Content uploaded by Meredith Bush Mock
Author content
All content in this area was uploaded by Meredith Bush Mock on Oct 11, 2023
Content may be subject to copyright.
Biotechnology and Bioprocess Engineering 27: 640-651 (2022)
DOI 10.1007/s12257-021-0301-0
pISSN 1226-8372
eISSN 1976-3816
Biocatalytic Production and Purification of the High-value
Biochemical Paraxanthine
Meredith B Mock, Shelby Brooks Mills, Ashley Cyrus, Hailey Campo, Tyler Dreischarf, Sydney Strock, and Ryan
M Summers
Received: 3 October 2021 / Revised: 21 December 2021 / Accepted: 17 January 2022
© The Korean Society for Biotechnology and Bioengineering and Springer 2022
Abstract Paraxanthine (1,7-dimethylxanthine), a purine
alkaloid derivative of caffeine (1,3,7-trimethylxanthine), is
a high-value biochemical with several applications in the
pharmaceutical and cosmetic industries. However, chemical
synthesis of paraxanthine requires harsh conditions and
frequently results in low yield mixtures of non-specifically
N-methylated compounds. We have recently demonstrated
that the mutant bacterial N-demethylase NdmA4 with its
partner reductase NdmD is capable of producing paraxanthine
as the major metabolite from caffeine. Here, we report the
construction and screening of several Escherichia coli
strains to produce paraxanthine from caffeine by means of
whole-cell biocatalysts using varying dosages of ndmA4,
ndmD, and the frmAB formaldehyde dehydrogenase genes.
Preliminary resting cell assay results with the best
paraxanthine-producing strain, MBM019, showed a 33%
molar conversion of caffeine, from 5 mM to 3.35 mM,
resulting in approximately 0.90 mM paraxanthine. However,
a small amount of 7-methylxanthine was unexpectedly
produced at a concentration of approximately 0.35 mM. After
optimizing reaction conditions to a cellular concentration
of OD
600
= 50 and a caffeine concentration of 5 mM, the
reaction was scaled-up to a volume of 620 mL, producing
1.02 mM paraxanthine and consuming 2.49 mM caffeine.
The purified paraxanthine was then isolated via preparatory
scale chromatography, resulting in 104.1 mg of product at
high purity. This is the first reported strain genetically
optimized for the biosynthetic production of paraxanthine.
Keywords: paraxanthine, caffeine, biocatalysis, N-demethylase
1. Introduction
Paraxanthine (1,7-dimethylxanthine) is a purine alkaloid
derivative of caffeine (1,3,7-trimethylxanthine), the world’s
most widely consumed psychoactive substance. In humans,
the liver converts approximately 80% of caffeine ingested into
84% paraxanthine, 12% theobromine, and 4% theophylline
before excreting the compounds. Methylxanthines as a
whole offer a variety of promising medical properties,
mostly due to their ability to act as an adenosine receptor
antagonist [1]. The ability of methylxanthines to cross the
blood-brain barrier can be exploited in the formation of
new drugs derived from methylxanthine scaffolds [2,3],
which can be used for the treatment of central nervous
system disorders [4-6]. Caffeine and its derivatives have
been tested for their potential use in the treatment of
neurological disease, such as Alzheimer’s and Parkinson’s
diseases [4-8]. They are also believed to reduce arterial
stiffness [9], act as antioxidants [10], and inhibit HIV-1
replication [11]. Methylxanthine derivatives have also been
investigated for possible anti-inflammatory effects on
inflammatory bowel disease [12].
Specifically, paraxanthine and its metabolite, 7-
methylxanthine [13], have been found to be less toxic [14-
16] and less addictive than caffeine [17]. Paraxanthine has a
higher potency and additional mechanism of action towards
A1 and A2a receptors than caffeine [15,16]. A significant
amount of research has confirmed that paraxanthine exhibits
preventative properties against dopaminergic neuron death
and may be used in the management of Parkinsonʼs disease
[5,8,18]. In addition, paraxanthine has been found to be
Meredith B Mock, Shelby Brooks Mills, Ashley Cyrus, Hailey Campo,
Tyler Dreischarf, Sydney Strock, Ryan M Summers
*
Department of Chemical and Biological Engineering, The University of
Alabama, Tuscaloosa, AL 35487, USA
Tel: +1-205-348-3169; Fax: +1-205-348-7558
E-mail: rmsummers@eng.ua.edu
RESEARCH PAPER
Biocatalytic Production of Paraxanthine 641
more effective than caffeine or its dimethylxanthine
counterparts at attenuating liver fibrosis [19] and is a
promising therapeutic in the treatment of the inflammation
seen in chronic obstructive pulmonary disease [20]. There
is even less research focused on the medical benefits of 7-
methylxanthine; however, it has been shown to prevent the
progression of myopia and slow axial eye growth in
children [21]. The effect of paraxanthine on nitric oxide
and cyclic guanosine monophosphate levels as an A1R
antagonist could open the door for a new class of therapeutic
drugs designed to alleviate drug addiction and basal
ganglia disorders [14,16]. Methylxanthines possess some
anticarcinogenic properties, however, this can be amplified
into potential anti-cancer agents when the structures are
used to construct N-heterocyclic carbene complexes [22-
26]. Thus, production of paraxanthine could diversify the
possible structures achievable for the design of new drugs.
Paraxanthine is a rare compound and is not readily
found in nature, with the exception of human metabolism
of caffeine [27]. Attempts have been made to circumvent
the lack of naturally produced paraxanthine by developing
chemical synthetic methods, but all current routes for the
synthesis of methylxanthines either involve a complicated
chemical process and/or require a variety of hazardous
chemicals [28]. Furthermore, attempts at direct N-substitution
of xanthine have encountered selectivity issues because the
acidity of N
3
-H and N
7
-H are nearly equal, followed closely
by N
1
-H. As expected from the acidity of the N-H groups,
substitution generally follows the pattern of N
3
≥ N
7
> N
1
[29]. In 2006, a library of (1,7) – disubstituted xanthines
was generated using a solid-phase synthesis method. No
yields were reported specifically for paraxanthine, but the
library yields ranged from 35% to as low as 6% [30].
Traube purine synthesis is considered the classical method
to chemically synthesize purines, such as substituted
methylxanthines, from 6-aminouracil derivatives. However,
even with updates and modifications, this method remains
limited, lengthy, and under the constraint of harsh conditions
[31]. In addition, this method can produce two isomers that
in some cases can be difficult to separate [29]. As a result,
it can be largely agreed that no general procedure has been
developed that is convenient for the production of
paraxanthine and its analogs [28,32].
Due to the low yield of paraxanthine through chemical
synthesis, a biochemical synthesis route may provide a
more efficient and robust approach. Despite the toxicity of
caffeine, some bacteria, primarily pseudomonads, have
developed the ability to degrade and even survive on the
compound as a sole source of carbon and nitrogen [33-37].
The bacterium Pseudomonas putida CBB5 uses enzymes
from the Rieske [2Fe-2S] nonheme monooxygenase family
encoded by the genes ndmABCDE on the Alx gene cluster
to metabolize caffeine to xanthine (Fig. 1) [38]. Caffeine is
first converted to theobromine by the N
1
-demethylase NdmA.
The N
3
-demethylase NdmB then converts theobromine to
7-methylxanthine, which is further converted to xanthine
by the N
7
-demethylase NdmC [34,36,38-40]. NdmD is a
Rieske-type reductase that is absolutely required for all
caffeine N-demethylation reactions by transferring electrons
from NADH to the N-demethylase. NdmE is required for
structural support in the formation of an NdmCDE complex
that allows all three proteins to be soluble and active [38,39].
We have constructed several strains of Escherichia coli
whole cell biocatalysts, metabolically engineered to
successfully interconvert between methylxanthines. We
first reported on the production of 3-methylxanthine from
theophylline using whole cells containing combinations of
NdmA and NdmD [41]. This was quickly followed by
demonstrations of converting caffeine to theobromine with
NdmA and NdmD [42], and theobromine to 7-methylxanthine
using NdmB and NdmD [43]. In these previous studies, we
identified some promiscuity of the N-demethylase enzymes
when assayed in vivo that was not initially identified for
purified enzymes in vitro. For example, NdmB primarily
carries out N
3
-demethylation, but has also been shown to
Fig. 1. Conversion of caffeine to xanthine by the N-demethylation
pathway characterized in Pseudomonas putida CBB5.
642 Biotechnology and Bioprocess Engineering 27: 640-651 (2022)
carry out N
1
-demethylation of paraxanthine to produce 7-
methylxanthine [44]. NdmA also exhibits a slight N
3
-
demethylation activity, with approximately 1-2% of caffeine
being converted to paraxanthine [42] and 13% of theophylline
being converted to 1-methylxanthine [41]. We recently
generated a mutant version of NdmA, initially named
NdmA
QL + loop
but hereafter referred to as NdmA4, which
harbors N282Q and F286L mutations at the active site and
the NdmB loop from residues 213 to 227 swapped in place
of the NdmA loop from residues 207 to 221 [45,46]. The
mutations to produce NdmA4 resulted in an enzyme capable
of carrying out N
3
-demethylation of caffeine to produce
paraxanthine as the primary metabolite.
The rarity of paraxanthine contributes significantly to the
cost, with retail prices of paraxanthine over 30,000 times the
retail price of caffeine on a mass basis, providing motivation
for optimizing the conversion of caffeine into paraxanthine
(Table S1). Here, we demonstrate a bench-scale process for
the microbial production, separation, and purification of
paraxanthine from caffeine using metabolically engineered
E. coli. We generated and screened several strains of E. coli
capable of producing paraxanthine from caffeine by varying
and optimizing gene dosage and incorporating an NADH
recycling system. The optimal strain was selected for final
paraxanthine production in a bench scale reaction, followed
by separation to produce a pure, crystallized product. This
is the first report to describe biocatalytic production of
paraxanthine at sufficient quantities to enable product
isolation.
2. Materials and Methods
2.1. Chemicals and reagents
Caffeine was purchased from J.T. Baker (Phillipsberg, NJ,
USA). Paraxanthine was procured from Sigma-Aldrich (St.
Louis, MO, USA). 7-Methylxanthine was acquired from
Alfa Aesar (Haverhill, MA, USA). Luria-Bertani media was
made according to the protocol described by MacWilliams
and Liao [47]. Isopropyl β-D-thiogalactopyranoside
(IPTG) was bought from INDOFINE Chemical Company
(Hillsborough, NJ, USA). Polymerase chain reaction (PCR)
reactions were performed using Phusion HF polymerase.
All restriction enzymes and PCR reagents were purchased
from New England BioLabs (Ipswich, MA, USA).
Antibiotics were obtained from AMRESCO (Solon, OH,
USA). Methanol used during chromatograph separations
was of high-performance liquid chromatography (HPLC)-
grade from J.T. Baker.
2.2. Plasmid construction
All plasmids used in this study are listed in Table 1, and a
list of all primers used can be found in the Supplementary
Information in Table S2. Plasmid maps for all constructed
Table 1. Complete list of plasmids used in this study
Plasmids Characteristics Source
pET-28a(+) Kan
R
, T7 promoter, N-terminal His
6
tag, pBR322 origin Novagen
pET-32a(+) Amp
R
, T7 promoter, C-terminal His
6
tag, pBR322 origin Novagen
pETDuet-1 Amp
R
, two T7 promoters, two MCS, pBR322 origin Novagen
pACYCDuet-1 Cm
R
, two T7 promoters, two MCS, p15A origin Novagen
pET28-His-ndmD pET-28a(+) with one copy of ndmD [38]
pDrbs1A4 pET-28a(+) with one copy of ndmD one copy of mutant ndmA4 connected by a ribosomal
binding site (1)
This study
pDP1 pET-28a(+) with one copy of truncated ndmD (ndmDP1) This study
pDrbs2A4 pET-28a(+) with one copy of ndmD and one ndmA4 connected by a ribosomal binding site (2) This study
pDP1rbs2A4 pET-28a(+) with one copy of ndmDP1 and one copy of ndmA4 connected by a ribosomal
binding site (2)
This study
p32D pET-32a(+) with one copy of ndmD This study
pA4 pETDuet with one copy of ndmA4 [45]
dA4 pACYCDuet-1 with one copy of ndmA4 spanning both MCS’s This study
dDA4 pACYCDuet-1 with one copy of ndmD and one copy of A4 [46]
dDD pACYCDuet-1 with two copies of ndmD [41]
dA4A4 pACYCDuet-1 with two copies of ndmA4 This study
dA4frmAB pACYCDuet-1 with one copy of ndmA4 and one copy of frmAB This study
dA0 pACYCDuet-1 with one copy of ndmA within MCS1, MCS2 empty [41]
dA pACYCDuet-1 with one copy of ndmA [41]
dAA pACYCDuet-1 with two copies of ndmA [41]
dAfrmAB pACYCDuet-1 with one copy of ndmA and one copy of frmAB This study
MCS: multiple cloning site.
Biocatalytic Production of Paraxanthine 643
plasmids are provided in Fig. S1. All genes were amplified
using Phusion HF polymerase. Based on literature, the
copy number of the pET-28a(+), pETDuet-1 and pET-
32a(+) vector backbones was assumed to be approximately
40, while that of the pACYCDuet-1 vector backbone was
assumed to be approximately 10 [48,49]. All plasmids
were constructed such that the genes are under the control
of the strong T7 promoter, allowing for selective induction
of expression via IPTG. When more than one gene was
incorporated into the pET-28a(+) vector, a synthetic
ribosomal binding site was designed and included between
the two genes. Construction of plasmids dDA4 [46], pD,
dDD, dA0, dA, and dAA have been previously described
[41].
Plasmid pDP1 was constructed by first amplifying the
NdmDP1 fragment, a truncated NdmD, using primers
NdmDP1-GA-F/NdmDP1-GA-R and inserting the fragment
via Gibson Assembly into a pET28a(+) backbone that had
been digested using the restriction enzymes NdeI and
BamHI. Plasmid pDrbs1A4 was constructed in a similar
manner as pDP1; however, NdmD and NdmA4 were first
amplified using the primers NdmD1-F/NdmD1-R and
Loop2-F/Loop2-R, respectively. The two fragments were
then inserted into the digested backbone via Gibson
Assembly, combining them into one fragment connected
by a ribosomal binding site (rbs1), 5ʹ-TCTAGAGAAAGA
GGAGAAATACTAG-3ʹ, that had been built into the
primers. Plasmids pDrbs2A4 and pDP1rbs2A4 were
constructed in the same manner using primers NdmD-GA-
F/NdmD-rbs-R and rbs-Loop-F/Loop-GA-R, and NdmDP1-
GA-F/NdmD-rbs-R and rbs-Loop-F/Loop-GA-R, respectively.
These fragments were linked by a second synthetic ribosomal
binding site (rbs2), 5ʹ-CGCGCAAGTCGTTACCAGGAA
ATTCTAT-3ʹ. All ribosomal binding sites were designed
using the De Novo DNA: RBS calculator [50].
In general, genes inserted into the first multiple cloning
site (MCS1) of pACYCDuet-1 were amplified by forward
primers containing an NcoI site and reverse primers
containing a BamHI site. These fragments, along with the
pACYCDuet-1 backbone, were then digested with NcoI
and BamHI and ligated together. In a like manner, the
second multiple cloning site (MCS2) required the use of
NdeI and KpnI as the unique restriction sites. Specifically,
plasmid dA4A4 was constructed by first digesting the
pACYCDuet-1 with NcoI and BamHI and amplifying
NdmA4 using the primers Loop-F-NcoI/NdmA-R-BamHI.
The fragment was digested with the corresponding restriction
enzymes and ligated into the backbone in MCS1. This
process was repeated using NdeI and KpnI, the primers
Loop-F-NdeI/NdmA-R-KpnI and MCS2. For genes spanning
both multiple cloning sites (MCSs), a forward primer
containing an NcoI site and a reverse primer containing a
KpnI site were used the amplify the fragments, followed by
digestion of the fragment and the backbone, and ligation.
Plasmid dA4 was constructed to span both MCSs using the
primer pair Loop-F-NcoI/NdmA-R-KpnI.
To construct the dAfrmAB and dA4frmAB plasmids,
dA0 [41] and dA4A4 plasmids were digested with NdeI
and KpnI to linearize the pACYCDuet-1 vector without
affecting MCS1. The formaldehyde degrading genes (frmA
and frmB) were then amplified from E. coli BL21(DE3)
genomic DNA using the primers frmA-F1/frmA-rbs-R1
(frmA) and rbs-frmB-F1/frmB-R1 (frmB). The two fragments
were then linked together during a Gibson Assembly by a
ribosomal binding site designed into the primers, generating
a single fragment (FrmAB). This FrmAB fragment was
incorporated into MCS2 of the linearized pACYCDuet-1
vector during the Gibson Assembly.
2.3. Strain construction
E. coli BL21(DE3) was used as the parent strain to
construct all of the strains used in this research. A complete
list of strains with their descriptions is located in Table 2.
Plasmids were transformed into chemically competent
E. coli BL21(DE3) and recombinant strains were plated on
Luria-Bertani (LB) agar plates [47] containing appropriate
antibiotics at the following concentrations: 100 µg/mL
ampicillin, 34 µg/mL chloramphenicol, and 30 µg/mL
kanamycin. If two plasmids needed to be incorporated into
one strain, one plasmid type was transformed into E. coli
BL12(DE3) and used to generate chemically competent
cells for transformation of the second plasmid.
2.4. Cell growth and protein expression
For initial strain comparison, all E. coli strains were grown
and protein expressed as described by Mock et al. [44].
Briefly, cells were grown in LB with appropriate antibiotics
at 37°C and shaking at 200 rpm. When the OD
600
of the
cells reached ~0.5, sterile iron chloride was added to a final
concentration of 10 μM and the culture was shifted to
18°C. IPTG was added to a final concentration of 0.1 mM
to induce gene expression when the OD
600
reached 0.8, and
the cells were grown 14-16 h post-induction at 18°C with
200 rpm shaking. Cells were harvested by centrifugation at
10,000 × g for 10 min at 4°C. Small scale cultures were
carried out in 50 mL media. Cultures designated for product
isolation were grown in four 2.8-L Fernbach flasks, each
containing 1 L of media.
2.5. Reaction conditions for paraxanthine production
Harvested cells were resuspended in ice cold 50 mM
potassium phosphate (KP
i
) buffer (pH 7.5). Unless otherwise
indicated, resting cell assays were conducted in test tubes
at a volume of 2 mL, cells at an OD
600
of 5, and a caffeine
644 Biotechnology and Bioprocess Engineering 27: 640-651 (2022)
concentration of 1 mM in KP
i
buffer. Reactions were
carried out at 30°C and 200 rpm shaking for 5 h, and
approximately 100 μL samples were taken periodically and
analyzed via HPLC to determine methylxanthine concentrations
using the appropriate standards.
The large-scale reaction for production and purification
of paraxanthine used a maximized volume of 620 mL
based on harvested cell density, with an OD
600
of 50 and a
caffeine concentration of 5 mM. The reaction was incubated
in a Fernbach flask at 30
o
C and 200 rpm shaking for 5 h.
At the end of the reaction, the cells were harvested by
centrifugation at 10,000 × g for 10 min at 4
o
C to separate
them from the product, and the supernatant was collected
for purification.
2.6. Preparatory HPLC
Prior to HPLC purification, the harvested supernatant was
filtered through a 0.2 μm filter, and the final volume of
supernatant collected measured 600 mL. Paraxanthine
purification was conducted using a ThermoScientific
Hypersil BDS C18 preparatory HPLC column (20 mm
diameter × 150 mm length). The column was connected to
a Shimadzu LC-20AT HPLC system equipped with a
photodiode array detector to detect and record the
ultraviolet-visible absorption spectra. A mobile phase of
7.5:92.5:0.5 (vol/vol/vol) methanol-water-acetic acid at a
flow rate of 2.5 mL/min. An isocratic program was
developed using two pumps operating at 2.5 mL/min so
that one pump would load the post reaction mixture for
20 min (50 mL total) and the second pump would deliver
the mobile phase. Methanol (~48 mL) was added to the
reaction supernatant to match the HPLC concentration of
7.5% MeOH and prevent a swing in MeOH concentration
from affecting the HPLC chromatograph. The supernatant-
methanol mixture was loaded onto the column at a rate of
2.5 mL/min for 20 min, resulting in a total of 50 mL of
supernatant loaded each round. After 12 rounds of
separation, 745 mL volume of paraxanthine solution was
collected. The solution was concentrated using a rotary
evaporator at 70
o
C and 200-220 mbar, reducing the volume
to 196 mL. The concentrated solution was finally dried at
140
o
C for 9 h to produce paraxanthine powder. A total of
5 h and 20 min was required for all of the compounds from
each round to exit the column before the next round could
be started.
2.7. Analytical procedures
Paraxanthine was identified and quantified using the same
HPLC system as described above. A ThermoScientific
Hypersil BDS C18 HPLC column (4.6 mm inner diameter
× 150 mm length) was used as the stationary phase. A
mobile phase of 15:85:0.5 (vol/vol/vol) methanol-water-
acetic acid at a flow rate of 0.5 mL/min. Purity of the
paraxanthine was confirmed using HPLC and nuclear
Table 2. Complete list of strains used in this study
Strain Characteristics Source
Escherichia coli BL21(DE3) F
-
ompT hsdS
B
(r
-
B
m
-
B
) gal dcm (DE3) Invitrogen
E. coli pDdA BL21(DE3) pET28-His-ndmD dA [41]
E. coli pDdAA BL21(DE3) pET28-His-ndmD dAA [41]
E. coli MBM001 BL21(DE3) pET32-ndmD-His dA0 This study
E. coli MBM002 BL21(DE3) dDA4 [46]
E. coli MBM003 BL21(DE3) pET28-His-ndmD dA4 This study
E. coli MBM004 BL21(DE3) pET28-His-ndmD dA4A4 This study
E. coli MBM005 BL21(DE3) pETDuet-ndmA4 dDD This study
E. coli MBM006 BL21(DE3) pDP1 dA4 This study
E. coli MBM007 BL21(DE3) pDP1 dA4A4 This study
E. coli MBM008 BL21(DE3) pDrbs1A4 This study
E. coli MBM009 BL21(DE3) pDrbs2A4 This study
E. coli MBM010 BL21(DE3) pDrbs1A4 dA4A4 This study
E. coli MBM011 BL21(DE3) pDrbs2A4 dA4A4 This study
E. coli MBM012 BL21(DE3) pDP1rbs2A4 This study
E. coli MBM013 BL21(DE3) pDP1rbs2A4 dA4 This study
E. coli MBM014 BL21(DE3) pDP1rbs2A4 dA4A4 This study
E. coli MBM015 BL21(DE3) pET28-His-ndmD dAfrmAB This study
E. coli MBM016 BL21(DE3) pET28-His-ndmD dA4frmAB This study
E. coli MBM017 BL21(DE3) pDrbs2A4 dA4frmAB This study
E. coli MBM018 BL21(DE3) pDP1 dA4frmAB This study
E. coli MBM019 BL21(DE3) pDP1rbs2A4 dA4frmAB This study
Biocatalytic Production of Paraxanthine 645
magnetic resonance (NMR). The NMR results were obtained
from the NMR facility in the Chemistry Department of the
University of Alabama. The spectrum was recorded in
dimethylsulfoxide-d
6
(DMSO-d
6
)
with a Bruker DRX 500
NMR spectrometer at 299 K. The chemical shifts were
relative to DMSO-d
6
using the standard δ notation in parts
per million.
3. Results and Discussion
3.1. Genetic optimization
To begin optimization of paraxanthine production, we
constructed and screened twenty-one strains of metabolically
engineered E. coli (Tables 2, S3, and S4) to study the effect of
gene dosage, ribosomal binding sites, and NADH recycling
on their ability to metabolize caffeine. The initial screening
was carried out in 2 mL reactions at 30°C with an initial
caffeine concentration of 1 mM and cells at an OD
600
of
5.0. A summary of caffeine consumed and paraxanthine
produced by each strain is provided in Table S4.
3.1.1. Gene dosage
Algharrawi et al. [41] previously demonstrated that N
1
-
demethylation of theophylline for the production of 3-
methylxanthine could be improved by increasing the ndm
gene copy number and ultimately increasing the amount of
soluble NdmD within recombinant E. coli cells. These
genetic modifications resulted in an optimized strain with
improved 3-methylxanthine yields and production rates.
Similarly, we observed that caffeine was consumed and
theobromine was produced at a faster rate by increasing the
ndmA copy number (Fig. S2). Thus, we evaluated the
ability of cells with different copy numbers of ndmA4 and
ndmD to produce paraxanthine from caffeine (Fig. 2A,
Table S4). NdmA4 is a mutant of NdmA that produces
paraxanthine as the major product from caffeine instead of
theobromine [45,46].
Our initial strain, MBM002, contained one copy of
ndmA4 and one copy of ndmD in the pACYCDuet-1
vector. This strain consumed 169 ± 38 µM caffeine and
produced 92 ± 4 µM paraxanthine over four hours. Previous
work demonstrated that increasing the ndmD reductase
gene copy number could increase overall activity of the
cells [41], thus we moved the ndmD gene to the pET28a(+)
vector, which has a copy number approximately four times
higher than the pACYCDuet-1 vector. The resulting strain,
MBM003, produced slightly more paraxanthine than
MBM002 (Fig. 2A, Table S4), indicating that the activity
may be limited by the amount of active NdmA4 in the
cells. Therefore, we doubled the gene dosage of ndmA4 by
expressing two copies from the pACYCDuet-1 vector in
MBM004 and observed a marked increase in paraxanthine
production, with 141 ± 9 µM paraxanthine produced from
256 ± 47 µM caffeine. When we increased ndmA4 dosage
further while lowering ndmD in MBM005, paraxanthine
production decreased to 73 ± 4 μM over five hours, indicating
that we still needed increased NdmD levels.
The NdmD reductase has an extra Rieske [2Fe-2S]
cluster that is not necessary for activity of NdmA [51],
which led us to test a truncated ndmD gene, ndmDP1,
which was created by removing the first 266 amino acids
from the ndmD sequence. We swapped out the ndmD gene
in strains MBM003 and MBM004 for the ndmDP1 gene,
creating strains MBM006 and MBM007. In both strains,
the initial rate of paraxanthine production was lower than
Fig. 2. Production of paraxanthine from caffeine by metabolically engineered Escherichia coli. The resting cell assays were conducted at
an OD
600
of 5 and a caffeine concentration of 1 mM. The collection of strains were used to test the effects of (A) varying the copy
numbers of ndmA4, ndmD, and ndmDP1 genes, (B) the synthetic ribosomal binding sites, and (C) an NADH recycle system through the
enhanced degradation of formaldehyde. Concentrations reported are means with standard deviations of triplicate results.
646 Biotechnology and Bioprocess Engineering 27: 640-651 (2022)
in their respective ndmD strain, but the paraxanthine
produced from MBM006 over five hours was similar to
MBM003, and MBM007 produced more paraxanthine
over MBM004 with a final concentration of 172 ± 5 µM
paraxanthine (Fig. 2A). Additionally, the increasing trend
for MBM007 appears to continue past 300 min, suggesting
that a greater production of paraxanthine may be possible
over longer times (Fig. 2A).
After observing the increase in activity from MBM003
to MBM004 by adding an extra copy of ndmA4 in the Duet
vector, we added an additional copy of ndmA4 downstream
of the reductase gene on the pET28a(+) vector. The genes
were expressed as a bicistronic fragment linked by one of
two synthetic ribosomal binding sites, rbs1 and rbs2. Using
the De Novo DNA: RBS calculator, rbs1 was determined
to have a translation rate of 2,867.02 au and rbs2 to have
a translational rate of 11,711.93 au. Strain MBM008, which
contained only the pET28a(+) vector with ndmD and
ndmA4 linked by rbs1, produced 23 ± 1 µM paraxanthine
over 5 h (Fig. 2B). By swapping out rbs1 with rbs2, we
observed a total of 48 ± 1 µM paraxanthine in the same
time period using strain MBM009. Intriguingly, addition of
the dA4A4 plasmid to MBM008 to create strain MBM010
greatly reduced the activity of the cells, which produced
only 7 ± 1 µM paraxanthine. In contrast, using rbs2 to give
strain MBM011 resulted in production of 120 ± 2 μM
paraxanthine from 187 ± 7 μM caffeine. These results
indicated the rbs2 is superior to rbs1, as suggested by
predicted translation rates, and that activity is significantly
improved when multiple copies of ndmA4 are present in
proportion to ndmD. To continue improving production,
we replaced the ndmD gene of MBM011 with the
truncated ndmDP1 to construct strain MBM014, which
produced 178 ± 7 μM paraxanthine over five hours. This
denotes a significant improvement in paraxanthine production,
in this case an increase of approximately 58 mM, when
ndmD is exchanged for ndmDP1. This strain, however, did
not differ significantly in its paraxanthine production from
strain MBM007, which led us to consider a different
strategy to improve paraxanthine production.
3.1.2. NADH recycle
During the N-demethylation process, one molecule of NADH
is oxidized to NAD
+
and one molecule of formaldehyde is
produced per methyl group removed [38]. Because of the
low rate of reaction and overall conversion previously
observed, we theorized that NADH availability may be a
potential limitation to the conversion of caffeine to
paraxanthine. NADH regeneration has been explored in
other microbial systems with the purpose of increasing
NADH availability and investigating the impact on cell
metabolism with some success in improving the metabolic
flux [52,53]. With this concept in mind, we designed an
NADH recycle system (Fig. 3) using the frmAB formaldehyde
dehydrogenase genes native to E. coli. These genes are part
of a detoxification stress response system used by E. coli to
protect the cell from the cytotoxicity of formaldehyde
[54,55]. During conversion of formaldehyde to formate,
one molecule of NAD
+
is reduced to NADH [56,57]. This
enables the potential for the circulation of NADH to NAD
+
and back to NADH, removing NADH as a limiting factor
in the demethylation of caffeine to paraxanthine.
To test the effectiveness of this NADH recycle system,
we constructed strain MBM015 from pDdA0, which was
previously established to convert 100% of caffeine to
theobromine within two hours [42], by placing the frmAB
genes under control of the T7 promoter in the empty
cloning site of dA0. Resting cells of MBM015 produced
theobromine from caffeine faster than did pDdA0 (Fig. S3),
suggesting that expression of frmAB could improve
reaction rates by improving NADH recycle. Therefore, we
constructed four strains containing both frmAB and ndmA4
genes to test their ability to produce paraxanthine. Addition
of frmAB to MBM003 to generate MBM016 did not
increase the amount of paraxanthine produced, but did
slightly increase the rate at which paraxanthine was formed
(Fig. S4A). This would suggest that, while critical to the
Fig. 3. Theoretical depiction of the NADH recycle pathway for
the N-demethylation of caffeine to paraxanthine and concomitant
conversion of formaldehyde to formate.
Biocatalytic Production of Paraxanthine 647
reaction, NADH concentration is not a limiting factor in
the conversion of caffeine to paraxanthine. Further
investigations can be done to study the extent to which a
second cofactor, O
2
availability, effects the overall yield of
paraxanthine. Oxygenation is not expected to be a limiting
factor when considering the efficiency of caffeine conversion
by pDdA0 under the same reaction conditions. For these
reasons, it is most likely low efficiency of ndmA4 itself,
either to bind caffeine, react, or release paraxanthine that
results in the low conversion. Identification of kinetic
parameters along with targeted optimization of ndmA4
would likely result in the most significant improvement to
paraxanthine yield.
When ndmD was replaced with ndmDP1, there was no
significant difference between paraxanthine production in
MBM006 and MBM018 (Fig. S4B). However, we did
observe an increase in both paraxanthine yield and rate
when comparing MBM013 with MBM019 (Fig. S4C).
Strain MBM019 includes the ndmDP1 reductase gene for
improved N-demethylase activity and the frmAB genes for
NADH recycle, and this strain produced a final concentration
of 181 ± 5 μM paraxanthine over five hours.
3.2. Paraxanthine production and reaction condition
optimization
Because our best paraxanthine-producing strains consumed
less than 30% of the caffeine in the resting cell assays over
five hours, we hypothesized that increasing cell concentration
in the reaction would increase caffeine conversion to
paraxanthine. To test this hypothesis, we performed resting
cell assays with strain MBM013 at OD
600
of 5, 10, 20, and
50 while maintaining initial concentration of caffeine at
1 mM. Caffeine consumption increased with increasing
cell density (Fig. 4A, Table 3). Surprisingly, paraxanthine
yields were similar at an OD
600
of 10 and 20, but were
greatly reduced at an OD
600
of 50. This decrease in
paraxanthine yield as cell concentrations increased can be
accounted for in the increased concentrations of 7-
methylxanthine (Table 3). The data demonstrate that
NdmA4 can carry out both N
3
-demethylation of caffeine
and N
1
-demethylation of paraxanthine and suggest that at a
certain point during the reaction, the equilibrium shifts in
favor of 7-methylxanthine. This would explain the
discrepancies between the reactions with an OD
600
of 20
and 50, as well as high concentration of 7-methylxanthine
seen in the OD
600
of 50 reaction (Table 3).
Strains MBM007, MBM014, and MBM019 produced
similar amounts of paraxanthine over five hours (Fig. 2,
Table S4). We selected MBM019 for further process
optimization prior to scaleup because the concentration of
paraxanthine produced by MBM019 appeared to be
increasing when reaching 5 h and, as a result, had the
potential to produce even more paraxanthine if given a
longer reaction time (Fig. 2C). Cell concentration (OD
600
of 20 and 50) and initial caffeine concentrations of 2 and
5 mM were evaluated for paraxanthine production by
Fig. 4. Resting cell assays of MBM013 at varying cell concentrations convert caffeine (A) to paraxanthine (B) and 7-methylxanthine (C).
The varying cell concentrations include OD
600
of 50 (
▼
), 20 (
▲
), 10 (
●
), and 5 (
■
).
Table 3. Conversion of caffeine to paraxanthine and 7-methylxanthine by strain MBM013 in resting cell assays at varying optical
densities with an initial 1 mM caffeine
OD
600
Caffeine consumed (µM) Paraxanthine produced (µM) 7-methylxanthine produced (µM)
5 367 ± 39 229 ± 7 81 ± 3
10 563 ± 64 267 ± 27 177 ± 16
20 678 ± 23 263 ± 8 250 ± 6
50 921 ± 21 153 ± 1 617 ± 12
All values reported are the averages of three replicates with standard deviations in samples taken after five hours of reaction.
648 Biotechnology and Bioprocess Engineering 27: 640-651 (2022)
MBM019. For both initial substrate concentrations, the
higher OD
600
of 50 showed a greater overall conversion of
caffeine than the OD
600
of 20 (Fig. 5, Table S5). The
maximum paraxanthine concentration observed was 1,163
± 49 μM from an assay with an initial caffeine concentration
of 5 mM and cell OD
600
of 50. Thus, we selected these
initial reaction conditions for scaleup to produce and isolate
paraxanthine.
3.3. Paraxanthine purification and recovery
For the production and purification of paraxanthine, strain
MBM019 was tested in a 15 mL reaction prior to the larger
scale reaction at an OD
600
of 50 and a caffeine concentration
of 5 mM. Upon reaction completion, the cells were
separated from the supernatant by centrifugation, and the
methylxanthines in the supernatant were quantified by
HPLC. This preliminary analysis revealed that 1,686 ±
121 μM of caffeine was consumed during the reaction, a
33% molar conversion from 5 mM to approximately
3.35 mM caffeine, resulting in 905 ± 26 μM paraxanthine,
350 ± 19 μM 7-methylxanthine, and a few other minor
unidentified products, some of which can be attributed to
the E. coli host (Fig. S5A and S5B). Although the overall
reaction conversion was slightly lower than in the 2 mL
reaction described above (Fig. 5), the concentration of
paraxanthine would be sufficient to purify easily given a
large reaction volume. Therefore, we proceeded to scale up
the reaction.
Strain MBM019 was grown in four 2.8 L Fernbach
flasks, producing 22.27 g wet cells, which was sufficient to
be used in a 620 mL reaction with an OD
600
of 50. The
cell-caffeine mixture was allowed to react for five hours to
ensure maximum conversion before harvesting. At the
conclusion of the large-scale reaction, caffeine was degraded
to a final concentration of 2.51 mM (corresponding to 300 mg
caffeine consumed), producing 1.02 mM paraxanthine and
0.60 mM 7-methylxanthine. Overall, 49.8 mol% of the
caffeine was consumed; however, only 21.2 mol% of the
initial caffeine was converted to paraxanthine with another
12.4 mol% caffeine used to generate 7-methylxanthine,
summing to a total of 33.6 mol% conversion of caffeine. It
is likely that some of the caffeine was converted to other
products indicated by the presence of a modest 1-
methylxanthine peak seen at 4.5 min and other unknown
peaks seen in the HPLC chromatograph in Fig. S5A.
Preparation of the reaction by filtration and addition of
methanol for HPLC purification resulted in 648 mL
supernatant. The purification conditions were optimized
using approximately 41 mL, leaving 607 mL of 7.5%
methanol-containing supernatant to be separated by HPLC
Fig. 5. Strain MBM019 was reacted with caffeine (A and D) to compare the overall conversion of substrate to paraxanthine (B and E)
and 7-methylxanthine (C and F) at varying cell concentrations and caffeine concentrations. Combinations include an OD
600
of 50 and
2 mM caffeine (
▼
), 50 and 5 mM (
■
), 20 and 2 mM (
●
), and 20 and 5 mM (
▲
).
Biocatalytic Production of Paraxanthine 649
with collection of the paraxanthine peak. 7-Methylxanthine
was partially purified but could not be fully separated and
purified under the conditions described due to overlap of
other compound peaks (Fig. S6).
Following HPLC purification (Fig. S6) and drying of
paraxanthine, we recovered 104.1 mg paraxanthine (Fig. S7).
Given that the final concentration of paraxanthine produced
was 1.02 mM, the theoretical maximum amount of
paraxanthine that could be recovered for this process was
114.5 mg. Compared to the actual collected mass, this
process gave a recovery of 90.9%. The low conversion of
caffeine to paraxanthine (21.2 mol%) remains the largest
barrier to overcome. Development of other mutant N-
demethylase enzymes with increased paraxanthine generation
and decreased 7-methylxanthine production could greatly
improve the yield of paraxanthine from caffeine. Nevertheless,
combination of the reaction and purification processes
described here could result in production of 168 mg
paraxanthine per liter of resting cell reaction.
3.4. Analytical characterization of paraxanthine
Paraxanthine purity was analyzed using authentic HPLC
standards and the retention times were confirmed to be the
same (Fig. S7).
1
H NMR was also used to confirm the
identity of the biologically produced paraxanthine (Fig. S8).
The presence of peaks was confirmed at δ 11.82 (1H)
corresponding to –NH, δ 7.92 (1H) corresponding to -C=CH,
δ 3.86 (3H) and 3.18 (3H) corresponding to both –CH
3
groups. The peaks at δ 3.32 and δ 2.51 have been
confirmed to correspond to water and DMSO, respectively.
There is a very small amount of contamination observed
just below δ 2 that is believed to be the presence of acetic
acid.
By utilizing both HPLC and NMR techniques to confirm
the identity of the produced compound, we have verified
that NdmA4 is capable of producing paraxanthine as the
primary product from caffeine. With so many potential
medical applications, the need for a reliable source of
paraxanthine production is of great importance. This need
has also encouraged the investigation of synthetic routes of
production. A direct methylation of xanthine by an alkylation/
deprotonation method has been proven to successfully
produce 3-methylxanthine, however, the product was not
pure and de-protecting the methylxanthine required a
catalyst, high pressures, high temperatures and long
reaction times [29,58]. Imidazole derivatives offer a new,
somewhat milder route of synthesis, but still require the use
of a strong solvent, such as tetrahydrofuran [29,31,59].
Solid-phase synthetic routes require high-cost CHO (carbon,
hydrogen, and oxygen) resins in organic solvents, which
limits the practicality of large-scale production [60]. Even
more mild routes of synthesis capable of producing 1-, 3-,
and 7-substituted xanthines are only moderately successful
and still required multiple, complicated steps [61].
In contrast, whole-cell biosynthetic production offers the
much-needed alternative to purely synthetic production of
paraxanthine. E. coli is known for the simplicity with
which it can be grown on low-cost, non-toxic resources,
such as LB, and at mild temperatures from 18-37
o
C. By
harnessing bacterial enzymes, methylxanthines like
paraxanthine can be produced to a much higher selectivity
and purity over the course of just a few days. In addition,
this process allows for the use of caffeine as a starting point
for conversion of a low-cost compound into a high-value
biochemical, paraxanthine, that will incentivize the collection
and recapture of waste caffeine. Overall, microbial production
is more scalable than a strictly chemical process, particularly
when resins and other catalysts are required, and offers a
platform for process optimization and controllable product
diversification [62].
4. Conclusion
We have presented the first report of a biocatalytic process
designed for the production and purification of the high-
value biochemical paraxanthine using E. coli strain
MBM019. The process described here produced 114.5 mg
paraxanthine from 300 mg caffeine under ambient
conditions using a simple biocatalytic reaction prior to
further purification steps. We further isolated 104.1 mg
paraxanthine powder via prep-scale HPLC with a
purification yield of 90.9%. This first demonstration of
biocatalytic paraxanthine production and purification will
provide the foundation for additional increases in conversion
and yield via enzyme, strain, and process improvements.
Acknowledgements
The authors thank Dr. Ken Belmore and the University of
Alabama Department of Chemistry and Biochemistry for
assistance with the NMR.
This work was supported by University of Alabama research
funds. M.B. Mock is supported by the U.S. Department of
Education as a GAANN Fellow (P200A180056).
Ethical Statements
The authors declare no competing interests.
Neither ethical approval nor informed consent was required
for this study.
650 Biotechnology and Bioprocess Engineering 27: 640-651 (2022)
Electronic Supplementary Material (ESM)
The online version of this article (doi: 10.1007/s12257-
021-0301-0) contains supplementary material, which is
available to authorized users.
References
1. Franco, R., A. Oñatibia-Astibia, and E. Martínez-Pinilla (2013)
Health benefits of methylxanthines in cacao and chocolate.
Nutrients. 5: 4159-4173.
2. Singh, N., A. K. Shreshtha, M. S. Thakur, and S. Patra (2018)
Xanthine scaffold: scope and potential in drug development.
Heliyon. 4: e00829.
3. Constantin, S., F. G. Lupascu, M. Apotrosoaei, I. M. Vasincu, D.
Lupascu, F. Buron, S. Routier, and L. Profire (2017) Synthesis
and biological evaluation of the new 1,3-dimethylxanthine
derivatives with thiazolidine-4-one scaffold. Chem. Cent. J. 11: 12.
4. Xu, K., Y.-H. Xu, J.-F. Chen, and M. A. Schwarzschild (2010)
Neuroprotection by caffeine: time course and role of its
metabolites in the MPTP model of Parkinson's disease.
Neuroscience. 167: 475-481.
5. Janitschke, D., A. A. Lauer, C. M. Bachmann, H. S. Grimm, T.
Hartmann, and M. O. Grimm (2021) Methylxanthines and
neurodegenerative diseases: an update. Nutrients. 13: 803.
6. Oñatibia-Astibia, A., R. Franco, and E. Martínez-Pinilla (2017)
Health benefits of methylxanthines in neurodegenerative diseases.
Mol. Nutr. Food Res. 61: 1600670.
7. Negida, A., M. Elfil, A. Attia, E. Farahat, M. Gabr, A. Essam, D.
Attia, and H. Ahmed (2017) Caffeine; the forgotten potential for
Parkinson's disease. CNS Neurol. Disord. Drug Targets. 16: 652-
657.
8. Ross, G. W., R. D. Abbott, H. Petrovitch, D. M. Morens, A.
Grandinetti, K.-H. Tung, C. M. Tanner, K. H. Masaki, P. L.
Blanchette, J. D. Curb, J. S. Popper, and L. R. White (2000)
Association of coffee and caffeine intake with the risk of
Parkinson disease. JAMA. 283: 2674-2679.
9. Ponte, B., M. Pruijm, D. Ackermann, G. Ehret, N. Ansermot, J.
A. Staessen, B. Vogt, A. Pechère-Bertschi, M. Burnier, P.-Y.
Martin, C. B. Eap, M. Bochud, and I. Guessous (2018)
Associations of urinary caffeine and caffeine metabolites with
arterial stiffness in a large population-based study. Mayo Clin.
Proc. 93: 586-596.
10. Scurachio, R. S., F. Mattiucci, W. G. Santos, L. H. Skibsted, and
D. R. Cardoso (2016) Caffeine metabolites not caffeine protect
against riboflavin photosensitized oxidative damage related to
skin and eye health. J. Photochem. Photobiol. B 163: 277-283.
11. Nunnari, G., E. Argyris, J. Fang, K. E. Mehlman, R. J. Pomerantz,
and R. Daniel (2005) Inhibition of HIV-1 replication by caffeine
and caffeine-related methylxanthines. Virology. 335: 177-184.
12. Lee, I.-A., A. Kamba, D. Low, and E. Mizoguchi (2014) Novel
methylxanthine derivative-mediated anti-inflammatory effects in
inflammatory bowel disease. World J. Gastroenterol. 20: 1127-1138.
13. Singh, H., N. S. Sahajpal, H. Singh, V. Vanita, P. Roy, S. Paul, S.
K. Singh, I. Kaur, and S. K. Jain (2021) Pre-clinical and cellular
toxicity evaluation of 7-methylxanthine: an investigational drug
for the treatment of myopia. Drug Chem. Toxicol. 44: 575-584.
14. Ferré, S., M. Orrú, and X. Guitart (2013) Paraxanthine: connecting
caffeine to nitric oxide neurotransmission. J. Caffeine Res. 3: 72-78.
15. Okuro, M., N. Fujiki, N. Kotorii, Y. Ishimaru, P. Sokoloff, and S.
Nishino (2010) Effects of paraxanthine and caffeine on sleep,
locomotor activity, and body temperature in orexin/ataxin-3
transgenic narcoleptic mice. Sleep. 33: 930-942.
16. Orrú, M., X. Guitart, M. Karcz-Kubicha, M. Solinas, Z.
Justinova, S. K. Barodia, J. Zanoveli, A. Cortes, C. Lluis, V.
Casado, F. G. Moeller, and S. Ferré (2013) Psychostimulant
pharmacological profile of paraxanthine, the main metabolite of
caffeine in humans. Neuropharmacology. 67: 476-484.
17. Negus, S. S. and L. L. Miller (2014) Intracranial self-stimulation
to evaluate abuse potential of drugs. Pharmacol. Rev. 66: 869-917.
18. Guerreiro, S., D. Toulorge, E. Hirsch, M. Marien, P. Sokoloff,
and P. P. Michel (2008) Paraxanthine, the primary metabolite of
caffeine, provides protection against dopaminergic cell death via
stimulation of ryanodine receptor channels. Mol. Pharmacol.
74: 980-989.
19. Klemmer, I., S. Yagi, and O. A. Gressner (2011) Oral application of
1,7-dimethylxanthine (paraxanthine) attenuates the formation of
experimental cholestatic liver fibrosis. Hepatol. Res. 41: 1094-1109.
20. Geraets, L., A. Haegens, A. R. Weseler, K. Brauers, J. H.
Vernooy, E. F. Wouters, A. Bast, and G. J. Hageman (2010)
Inhibition of acute pulmonary and systemic inflammation by 1,7-
dimethylxanthine. Eur. J. Pharmacol. 629: 132-139.
21. Trier, K., S. Munk Ribel-Madsen, D. Cui, and S. Brøgger
Christensen (2008) Systemic 7-methylxanthine in retarding axial
eye growth and myopia progression: a 36-month pilot study. J.
Ocul. Biol. Dis. Infor. 1: 85-93.
22. Bertrand, B., G. M. M. Groothuis, A. Casini, S. Loic, E. Bodio,
R. Philippe, P. Legendre, D. Monchaud, and M. Picquet (2014)
Xanthine-based gold(I) N-heterocyclic carbene complexes:
synthesis and anticancer evaluation. J. Biol. Inorg. Chem. 19: S589.
23. Mohamed, H. A., B. R. Lake, T. Laing, R. M. Phillips, and C. E.
Willans (2015) Synthesis and anticancer activity of silver(I)-N-
heterocyclic carbene complexes derived from the natural xanthine
products caffeine, theophylline and theobromine. Dalton Trans.
44: 7563-7569.
24. Szadkowska, A., S. Staszko, E. Zaorska, and R. Pawłowski
(2016) A theophylline based copper N-heterocyclic carbene
complex: synthesis and activity studies in green media. RSC Adv.
6: 44248-44253.
25. Valdés, H., D. Canseco-González, J. M. Germán-Acacio, and D.
Morales-Morales (2018) Xanthine based N-heterocyclic carbene
(NHC) complexes. J. Organomet. Chem. 867: 51-54.
26. Zhang, J.-J., C.-M. Che, and I. Ott (2015) Caffeine derived
platinum(II) N-heterocyclic carbene complexes with multiple
anti-cancer activities. J. Organomet. Chem. 782: 37-41.
27. Stavric, B. (1988) Methylxanthines: toxicity to humans. 3.
Theobromine, paraxanthine and the combined effects of
methylxanthines. Food Chem. Toxicol. 26: 725-733.
28. Müller, C. E., D. Deters, A. Dominik, and M. Pawlowski (1998)
Synthesis of paraxanthine and isoparaxanthine analogs (1,7- and
1,9-substituted xanthine derivatives). Synthesis (Stuttg.) 1998:
1428-1436.
29. Gulevskaya, A. V. and A. F. Pozharskii (1991) Synthesis of N-
substituted xanthines (review). Chem. Heterocycl. Compd. (N. Y.)
27: 1-23.
30. He, R., S. M. Ching, and Y. Lam (2006) Traceless solid-phase
synthesis of substituted xanthines. J. Comb. Chem. 8: 923-928.
31. Zavialov, I. A., V. H. Dahanukar, H. Nguyen, C. Orr, F. Zhang,
and D. R. Andrews (2004) New and practical method for
synthesis of 1- and 1,3-substituted xanthines. Org. Lett. 6: 2237-
2240. (Erratum published 2004, Org. Lett. 6: 3017)
32. Müller, C. E., D. Shi, M. Manning Jr., and J. W. Daly (1993)
Synthesis of paraxanthine analogs (1,7-disubstituted xanthines)
and other xanthines unsubstituted at the 3-position: structure-
activity relationships at adenosine receptors. J. Med. Chem. 36:
3341-3349.
33. Ashengroph, M. (2017) Salinivibrio costicola GL6, a novel
isolated strain for biotransformation of caffeine to theobromine
Biocatalytic Production of Paraxanthine 651
under hypersaline conditions. Curr. Microbiol. 74: 34-41.
34. Summers, R. M., T. M. Louie, C. L. Yu, and M. Subramanian
(2011) Characterization of a broad-specificity non-haem iron N-
demethylase from Pseudomonas putida CBB5 capable of utilizing
several purine alkaloids as sole carbon and nitrogen source.
Microbiology (Reading). 157: 583-592.
35. Dash, S. S. and S. N. Gummadi (2010) Biodegradation of caffeine
by Pseudomonas sp. NCIM 5235. Res. J. Microbiol. 5: 745-753.
36. Yu, C. L., T. M. Louie, R. Summers, Y. Kale, S. Gopishetty, and
M. Subramanian (2009) Two distinct pathways for metabolism
of theophylline and caffeine are coexpressed in Pseudomonas
putida CBB5. J. Bacteriol. 191: 4624-4632.
37. Quandt, E. M., R. M. Summers, M. V. Subramanian, and J. E.
Barrick (2015) Draft genome sequence of the bacterium
Pseudomonas putida CBB5, which can utilize caffeine as a sole
carbon and nitrogen source. Genome Announc. 3: e00640-15.
38. Summers, R. M., T. M. Louie, C.-L. Yu, L. Gakhar, K. C. Louie,
and M. Subramanian (2012) Novel, highly specific N-
demethylases enable bacteria to live on caffeine and related
purine alkaloids. J. Bacteriol. 194: 2041-2049.
39. Summers, R., S. Gopishetty, S. Mohanty, and M. Subramanian
(2014) New genetic insights to consider coffee waste as
feedstock for fuel, feed, and chemicals. Cent. Eur. J. Chem. 12:
1271-1279.
40. Summers, R. M., J. L. Seffernick, E. M. Quandt, C. L. Yu, J. E.
Barrick, and M. V. Subramanian (2013) Caffeine junkie: an
unprecedented glutathione S-transferase-dependent oxygenase
required for caffeine degradation by Pseudomonas putida CBB5.
J. Bacteriol. 195: 3933-3939.
41. Algharrawi, K. H., R. M. Summers, S. Gopishetty, and M.
Subramanian (2015) Direct conversion of theophylline to 3-
methylxanthine by metabolically engineered E. coli. Microb. Cell
Fact. 14: 203.
42. Algharrawi, K. H. R., R. M. Summers, and M. Subramanian
(2017) Production of theobromine by N-demethylation of
caffeine using metabolically engineered E. coli. Biocatal. Agric.
Biotechnol. 11: 153-160.
43. Algharrawi, K. H. R. and M. Subramanian (2020) Production of
7-methylxanthine from theobromine by metabolically engineered
E. coli. Iraqi J. Chem. Pet. Eng. 21: 19-27.
44. Mock, M. B., S. Zhang, B. Pniak, N. Belt, M. Witherspoon, and
R. M. Summers (2021) Substrate promiscuity of the NdmCDE
N
7
-demethylase enzyme complex. Biotechnol. Notes. 2: 18-25.
45. Kim, J. H., B. H. Kim, S. Brooks, S. Y. Kang, R. M. Summers,
and H. K. Song (2019) Structural and mechanistic insights into
caffeine degradation by the bacterial N-demethylase complex. J.
Mol. Biol. 431: 3647-3661.
46. Mills, S. B., M. B. Mock, and R. M. Summers (2021) Rational
protein engineering of bacterial N-demethylases to create
biocatalysts for the production of methylxanthines. https://
doi.org/10.1101/2021.12.17.472166
47. MacWilliams, M. P. and M.-K. Liao (2006) Luria Broth (LB) and
Luria Agar (LA) Media and Their Uses Protocol. American
Society for Microbiology, Washington, DC, USA.
48. Held, D., K. Yaeger, and R. Novy (2003) New coexpression
vectors for expanded compatibilities in E. coli. Innovations. 18: 4-6.
49. Sektas, M. and W. Szybalski (2002) Novel single-copy pETcoco
vector with dual controls for amplification and expression.
Innovations. 14: 10-12.
50. Salis, H. M. (2011) The ribosome binding site calculator.
Methods Enzymol. 498: 19-42.
51. Retnadhas, S. and S. N. Gummadi (2018) Identification and
characterization of oxidoreductase component (NdmD) of
methylxanthine oxygenase system in Pseudomonas sp. NCIM
5235. Appl. Microbiol. Biotechnol. 102: 7913-7926.
52. Zhang, Y., Z. Huang, C. Du, Y. Li, and Z. Cao (2009) Introduction
of an NADH regeneration system into Klebsiella oxytoca leads to
an enhanced oxidative and reductive metabolism of glycerol.
Metab. Eng. 11: 101-106.
53. Berríos-Rivera, S. J., G. N. Bennett, and K.-Y. San (2002) The
effect of increasing NADH availability on the redistribution of
metabolic fluxes in Escherichia coli chemostat cultures. Metab.
Eng. 4: 230-237.
54. Denby, K. J., J. Iwig, C. Bisson, J. Westwood, M. D. Rolfe, S. E.
Sedelnikova, K. Higgins, M. J. Maroney, P. J. Baker, P. T.
Chivers, and J. Green (2016) The mechanism of a formaldehyde-
sensing transcriptional regulator. Sci. Rep. 6: 38879.
55. Chen, N. H., K. Y. Djoko, F. J. Veyrier, and A. G. McEwan
(2016) Formaldehyde stress responses in bacterial pathogens.
Front. Microbiol. 7: 257.
56. Gutheil, W. G., E. Kasimoglu, and P. C. Nicholson (1997)
Induction of glutathione-dependent formaldehyde dehydrogenase
activity in Escherichia coli and Hemophilus influenza. Biochem.
Biophys. Res. Commun. 238: 693-696.
57. Herring, C. D. and F. R. Blattner (2004) Global transcriptional
effects of a suppressor tRNA and the inactivation of the regulator
frmR. J. Bacteriol. 186: 6714-6720.
58. Szolomájer, J., G. Paragi, G. Batta, C. F. Guerra, F. M. Bickelhaupt,
Z. Kele, P. Pádár, Z. Kupihár, and L. Kovács (2011) 3-Substituted
xanthines as promising candidates for quadruplex formation:
computational, synthetic and analytical studies. New J. Chem.
35: 476-482.
59. Hergueta, A. R., M. J. Figueira, C. López, O. Caamaño, F.
Fernández, and J. E. Rodríguez-Borges (2002) Synthesis of series
of 1- and 3-differently substituted xanthines from imidazoles.
Chem. Pharm. Bull. (Tokyo) 50: 1379-1382.
60. He, R. and Y. Lam (2005) A highly efficient solid-phase synthesis
of 1,3-substituted xanthines. J. Comb. Chem. 7: 916-920.
61. Müller, C. E. and J. Sandoval-Ramírez (1995) A new versatile
synthesis of xanthines with variable substituents in the 1-, 3-, 7-
and 8-positions. Synthesis (Stuttg.) 1995: 1295-1299.
62. McKeague, M., Y.-H. Wang, A. Cravens, M. N. Win, and C. D.
Smolke (2016) Engineering a microbial platform for de novo
biosynthesis of diverse methylxanthines. Metab. Eng. 38: 191-203.
Publisher’s Note Springer Nature remains neutral with regard to
jurisdictional claims in published maps and institutional affiliations.
