2341–50.
1618 [61] Junker BH, Lonien J, Heady LE, Rogers A, Schwender J. Parallel determination
1619 of enzyme activities and in vivo fluxes in Brassica napus embryos grown on
1620 organic or inorganic nitrogen source. Phytochemistry 2007;68:2232–42.
1621 [62] Williams TCR, Miguet L, Masakapalli SK, Kruger NJ, Sweetlove LJ, Ratcliffe RG.
1622 Metabolic network fluxes in heterotrophic Arabidopsis cells: stability of the
1623 flux distribution under different oxygenation conditions. Plant Physiol
1624 2008;148:704–18.
1625 [63] Masakapalli SK, Kruger NJ, Ratcliffe RG. The metabolic flux phenotype of
1626 heterotrophic Arabidopsis cells reveals a complex response to changes in
1627 nitrogen supply. Plant J 2013;74:569–82.
1628 [64] Truong Q, Koch K, Yoon JM, Everard JD, Shanks JV. Influence of carbon to
1629 nitrogen ratios on soybean somatic embryo (cv. Jack) growth and
1630 composition. J Exp Bot 2013;64:2985–95.
1631 [65] Shastri AA, Morgan JA. A transient isotopic labeling methodology for
13
C
1632 metabolic flux analysis of photoautotrophic microorganisms. Phytochemistry
1633 2007;68:2302–12.
1634 [66] Szecowka M, Heise R, Tohge T, Nunes-Nesi A, Vosloh D, Huege J, et al.
1635 Metabolic fluxes in an illuminated Arabidopsis rosette. Plant Cell
1636 2013;25:694–714.
1637 [67] Hasunuma T, Harada K, Miyazawa S, Kondo A, Fukusaki E, Miyake C.
1638 Metabolic turnover analysis by a combination of in vivo
13
C-labelling from
1639
13
CO
2
and metabolic profiling with CE-MS/MS reveals rate-limiting steps of
1640 the C3 photosynthetic pathway in Nicotiana tabacum leaves. J Exp Bot
1641 2010;61:1041–51.
1642 [68] Jazmin LJ, O’Grady JP, Ma F, Allen DK, Morgan JA, Young JD. Isotopically
1643 nonstationary MFA (INST-MFA) of autotrophic metabolism. Methods Mol Biol
1644 2014;1090:181–210.
1645 [69] Ma F, Jazmin LJ, Young JD, Allen DK. Isotopically nonstationary
13
C flux
1646 analysis of changes in Arabidopsis thaliana leaf metabolism due to high light
1647 acclimation. Proc Natl Acad Sci U S A 2014;111:16967–72.
1648 [70] Young JD, Shastri AA, Stephanopoulos G, Morgan JA. Mapping
1649 photoautotrophic metabolism with isotopically nonstationary
13
C flux
1650 analysis. Metab Eng 2011;13:656–65.
1651
[71] Segel IH. Biochemical calculations. John Wiley and Sons, Inc.; 1976.
1652
[72] Libourel IG, Gehan JP, Shachar-Hill Y. Design of substrate label for steady state
1653
flux measurements in plant systems using the metabolic network of Brassica
1654
napus embryos. Phytochemistry 2007;68:2211–21.
1655
[73] Leighty RW, Antoniewicz MR. COMPLETE-MFA: complementary parallel
1656
labeling experiments technique for metabolic flux analysis. Metab Eng
1657
2013;20:49–55.
1658
[74] Mollney M, Wiechert W, Kownatzki D, de Graaf AA. Bidirectional reaction
1659
steps in metabolic networks: IV. Optimal design of isotopomer labeling
1660
experiments. Biotechnol Bioeng 1999;66:86–103.
1661
[75] Nargund S, Sriram G. Designer labels for plant metabolism: statistical design
1662
of isotope labeling experiments for improved quantification of flux in
1663
complex plant metabolic networks. Mol BioSyst 2013;9:99–112.
1664
[76] Walther JL, Metallo CM, Zhang J, Stephanopoulos G. Optimization of
13
C
1665
isotopic tracers for metabolic flux analysis in mammalian cells. Metab Eng
1666
2012;14:162–71.
1667
[77] Metallo CM, Walther JL, Stephanopoulos G. Evaluation of
13
C isotopic tracers
1668
for metabolic flux analysis in mammalian cells. J Biotechnol
1669
2009;144:167–74.
1670
[78] Crown SB, Antoniewicz MR. Selection of tracers for
13
C-metabolic flux
1671
analysis using elementary metabolite units (EMU) basis vector methodology.
1672
Metab Eng 2012;14:150–61.
1673
[79] Millard P, Sokol S, Letisse F, Portais JC. IsoDesign: a software for optimizing
1674
the design of
13
C-metabolic flux analysis experiments. Biotechnol Bioeng
1675
2014;111:202–8.
1676
[80] Ruben S, Randall M, Kamen M, Hyde JL. Heavy oxygen (O
18
) as a tracer in the
1677
study of photosynthesis. J Am Chem Soc 1941;63:877–9.
1678
[81] Ruben S, Hassid WZ, Kamen MD. Radioactive carbon in the study of
1679
photosynthesis. J Am Chem Soc 1939;61:661–3.
1680
[82] Bassham JA, Benson AA, Calvin M. The path of carbon in photosynthesis. J Biol
1681
Chem 1950;185:781–7.
1682
[83] Hatch MD, Slack CR. Photosynthesis by sugar-cane leaves. A new
1683
carboxylation reaction and the pathway of sugar formation. Biochem J
1684
1966;101:103–11.
1685
[84] Schoenheimer R, Rittenberg D. Deuterium as an indicator in the study of
1686
intermediary metabolism. Science 1935;82:156–7.
1687
[85] Rittenberg D, Schoenheimer R. Deuterium as an indicator in the study of
1688
intermediary metabolism: VIII. Hydrogenation of fatty acids in the animal
1689
organism. J Biol Chem 1937;117:485–90.
1690
[86] Bligh EG, Dyer WJ. A rapid method of total lipid extraction and purification.
1691
Can J Biochem Physiol 1959;37:911–7.
1692
[87] Roughan PG, Slack CR. Cellular organization of glycerolipid metabolism. Annu
1693
Rev Plant Physiol 1982;33:97–132.
1694
[88] Harwood JL. Recent advances in the biosynthesis of plant fatty acids. Biochim
1695
Biophys Acta 1996;1301:7–56.
1696
[89] Bao X, Focke M, Pollard M, Ohlrogge J. Understanding in vivo carbon
1697
precursor supply for fatty acid synthesis in leaf tissue. Plant J 2000;22:39–50.
1698
[90] Harwood JL, Stumpf PK. Fat metabolism in higher plants. LI. Palmitic and
1699
stearic synthesis by an avocado supernatant system. Arch Biochem Biophys
1700
1972;148:282–90.
1701
[91] Roughan PG, Slack CR, Holland R. High rates of [1-
14
C]acetate incorporation
1702
into the lipid of isolated spinach chloroplasts. Biochem J 1976;158:
1703
593–601.
1704
[92] Slack CR, Roughan PG. The kinetics of incorporation in vivo of [
14
C]acetate and
1705
[
14
C]carbon dioxide into the fatty acids of glycerolipids in developing leaves.
1706
Biochem J 1975;152:217–28.
1707
[93] Bates PD, Ohlrogge JB, Pollard M. Incorporation of newly synthesized fatty
1708
acids into cytosolic glycerolipids in pea leaves occurs via acyl editing. J Biol
1709
Chem 2007;282:31206–16.
1710
[94] Williams JP, Imperial V, Khan MU, Hodson JN. The role of phosphatidylcholine
1711
in fatty acid exchange and desaturation in Brassica napus L. leaves. Biochem J
1712
2000;349:127–33.
1713
[95] Browse J, Roughan PG, Slack CR. Light control of fatty acid synthesis and
1714
diurnal fluctuations of fatty acid composition in leaves. Biochem J
1715
1981;196:347–54.
1716
[96] Slack CR, Roughan PG, Balasingham N. Labelling studies in vivo on the
1717
metabolism of the acyl and glycerol moieties of the glycerolipids in the
1718
developing maize leaf. Biochem J 1977;162:289–96.
1719
[97] Gurr MI, Blades J, Appleby RS, Smith CG, Robinson MP, Nichols BW. Studies on
1720
seed-oil triglycerides. Triglyceride biosynthesis and storage in whole seeds
1721
and oil bodies of Crambe abyssinica. Eur J Biochem 1974;43:281–90.
1722
[98] James AT. The biosynthesis of long-chain saturated and unsaturated fatty
1723
acids in isolated plant leaves. Biochim Biophys Acta 1963;70:9–19.
1724
[99] Murphy DJ, Stumpf PK. In vivo pathway of oleate and linoleate desaturation
1725
in developing cotyledons of Cucumis sativus L. seedlings. Plant Physiol
1726
1980;66:666–71.
1727
[100] Hawke JC, Stumpf PK. The incorporation of oleic and linoleic acids and their
1728
desaturation products into the glycerolipids of maize leaves. Arch Biochem
1729
Biophys 1980;203:296–306.
1730
[101] Gurr MI, Robinson MP, James AT. The mechanism of formation of
1731
polyunsaturated fatty acids by photosynthetic tissue. The tight coupling of
1732
oleate desaturation with phospholipid synthesis in Chlorella vulgaris. Eur J
1733
Biochem 1969;9:70–8.
1734
[102] Pollard MR, Stumpf PK. Biosynthesis of C
20
and C
22
fatty acids by developing
1735
seeds of Limnanthes alba: chain elongation and delta 5 desaturation. Plant
1736
Physiol 1980;66:649–55.
D.K. Allen et al. / Progress in Lipid Research xxx (2015) xxx–xxx 19
JPLR 873 No. of Pages 24, Model 5G
16 March 2015
Please cite this article in press as: Allen DK et al. Tracking the metabolic pulse of plant lipid production with isotopic labeling and flux analyses: Past, pre-
sent and future. Prog Lipid Res (2015), http://dx.doi.org/10.1016/j.plipres.2015.02.002

1737 [103] Pollard MR, Stumpf PK. Long chain (C
20
and C
22
) fatty acid biosynthesis in
1738 developing seeds of Tropaeolum majus:anin vivo study. Plant Physiol
1739 1980;66:641–8.
1740 [104] Ohlrogge JB, Kuhn DN, Stumpf PK. Subcellular localization of acyl carrier
1741 protein in leaf protoplasts of Spinacia oleracea. Proc Natl Acad Sci U S A
1742 1979;76:1194–8.
1743 [105] Weiss SB, Kennedy EP, Kiyasu JY. The enzymatic synthesis of triglycerides. J
1744 Biol Chem 1960;235:40–4.
1745 [106] Barron EJ, Stumpf PK. Fat metabolism in higher plants. XIX. The biosynthesis
1746 of triglycerides by avocado-mesocarp enzymes. Biochim Biophys Acta
1747 1962;60:329–37.
1748 [107] Kornberg A, Pricer WE. Enzymatic esterification of
a
-glycerophosphate by
1749 long chain fatty acids. J Biol Chem 1953;204:345–57.
1750 [108] Kornberg A, Pricer WE. Enzymatic synthesis of the coenzyme A derivatives of
1751 long chain fatty acids. J Biol Chem 1953;204:329–43.
1752 [109] Kennedy EP, Weiss SB. The function of cytidine coenzymes in the
1753 biosynthesis of phospholipides. J Biol Chem 1956;222:193–214.
1754 [110] Stymne S, Stobart AK. Oil synthesis in vitro in microsomal membranes from
1755 developing cotyledons of Linum usitatissimum L. Planta 1985;164:101–4.
1756 [111] Mazelis M, Stumpf PK. Fat metabolism in higher plants. VI. Incorporation of P
1757 into peanut mitochondrial phospholipids. Plant Physiol 1955;30:237–43.
1758 [112] Bradbeer C, Stumpf PK. Fat metabolism in higher plants. XIII. Phosphatidic
1759 acid synthesis and diglyceride phosphokinase activity in mitochondria from
1760 peanut cotyledons. J Lipid Res 1960;1:214–20.
1761 [113] Sastry PS, Kates M. Biosynthesis of lipids in plants. II. Incorporation of
1762 glycerophosphate-
32
-P into phosphatides by cell-free preparations from
1763 spinach leaves. Can J Biochem 1966;44:459–67.
1764 [114] Cheniae GM. Phosphatidic acid and glyceride synthesis by particles from
1765 spinach leaves. Plant Physiol 1965;40:235–43.
1766 [115] Banas W, Sanchez Garcia A, Banas A, Stymne S. Activities of acyl-
1767 CoA:diacylglycerol acyltransferase (DGAT) and phospholipid:diacylglycerol
1768 acyltransferase (PDAT) in microsomal preparations of developing sunflower
1769 and safflower seeds. Planta 2013;237:1627–36.
1770 [116] Lager I, Yilmaz JL, Zhou XR, Jasieniecka K, Kazachkov M, Wang P, et al. Plant
1771 acyl-CoA:lysophosphatidylcholine acyltransferases (LPCATs) have different
1772 specificities in their forward and reverse reactions. J Biol Chem
1773 2013;288:36902–14.
1774 [117] Lung SC, Weselake RJ. Diacylglycerol acyltransferase: a key mediator of plant
1775 triacylglycerol synthesis. Lipids 2006;41:1073–88.
1776 [118] Eccleston V, Cranmer A, Voelker T, Ohlrogge J. Medium-chain fatty acid
1777 biosynthesis and utilization in Brassica napus plants expressing lauroyl-acyl
1778 carrier protein thioesterase. Planta 1996;198:46–53.
1779 [119] Stahl U, Stalberg K, Stymne S, Ronne H. A family of eukaryotic
1780 lysophospholipid acyltransferases with broad specificity. FEBS Lett
1781 2008;582:305–9.
1782 [120] Bates PD, Fatihi A, Snapp AR, Carlsson AS, Browse J, Lu C. Acyl editing and
1783 headgroup exchange are the major mechanisms that direct polyunsaturated
1784 fatty acid flux into triacylglycerols. Plant Physiol 2012;160:1530–9.
1785 [121] Lu C, Xin Z, Ren Z, Miquel M, Browse J. An enzyme regulating triacylglycerol
1786 composition is encoded by the ROD1 gene of Arabidopsis. Proc Natl Acad Sci
1787 U S A 2009;106:18837–42.
1788 [122] Lager I, Glab B, Eriksson L, Chen G, Banas A, Stymne S. Novel reactions in acyl
1789 editing of phosphatidylcholine by lysophosphatidylcholine transacylase
1790 (LPCT) and acyl-CoA:glycerophosphocholine acyltransferase (GPCAT)
1791 activities in microsomal preparations of plant tissues. Planta 2014.
1792 [123] Dahlqvist A, Stahl U, Lenman M, Banas A, Lee M, Sandager L, et al.
1793 Phospholipid:diacylglycerol acyltransferase: an enzyme that catalyzes the
1794 acyl-CoA-independent formation of triacylglycerol in yeast and plants. Proc
1795 Natl Acad Sci U S A 2000;97:6487–92.
1796 [124] Stymne S, Stobart A. Triacylglycerol biosynthesis. Biochem Plants
1797 1987;9:175–214.
1798 [125] Inkpen JA, Quackenbush FW. Desaturation of palmitate and stearate by cell-
1799 free fractions from soybean cotyledons. Lipids 1969;4:539–43.
1800 [126] Simmons RO, Quackenbush FW. The sequence of formation of fatty acids in
1801 developing soybean seeds. J Am Oil Chem Soc 1954;31:441–3.
1802 [127] Dutton HJ, Mounts TL. Desaturation of fatty acids in seeds of higher plants. J
1803 Lipid Res 1966;7:221–5.
1804 [128] Harris RV, James AT. Conversion of oleic acid into linoleic acid by a
1805 subcellular system of Chlorella vulgaris. Biochem J 1965;94:15C–6C.
1806 [129] Stymne S, Appelqvist LA. The biosynthesis of linoleate from oleoyl-CoA via
1807 oleoyl-phosphatidylcholine in microsomes of developing safflower seeds. Eur
1808 J Biochem 1978;90:223–9.
1809 [130] Slack CR, Roughan PG, Browse J. Evidence for an oleoyl phosphatidylcholine
1810 desaturase in microsomal preparations from cotyledons of safflower
1811 (Carthamus tinctorius) seed. Biochem J 1979;179:649–56.
1812 [131] Stymne S, Stobart AK, Glad G. The role of the acyl-CoA pool in the synthesis of
1813 polyunsaturated 18-carbon fatty acids and triacylglycerol production in the
1814 microsomes of developing safflower seeds. Biochim Biophys Acta
1815 1983;752:198–208.
1816 [132] Roughan PG. Phosphatidyl choline: donor of 18-carbon unsaturated fatty
1817 acids for glycerolipid biosynthesis. Lipids 1975;10:609–14.
1818 [133] Slack CR, Roughan PG, Balasingham N. Labelling of glycerolipids in the
1819 cotyledons of developing oilseeds by [1-
14
C] acetate and [2-
3
H] glycerol.
1820 Biochem J 1978;170:421–33.
1821 [134] Williams JP, Watson GR, Leung SPK. Galactolipid synthesis in Vicia faba
1822 leaves: II. Formation and desaturation of long chain fatty acids in
1823
phosphatidylcholine, phosphatidylglycerol, and the galactolipids. Plant
1824
Physiol 1976;57:179–84.
1825
[135] Sperling P, Linscheid M, Stöcker S, Mühlbach HP, Heinz E. In vivo desaturation
1826
of cis-delta 9-monounsaturated to cis-delta 9,12-diunsaturated alkenylether
1827
glycerolipids. J Biol Chem 1993;268:26935–40.
1828
[136] Ohlrogge J, Browse J. Lipid biosynthesis. Plant Cell 1995;7:957–70.
1829
[137] Frentzen M, Heinz E, McKeon TA, Stumpf PK. Specificities and selectivities of
1830
glycerol-3-phosphate acyltransferase and monoacylglycerol-3-phosphate
1831
acyltransferase from pea and spinach chloroplasts. Eur J Biochem
1832
1983;129:629–36.
1833
[138] Browse J, Warwick N, Somerville CR, Slack CR. Fluxes through the prokaryotic
1834
and eukaryotic pathways of lipid synthesis in the ‘16:3’ plant Arabidopsis
1835
thaliana. Biochem J 1986;235:25–31.
1836
[139] Heinz E, Roughan PG. Similarities and differences in lipid metabolism of
1837
chloroplasts isolated from 18:3 and 16:3 plants. Plant Physiol
1838
1983;72:273–9.
1839
[140] Mongrand S, Bessoule J-J, Cabantous F, Cassagne C. The C16:3nC18:3 fatty
1840
acid balance in photosynthetic tissues from 468 plant species.
1841
Phytochemistry 1998;49:1049–64.
1842
[141] Roughan PG. Turnover of the glycerolipids of pumpkin leaves. The
1843
importance of phosphatidylcholine. Biochem J 1970;117:1–8.
1844
[142] Browse J, Somerville C. Glycerolipid synthesis – biochemistry and regulation.
1845
Annu Rev Plant Physiol Plant Mol Biol 1991;42:467–506.
1846
[143] Wallis JG, Browse J. Mutants of Arabidopsis reveal many roles for membrane
1847
lipids. Prog Lipid Res 2002;41:254–78.
1848
[144] Somerville C, Browse J. Plant lipids: metabolism, mutants, and membranes.
1849
Science 1991;252:80–7.
1850
[145] Xu C, Fan J, Riekhof W, Froehlich JE, Benning C. A permease-like protein
1851
involved in ER to thylakoid lipid transfer in Arabidopsis. EMBO J
1852
2003;22:2370–9.
1853
[146] Kunst L, Browse J, Somerville C. Altered regulation of lipid biosynthesis in a
1854
mutant of Arabidopsis deficient in chloroplast glycerol-3-phosphate
1855
acyltransferase activity. Proc Natl Acad Sci U S A 1988;85:4143–7.
1856
[147] Kunst L, Browse J, Somerville C. Altered chloroplast structure and function in
1857
a mutant of Arabidopsis deficient in plastid glycerol-3-phosphate
1858
acyltransferase activity. Plant Physiol 1989;90:846–53.
1859
[148] Shockey JM, Fulda MS, Browse J. Arabidopsis contains a large superfamily of
1860
acyl-activating enzymes. Phylogenetic and biochemical analysis reveals a
1861
new class of acyl-coenzyme a synthetases. Plant Physiol 2003;132:
1862
1065–76.
1863
[149] Schnurr JA, Shockey JM, de Boer GJ, Browse JA. Fatty acid export from the
1864
chloroplast. Molecular characterization of a major plastidial acyl-coenzyme A
1865
synthetase from Arabidopsis. Plant Physiol 2002;129:1700–9.
1866
[150] Jessen D, Roth C, Wiermer M, Fulda M. Two activities of long-chain acyl-CoA
1867
synthetase are involved in lipid trafficking between the endoplasmic
1868
reticulum and the plastid in Arabidopsis. Plant Physiol 2014.
1869
[151] Koo AJ, Ohlrogge JB, Pollard M. On the export of fatty acids from the
1870
chloroplast. J Biol Chem 2004;279:16101–10.
1871
[152] Tjellstrom H, Yang Z, Allen DK, Ohlrogge JB. Rapid kinetic labeling of
1872
Arabidopsis cell suspension cultures: implications for models of lipid export
1873
from plastids. Plant Physiol 2012;158:601–11.
1874
[153] Andersson MX, Goksor M, Sandelius AS. Optical manipulation reveals strong
1875
attracting forces at membrane contact sites between endoplasmic reticulum
1876
and chloroplasts. J Biol Chem 2007;282:1170–4.
1877
[154] Larsson KE, Kjellberg JM, Tjellstrom H, Sandelius AS. LysoPC acyltransferase/
1878
PC transacylase activities in plant plasma membrane and plasma membrane-
1879
associated endoplasmic reticulum. BMC Plant Biol 2007;7:64.
1880
[155] Xu C, Fan J, Froehlich JE, Awai K, Benning C. Mutation of the TGD1 chloroplast
1881
envelope protein affects phosphatidate metabolism in Arabidopsis. Plant Cell
1882
2005;17:3094–110.
1883
[156] Xu C, Fan J, Cornish AJ, Benning C. Lipid trafficking between the endoplasmic
1884
reticulum and the plastid in Arabidopsis requires the extraplastidic TGD4
1885
protein. Plant Cell 2008;20:2190–204.
1886
[157] Wang Z, Anderson NS, Benning C. The phosphatidic acid binding site of the
1887
Arabidopsis trigalactosyldiacylglycerol 4 (TGD4) protein required for lipid
1888
import into chloroplasts. J Biol Chem 2013;288:4763–71.
1889
[158] Wang Z, Benning C. Chloroplast lipid synthesis and lipid trafficking through
1890
ER-plastid membrane contact sites. Biochem Soc Trans 2012;40:457–63.
1891
[159] Wang Z, Xu C, Benning C. TGD4 involved in endoplasmic reticulum-to-
1892
chloroplast lipid trafficking is a phosphatidic acid binding protein. Plant J
1893
2012;70:614–23.
1894
[160] Awai K, Xu C, Tamot B, Benning C. A phosphatidic acid-binding protein of the
1895
chloroplast inner envelope membrane involved in lipid trafficking. Proc Natl
1896
Acad Sci U S A 2006;103:10817–22.
1897
[161] Lu B, Xu C, Awai K, Jones AD, Benning C. A small ATPase protein of
1898
Arabidopsis, TGD3, involved in chloroplast lipid import. J Biol Chem
1899
2007;282:35945–53.
1900
[162] Bessoule JJ, Testet E, Cassagne C. Synthesis of phosphatidylcholine in the
1901
chloroplast envelope after import of lysophosphatidylcholine from
1902
endoplasmic reticulum membranes. Eur J Biochem 1995;228:490–7.
1903
[163] Mongrand S, Cassagne C, Bessoule JJ. Import of lyso-phosphatidylcholine into
1904
chloroplasts likely at the origin of eukaryotic plastidial lipids. Plant Physiol
1905
2000;122:845–52.
1906
[164] Mongrand S, Bessoule JJ, Cassagne C. A re-examination in vivo of the
1907
phosphatidylcholine-galactolipid metabolic relationship during plant lipid
1908
biosynthesis. Biochem J 1997;327(Pt 3):853–8.
20 D.K. Allen et al. / Progress in Lipid Research xxx (2015) xxx–xxx
JPLR 873 No. of Pages 24, Model 5G
16 March 2015
Please cite this article in press as: Allen DK et al. Tracking the metabolic pulse of plant lipid production with isotopic labeling and flux analyses: Past, pre-
sent and future. Prog Lipid Res (2015), http://dx.doi.org/10.1016/j.plipres.2015.02.002

1909 [165] Sperling P, Linscheid M, Stocker S, Muhlbach HP, Heinz E. In vivo desaturation
1910 of cis-delta 9-monounsaturated to cis-delta 9,12-diunsaturated alkenylether
1911 glycerolipids. J Biol Chem 1993;268:26935–40.
1912 [166] van de Loo FJ, Broun P, Turner S, Somerville C. An oleate 12-hydroxylase from
1913 Ricinus communis L. is a fatty acyl desaturase homolog. Proc Natl Acad Sci U S
1914 A 1995;92:6743–7.
1915 [167] Stymne S, Stobart AK. Evidence for the reversibility of the acyl-
1916 CoA:lysophosphatidylcholine acyltransferase in microsomal preparations
1917 from developing safflower (Carthamus tinctorius L.) cotyledons and rat liver.
1918 Biochem J 1984;223:305–14.
1919 [168] Stymne S, Glad G. Acyl exchange between oleoyl-CoA and
1920 phosphatidylcholine in microsomes of developing soya bean cotyledons
1921 and its role in fatty acid desaturation. Lipids 1981;16:298–305.
1922 [169] Bates PD, Stymne S, Ohlrogge J. Biochemical pathways in seed oil synthesis.
1923 Curr Opin Plant Biol 2013;16:358–64.
1924 [170] Slack CR, Campbell LC, Browse JA, Roughan PG. Some evidence for the
1925 reversibility of the cholinephosphotransferasecatalysed reaction in
1926 developing linseed cotyledons in vivo. Biochim Biophys Acta Lipids Lipid
1927 Metab 1983;754:10–20.
1928 [171] Slack CR, Roughan PG, Browse JA, Gardiner SE. Some properties of
1929 cholinephosphotransferase from developing safflower cotyledons. Biochim
1930 Biophys Acta Lipids Lipid Metab 1985;833:438–48.
1931 [172] Bates PD, Durrett TP, Ohlrogge JB, Pollard M. Analysis of acyl fluxes through
1932 multiple pathways of triacylglycerol synthesis in developing soybean
1933 embryos. Plant Physiol 2009;150:55–72.
1934 [173] Cahoon EB, Ohlrogge JB. Apparent role of phosphatidylcholine in the
1935 metabolism of petroselinic acid in developing Umbelliferae endosperm.
1936 Plant Physiol 1994;104:845–55.
1937 [174] Bates PD, Browse J. The pathway of triacylglycerol synthesis through
1938 phosphatidylcholine in Arabidopsis produces a bottleneck for the
1939 accumulation of unusual fatty acids in transgenic seeds. Plant J
1940 2011;68:387–99.
1941 [175] Griffiths G, Stymne S, Stobart AK. The utilisation of fatty-acid substrates in
1942 triacylglycerol biosynthesis by tissue-slices of developing safflower
1943 (Carthamus tinctorius L.) and sunflower (Helianthus annuus L.) cotyledons.
1944 Planta 1988;173:309–16.
1945 [176] Griffiths G, Harwood JL. The regulation of triacylglycerol biosynthesis in
1946 cocoa (Theobroma cacao) L. Planta 1991;184:279–84.
1947 [177] Cahoon EB, Ohlrogge JB. Metabolic evidence for the involvement of a
1948 [delta]4-palmitoyl-acyl carrier protein desaturase in petroselinic acid
1949 synthesis in Coriander endosperm and transgenic tobacco cells. Plant
1950 Physiol 1994;104:827–37.
1951 [178] Malloy CR, Sherry AD, Jeffrey FM. Carbon flux through citric acid cycle
1952 pathways in perfused heart by
13
C NMR spectroscopy. FEBS Lett
1953 1987;212:58–62.
1954 [179] Christensen B, Nielsen J. Isotopomer analysis using GC–MS. Metab Eng
1955 1999;1:282–90.
1956 [180] Sauer U, Hatzimanikatis V, Bailey JE, Hochuli M, Szyperski T, Wuthrich K.
1957 Metabolic fluxes in riboflavin-producing Bacillus subtilis. Nat Biotechnol
1958 1997;15:448–52.
1959 [181] Park SM, Shaw-Reid C, Sinskey AJ, Stephanopoulos G. Elucidation of
1960 anaplerotic pathways in Corynebacterium glutamicum via
13
C-NMR
1961 spectroscopy and GC–MS. Appl Microbiol Biotechnol 1997;47:430–40.
1962 [182] Dauner M, Bailey JE, Sauer U. Metabolic flux analysis with a comprehensive
1963 isotopomer model in Bacillus subtilis. Biotechnol Bioeng 2001;76:
1964 144–56.
1965 [183] van Winden WA, van Gulik WM, Schipper D, Verheijen PJ, Krabben P, Vinke
1966 JL, et al. Metabolic flux and metabolic network analysis of Penicillium
1967 chrysogenum using 2D [
13
C,
1
H] COSY NMR measurements and cumulative
1968 bondomer simulation. Biotechnol Bioeng 2003;83:75–92.
1969 [184] Schwender J. Plant metabolic networks. Springer; 2009.
1970 [185] Dieuaide-Noubhani M, Alonso A. Plant metabolic flux analysis. Springer;
1971 2014.
1972 [186] Sriram G. Plant metabolism. Springer; 2014.
1973 [187] Kruger NJ, Ratcliffe RG. Dynamic metabolic networks: going with the flow.
1974 Phytochemistry 2007;68:2136–8.
1975 [188] Kruger NJ, Ratcliffe RG. Pathways and fluxes: exploring the plant metabolic
1976 network. J Exp Bot 2012;63:2243–6.
1977 [189] Wiechert W.
13
C metabolic flux analysis. Metab Eng 2001;3:195–206.
1978 [190] Klapa MI, Park SM, Sinskey AJ, Stephanopoulos G. Metabolite and isotopomer
1979 balancing in the analysis of metabolic cycles: I. Theory. Biotechnol Bioeng
1980 1999;62:375–91.
1981 [191] Park SM, Klapa MI, Sinskey AJ, Stephanopoulos G. Metabolite and isotopomer
1982 balancing in the analysis of metabolic cycles: II. Applications. Biotechnol
1983 Bioeng 1999;62:392–401.
1984 [192] Press WH, Vetterling WT. Numerical recipes in C the art of scientific
1985 computing. Cambridge: Cambridge University Press; 1992.
1986 [193] Antoniewicz MR, Kelleher JK, Stephanopoulos G. Determination of confidence
1987 intervals of metabolic fluxes estimated from stable isotope measurements.
1988 Metab Eng 2006;8:324–37.
1989 [194] Wiechert W, Siefke C, de Graaf AA, Marx A. Bidirectional reaction steps in
1990 metabolic networks: II. Flux estimation and statistical analysis. Biotechnol
1991 Bioeng 1997;55:118–35.
1992 [195] Dauner M, Sauer U. GC–MS analysis of amino acids rapidly provides rich
1993 information for isotopomer balancing. Biotechnol Prog 2000;16:642–9.
1994
[196] Antoniewicz MR, Kelleher JK, Stephanopoulos G. Accurate assessment of
1995
amino acid mass isotopomer distributions for metabolic flux analysis. Anal
1996
Chem 2007;79:7554–9.
1997
[197] Wiechert W, Mollney M, Isermann N, Wurzel M, de Graaf AA. Bidirectional
1998
reaction steps in metabolic networks: III. Explicit solution and analysis of
1999
isotopomer labeling systems. Biotechnol Bioeng 1999;66:69–85.
2000
[198] Antoniewicz MR, Kelleher JK, Stephanopoulos G. Elementary metabolite units
2001
(EMU): a novel framework for modeling isotopic distributions. Metab Eng
2002
2007;9:68–86.
2003
[199] Young JD, Walther JL, Antoniewicz MR, Yoo H, Stephanopoulos G. An
2004
elementary metabolite unit (EMU) based method of isotopically
2005
nonstationary flux analysis. Biotechnol Bioeng 2008;99:686–99.
2006
[200] Sriram G, Shanks JV. Improvements in metabolic flux analysis using carbon
2007
bond labeling experiments: bondomer balancing and Boolean function
2008
mapping. Metab Eng 2004;6:116–32.
2009
[201] van Winden WA, Heijnen JJ, Verheijen PJ. Cumulative bondomers: a new
2010
concept in flux analysis from 2D [
13
C,
1
H] COSY NMR data. Biotechnol Bioeng
2011
2002;80:731–45.
2012
[202] Rainbird RM, Thorne JH, Hardy RW. Role of amides, amino acids, and ureides
2013
in the nutrition of developing soybean seeds. Plant Physiol 1984;74:329–34.
2014
[203] Hsu FC, Bennett AB, Spanswick RM. Concentrations of sucrose and
2015
nitrogenous compounds in the apoplast of developing soybean seed coats
2016
and embryos. Plant Physiol 1984;75:181–6.
2017
[204] Nelson DR, Rinne RW. Citrate cleavage enzymes from developing soybean
2018
cotyledons: incorporation of citrate carbon into fatty acids. Plant Physiol
2019
1975;55:69–72.
2020
[205] Nelson DR, Rinne RW. The role of citrate in lipid synthesis in developing
2021
soybean cotyledons. Plant Cell Physiol 1977;18:1021–7.
2022
[206] Allen DK, Ohlrogge JB, Shachar-Hill Y. The role of light in soybean seed filling
2023
metabolism. Plant J 2009;58:220–34.
2024
[207] Schwender J, Shachar-Hill Y, Ohlrogge JB. Mitochondrial metabolism in
2025
developing embryos of Brassica napus. J Biol Chem 2006;281:34040–7.
2026
[208] Alonso AP, Val DL, Shachar-Hill Y. Understanding fatty acid synthesis in
2027
developing maize embryos using metabolic flux analysis. Metab Eng
2028
2010;12:488–97.
2029
[209] Liedvogel B, Stumpf PK. Origin of acetate in spinach leaf cell. Plant Physiol
2030
1982;69:897–903.
2031
[210] Kruger NJ, von Schaewen A. The oxidative pentose phosphate pathway:
2032
structure and organisation. Curr Opin Plant Biol 2003;6:236–46.
2033
[211] O’Grady J, Schwender J, Shachar-Hill Y, Morgan JA. Metabolic cartography:
2034
experimental quantification of metabolic fluxes from isotopic labelling
2035
studies. J Exp Bot 2012;63:2293–308.
2036
[212] Ruuska SA, Schwender J, Ohlrogge JB. The capacity of green oilseeds to utilize
2037
photosynthesis to drive biosynthetic processes. Plant Physiol
2038
2004;136:2700–9.
2039
[213] Goffman FD, Ruckle M, Ohlrogge J, Shachar-Hill Y. Carbon dioxide
2040
concentrations are very high in developing oilseeds. Plant Physiol Biochem
2041
2004;42:703–8.
2042
[214] Schwender J, Goffman F, Ohlrogge JB, Shachar-Hill Y. Rubisco without the
2043
Calvin cycle improves the carbon efficiency of developing green seeds. Nature
2044
2004;432:779–82.
2045
[215] Goffman FD, Alonso AP, Schwender J, Shachar-Hill Y, Ohlrogge JB. Light
2046
enables a very high efficiency of carbon storage in developing embryos of
2047
rapeseed. Plant Physiol 2005;138:2269–79.
2048
[216] Borisjuk L, Nguyen TH, Neuberger T, Rutten T, Tschiersch H, Claus B, et al.
2049
Gradients of lipid storage, photosynthesis and plastid differentiation in
2050
developing soybean seeds. New Phytol 2005;167:761–76.
2051
[217] Rolletschek H, Radchuk R, Klukas C, Schreiber F, Wobus U, Borisjuk L.
2052
Evidence of a key role for photosynthetic oxygen release in oil storage in
2053
developing soybean seeds. New Phytol 2005;167:777–86.
2054
[218] Borisjuk L, Neuberger T, Schwender J, Heinzel N, Sunderhaus S, Fuchs J, et al.
2055
Seed architecture shapes embryo metabolism in oilseed rape. Plant Cell
2056
2013;25:1625–40.
2057
[219] Willms JR, Salon C, Layzell DB. Evidence for light-stimulated fatty acid
2058
synthesis in soybean fruit. Plant Physiol 1999;120:1117–28.
2059
[220] Furbank RT, White R, Palta JA, Turner NC. Internal recycling of respiratory CO
2
2060
in pods of chickpea (Cicer arietinum L.): the role of pod wall, seed coat, and
2061
embryo. J Exp Bot 2004;55:1687–96.
2062
[221] Flinn AM, Atkins CA, Pate JS. Significance of photosynthetic and respiratory
2063
exchanges in the carbon economy of the developing pea fruit. Plant Physiol
2064
1977;60:412–8.
2065
[222] Masakapalli SK, Le Lay P, Huddleston JE, Pollock NL, Kruger NJ, Ratcliffe RG.
2066
Subcellular flux analysis of central metabolism in a heterotrophic Arabidopsis
2067
cell suspension using steady-state stable isotope labeling. Plant Physiol
2068
2010;152:602–19.
2069
[223] Sweetlove LJ, Williams TC, Cheung CY, Ratcliffe RG. Modelling metabolic CO
2
2070
evolution – a fresh perspective on respiration. Plant, Cell Environ
2071
2013;36:1631–40.
2072
[224] Egli DB. Seed biology and the yield of grain crops. New York: CAB
2073
International; 1998.
2074
[225] Sweetlove LJ, Beard KF, Nunes-Nesi A, Fernie AR, Ratcliffe RG. Not just a
2075
circle: flux modes in the plant TCA cycle. Trends Plant Sci 2010;15:462–70.
2076
[226] Iyer VV, Sriram G, Fulton DB, Zhou R, Westgate ME, Shanks JV. Metabolic flux
2077
maps comparing the effect of temperature on protein and oil biosynthesis in
2078
developing soybean cotyledons. Plant Cell Environ 2008;31:506–17.
D.K. Allen et al. / Progress in Lipid Research xxx (2015) xxx–xxx 21
JPLR 873 No. of Pages 24, Model 5G
16 March 2015
Please cite this article in press as: Allen DK et al. Tracking the metabolic pulse of plant lipid production with isotopic labeling and flux analyses: Past, pre-
sent and future. Prog Lipid Res (2015), http://dx.doi.org/10.1016/j.plipres.2015.02.002

2079 [227] Alonso AP, Raymond P, Rolin D, Dieuaide-Noubhani M. Substrate cycles in the
2080 central metabolism of maize root tips under hypoxia. Phytochemistry
2081 2007;68:2222–31.
2082 [228] Masakapalli SK, Bryant FM, Kruger NJ, Ratcliffe RG. The metabolic flux
2083 phenotype of heterotrophic Arabidopsis cells reveals a flexible balance
2084 between the cytosolic and plastidic contributions to carbohydrate oxidation
2085 in response to phosphate limitation. Plant J 2014;78:964–77.
2086 [229] Spielbauer G, Margl L, Hannah LC, Romisch W, Ettenhuber C, Bacher A, et al.
2087 Robustness of central carbohydrate metabolism in developing maize kernels.
2088 Phytochemistry 2006;67:1460–75.
2089 [230] Alonso AP, Raymond P, Hernould M, Rondeau-Mouro C, de Graaf A, Chourey
2090 P, et al. A metabolic flux analysis to study the role of sucrose synthase in the
2091 regulation of the carbon partitioning in central metabolism in maize root
2092 tips. Metab Eng 2007;9:419–32.
2093 [231] Focks N, Benning C. Wrinkled1: a novel, low-seed-oil mutant of Arabidopsis
2094 with a deficiency in the seed-specific regulation of carbohydrate metabolism.
2095 Plant Physiol 1998;118:91–101.
2096 [232] Cernac A, Benning C. WRINKLED1 encodes an AP2/EREB domain protein
2097 involved in the control of storage compound biosynthesis in Arabidopsis.
2098 Plant J 2004;40:575–85.
2099 [233] Baud S, Mendoza MS, To A, Harscoet E, Lepiniec L, Dubreucq B. WRINKLED1
2100 specifies the regulatory action of LEAFY COTYLEDON2 towards fatty acid
2101 metabolism during seed maturation in Arabidopsis. Plant J 2007;50:
2102 825–38.
2103 [234] Baud S, Wuilleme S, Dubreucq B, de Almeida A, Vuagnat C, Lepiniec L, et al.
2104 Function of plastidial pyruvate kinases in seeds of Arabidopsis thaliana. Plant J
2105 2007;52:405–19.
2106 [235] Baud S, Graham IA. A spatiotemporal analysis of enzymatic activities
2107 associated with carbon metabolism in wild-type and mutant embryos of
2108 Arabidopsis using in situ histochemistry. Plant J 2006;46:155–69.
2109 [236] Ruuska SA, Girke T, Benning C, Ohlrogge JB. Contrapuntal networks of gene
2110 expression during Arabidopsis seed filling. Plant Cell 2002;14:1191–206.
2111 [237] Stalberg K, Stahl U, Stymne S, Ohlrogge J. Characterization of two Arabidopsis
2112 thaliana acyltransferases with preference for lysophosphatidylethanolamine.
2113 BMC Plant Biol 2009;9:60.
2114 [238] Wang L, Shen W, Kazachkov M, Chen G, Chen Q, Carlsson AS, et al. Metabolic
2115 interactions between the Lands cycle and the Kennedy pathway of
2116 glycerolipid synthesis in Arabidopsis developing seeds. Plant Cell
2117 2012;24:4652–69.
2118 [239] Badami RC, Patil KB. Structure and occurrence of unusual fatty acids in minor
2119 seed oils. Prog Lipid Res 1980;19:119–53.
2120 [240] Dyer JM, Stymne S, Green AG, Carlsson AS. High-value oils from plants. Plant J
2121 2008;54:640–55.
2122 [241] Cahoon EB, Shockey JM, Dietrich CR, Gidda SK, Mullen RT, Dyer JM.
2123 Engineering oilseeds for sustainable production of industrial and
2124 nutritional feedstocks: solving bottlenecks in fatty acid flux. Curr Opin
2125 Plant Biol 2007;10:236–44.
2126 [242] Napier JA. The production of unusual fatty acids in transgenic plants. Annu
2127 Rev Plant Biol 2007;58:295–319.
2128 [243] Napier JA, Haslam RP, Beaudoin F, Cahoon EB. Understanding and
2129 manipulating plant lipid composition: metabolic engineering leads the
2130 way. Curr Opin Plant Biol 2014;19:68–75.
2131 [244] Pollard MR, Anderson L, Fan C, Hawkins DJ, Davies HM. A specific acyl-ACP
2132 thioesterase implicated in medium-chain fatty acid production in immature
2133 cotyledons of Umbellularia californica. Arch Biochem Biophys
2134 1991;284:306–12.
2135 [245] Voelker TA, Worrell AC, Anderson L, Bleibaum J, Fan C, Hawkins DJ, et al. Fatty
2136 acid biosynthesis redirected to medium chains in transgenic oilseed plants.
2137 Science 1992;257:72–4.
2138 [246] Eccleston VS, Ohlrogge JB. Expression of lauroyl-acyl carrier protein
2139 thioesterase in Brassica napus seeds induces pathways for both fatty acid
2140 oxidation and biosynthesis and implies a set point for triacylglycerol
2141 accumulation. Plant Cell 1998;10:613–22.
2142 [247] Voelker TA, Hayes TR, Cranmer AM, Turner JC, Davies HM. Genetic
2143 engineering of a quantitative trait: metabolic and genetic parameters
2144 influencing the accumulation of laurate in rapeseed. Plant J 1996;9:229–41.
2145 [248] Broun P, Boddupalli S, Somerville C. A bifunctional oleate 12-hydroxylase:
2146 desaturase from Lesquerella fendleri. Plant J 1998;13:201–10.
2147 [249] Stahl U, Banas A, Stymne S. Plant microsomal phospholipid acyl hydrolases
2148 have selectivities for uncommon fatty acids. Plant Physiol 1995;107:953–62.
2149 [250] Bafor M, Smith MA, Jonsson L, Stobart K, Stymne S. Ricinoleic acid
2150 biosynthesis and triacylglycerol assembly in microsomal preparations from
2151 developing castor-bean (Ricinus communis) endosperm. Biochem J
2152 1991;280(Pt 2):507–14.
2153 [251] van Erp H, Bates PD, Burgal J, Shockey J, Browse J. Castor
2154 phospholipid:diacylglycerol acyltransferase facilitates efficient metabolism
2155 of hydroxy fatty acids in transgenic Arabidopsis. Plant Physiol
2156 2011;155:683–93.
2157 [252] Broun P, Somerville C. Accumulation of ricinoleic, lesquerolic, and densipolic
2158 acids in seeds of transgenic Arabidopsis plants that express a fatty acyl
2159 hydroxylase cDNA from castor bean. Plant Physiol 1997;113:933–42.
2160 [253] Lu C, Fulda M, Wallis JG, Browse J. A high-throughput screen for genes from
2161 castor that boost hydroxy fatty acid accumulation in seed oils of transgenic
2162 Arabidopsis. Plant J 2006;45:847–56.
2163 [254] Burgal J, Shockey J, Lu C, Dyer J, Larson T, Graham I, et al. Metabolic
2164 engineering of hydroxy fatty acid production in plants: RcDGAT2 drives
2165
dramatic increases in ricinoleate levels in seed oil. Plant Biotechnol J
2166
2008;6:819–31.
2167
[255] Smith MA, Moon H, Chowrira G, Kunst L. Heterologous expression of a fatty
2168
acid hydroxylase gene in developing seeds of Arabidopsis thaliana. Planta
2169
2003;217:507–16.
2170
[256] Dauk M, Lam P, Kunst L, Smith MA. A FAD2 homologue from Lesquerella
2171
lindheimeri has predominantly fatty acid hydroxylase activity. Plant Sci
2172
2007;173:43–9.
2173
[257] Bates PD, Johnson SR, Cao X, Li J, Nam JW, Jaworski JG, et al. Fatty acid
2174
synthesis is inhibited by inefficient utilization of unusual fatty acids for
2175
glycerolipid assembly. Proc Natl Acad Sci U S A 2014;111:1204–9.
2176
[258] Li R, Yu K, Wu Y, Tateno M, Hatanaka T, Hildebrand DF. Vernonia DGATs can
2177
complement the disrupted oil and protein metabolism in epoxygenase-
2178
expressing soybean seeds. Metab Eng 2012;14:29–38.
2179
[259] Snapp AR, Kang J, Qi X, Lu C. A fatty acid condensing enzyme from Physaria
2180
fendleri increases hydroxy fatty acid accumulation in transgenic oilseeds of
2181
Camelina sativa. Planta 2014.
2182
[260] Eastmond PJ, Rawsthorne S. Coordinate changes in carbon partitioning and
2183
plastidial metabolism during the development of oilseed rape embryos. Plant
2184
Physiol 2000;122:767–74.
2185
[261] Fell DA. Increasing the flux in metabolic pathways: a metabolic control
2186
analysis perspective. Biotechnol Bioeng 1998;58:121–4.
2187
[262] Kruger NJ, Ratcliffe RG. Insights into plant metabolic networks from steady-
2188
state metabolic flux analysis. Biochimie 2009;91:697–702.
2189
[263] Ohlrogge JB, Jaworski JG. Regulation of fatty acid synthesis. Annu Rev Plant
2190
Physiol Plant Mol Biol 1997;48:109–36.
2191
[264] Kacser H, Burns JA. The control of flux. Symp Soc Exp Biol 1973;27:65–104.
2192
[265] Heinrich R, Rapoport TA. A linear steady-state treatment of enzymatic chains.
2193
General properties, control and effector strength. Eur J Biochem
2194
1974;42:89–95.
2195
[266] Fell D. Understanding the control of metabolism. London: Portland Press Ltd.;
2196
1994.
2197
[267] Brand MD. Top down metabolic control analysis. J Theor Biol
2198
1996;182:351–60.
2199
[268] Ramli US, Baker DS, Quant PA, Harwood JL. Control analysis of lipid
2200
biosynthesis in tissue cultures from oil crops shows that flux control is
2201
shared between fatty acid synthesis and lipid assembly. Biochem J
2202
2002;364:393–401.
2203
[269] Ramli US, Baker DS, Quant PA, Harwood JL. Control mechanisms operating for
2204
lipid biosynthesis differ in oil-palm (Elaeis guineensis Jacq.) and olive (Olea
2205
europaea L.) callus cultures. Biochem J 2002;364:385–91.
2206
[270] Ramli US, Salas JJ, Quant PA, Harwood JL. Metabolic control analysis reveals
2207
an important role for diacylglycerol acyltransferase in olive but not in oil
2208
palm lipid accumulation. FEBS J 2005;272:5764–70.
2209
[271] Ramli US, Salas JJ, Quant PA, Harwood JL. Use of metabolic control analysis to
2210
give quantitative information on control of lipid biosynthesis in the
2211
important oil crop, Elaeis guineensis (oil palm). New Phytol 2009;184:330–9.
2212
[272] Weselake RJ, Shah S, Tang M, Quant PA, Snyder CL, Furukawa-Stoffer TL, et al.
2213
Metabolic control analysis is helpful for informed genetic manipulation of
2214
oilseed rape (Brassica napus) to increase seed oil content. J Exp Bot
2215
2008;59:3543–9.
2216
[273] Tang M, Guschina IA, O’Hara P, Slabas AR, Quant PA, Fawcett T, et al.
2217
Metabolic control analysis of developing oilseed rape (Brassica napus cv
2218
Westar) embryos shows that lipid assembly exerts significant control over oil
2219
accumulation. New Phytol 2012;196:414–26.
2220
[274] Guschina IA, Everard JD, Kinney AJ, Quant PA, Harwood JL. Studies on the
2221
regulation of lipid biosynthesis in plants: application of control analysis to
2222
soybean. Biochim Biophys Acta 2014;1838:1488–500.
2223
[275] Page RA, Okada S, Harwood JL. Acetyl-CoA carboxylase exerts strong flux
2224
control over lipid synthesis in plants. Biochim Biophys Acta
2225
1994;1210:369–72.
2226
[276] Fenn JB, Mann M, Meng CK, Wong SF, Whitehouse CM. Electrospray
2227
ionization for mass spectrometry of large biomolecules. Science
2228
1989;246:64–71.
2229
[277] Bleijerveld OB, Houweling M, Thomas MJ, Cui Z. Metabolipidomics: profiling
2230
metabolism of glycerophospholipid species by stable isotopic precursors and
2231
tandem mass spectrometry. Anal Biochem 2006;352:1–14.
2232
[278] Cui Z, Thomas MJ. Phospholipid profiling by tandem mass spectrometry. J
2233
Chromatogr B 2009;877:2709–15.
2234
[279] Postle AD, Wilton DC, Hunt AN, Attard GS. Probing phospholipid dynamics by
2235
electrospray ionisation mass spectrometry. Prog Lipid Res 2007;46:200–24.
2236
[280] Wolf C, Quinn PJ. Lipidomics: practical aspects and applications. Prog Lipid
2237
Res 2008;47:15–36.
2238
[281] Han X, Gross RW. Shotgun lipidomics: electrospray ionization mass
2239
spectrometric analysis and quantitation of cellular lipidomes directly from
2240
crude extracts of biological samples. Mass Spectrom Rev 2005;24:367–412.
2241
[282] Han X, Gross RW. Shotgun lipidomics: multidimensional MS analysis of
2242
cellular lipidomes. Expert Rev Proteomics 2005;2:253–64.
2243
[283] Hsu FF, Turk J. Electrospray ionization with low-energy collisionally activated
2244
dissociation tandem mass spectrometry of glycerophospholipids:
2245
mechanisms of fragmentation and structural characterization. J Chromatogr
2246
B 2009;877:2673–95.
2247
[284] Zehethofer N, Pinto DM. Recent developments in tandem mass spectrometry
2248
for lipidomic analysis. Anal Chim Acta 2008;627:62–70.
2249
[285] Luo B, Groenke K, Takors R, Wandrey C, Oldiges M. Simultaneous
2250
determination of multiple intracellular metabolites in glycolysis, pentose
22 D.K. Allen et al. / Progress in Lipid Research xxx (2015) xxx–xxx
JPLR 873 No. of Pages 24, Model 5G
16 March 2015
Please cite this article in press as: Allen DK et al. Tracking the metabolic pulse of plant lipid production with isotopic labeling and flux analyses: Past, pre-
sent and future. Prog Lipid Res (2015), http://dx.doi.org/10.1016/j.plipres.2015.02.002

2251 phosphate pathway and tricarboxylic acid cycle by liquid chromatography–
2252 mass spectrometry. J Chromatogr A 2007;1147:153–64.
2253 [286] van Winden WA, van Dam JC, Ras C, Kleijn RJ, Vinke JL, van Gulik WM, et al.
2254 Metabolic-flux analysis of Saccharomyces cerevisiae CEN.PK113-7D based on
2255 mass isotopomer measurements of C-
13
-labeled primary metabolites. FEMS
2256 Yeast Res 2005;5:559–68.
2257 [287] Ecker J, Liebisch G. Application of stable isotopes to investigate the
2258 metabolism of fatty acids, glycerophospholipid and sphingolipid species.
2259 Prog Lipid Res 2014;54:14–31.
2260 [288] Neuberger T, Rolletschek H, Webb A, Borisjuk L. Non-invasive mapping of
2261 lipids in plant tissue using magnetic resonance imaging. Methods Mol Biol
2262 2009;579:485–96.
2263 [289] Fuchs J, Neuberger T, Rolletschek H, Schiebold S, Nguyen TH, Borisjuk N, et al.
2264 A noninvasive platform for imaging and quantifying oil storage in
2265 submillimeter tobacco seed. Plant Physiol 2013;161:583–93.
2266 [290] Horn PJ, Silva JE, Anderson D, Fuchs J, Borisjuk L, Nazarenus TJ, et al. Imaging
2267 heterogeneity of membrane and storage lipids in transgenic Camelina sativa
2268 seeds with altered fatty acid profiles. Plant J 2013;76:138–50.
2269 [291] Rolletschek H, Melkus G, Grafahrend-Belau E, Fuchs J, Heinzel N, Schreiber F,
2270 et al. Combined noninvasive imaging and modeling approaches reveal
2271 metabolic compartmentation in the barley endosperm. Plant Cell
2272 2011;23:3041–54.
2273 [292] Horn PJ, Korte AR, Neogi PB, Love E, Fuchs J, Strupat K, et al. Spatial mapping
2274 of lipids at cellular resolution in embryos of cotton. Plant Cell
2275 2012;24:622–36.
2276 [293] Lee YJ, Perdian DC, Song Z, Yeung ES, Nikolau BJ. Use of mass spectrometry for
2277 imaging metabolites in plants. Plant J 2012;70:81–95.
2278 [294] Horn PJ, Chapman KD. Lipidomics in situ: insights into plant lipid metabolism
2279 from high resolution spatial maps of metabolites. Prog Lipid Res
2280 2014;54:32–52.
2281 [295] Yang Z, Ohlrogge JB. Turnover of fatty acids during natural senescence of
2282 Arabidopsis, Brachypodium, and switchgrass and in Arabidopsis beta-
2283 oxidation mutants. Plant Physiol 2009;150:1981–9.
2284 [296] Sanjaya, Durrett TP, Weise SE, Benning C. Increasing the energy density of
2285 vegetative tissues by diverting carbon from starch to oil biosynthesis in
2286 transgenic Arabidopsis. Plant Biotechnol J 2011;9:874–83.
2287 [297] Andrianov V, Borisjuk N, Pogrebnyak N, Brinker A, Dixon J, Spitsin S, et al.
2288 Tobacco as a production platform for biofuel: overexpression of Arabidopsis
2289 DGAT and LEC2 genes increases accumulation and shifts the composition of
2290 lipids in green biomass. Plant Biotechnol J 2010;8:277–87.
2291 [298] James CN, Horn PJ, Case CR, Gidda SK, Zhang D, Mullen RT, et al. Disruption of
2292 the Arabidopsis CGI-58 homologue produces Chanarin-Dorfman-like lipid
2293 droplet accumulation in plants. Proc Natl Acad Sci U S A 2010;107:
2294 17833–8.
2295 [299] Sanjaya, Miller R, Durrett TP, Kosma DK, Lydic TA, Muthan B, et al. Altered
2296 lipid composition and enhanced nutritional value of Arabidopsis leaves
2297 following introduction of an algal diacylglycerol acyltransferase 2. Plant Cell
2298 2013;25:677–93.
2299 [300] Shachar-Hill Y. Metabolic network flux analysis for engineering plant
2300 systems. Curr Opin Biotechnol 2013;24:247–55.
2301 [301] Graham JWA, Williams TCR, Morgan M, Fernie AR, Ratcliffe RG, Sweetlove LJ.
2302 Glycolytic enzymes associate dynamically with mitochondria in response to
2303 respiratory demand and support substrate channeling. Plant Cell
2304 2007;19:3723–38.
2305 [302] Giege P, Heazlewood JL, Roessner-Tunali U, Millar AH, Fernie AR, Leaver CJ,
2306 et al. Enzymes of glycolysis are functionally associated with the
2307 mitochondrion in Arabidopsis cells. Plant Cell 2003;15:2140–51.
2308 [303] Heinig U, Gutensohn M, Dudareva N, Aharoni A. The challenges of cellular
2309 compartmentalization in plant metabolic engineering. Curr Opin Biotechnol
2310 2013;24:239–46.
2311 [304] Niklas J, Schneider K, Heinzle E. Metabolic flux analysis in eukaryotes. Curr
2312 Opin Biotechnol 2010;21:63–9.
2313 [305] Wahrheit J, Nicolae A, Heinzle E. Eukaryotic metabolism: measuring
2314 compartment fluxes. Biotechnol J 2011;6:1071–85.
2315 [306] Zamboni N.
13
C metabolic flux analysis in complex systems. Curr Opin
2316 Biotechnol 2011;22:103–8.
2317 [307] Sweetlove LJ, Fernie AR. The spatial organization of metabolism within the
2318 plant cell. Annu Rev Plant Biol 2013;64:723–46.
2319 [308] Fernie AR, Morgan JA. Analysis of metabolic flux using dynamic labelling and
2320 metabolic modelling. Plant Cell Environ 2013.
2321 [309] Gerhardt R, Heldt HW. Measurement of subcellular metabolite levels in
2322 leaves by fractionation of freeze-stopped material in nonaqueous media.
2323 Plant Physiol 1984;75:542–7.
2324 [310] Stitt M, Lilley RM, Gerhardt R, Heldt HW. Metabolite levels in specific cells
2325 and subcellular compartments of plant leaves. Methods Enzymol
2326 1989;174:518–52.
2327 [311] Farré EM, Tiessen A, Roessner U, Geigenberger P, Trethewey RN, Willmitzer L.
2328 Analysis of the compartmentation of glycolytic intermediates, nucleotides,
2329 sugars, organic acids, amino acids, and sugar alcohols in potato tubers using a
2330 nonaqueous fractionation method. Plant Physiol 2001;127:685–700.
2331 [312] Benkeblia N, Shinano T, Osaki M. Metabolite profiling and assessment of
2332 metabolome compartmentation of soybean leaves using non-aqueous
2333 fractionation and GC–MS analysis. Metabolomics 2007;3:297–305.
2334 [313] Ebert B, Zöller D, Erban A, Fehrle I, Hartmann J, Niehl A, et al. Metabolic
2335 profiling of Arabidopsis thaliana epidermal cells. J Exp Bot 2010;61:
2336 1321–35.
2337
[314] Oikawa A, Matsuda F, Kikuyama M, Mimura T, Saito K. Metabolomics of a
2338
single vacuole reveals metabolic dynamism in an alga Chara australis. Plant
2339
Physiol 2011;157:544–51.
2340
[315] Kueger S, Steinhauser D, Willmitzer L, Giavalisco P. High-resolution plant
2341
metabolomics: from mass spectral features to metabolites and from whole-
2342
cell analysis to subcellular metabolite distributions. Plant J 2012;70:39–50.
2343
[316] Allen DK, Laclair RW, Ohlrogge JB, Shachar-Hill Y. Isotope labelling of Rubisco
2344
subunits provides in vivo information on subcellular biosynthesis and
2345
exchange of amino acids between compartments. Plant, Cell Environ
2346
2012;35:1232–44.
2347
[317] Schwender J, Ohlrogge JB. Probing in vivo metabolism by stable isotope
2348
labeling of storage lipids and proteins in developing Brassica napus embryos.
2349
Plant Physiol 2002;130:347–61.
2350
[318] Allen DK, Evans BS, Libourel IGL. Analysis of isotopic labeling in peptide
2351
fragments by tandem mass spectrometry. PLoS One 2014;9:e91537.
2352
[319] Allen DK, Goldford J, Gierse JK, Mandy D, Diepenbrock C, Libourel IGL.
2353
Quantification of peptide m/zdistributions from
13
C-labeled cultures with
2354
high-resolution mass spectrometry. Anal Chem 2014;86:1894–901.
2355
[320] Mandy DE, Goldford JE, Yang H, Allen DK, Libourel IGL. Metabolic flux
2356
analysis using
13
C peptide label measurements. Plant J 2014;77:476–86.
2357
[321] Noack S, Noh K, Moch M, Oldiges M, Wiechert W. Stationary versus non-
2358
stationary
13
C-MFA: a comparison using a consistent dataset. J Biotechnol
2359
2011;154:179–90.
2360
[322] Nöh K, Wiechert W. The benefits of being transient: isotope-based metabolic
2361
flux analysis at the short time scale. Appl Microbiol Biotechnol
2362
2011;91:1247–65.
2363
[323] Nöh K, Grönke K, Luo B, Takors R, Oldiges M, Wiechert W. Metabolic flux
2364
analysis at ultra short time scale: isotopically non-stationary
13
C labeling
2365
experiments. J Biotechnol 2007;129:249–67.
2366
[324] Ettenhuber C, Spielbauer G, Margl L, Hannah LC, Gierl A, Bacher A, et al.
2367
Changes in flux pattern of the central carbohydrate metabolism during kernel
2368
development in maize. Phytochemistry 2005;66:2632–42.
2369
[325] Huege J, Sulpice R, Gibon Y, Lisec J, Koehl K, Kopka J. GC-EI-TOF-MS analysis
2370
of in vivo carbon-partitioning into soluble metabolite pools of higher plants
2371
by monitoring isotope dilution after
13
CO
2
labelling. Phytochemistry
2372
2007;68:2258–72.
2373
[326] Romisch-Margl W, Schramek N, Radykewicz T, Ettenhuber C, Eylert E, Huber
2374
C, et al.
13
CO
2
as a universal metabolic tracer in isotopologue perturbation
2375
experiments. Phytochemistry 2007;68:2273–89.
2376
[327] Arrivault S, Guenther M, Ivakov A, Feil R, Vosloh D, Van Dongen JT, et al. Use
2377
of reverse-phase liquid chromatography, linked to tandem mass
2378
spectrometry, to profile the Calvin cycle and other metabolic intermediates
2379
in Arabidopsis rosettes at different carbon dioxide concentrations. Plant J
2380
2009;59:824–39.
2381
[328] Huege J, Goetze J, Schwarz D, Bauwe H, Hagemann M, Kopka J. Modulation of
2382
the major paths of carbon in photorespiratory mutants of Synechocystis. PLoS
2383
One 2011;6:e16278.
2384
[329] Jorda J, Suarez C, Carnicer M, ten Pierick A, Heijnen JJ, van Gulik W, et al.
2385
Glucose-methanol co-utilization in Pichia pastoris studied by metabolomics
2386
and instationary
13
C flux analysis. BMC Syst Biol 2013;7:17.
2387
[330] Schaub J, Mauch K, Reuss M. Metabolic flux analysis in Escherichia coli by
2388
integrating isotopic dynamic and isotopic stationary
13
C labeling data.
2389
Biotechnol Bioeng 2008;99:1170–85.
2390
[331] Noh K, Wahl A, Wiechert W. Computational tools for isotopically instationary
2391
13
C labeling experiments under metabolic steady state conditions. Metab Eng
2392
2006;8:554–77.
2393
[332] Maier K, Hofmann U, Reuss M, Mauch K. Identification of metabolic fluxes in
2394
hepatic cells from transient
13
C-labeling experiments: Part II. Flux
2395
estimation. Biotechnol Bioeng 2008;100:355–70.
2396
[333] Zhao Z, Ten Pierick A, de Jonge L, Heijnen JJ, Wahl SA. Substrate cycles in
2397
Penicillium chrysogenum quantified by isotopic non-stationary flux analysis.
2398
Microb Cell Fact 2012;11:140.
2399
[334] Choi J, Antoniewicz MR. Tandem mass spectrometry: a novel approach for
2400
metabolic flux analysis. Metab Eng 2011;13:225–33.
2401
[335] Moreau P, Bessoule JJ, Mongrand S, Testet E, Vincent P, Cassagne C. Lipid
2402
trafficking in plant cells. Prog Lipid Res 1998;37:371–91.
2403
Glossary of abbreviations
2404
AA: amino acid
2405
ACP: acyl carrier protein
2406
ADPG: adenosine diphosphoglucose
2407
AKG: alpha-ketoglutarate
2408
ATP: adenosine triphosphate
2409
ACL: ATP-citrate lyase
2410
CoA: coenzyme A
2411
CPT: CDP-choline:diacylglycerol cholinephosphotransferase
2412
DAG: diacylglycerol
2413
DGAT: acyl-CoA:diacylglycerol acyltransferase
2414
DGDG: digalactosyldiacylglycerol
2415
DHAP: dihydroxyacetonephosphate
2416
ER: endoplasmic reticulum
2417
ESI: electrospray ionization
2418
E4P: erythrose-4-phosphate
D.K. Allen et al. / Progress in Lipid Research xxx (2015) xxx–xxx 23
JPLR 873 No. of Pages 24, Model 5G
16 March 2015
Please cite this article in press as: Allen DK et al. Tracking the metabolic pulse of plant lipid production with isotopic labeling and flux analyses: Past, pre-
sent and future. Prog Lipid Res (2015), http://dx.doi.org/10.1016/j.plipres.2015.02.002

2419 FA: fatty acid
2420 FAT A/B: fatty acid thioesterase A and B
2421 F6P: fructose-6-phosphate
2422 G6P: glucose-6-phosphate
2423 G3P: glycerol-3-phosphate
2424 GC: gas chromatography
2425 HP: hexose phosphate
2426 ICITDH: isocitrate dehydrogenase
2427 LACS: long chain acyl-CoA synthetase
2428 LC: liquid chromatography
2429 LPA: lyso-phosphatidic acid
2430 LPAAT: acyl-CoA:lyso-phosphatidic acid acyltransferase
2431 LPCAT: lysophosphatidylcholine acyltransferase
2432 ME: malic enzyme
2433 MFA: metabolic flux analysis
2434 MGDG: monogalactosyldiacylglycerol
2435 MRI: magnetic resonance imaging
2436 MS: mass spectrometry
2437 NADH: nicotinamide adenine dinucleotide
2438 NADPH: nicotinamide adenine dinucleotide phosphate
2439 NEFA: non-esterified fatty acid
2440
NMR: nuclear magnetic resonance
2441
OAA: oxaloacetate
2442
OPPP: oxidative pentose phosphate pathway
2443
PA: phosphatidic acid
2444
PAP: phosphatidic acid phosphatase
2445
PC: phosphatidylcholine
2446
PDAT: phospholipid:diacylglycerol acyltransferase
2447
PDCT: phosphatydlcholine:diacylglycerol cholinephosphotransferase
2448
PG: phosphatidylglycerol
2449
PUFA: polyunsaturated fatty acid
2450
PYR: pyruvate
2451
R5P: ribose-5-phosphate
2452
RuBP: ribulose 1,5-bisphosphate
2453
RuBisCO: ribulose bis-phosphate carboxylase/oxygenase
2454
S7P: sedoheptulose-7-phosphate
2455
TAG: triacylglycerol
2456
TCA: tricarboxylic acid cycle
2457
TGD: trigalactosyldiacylglycerol
2458
TP: triose phosphate
2459
UDPG: uridine diphosphoglucose
2460
3-PGA: 3-phosphoglyceric acid
2461
24 D.K. Allen et al. / Progress in Lipid Research xxx (2015) xxx–xxx
JPLR 873 No. of Pages 24, Model 5G
16 March 2015
Please cite this article in press as: Allen DK et al. Tracking the metabolic pulse of plant lipid production with isotopic labeling and flux analyses: Past, pre-
sent and future. Prog Lipid Res (2015), http://dx.doi.org/10.1016/j.plipres.2015.02.002