Content uploaded by Michael J. Benedik
Author content
All content in this area was uploaded by Michael J. Benedik
Content may be subject to copyright.
Helical structure of unidirectionally shadowed metal replicas
of cyanide hydratase from Gloeocercospora sorghi
J.D. Woodward
a,b
, B.W. Weber
a
, M.P. Scheffer
a,e
, M.J. Benedik
c
,
A. Hoenger
d
, B.T. Sewell
a,b,
*
a
Electron Microscope Unit, University of Cape Town, 7701 Cape Town, South Africa
b
Department of Biotechnology, University of the Western Cape, Cape Town, South Africa
c
Department of Biology, Texas A&M University, TX, USA
d
Department of Molecular, Cellular and Developmental Biology, University of Colorado at Boulder, USA
e
EMBL Heidelberg, Meyerhofstrasse 1, Heidelberg, Germany
Received 18 April 2007; received in revised form 26 September 2007; accepted 27 September 2007
Available online 1 October 2007
Abstract
The helical filaments of the cyanide hydratase from Gloeocercospora sorghi have been reconstructed in three dimensions from freeze
dried, unidirectionally shadowed specimens using iterative real-space helical reconstruction. The average power spectrum of all selected
images has three clear reflections on different layer lines. The reconstruction is complicated by the fact that three possible indexing
schemes are possible and reconstructions using the starting symmetries based on each of these indexing schemes converge on three-
dimensional volumes which appear plausible. Because only one side is visible in shadowed specimens, it is necessary to examine the
phases from a single filament by cryo-electron microscopy in order to make an unequivocal assignment of the symmetry. Because of
the novel nature of the reconstruction method used here, conventional cryo-EM methods were also used to determine a second recon-
struction, allowing us to make comparisons between the two. The filament is shown to have a left-handed one-start helix with D
1
sym-
metry, 5.46 dimers per turn and a pitch of 7.15 nm. The reconstruction suggests the presence of an interaction across the groove not
previously seen in nitrilase helical fibres.
2007 Elsevier Inc. All rights reserved.
Keywords: Cyanide hydratase; Gloeocercospora sorghi; Nitrilase; Helical reconstruction; Shadowing; Midilab; IHRSR
1. Introduction
Cyanide hydratase (CHT) is a substrate-specific member
of the nitrilase family of enzymes. Nitrilases catalyse the
hydrolysis of nitriles to amides or to acids and ammonia
with a broad range of specificities ranging from hydrogen
cyanide to various aliphatic and aromatic nitriles (Pace
and Brenner, 2001). In general, their natural substrates
have not been identified. Because of the inherent enantio-
specific nature and selectivity of enzymatic catalysis, nitri-
lases are used in a variety of industrial processes (Brady
et al., 2004). For instance, nitrilases are used to manufac-
ture the active enantiomers, (R)-mandelic acid, (R)-3-chlo-
romandelic acid (Brady et al., 2004), (S)-phenyllactic acid
and (R)-3-hydroxy-4-cyano-butyric acid, an important
intermediate in the synthesis of the cholesterol-lowering
drug atorvastatin calcium (Banerjee et al., 2002; O’Reilly
and Turner, 2003). Although microbial treatment of toxic
industrial effluent is hampered by the presence of condi-
tions that inhibit microbial growth (reviewed by Baxter
and Cummings, 2006), the potential exists for the use of
purified microbial nitrilases for on-site cyanide remediation
since they may be engineered to tolerate varying levels of
pH and temperature, which the organisms themselves
1047-8477/$ - see front matter 2007 Elsevier Inc. All rights reserved.
doi:10.1016/j.jsb.2007.09.019
*
Corresponding author. Address: Electron Microscope Unit, University
of Cape Town, 7701 Cape Town, South Africa. Fax: +27 21 6891528.
E-mail address: Trevor.Sewell@uct.ac.za (B.T. Sewell).
www.elsevier.com/locate/yjsbi
Available online at www.sciencedirect.com
Journal of Structural Biology 161 (2008) 111–119
Journal of
Structural
Biology
may be sensitive to (Jandhyala et al., 2003). CHTs have
been identified in a variety of fungal species, but appear
not to occur in bacteria or plants (Wang et al., 1999). They
catalyse the conversion of cyanide to formamide with a
small fraction (up to 0.4%, Nolan et al., 2003) being con-
verted to formic acid. The different reaction pathways
result from the breakage of a different bond in the tetrahe-
dral thioimidate intermediate form ed with the active site
cysteine (Jandhyala et al., 2003; reviewed by O’Reilly and
Turner, 2003).
Gloeocercospora sorghi is a fungus that infects sorghum.
When lesions are formed on sorghum, cyanoglycocides
degrade to form hydrogen cyanide (HCN) . Expression of
CHT is induced in G. sorghi by HCN and this enzyme
can account for up to 18% of the protein in the organism
(Wang et al., 1992). It was therefore postulated that the
purpose of the enzyme was to protect the pathogen in a
high cyanide environment; however, expression of CHT
is not necessary for infectivity and thus its role in the cell
remains uncertain ( Wang et al., 1999). G. sorghi CHT has
about 30% identity to the bacterial cyanide dihydratases
(cynD) from Pseudomonas stutzeri AK61 and Bacillus
pumilus C1. Both of these enzymes form self-terminating,
homo-oligomeric spirals of 14 and 18 subunits, respect ively
(Sewell et al., 2005). However, the CHT from G. sorghi has
a significantly higher molecular weight (2–10 MDa) (Fry
and Millar, 1972), despite having a similar subunit molec-
ular weight of 40.9 kDa (Wang et al., 1992).
Freeze-drying and unidirectional heavy metal shadow-
ing are an established procedure for investigating biologi-
cal macro-molecular surface structure (Gross et al., 1990)
allowing for unambiguous determination of the handed-
ness. Micrographs of samples prepared using these meth-
ods display excellent signal to noise ratio, allowing direct
observation of structures at resolutions down to 20 A
˚
(2 nm) (Hoenger et al., 2000) without making replicas.
The Midilab instrument in which the freeze-drying and
shadowing occur in a chamber mounted on the microscope
column allows optimal preservation of structural detail by
maintaining the sample under vacuum and cryo-conditions
during shadowing, transfer into the microscope and view-
ing (Gross et al., 1990).
Micrographs of unidirectionally shadowed material
contain three-dimensional information in the form of
pixel density, which is dependent on the angle between
the sample surface normal, and the shadowing direction.
The thickness of the metal film deposited is direct ly
related to the topography of the sample when viewed
from a direction other than the original evaporation
source (Gross, 1987). The contras t in the resulting micro-
graph is therefore related to the surface topography
(Guckenberger, 1985). This informat ion, combined with
image processing methods, has been used to derive
three-dimensional surface reconstructions in the form of
digital elevation maps (DEMs) by the method of surface
relief reconstruction (Smith and Kistler, 1977; Smith and
Ivanov, 1980).
The limitation of these methods is that the resulting map
consists of a 2.5-dimensional (2.5D) representation of a
three-dimensional object. This limitation has been over-
come by applying conventional Fourier–Bessel methods
to rotary shadowed filaments which produces a three-
dimensional reconstruction of the metal cast, from which
the filament can be extracted (Morris et al., 1994). More
recently reconstructions have been produced by applyi ng
tomographic and single-particle approaches to rotationally
shadowed samples (Lanzavecchia et al., 1998, 2005; Lupetti
et al., 2005). This allows a complete reconstruction of the
metal cast surrounding the object to be made. The advan-
tage of this approach is that there is no topological con-
straint on the resulting reconstruction.
Although several superb visualizations of helical objects
have been made with the Midilab instrument, no attempt
has been made to reconstruct such images in three dimen-
sions. Such reconstruction would require that the metal
coating be sufficiently thin to allow proportionality
between the thickness of the coating and the projected
density. The coating applied in the Midilab instrument
has a thickness of 0.3–0.4 nm and is thus thin enough to
allow reconstruction as the linear relationship between
the metal thickness and the absorbance is maintained.
Since the shadowing is unidirectional, the amplitudes of
the various layer lines are dependent on the orientation
of the fibre relative to the shadowing direction. This
means that Fourier–Bessel reconstruction cannot be used.
However, since the fibres lie at all angles to the shadowing
direction a reconstruction method based on averaging
techniques could conceivably be used and this should
result in a reconstruction of the metal component of the
specimen. The ability to average successfully would
require that there is sufficient overlap of information in
fibres shadowed from different directions to allow for
accurate alignment.
We have applied the iterative helical real-space recon-
struction method (IHRSR) (
Egelman, 2000) here to micro-
graphs of unidirectionally shadowed G. sorghi filaments.
IHRSR allows the determination of helical symmetry by
iterative refinement. This method applies a single-particle
approach, including reference-based alignment and back-
projection to boxed segments of the continuous extended
fibres. The advantage of IHRSR is that the method can
deal with disordered or heterogeneous filaments and is
unaffected by the problem of Bessel overlap (Egelman,
2000). Applying IHRSR to metal-shadowed data resulted
in the determination of the low-resolution structure of G.
sorghi CHT. We show here that the CH from G. sorghi
forms extended left-handed helices, built on similar princi-
ples to those described for the cyanide dihydratases (Sewell
et al., 2005). In parallel, for the purposes of comparison, we
have used IHRS R to produce a low-resolution reconstruc-
tion from images of the CHT embedded in ice and recorded
on a CCD camera. Comparisons between the two recon-
structions give considerable insight into the nature of
images made with Midilab.
112 J.D. Woodward et al. / Journal of Structural Biology 161 (2008) 111–119
2. Materials and methods
2.1. Expression and purificatio n
The cyanide hydratase was recombinantly expressed in
Escherichia coli BL21 pLysS by IPTG induction from the
plasmid MB2313 (Jand hyala et al., 2005). The cells were
pelleted by centrifugation at 4000g at 4 C for 20 min and
resuspended in 50 mM Tris–HCl, pH 8.0 containing a pro-
tease inhibitor cocktail (Roche). Cells were disrupted by
sonication (Misonix 3000, US) and the soluble fraction
was clarified by centrifugation at 20,000g at 4 C for
30 min. The soluble lysate was subjected to ammonium sul-
phate precipitation at 50% saturation on ice. The resultant
pellet was resuspended in 50 mM Tris–HCl, pH 8.0 and fil-
tered through a 0.45-lm acetate filter (GE-Healthcare).
Anion exchange chromatography was performed on a
Hi-prep 16/10 Q XL Column (GE-Healthcare) equilibrated
with 50 mM Tris–HCl, 100 mM NaCl pH 8.0 and eluted
with 50 mM Tris–HCl, 1 M NaCl pH 8.0 at 5 ml min
1
.
Fractions containing the cyanide hydratase were identi-
fied by performing the picric acid colorimetric assay.
Twenty microliters of a 100 mM KCN solution (in
50 mM Tris–HCl pH 8.0) was added to 80 ll aliquots of
fractionated protein and allowed to incubate at room tem-
perature for 1 h. Equal volumes of 500 mM Na
2
CO
3
and
1.5% picric acid in water were mixed and 80 ll added to
the protein samples. Samples were placed into boiling
water and after 5 min cyanide hydratase containing frac-
tions turned a yellow colour. These fractions were pooled
and concentrated using an amicon ultrafiltration mem-
brane with a molecular weight cut off of 10 kDa (Millipore,
USA). The concentrated protein fraction was separated on
a Sephacryl S-300 HR gel filtration column (GE-Health-
care) equilibrated with 50 mM Tris–HCl, 200 mM NaCl
pH 8.0 at 0.5 ml min
1
.
2.2. Electron microscopy
Micrographs of unidirectionally shadowed sample were
obtained on the Midilab instrument (Gross et al., 1990).
Briefly, a drop of purified sample was pipetted onto
glow-discharged, carbon-coated grids and allowed to
absorb for two to three minutes before being rinsed in dis-
tilled water, blotted and rapidly vitrified by plunging into
liquid nitr ogen. Samples were cryo-transferred into the
Midilab instrument, mounted onto a Philips-CM12 elec-
tron microscope, and freeze-dried for two hours at a tem-
perature of 193 K (80 C) and pressure of 1027 mbar
before being unidirectionally shadowed at 45 with Tanta-
lum/tungsten (Ta/W). The thickness of the coating was
monitored with a quartz crystal oscillator. The shadowing
was stopped when the frequency changed by 100 Hz, which
corresponded to a thickness of 0.3–0.4 nm at the specimen.
Twenty-five micrographs were recorded with a GATAN-
794 Multiscan CCD camera at an elect ron dose of 0.5 to
1 · 10
3
e
/nm
2
at a magnification of 52,500· and sampling
of 5.3 A
˚
/pixel (Fig. 1a).
Samples for cryo-microscopy were produced by pipet-
ting a drop of 0.2–0.3 mg ml
1
purified protein in buffer
onto holey-carbon grids, blotting these with filter paper
and plunging into liquid ethane. Electron microscopy was
performed at low dose on a Phillips CM 200 FEG operated
at 200 kV. Two hundred and twenty images were captured
at a magnification of 38,000· using a Tietz 2k · 2k CCD
camera (Fig. 1b). Calibration of the images was performed
by scaling the calculated power spectrum to match the cor-
rectly calibrated Midilab micrographs, yielding a sampling
of 3.40 A
˚
/pixel.
2.3. Image processing
Helical segments (526) of 128 · 128 pixels with 92%
overlap were selected and aligned using BOXER (Ludtke
et al., 1999) from 49 Midilab helices. Four thousand five
hundred and twenty-seven segments were selected from
400 helices from the cryo-dataset. In both cases the seg-
ments were band-pass filtered between 270 A
˚
and 25 A
˚
and normalised to a mean of 0 and standard deviation of
1. Boxed helical segments were aligned in a reference-free
procedure (Penczek et al., 1992) and averaged in order to
improve the SNR; power spectra were calculated using Spi-
der V9.05 (Frank et al., 1996). The IHRSR procedure as
described by Egelman (2000) was initiated with a feature-
less cylinder, using starting values for the helical symmetry
predicted by indexing. The cryo-dataset contained a large
proportion of out-of-plane filaments; these were identified
during reference-based alignment and eliminated. Images
were excluded if they fell below a correlation threshold,
which was increased every iteration. The final reconstruc-
tions were generated with 84 and 507 Midilab and cryo-
helical segments, respectively.
Molecular graphics images were produced using the
UCSF Chimera package from the Resource for Biocom-
puting, Visualization, and Informatics at the University
of California, San Francisco (supported by NIH P41
RR-01081) (Pettersen et al., 2004).
2.4. Resolution
The protocol for resolution estimation by Fourier shell
correlation (FSC) was provided by E.H. Egelman. Briefly,
two independent models were created from featureless cyl-
inders by starting at different angles (u
1
= 65.7,
u
2
= 66.2, z
1,2
=13A
˚
), the radial mask function was
uncorrelated by using two different maximum helical diam-
eters (90 A
˚
and 93 A
˚
, respectively). The two models were
iterated 50 times and converged on the same stabl e solu-
tion. The two volumes were aligned in z and u and win-
dowed (x
1
= 128, y
1
= 128, z
1
= 110; x
2
= 128, y
2
= 128,
z
2
= 116) and subsequently padded to the same volume
(x = 128, y = 128, z = 128). The Fourier shell correlation
was calculated between the two volumes with a ring size
J.D. Woodward et al. / Journal of Structural Biology 161 (2008) 111–119 113
of 1 and plotted in Matlab V6.5 (
Mathworks Ltd.). For a
discussion of the various issues involved with resolution
determination by FSC see Yang et al. (2003). In the case
of the cryo-reconstruction the image stack was split in half.
3. Results
The unprocessed, unidirectionally shadowed micro-
graphs of G. sorghi nitr ilase reveal that the enzyme forms
a filament with a diameter of 12 nm. The height of the
fibre can be approximated from the length of the shadow
cast by the fibre and the shadow ing angle (Fig. 1a). The
height is similar to the estimated diameter, suggesting that
the fibre is not flattened. Prominent one-start left-handed
helical striations as well as right-handed apparently four-
start helical striations are visible in unprocessed images
(Fig. 1a). The fibres seen embedded in vitreous ice in the
holes of the carbon film were generally short and not all
of them lay in the plane of the carbon film (Fig. 1b).
3.1. Symmetry determination
The power spectrum derived from unidirectionally shad-
owed helical segments was indexed by comparing the ratios
of different principal maxima of different Bessel orders with
the experimentally determined ratios obtained from the
power spectrum. At the available resolution the data were
consistent with three different indexing schemes; each pre-
dicting a different order for the l = 3 layer-line (Fig. 2).
The IHRSR algorithm converged on a different reconstruc-
tion, when initiated from each of the three symmetries pre-
dicted by the indexing (Fig. 3) and thus failed to resolve the
ambiguity. The correct indexing scheme, and therefore the
correct reconstruction, was identified by calculating the
phases and amplitudes of the Fourier transform of a single
cryo-fibre. The uncertain layer-line arising from the front
and back of the filament are in phase, indicating that the
layer-line order is even (Fig. 4). The correct indexing is
therefore the one that results from the unknown layer-line
being of order four. The filament has an axial rise of 13 A
˚
and 65 rotation per subunit.
Fig. 1. (a) Unprocessed micrograph of cryo-metal-shadowed G. sorghi nitrilase. The shadowing is at an elevation of 45. The majority of the metal falls
onto the front surface of the helix, allowing unambiguous determination of handedness (left-handed). Arrow indicates approximate shadowing direction.
(b) Filtered and contrast-enhanced cryo-micrograph of G. Sorghi nitrilase. Asterisk indicates an end-on view of a short filament. Scale-bars indicate 50 nm.
Fig. 2. Power spectrum calculated from vertically aligned helical seg-
ments; layer lines arise from the front surface of the filament only,
therefore phase information usually used to eliminate the ambiguity
associated with indexing helical patterns is unavailable. Layer lines occur
at 1/74 A
˚
, 1/49 A
˚
and 1/37 A
˚
. Three indexing schemes are compatible with
the data at the resolution available. Red spots indicate the resulting
indexing scheme if the unknown layer-line is of order 5 (hence 13 dimers in
2 turns or 13:2), blue spots indicate the indexing scheme that results from
the unknown layer-line being order 4 (hence 11:2) and green spots indicate
the resulting indexing scheme if the layer-line is of order 3 (hence 9:2). The
pitch is 7 nm.
114 J.D. Woodward et al. / Journal of Structural Biology 161 (2008) 111–119
3.2. Iterative helical real-space reconstruction
The reconstructions converged on stable models, which
closely approximated the predicted symmetries. Fig. 5
shows the convergence of the algorithm in the case of a
starting model having 11 dimers in two turns for the Mid-
ilab dataset.
3.3. Analysis of the Midilab model
The model has an axial rise of 13 A
˚
and 66 rotation
per subunit (Fig. 5). The one-start helix has a pitch of
7.15 nm and there are 5.46 dimers per turn. The filament
has a diameter of 11.5 nm, a central channel of diame ter
6.2 nm. Sites of interaction homologous to those
described in earlier reconstructions of helical and oligo-
meric nitrilases (e.g. Thuku et al., 2007) were identified
(Fig. 6a). The filament has two dyad axes (Fig. 6b), the first
passes through the A-surface, marked by a depression indi-
cating the dimer interface, and the hole. The second, lies
through the F- and C-surfaces. The C-surf ace is the site
at which adjacent dimers interact. The F-surface represents
an interaction occurring across the groove of the one start
helix, in this case occurring between dimers n and n +5
however; this interaction is in a different position to that
normally seen in nitrilase fibres (Fig. 6a). The helical sym-
metry is D
1
S
5.5
. The resolution was estimated to lie
between 33 A
˚
and 38 A
˚
(Fig. 7).
3.4. Comparison of the models
The metal shadowed- and cryo-reconstructions corre-
lated to a resolution of 38 A
˚
(FSC = 0.5) (Fig. 7). This
value approximates the resolution at which the highest fre-
quency layer-line is visible (37 A
˚
) as well as the estimated
resolution of the Midilab reconstruction (33 A
˚
). This simi-
larity is immediately apparent when the aligned metal-
shadowed images are compared to the Midilab and cryo-
reconstructions (Fig. 8). The helical striations and features,
as well as the diameters of the two reconstructions closely
resemble the aligned shadowed images as well as each
other. In cross-section, the metal component of the Midilab
Fig. 3. The IHRSR algorithm was initiated at the three helical symme-
tries, predicted by indexing and converged on three stable reconstructions.
All the reconstructions are plausible in the sense that the dimeric structure
of the repeating unit can be visualised and the interactions appear
reasonable based on our limited biochemical knowledge.
Fig. 4. By computing the amplitudes and phases of the power spectrum
from a single cryo-filament, the correct indexing scheme can be
determined. The amplitudes of the Fourier transform are indicated by
intensity, the phases by hue. The scale-bar indicates the relative phase
difference between the layer lines and represents a total of 2p radians. In
the case of the uncertain layer-line: the helix arising from the front is in
phase with that arising from the back and therefore the order of the layer-
line is even. According to the three alternate indexing schemes in Fig. 2,
this layer-line is of order 3, 4 or 5 and since it is even, it must be of order 4.
FFT calculated with 2D FFT/iFFT Adobe plugin (http://www.pages.
drexel.edu/avc25/archive.htm#FFT) by Alex Chirokov.
Fig. 5. Convergence of the IHRSR algorithm (Egelman, 2000). Indexing
predicted 11 subunits per 2 rotations (11:2), the pitch was measured from
the power spectrum to be 7 nm and therefore with this indexing scheme,
rotation per subunit (Du) was predicted to be 65 and axial rise (Dz) was
predicted to be 13 A
˚
. The final 25 iterations represent a convergence on a
stable solution of Du = 65.921 ± 0.004 (95% confidence interval) and
Dz = 13.019 ± 0.002 A
˚
(95% confidence interval). Statistics were calcu-
lated using Statistica 7.0 Statsoft Inc.
J.D. Woodward et al. / Journal of Structural Biology 161 (2008) 111–119 115
reconstruction is visible as a hollow cylinder (Fig. 9). When
superimposed, the regions of highest protein density of the
cryo-reconstruction correlate well with the highest density
regions of the metal-shadowed reconstruction (Fig. 9).
The reconstructions differ in some important respects how-
ever; the diameter of the central channel is smaller in the
case of the cryo-reconstruction and the individual subunits
of the metal-shadowed reconstruction are less distinct than
the equivalent features of the cryo-reconstruction (Figs. 8
and 9).
4. Discussion
4.1. Reconstruction strategies
Surface relief reconstruction (Guckenberger, 1985) pro-
vides a 2.5D digital elevation map of the structure of metal-
shadowed surfaces. This method has been successfully
applied (e.g. Rockel et al., 2000; Walz et al., 1996; Dimm-
eler et al., 2001) to specimens of appropriate topo logy.
However, the majority of biological structure is incompat-
ible with the digital elevation map (Lupetti et al., 2005).
The alternate method is to treat metal-shadowed surfaces
as two-dimensional projections of the metal cast. Lanza-
vecchia et al. (1998) used the method of random conical tilt
to reconstruct projections of rotationally shadowed metal
replicas. Lanzavecchia et al. (2005) applied the method of
conical-tilt tomography to rotationally shadowed speci-
mens of integral membr ane proteins. Thi s method allows
proteins to be studied, at high resolution, in the cellular
milieu. Lupetti et al. (2005) applied a similar method to
study dynein arms in situ.
The power spectra of metal-shadowed helices resemble
those obtained from negative-stain- and cryo-reconstruc-
tions and Fourier–Bessel helic al reconstruction methods
have been applied to rotationally shadowed actin filaments
(Morris et al., 1994). However, the same approach could
not be taken in this case because the contrast arising from
each fibre provided only parti al structural information, as a
function of the shadowing direction. In a shadowing exper-
iment, the fibres are randomly oriented on the grid and the
relative shadowing direction is different in each case. This
Fig. 6. (a) The reconstruction has been contoured to correspond to a subunit molecular mass of 45 kDa. The filament has a diameter of 11.5 nm and a pitch of
7.15 nm. The reconstruction has sufficient structural detail to allow us to propose positions for the various interacting surfaces which stabilize the filament.
These have been marked on the figure. The A- and C-surfaces closely resemble those previously seen in the cyanide dihydratases (e.g. Sewell et al., 2003).
However, the surface labelled F is an interaction not previously seen in nitrilase fibres. (b) The filament has been tilted by 15 as shown. The first dyad axis
passes through the hole (n) and the A-surface (n + 2) (green arrow). The second passes through the F-surface (n) and C-surface (n + 2) (blue arrow).
Fig. 7. Resolution estimation; the FSC calculated between reconstructions from the shadowed dataset (dashed line), at the 0.5 cut-off level is 33 A
˚
. Cryo-
dataset (dotted line): 27 A
˚
and between independent reconstructions from the metal shadowed and cryo-datasets (solid line): 38 A
˚
.
116 J.D. Woodward et al. / Journal of Structural Biology 161 (2008) 111–119
means that different fibre segments do not strictly represent
equivalent views. Contrast arises exclusively from the metal
coating (Walz et al., 1996), which is on the front side of the
filament: this allows unambiguous determination of the
handedness of the helix, but provides only half the ne ces-
sary information required to reconstruct the filament.
4.2. Iterative helical real-space reconstruction
The IHRSR algorithm, in which views of the helix that
have been shadowed from different directions are averaged
and helical symmetry is imposed during reconstruction,
overcomes these limitations. Instead of physically rotating
the source; images of the filaments, which have been unidi-
rectionally shadowed from all possible directions, are com-
bined and averaged in the computer. Back-projection of
these symmetry related views produces a reconstruction
of the average metal distribution surrounding the entire fil-
ament—or more accurately, an asymmetric unit within the
filament. This, however, requires that the helical segmen ts
can be alig ned correctly with the reconstruction during
projection matching. As evidenced by the similarity of
the Midilab reconstruction to the cryo-reconstruction, the
method appears to be successful.
It is important to consider what the basis of the appar-
ently successful alignment of filaments from different shad-
owing directions really is. It can be seen from reference-free
alignment of filaments shadowed from different directions
(Fig. 10), that even though shadowing from different direc-
tions accentuates different features, some common infor-
mation is preserved, this information is clearly sufficient
for the alignment of orthogonal views. A possible alterna-
tive explanation is that the protein component of the fila-
ment facilitates alignment. How ever, this seems unlikely
because the calculated power spectrum of the metal-shad-
owed images show that the contrast arises exclusively from
the metal covered parts, reflecting the typically asymmetric
density distribution expected from unidirectional shadow-
ing i.e. only helical reflections arising from one side of
the filament are visible.
In an ideal shadowing experiment, the distribution of
metal clusters differs as a function of shadowing direction.
Fig. 9. Successive transverse slices along the reconstructions; each slice represents 5.3 A
˚
. The metal-shadowed reconstruction is contoured in grey.
Corresponding slices through the cryo-reconstruction are superimposed and shown in blue. The protein component of the cryo-reconstruction and metal
component of the Midilab reconstruction correlate well.
Fig. 8. (a) 2D classified average of the metal-shadowed helical segments, the left-handed one-start helix is indicated by black lines, white lines indicate the
right-handed four-start helix. (b) The metal-shadowed- and (c) cryo-reconstructions, thresholded and oriented to approximate (a). Corresponding features
can be identified between all three images.
J.D. Woodward et al. / Journal of Structural Biology 161 (2008) 111–119 117
In reality, however, metal may accumulate at preferential
nucleation sites on the surface of the protein; this effect is
referred to as ‘‘decoration’’ (Weinkauf et al., 1991).
Because the position of nucleation sites is independent of
shadowing direction, the contrast arising from decoration
would presumably aid in alignment. However, with the
amount of metal used here, the pure decoration component
is likely to be very small. The calculated power spectra
(Fig. 10) from individual fibres show a layer-line amplitude
dependence on shadowing direction, meaning that the data
are predominately unidirectional and precluding Fourier–
Bessel type reconstruction strategies.
4.3. Interpretation of the Midilab reconstruction
This is the first attempt to reconstruct unidirectionally
metal-shadowed data in three dimensions. Perhaps the
most surprising feature of the resulting reconstruction is
the unexpected sim ilarity between the Midilab- and cryo-
volumes. Rather than surrounding the filament with a thin
film, the metal appears to be localized within the protein.
There are a number of possible explanations for this obser-
vation. One is that it may be an artefact of the process; the
thin layer of metal deposited on the surface of the filament
is convoluted with a Gaussian function arising from resolu-
tion limitations and alignment errors, leading to an appar-
ent thickening of the metal layer. If this were the case it
would be expected that the highest density regions of the
Midilab reconstruction would correspond to the outside
edge of the cryo-reconstruction. This is obviously not the
case (Fig. 9). Furthermore, the diameter of the Midilab
reconstruction corresponds closely to the diameter of the
classified average (Fig. 8).
An alternative explanation is that metal atoms partially
penetrate the surface of the protein matrix during the shad-
owing process and that contrast arising from these atoms
allows visualisation of the protein volume. This hypothesis
is consistent with the observation that contrast arises exclu-
sively from the front of the filament and yet in cross-section
the metal density of the Midilab reconstructio n appears to
encroach on the protein density of the cryo-reconstruction.
This mechanism woul d maintain the relationshi p between
the sample surface normal and the corresponding pixel
density used to calculate the digital elevation map. A fur-
ther alternative explanation is that the internal density is
an artefact of the back projection algorithm. In any case,
the surface structure is preserved .
4.4. Biological Interpretation
The interaction occurring across the groove, termed the
D-surface in the cyanide dihydrat ases (Sewell et al., 2003;
Jandhyala et al., 2003)andRhodococcus rhodochrous J1
nitrilase (Thuku et al., 2007) is found in a different location
in the G. sorghi cyanide hydratase metal shadowed- and
cryo-reconstructions. The position of this interaction,
termed the F-surface here, has implications for the stabil-
ization of the filament; an important consideration in
industrial applications. In purely geometric terms; this
change in interaction position is necessary to accommodate
the decreased twist of the G. sorghi filament which has 5.6
dimers per turn compared to 4.9 dimers per turn in the
nitrilase from R. rhodochrous J1. Both of these interactions
occur between dimers n and n + 5. We have shown here
that the G. sorghi cyanide hydratase is unambiguously
left-handed.
5. Conclusions
The helical structure of G. sorghi CHT has been deter-
mined at a resolution of 33 A
˚
by applying the IHRSR
algorithm to freeze-dried unidirectionally shadowed fila-
ments. These filaments have been unambiguously deter-
mined to be left-handed. The reconstruction is plausible,
but so are those which converge on the incorrect helical
symmetry. Therefore considerable care needs to be taken
Fig. 10. (a) Reference-free aligned average of a single metal-shadowed filament, the right-handed four-start helix is clearly visible from this shadowing
direction. (b) Calculated power spectrum from the filament used to create (a), the amplitude of the layer-line corresponding to the right-handed four-start
helix is greater than that corresponding to the left-handed one-start helix. (c) Calculated power spectrum from a single filament, the layer-line
corresponding to the left-handed one start has a greater amplitude than the right-handed four-start layer line. (d) 2D reference-free aligned average of the
filament from (c). The shadowing direction has enhanced the left-handed one-start helix.
118 J.D. Woodward et al. / Journal of Structural Biology 161 (2008) 111–119
when indexing the power spectrum. Surprisingly, the result-
ing reconstruction closely resembles that obtained using
conventional cryo-EM methods. The final reconstruction
suggests interactions across the groove of the one start
helix in the case of the CHT differ from those previously
seen in negative stain reconstructions of the cyanide dihy-
dratases from P. stutzeri and B. pumilus and the nitrilase
from R. rhodococcus J1.
Acknowledgments
We thank Heinz Gross and Peter Tittmann for gener-
ously giving us access to the Midilab instrument; Edward
H. Egelman for his considerable assistance with IHRSR;
the National Research Foundation and the Carnegie Cor-
poration of New York for their financial support.
References
Banerjee, A., Sharma, R., Banerjee, U.C., 2002. The nitrile-degrading
enzymes: current status and future prospects. Appl. Microbiol.
Biotechnol. 60, 33–44.
Baxter, J., Cummings, S.P., 2006. The current and future applications of
microorganisms in the bioremediation of cyanide contamination.
Antonie van Leeuwenhoek 90 (1), 1–17.
Brady, D., Beeton, A., Zeevaart, J., Kgaje, C., van Rantwijk, F., Sheldon,
R.A., 2004. Characterisation of nitrilase and nitrile hydratase biocata-
lytic systems. Appl. Microbiol. Biotechnol. 64, 76–85.
Dimmeler, E., Marabini, R., Tittman, P., Gross, H., 2001. Correlation of
topographic surface and volume data from three-dimensional electron
microscopy. J. Struct. Biol. 136, 20–29.
Egelman, E.H., 2000. A robust algorithm for the reconstruction of
helical filaments using single-particle methods. Ultramicroscopy 85,
225–234.
Frank, J., Radermacher, M., Penczek, P., Zhu, J., Li, Y., Ladjadj, M.,
Leith, A., 1996. SPIDER and WEB: processing and visualization of
images in 3D electron microscopy and related fields. J. Struct. Biol.
116, 190–199.
Fry, W.E., Millar, R.L., 1972. Cyanide degredation by an enzyme from
Stemphylium loti. Arch. Biochem. Biophys. 151, 468–474.
Gross, H., 1987. High resolution metal replication of freeze-dried
specimens. In: Steinbrecht, R.A., Zierold, K. (Eds.), Cryotechniques
in Biological Electron Microscopy. Springer-Verlag, Berlin, Heidel-
berg, pp. 205–215.
Gross, H., Krusche, K., Tittmann, P., 1990. Recent progress in high
resolution shadowing for biological samples. In: Proceedings of the
XIIth International Congress for Electron Microscopy. San Francisco
Press, Inc., San Francisco, CA, pp. 510–511.
Guckenberger, R., 1985. Surface reliefs derived from heavy-metal shad-
owed specimens: Fourier space technique applied to periodic objects.
Ultramicroscopy 16, 357–370.
Hoenger, A., Doerhoefer, M., Woehlke, G., Tittmann, P., Gross, H.,
Song, Y.H., Mandelkow, E., 2000. Surface topography of microtubule
walls decorated with monomeric and dimeric kinesin constructs. Biol.
Chem. 381, 1001–1011.
Jandhyala, D., Berman, M., Meyers, P.R., Sewell, B.T., Willson, R.C.,
Benedik, M.J., 2003. CynD, the cyanide dihydratase from Bacillus
pumilus: gene cloning and structural studies. Appl. Environ. Microbiol.
69, 4794–4805.
Jandhyala, D.M., Willson, R.C., Sewell, B.T., Benedik, M.J., 2005.
Comparison of cyanide-degrading nitrilases. Appl. Microbiol. Bio-
technol. 68, 327–335.
Lanzavecchia, S., Bellon, P.L., Lupetti, P., Dallai, R., Rappuoli, R.,
Telford, J.L., 1998. Three-dimensional reconstruction of metal replicas
of the Helicobacter pylori vacuolating cytotoxin. J. Struct. Biol. 121,
9–18.
Lanzavecchia, S., Cantele, F., Bellon, P.L., Zampighi, L., Kreman, M.,
Wright, E., Zampighi, G.A., 2005. Conical tomography of freeze-
fracture replicas: a method for the study of integral membrane proteins
inserted in phospholipid bilayers. J. Struct. Biol. 149, 87–98.
Lupetti, P., Lanzavecchia, S., Mercati, D., Cantele, F., Dallai, R.,
Mencarelli, C., 2005. Three-dimensional reconstruction of axonemal
outer dynein arms in situ by electron tomography. Cell Motil.
Cytoskeleton 62, 69–83.
Ludtke, S.J., Baldwin, P.R., Chiu, W., 1999. EMAN: semiautomated
software for high-resolution single-particle reconstructions. J. Struct.
Biol. 128, 82–97.
Morris, E.P., Katayama, E., Squire, J.M., 1994. Evaluation of high-
resolution shadowing applied to freeze-fractured, deep-etched parti-
cles: 3D helical reconstruction of shadowed actin filaments. J. Struct.
Biol. 113, 47–55.
Nolan, L.M., Harnedy, P.A., Turner, P., Hearne, A.B., O’Reilly, C., 2003.
The cyanide hydratase enzyme of Fusarium lateritium also has
nitrilase activity. FEMS Microbiol. Lett. 221, 161–165.
O’Reilly, C., Turner, P.D., 2003. The nitrilase family of CN hydrolysing
enzymes—a comparative study. J. Appl. Microbiol. 95, 1161–1174.
Pace, H.C., Brenner, C., 2001. The nitrilase superfamily: classification,
structure and function. Genome Biol. 2 (1), REVIEWS0001.
Penczek, P., Radermacher, M., Frank, J., 1992. Three-dimensional
reconstruction of single particles embedded in ice. Ultramicroscopy
40, 33–53.
Pettersen, E.F., Goddard, T.D., Huang, C.C., Couch, G.S., Greenblatt,
D.M., Meng, E.C., Ferrin, T.E., 2004. UCSF Chimera—a visualiza-
tion system for exploratory research and analysis. J. Comput. Chem.
25, 1605–1612.
Rockel, B., Guckenberger, R., Gross, H., Tittmann, P., Baumeister, W.,
2000. Rotary and unidirectional metal shadowing of VAT: localization
of the substrate-binding domain. J. Struct. Biol. 132, 162–168.
Sewell, B.T., Thuku, R.N., Zhang, X., Benedik, M.J., 2005. Oligomeric
structure of nitrilases: effect of mutating interfacial residues on activity.
Ann. N. Y. Acad. Sci. 1056, 153–159.
Sewell, B.T., Berman, M.N., Meyers, P.R., Jandhyala, D., Benedik, M.J.,
2003. The cyanide degrading nitrilase from Pseudomonas stutzeri
AK61 is a twofold symmetric, 14-subunit spiral. Structure 11, 1413–
1422.
Smith, P.R., Kistler, J., 1977. Surface reliefs computed from electron
micrographs of heavy metal shadowed specimens. J. Ultrastruct. Res.
61, 124–133.
Smith, P.R., Ivanov, I.E., 1980. Surface reliefs computed from micro-
graphs of isolated heavy metal shadowed particles. J. Ultrastruct. Res.
71 (1), 25–36.
Thuku, R.N., Weber, B.W., Varsani, A., Sewell, B.T., 2007. Post-
translational cleavage of recombinantly expressed nitrilase from
Rhodococcus rhodochrous J1 yields a stable, active helical form. FEBS
J. 274 (8), 2099–2108.
Walz, T., Tittmann, P., Fuchs, K.H., Mu
¨
ller, D.J., Smith, B.L., Agre, P.,
Gross, H., Engel, A., 1996. Surface topographies at subnanometer-
resolution reveal asymmetry and sidedness of aquaporin-1. Mol. Biol.
264, 907–918.
Wang, P., Matthews, D.E., VanEtten, H.D., 1992. Purification and
characterization of cyanide hydratase from the phytopathogenic
fungus Gloeocercospora sorghi. Arch. Biochem. Biophys. 298, 569–575.
Wang, P., Sandrock, R.W., VanEtten, H.D., 1999. Disruption of the
cyanide hydratase gene in Gloeocercospora sorghi increases its sensi-
tivity to the phytoanticipin cyanide but does not affect its pathogenicity
on the cyanogenic plant sorghum. Fungal. Genet. Biol. 28, 126–134.
Weinkauf, S., Bacher, A., Baumeister, W., Ladenstein, R., Huber, R.,
Bachmann, L., 1991. Correlation of metal decoration and topochem-
istry on protein surfaces. J. Mol. Biol. 221, 637–645.
Yang, S., Yu, X., Galkin, V.E., Egelman, E.H., 2003. Issues of resolution
and polymorphism in single-particle reconstruction. J. Struct. Biol. 144
(1), 162–171.
J.D. Woodward et al. / Journal of Structural Biology 161 (2008) 111–119 119