Investigation of the enzymes required for the biosynthesis of an unusual formylated sugar in the emerging human pathogen Helicobacter canadensis

Article abstract of DOI:10.1002/pro.4169

It is now well established that the Gram-negative bacterium, Helicobacter pylori, causes gastritis in humans. In recent years, it has become apparent that the so-called non-pylori Helicobacters, normally infecting pigs, cats, and dogs, may also be involved in human pathology via zoonotic transmission. Indeed, more than 30 species of non-pylori Helicobacters have been identified thus far. One such organism is Helicobacter canadensis, an emerging pathogen whose genome sequence was published in 2009. Given our long-standing interest in the biosynthesis of N-formylated sugars found in the O-antigens of some Gram-negative bacteria, we were curious as to whether H. canadensis produces such unusual carbohydrates. Here, we demonstrate using both biochemical and structural techniques that the proteins encoded by the HCAN_0198, HCAN_0204, and HCAN_0200 genes in H. canadensis, correspond to a 3,4-ketoisomerase, a pyridoxal 5′-phosphate aminotransferase, and an N-formyltransferase, respectively. For this investigation, five high-resolution X-ray structures were determined and the kinetic parameters for the isomerase and the N-formyltransferase were measured. Based on these data, we suggest that the unusual sugar, 3-formamido-3,6-dideoxy-d-glucose, will most likely be found in the O-antigen of H. canadensis. Whether N-formylated sugars found in the O-antigen contribute to virulence is presently unclear, but it is intriguing that they have been observed in such pathogens as Francisella tularensis, Mycobacterium tuberculosis, and Brucella melitensis.

Full text of DOI:10.1002/pro.4169

Received: 9 July 2021
DOI: 10.1002/pro.4169
Revised: 2 August 2021
Accepted: 4 August 2021
F U L L - L E N G T H P A P E R
Investigation of the enzymes required for the biosynthesis
of an unusual formylated sugar in the emerging human
pathogen Helicobacter canadensis
Colton J. Heisdorf | William A. Griffiths | James B. Thoden | Hazel M. Holden
Department of Biochemistry, University of
Abstract
Wisconsin, Madison, Wisconsin, USA
It is now well established that the Gram-negative bacterium, Helicobacter
Correspondence
Hazel M. Holden, Department of
pylori, causes gastritis in humans. In recent years, it has become apparent that
the so-called non-pylori Helicobacters, normally infecting pigs, cats, and dogs,
Biochemistry, University of Wisconsin,
Madison, WI 53706, USA.
may also be involved in human pathology via zoonotic transmission. Indeed,
Email: hazel_holden@biochem.wisc.edu
more than 30 species of non-pylori Helicobacters have been identified thus far.
One such organism is Helicobacter canadensis, an emerging pathogen whose
genome sequence was published in 2009. Given our long-standing interest in
the biosynthesis of N-formylated sugars found in the O-antigens of some
Gram-negative bacteria, we were curious as to whether H. canadensis produces
such unusual carbohydrates. Here, we demonstrate using both biochemical
and structural techniques that the proteins encoded by the HCAN_0198,
HCAN_0204, and HCAN_0200 genes in H. canadensis, correspond to a
Funding information
National Institutes of Health, Grant/
Award Number: R35 GM134643
3,4-ketoisomerase,
a
pyridoxal 50-phosphate aminotransferase, and an
N-formyltransferase, respectively. For this investigation, five high-resolution
X-ray structures were determined and the kinetic parameters for the isomerase
and the N-formyltransferase were measured. Based on these data, we suggest
that the unusual sugar, 3-formamido-3,6-dideoxy-D-glucose, will most likely be
found in the O-antigen of H. canadensis. Whether N-formylated sugars found
in the O-antigen contribute to virulence is presently unclear, but it is intrigu-
ing that they have been observed in such pathogens as Francisella tularensis,
Mycobacterium tuberculosis, and Brucella melitensis.
K E Y W O R D S
aminotransferase, carbohydrate, enzyme kinetics, folate, ketoisomerase,
N-formyltransferase, pyridoxal 50-phosphate, structure–function, tertiary structure,
X-ray crystallography
Abbreviations: dTDP, thymidine diphosphate; HEPES, N-
2-hydroxyethylpiperazine-N0-2-ethanesulfonic acid; HEPPS, N-
2-hydroxyethylpiperazine-N0-3-propanesulfonic acid; Homopipes,
homopiperazine-1,4-bis(2-ethanesulfonic acid); HPLC, high-
performance liquid chromatography; MES, 2-(N-morpholino)
ethanesulfonic acid; Ni-NTA, nickel-nitrilotriacetic acid; PDB, Protein
Data Bank; rTEV, recombinant Tobacco Etch Virus; Tris, tris-
(hydroxymethyl)aminomethane.
1 | INTRODUCTION
In the early 1980s, the field of gastroduodenal pathology
was forever changed by the isolation of Helicobacter
pylori, a Gram-negative bacterium which lives in the
mucus layer surrounding the gastric epithelial cells of
2144
© 2021 The Protein Society.
wileyonlinelibrary.com/journal/pro
Protein Science. 2021;30:2144–2160.

HEISDORF ET AL.
2145
humans and other primates.1,2 The prevailing wisdom at
the time was that the acidic nature of the stomach could
not support bacterial growth and that peptic ulcers
resulted from stress. The discovery was of such medical
relevance that the scientists who isolated H. pylori, Drs.
Barry Marshall and Robin Warren, were eventually
awarded the Nobel Prize in Physiology and Medicine in
2005 for deciphering the role of this organism in gastritis,
peptic ulcer disease, and gastric adenocarcinoma.3 Strik-
ingly, it is thought that ~50% of the world's human popu-
lation is infected by H. pylori.4
Following the identification of H. pylori in 1984, addi-
tional Helicobacter strains were identified and referred to
as non-pylori Helicobacter species or NPHS. These bacte-
ria inhabit humans as well as other mammals and birds.
At least 36 species have been identified thus far.5
Throughout the last 20 years, clinical evidence has accu-
mulated to suggest that some NPHS bacteria cause
human disease via zoonotic infections.
patients with diarrhea.6–8 Both barnacle and Canada geese
are thought to serve as the natural reservoirs for the bacte-
rium.9 It has also been isolated from wild rodents, Guinea
fowl, and laboratory rabbits, among others.10–12
The sequence of the H. canadensis genome was first
reported in 2009 and revealed the presence of both genes
encoding proteins required for capsular polysaccharide
export and for an N-linked glycosylation pathway, similar
to those systems found in Campylobacter jejuni.13
Given our interest in the biosynthesis of bacterial
N-formylated sugars in pathogenic Gram-negative bacteria
such as C. jejuni, we were curious as to whether
H. canadensis produces such sugars.14–24 Analysis of the
H. canadensis MIT 98-5491 genome sequence suggested that
HCAN_0197, HCAN_0198, HCAN_0204, and HCAN_0200
encode for 4,6-dehydratase, a 3,4-ketoisomerase, a pyridoxal
50-phosphate or PLP-dependent aminotransferase, and an
N-formyltransferase, respectively. These are the enzymes
required for the biosynthesis of either dTDP-3-formamido-
3,6-dideoxy-D-glucose or dTDP-3-formamido-3,6-dideoxy-D-
galactose as indicated in Figure 1
One such emerging pathogen is Helicobacter can-
adensis, which was first isolated from four Canadian
FIGURE 1 Predicted pathway for the
biosynthesis of dTDP-3-formamido-3,6-dideoxy-
D-glucose (or dTDP-3-formamido-3,6-dideoxy-D-
galactose) in H. canadensis. When this
investigation was initiated, the stereochemistry
about the C-4'carbon following the
ketoisomerase reaction was unknown

2146
HEISDORF ET AL.
Here, we describe a biochemical analysis of the pro-
“cupin” superfamily with some members demonstrating
enzymatic activities, whereas others serving as storage
proteins.25 Until now, those sugar ketoisomerases belonging
to the cupin superfamily were thought to function as
dimers.26–30 Interestingly, the two H. canadensis subunits in
the asymmetric unit did not display the typical dimeric qua-
ternary structure. Upon inspection of the crystalline pack-
ing interactions, it became clear that the H. canadensis
ketoisomerase adopted a tetrameric quaternary struc-
ture with D2 symmetry. To confirm that the enzyme
exists as a tetramer in solution, a gel filtration experi-
ment was subsequently conducted with the results
shown in Figure 2. The ketoisomerase migrates as a tet-
ramer with no evidence of dimer formation. Note that
the α-carbons for these two individual subunits in the
asymmetric unit correspond with a root-mean-square
deviation of 0.4 Å.
Shown in Figure 3a is a ribbon representation of the
ketoisomerase tetramer. Each subunit is composed of
10 β-strands that form a flattened β-barrel and one
α-helix that extends from Asp 122 to the C-terminus. The
first two β-strands, formed by Asp 2 to Asn 12 and Ser 17 to
Ile 19, participate in domain swapping where they extend
into another subunit of the tetramer across the A/C and
teins encoded by HCAN_0198, HCAN_0204, and
HCAN_0200, hereafter referred to as the ketoisomerase,
the aminotransferase, and the N-formyltransferase. The
first protein cloned, expressed, and purified for this
investigation was the N-formyltransferase. The mea-
sured kinetic parameters revealed that the enzyme could
function on either dTDP-3-amino-3,6-dideoxy-D-glucose
or dTDP-3-amino-3,6-dideoxy-D-galactose using N10-
formyltetrahydrofolate (N10-formyl-THF) as the carbon
source (Figure 1). This ability to utilize substrates differing
in stereochemistry about the C-40 carbon position was pre-
viously observed for the C. jejuni N-formyltransferase.14
Given that the O-antigen sequence for H. canadensis
is presently unknown, we subsequently cloned,
expressed, and purified the 3,4-ketoisomerase and the
aminotransferase, which were used to produce in vitro
the physiological substrate for the N-formyltransferase. A
subsequent X-ray crystallographic analysis of the N-
formyltransferase/ligand complex revealed the presence
of dTDP-3-amino-3,6-dideoxy-D-glucose bound in the
active site thereby proving that H. canadensis produces
dTDP-3-formamido-3,6-dideoxy-D-glucose rather than
dTDP-3-formamido-3,6-dideoxy-D-galactose.
We also describe in this report the crystal structures
of the 3,4-ketoisomerase and the aminotransferase.
Taken together, our results suggest that N-formylated
sugars may be more prevalent in the O-antigens of
Helicobacter species than previously thought. Indeed, a
BLAST search suggests that Helicobacter ganmani,
Helicobacter hepaticus, Helicobacter burdigaliensis,
Helicobacter sanguini, Helicobacter bilis, and Helicobacter
pullorum, among others, may also produce N-formylated
sugars. Whereas the physiological role of these unusual
sugars is still not well understood, it is possible that they
play a role in virulence.
ketoisomerase
1950
1450
carbonic anhydrase
950
cytochrome c
2 | RESULTS
bovine serum albumin
450
2.1 | Molecular architecture of the
ketoisomerase
7
9
11
13
15
17
19
21
23
Retention volume (ml)
Although the first structure solved in this investigation
was that of the N-formyltransferase, for the sake of clar-
ity, the models of the two other enzymes will be
described in the order shown in Figure 1. The
ketoisomerase crystallized in the space group P3121 with
two subunits in the asymmetric unit (133 amino acid resi-
dues per monomer). The model was refined to an overall
R-factor of 20.4% at 2.5 Å resolution.
FIGURE 2 Quaternary structure of the ketoisomerase. Gel
filtration chromatography was utilized to probe the quaternary
structure of enzyme. Shown in this figure is a graph of retention
time versus milliabsorbance units. The orange trace corresponds to
the H. canadensis ketoisomerase (calculated subunit MW 15,740).
The molecular weight standards were bovine serum albumin
(MW 66,000), carbonic anhydrase (MW 29,000), and cytochrome c
(MW 12,400). Note that the commercially available bovine serum
albumin contains a contaminant, hence the double peak
The overall subunit architecture of the H. canadensis
ketoisomerase places it into the biologically diverse

HEISDORF ET AL.
2147
FIGURE 3 Quaternary
structure of the ketoisomerase.
A ribbon representation of the
tetramer, which assumes D2
symmetry, is shown in (a). Two
of the dyads are indicated by the
black arrows. The third twofold
is perpendicular to the plane of
the paper. A close-up stereo
view of the interface between
subunits A and C is displayed in
(b) whereas that between
subunits A and B is presented in
(c). This figure and Figures 4,
6, 8, and 9 were prepared with
PyMOL31
B/D interfaces shown in Figure 3a. The total buried surface
areas between the A/C or B/D interfaces are extensive at
~3,700 Å2. The total buried surface areas between the A/B
and C/D interfaces are rather small at ~950 Å2.
A close-up stereo view of the B/D interface is provided in
Figure 3b, and as can be seen it is primarily formed by one
β-strand in subunit B (Ser 17 to Ile 19) and another in sub-
unit D (Arg 33 to Ile 37), which run anti-parallel. The sub-
unit: subunit interface is lined with various hydrophobic
residues. This is in sharp contrast to that observed at the A/B
interface shown in Figure 3c. Here, the interface is domi-
nated by hydrogen bonding interactions and salt bridges.

2148
HEISDORF ET AL.
FIGURE 4 Active site of
the ketoisomerase. A stereo view
of the electron density
corresponding to the bound
dTDP ligand is presented in (a).
The map was calculated with
(Fo–Fc) coefficients and
contoured at 3σ. The ligand was
not included in the X-ray
coordinate file used to calculate
the omit map, and thus there is
no model bias. Shown in stereo
in (b) is a close-up view of the
active site. Ordered water
molecules are represented as
spheres. Potential hydrogen
bonds (within 3.2 Å) are
indicated by the dashed lines.
Side chains highlighted in violet
belong to subunit A whereas
those colored in blue are
contributed by subunit C
The ketoisomerase was crystallized in the presence of
parameters were determined via a continuous assay using
dTDP-4-keto-6-deoxy-D-glucose as the substrate. A plot of
initial velocity versus substrate concentration is provided
in Figure 5, and the kinetic constants are listed in
Table 1. The enzyme demonstrates a catalytic efficiency
5 mM dTDP. Shown in Figure 4a is the observed electron
density corresponding to the bound nucleotide in subunit
A. A close-up view of the protein region surround the
ligand is provided in Figure 4b. The thymine ring is posi-
tioned into the active site via a hydrogen bond with Arg
33 and a parallel stacking interaction with Tyr 114. There
is no side chains located within 3.2 Å of the ribose, which
adopts the C2 endo pucker. The negative charge of the
pyrophosphoryl group is neutralized by the side chains of
Arg 46 and Arg 121 from one subunit and Arg 15 from
the other subunit. There are additional hydrogen bonding
interactions between the pyrophosphoryl oxygens,
five water molecules, and the backbone amide group
of Gly 47. Also shown in Figure 4b are the positions
of His 49, His 51, and Tyr 35, which were originally
thought to be critical for catalysis.26,28 To date, all the
ketoisomerases that have been studied have a conserved
histidine, which is His 49 in the H. canadensis enzyme.
In some, however, His 51 is an asparagine and Tyr 35 is a
phenylalanine.29,30
of 7.40 ( 0.93) ꢀ 103 Mꢁ1
s
ꢁ1. This value is similar to
that observed for Tyl1a from Streptomyces fradiae
(3.8 ꢀ 103 Mꢁ1
s
ꢁ1),32 but lower for WlaRA and WlaRB
from C. jejuni 81116 (2.9 ꢀ 106 and 4.8 ꢀ 105 Mꢁ1 sꢁ1
,
respectively).30
2.2 | Molecular architecture of the
aminotransferase
Crystals of the H. canadensis aminotransferase were
grown in the presence of PLP and dTDP-3-amino-3,
6-dideoxy-D-glucose, which resulted in the formation of
the external aldimine as can be seen from the electron
density presented in Figure 6a. The crystals belonged to
the orthorhombic space group C2221 with one subunit in
the asymmetric unit. The H. canadensis enzyme belongs
to the Fold Type I superfamily of aminotransferases,
To further verify that the protein encoded by
HCAN_0198 is, indeed, a ketoisomerase, its kinetic

HEISDORF ET AL.
2149
members of which are known to function as dimers.33
Thus, in the crystals utilized in this investigation, the
local twofold of the dimer packed along a crystallo-
graphic dyad. The aminotransferase model was refined to
an overall R-factor of 17.2% at 1.4 Å resolution.
phosphoryl group of the cofactor is surrounded by
the backbone amides of Gly 59 and Leu 60, and the side
chain of Ser 180, all of which are contributed by subunit
A. Additionally, the side chains of Tyr 213 and Asn
227 from subunit B participate in hydrogen bonding
interactions with the phosphoryl group. The pyranosyl
Shown in Figure 6b is a ribbon representation of the
dimer. The total buried surface area is extensive at
~4,500 Å2 with the active sites located between the sub-
units. Each subunit contains 11 α-helices and 11 β-strands
that form two β-sheets, one of which is a mixed seven-
stranded and the other a four-stranded anti-parallel. A
close-up stereo view of the active site harboring the exter-
nal aldimine in subunit A is displayed in Figure 6c. The
carboxylate group of Asp 156 lies within 3.2 Å of the pyri-
dine nitrogen. This residue is absolutely conserved
among members of the Fold Type I superfamily where it
is believed to maintain the nitrogen in the protonated
state, thereby enhancing the cofactor's reactivity.34 In
addition to Asp 156, the carboxamide side chain of Gln
159 hydrogen bonds to the ring carbonyl oxygen. The
4
group of the external aldimine, which adopts C1 pucker,
is anchored into the active site by two water molecules
and the side chain of Tyr 309 that assumes a cis-peptide
conformation. The electron density for the external
aldimine suggests that the α-phosphoryl group of the pyr-
ophosphoryl moiety adopts two distinct conformations as
shown in Figure 6c. Numerous water molecules lie
within 3.2 Å of the pyrophosphoryl moiety as well as the
side chains of Tyr 182 and His 308 from subunit A and
Lys 218 from subunit B. The thymine ring of the external
aldimine is positioned into the active site via a water mol-
ecule, the backbone carbonyl group of Gly 29 and the
backbone amide of Tyr 31, contributed by subunit B. In
addition, a parallel stacking interaction occurs between
the thymine ring and the indole side chain of Trp 30 (sub-
unit B). The ribose adopts the C2 endo conformation.
Attempts to measure the kinetic parameters of the
H. canadensis aminotransferase were unsuccessful given
the inherent instability of its substrate, dTDP-3-keto-3,
6-dideoxy-D-glucose. Note that the product of the reac-
tion, dTDP-3-amino-3,6-dideoxy-D-glucose, is stable
which allowed for the formation of the external aldimine
described in this investigation.
10
8
6
4
2
0
2.3 | Molecular architecture of the N-
formyltransferase
0
1
2
3
4
[dTDP-4-keto-6-deoxyglucose
concentration] (millimolar)
The identity of the sugars in the O-antigen of H. canadensis
is presently unknown. It is clear, however, from searching
the genome, that there is a gene encoding for a putative N-
formyltransferase. Given that both 3-formamido-3,6-dideoxy-
D-glucose and 3-formamido-3,6-dideoxy-D-galactose have
been observed in various O-antigens, kinetic analyses were
conducted to determine the physiologically relevant substrate
for the H. canadensis N-formyltransferase.14,35,36 A discontin-
uous HPLC assay was utilized to determine kcat and KM.
Shown in Figure 7a,b are the plots of initial velocities versus
dTDP-3-amino-3,6-dideoxy-D-glucose or dTDP-3-amino-3,
FIGURE 5 Plot of initial velocities versus substrate
concentrations for the ketoisomerase. In presenting the data as we
do, we are adhering to standard conventions in enzymology.
Measuring velocities over a wide range of substrate concentrations
allows us to obtain data that define both kcat and kcat/KM well,
which is not accomplished by measuring replicates at fewer
different concentrations. The graph shown allows for a qualitative
appreciation of the quality of the data; the quantitative goodness-of-
fit to the Michaelis–Menten equation is given by the standard
errors as described in Section 4
TABLE 1 Kinetic data for the N-formyltransferase and the 3,4-ketoisomerase
Enzyme
Substrate
KM (mM)
kcat (sꢁ1
)
kcat/KM (Mꢁ1 sꢁ1
7.40 ( 0.93) ꢀ 103
1.20 ( 0.15) ꢀ 104
5.90 ( 0.74) ꢀ 103
)
3,4-ketoisomerase
N-formyltransferase
N-formyltransferase
dTDP-4-keto-6-deoxy-D-glucose
dTDP-3-amino-3,6-dideoxy-D-glucose
dTDP-3-amino-3,6-dideoxy-D-galactose
0.035 0.004
0.075 0.006
0.051 0.007
0.26 0.04
0.90 0.11
0.30 0.04

2150
HEISDORF ET AL.
FIG URE 6 The structure of
the amidotransferase. A stereo
view of the observed electron
density corresponding to the
bound external aldimine is
shown in (a). The map was
calculated as described in the
legend to Figure 4 and
contoured at 3σ. A ribbon
representation of the dimer is
depicted in (b). Shown in stereo
in (c) is a close-up view of the
active site. Ordered water
molecules are represented as
spheres. Potential hydrogen
bonds (within 3.2 Å) are
indicated by the dashed lines.
Side chains highlighted in wheat
belong to subunit A whereas
those colored in light blue are
contributed by subunit B
6-dideoxy-D-galactose concentrations, respectively. The
kinetic constants are listed in Table 1, and as can be seen,
the catalytic efficiency of the H. canadensis N-
formyltransferase for dTDP-3-amino-3,6-dideoxy-D-glucose
versus dTDP-3-amino-3,6-dideoxy-D-galactose is only ~2ꢀ
greater, which does not unambiguously address which sugar
is synthesized by the organism.
presented in Figure 1. Starting from dTDP-D-glucose, dTDP-
4-keto-6-deoxy-D-glucose was prepared using DesIV, a
4,6-dehydratase from Streptomyces venezuelae that has been
previously studied in the laboratory.37 This compound then
served as the substrate for the H. canadensis isomerase,
followed by reaction with the H. canadensis aminotransfer-
ase, and finally with the N-formyltransferase.
Thus, to clarify this issue, a dTDP-amino-containing
sugar was synthesized in vitro according to the pathway
The crystal structure of the H. canadensis
N-formyltransferase in complex with this ligand and

HEISDORF ET AL.
2151
side by a short α-helix (Leu 48 to Asn 54) and on the
other by three α-helices, one of which is quite long (Ala
150 to Asn 174). There is an additional α-helix, formed by
Thr 108 to Leu 114, which runs anti-parallel with the sin-
gle α-helix in the C-terminal region defined by Ser 207 to
Ala 216. The C-terminal region also contains a four-
stranded anti-parallel β-sheet. Pro 101, located in a tight
turn connecting β-strand 5 to α-helix 4, adopts a cis-con-
formation. This cis-peptide positions the side chain of
Leu 100 near that of His 96, the presumed catalytic base.
A close-up view of the active site is displayed in
Figure 8c. The thymine ring of the ligand is positioned
into the active site by the side chain of Gln 223, two
water molecules, and both a T-shaped and a parallel aro-
matic interaction, with the side chains of Phe 107 and
Tyr 222, respectively. The ribose adopts the C3 endo
pucker, and its C-3 hydroxyl group lies within hydrogen
bonding distance of a water molecule that serves as a
bridge to an α-phosphoryl oxygen. Both the side chains of
Lys 9 and Arg 192 provide electrostatic interactions with
the pyrophosphoryl group of the dTDP-sugar. Four water
molecules lie within 3.2 Å of the phosphoryl oxygens.
(a)
50
40
30
20
10
0
0.0
0.5
1.0
1.5
2.0
[dTDP-3-amino-3,6-dideoxy-D-glucose
concentration] (millimolar)
(b) 15
10
5
4
The pyranosyl group, which assumes the C1 pucker, is
anchored into the active site by four water molecules and
the backbone carbonyl oxygens of Phe 78 and Gly 105 that
hydrogen bond to C-40 and the C-20 hydroxyl groups,
respectively. Importantly, no side chains lie within hydro-
gen bonding distance of the C-40 hydroxyl group, which
partially explains why the enzyme can function on both
the glucose- and galactose-based forms of the substrate.
From the residual Fo–Fc electron density, the geome-
try, and the bond lengths, it appears that a potassium ion
binds in the active site where it is coordinated by the side
chains of His 96, Asp 132, the carbonyl group of N5-for-
myl-THF, a water molecule, and the C-20 hydroxyl and
C-30 amino group of the pyranosyl moiety. Asn 94, His
96, and Asp 132, as shown in Figure 8b, are strictly con-
served amongst the sugar N-formyltransferases. His 96 is
thought to serve as an active site base for abstraction of a
proton from the sugar amino group as it attacks the car-
bonyl carbon of N10-formyl-THF.14
0
0.0
0.5
1.0
1.5
[dTDP-3-amino-3,6-dideoxy-D-galactose
concentration] (millimolar)
FIGURE 7 Plot of initial velocities versus substrate
concentrations for the N-formyltransferase. Shown in (a) and
(b) are the initial velocity experiments using either dTDP-3-amino-
3,6-dideoxy-D-glucose or dTDP-3-amino-3,6-dideoxy-D-galactose,
respectively
N5-formyl-tetrahydrofolate (N5-formyl-THF) clearly indi-
cates that dTDP-3-amino-3,6-dideoxy-D-glucose was syn-
thesized as can be seen in the electron density presented
in Figure 8a. The crystals employed for this analysis
belonged to the space group C2 with a dimer in the asym-
metric unit. The model was refined to an overall R-factor
of 18.4% at 2.0 Å resolution. In addition to this X-ray
model, the structures of the protein complexed with
dTDP-3-amino-3,6-dideoxy-D-galactose and N5-formyl-
THF and that of the apoenzyme were also determined.
A ribbon representation of the N-formyltransferase
dimer is displayed in Figure 8b. Each subunit adopts a
bilobal-type structure with the N-terminal and
C-terminal domains defined by Met 1 to Ile 197 and Asn
198 to Ala 270, respectively. Most of the side chains
involved in ligand binding are contributed by the
N-terminal domain that is dominated by a seven-
stranded mixed β-sheet. This β-sheet is flanked on one
The structure of the enzyme complexed with dTDP-
3-amino-3,6-dideoxy-D-galactose and N5-formyl-THF was
solved to 2.0 Å resolution, and the model was refined to
an overall R-factor of 19.5%. Shown in Figure 9a is the
electron density corresponding to the dTDP-sugar. The
α-carbons for the two models of the enzyme, with either
bound dTDP-3-amino-3,6-dideoxy-D-glucose or dTDP-
3-amino-3,6-dideoxy-D-galactose, correspond with
a
root-mean-square deviation of 0.3 Å. Indeed, there are
minimal changes when the galactose version of the ligand
occupies the active site as can be seen in Figure 9b. The
only significant difference is that the C-40 hydroxyl group

2152
HEISDORF ET AL.
FIGURE 8 The structure of
the N-formyltransferase/dTDP-
3-amino-3,6-dideoxy-D-glucose
complex. A stereo view of the
observed electron density
corresponding to the bound
dTDP-3-amino-3,6-dideoxy-D-
glucose is provided in (a). The
map was calculated as described
in the legend to Figure 4 and
contoured at 3σ. A ribbon
representation of the dimer,
viewed down the local twofold
rotational axis, is presented in
(b). Shown in stereo in (c) is a
close-up view of the active site.
Ordered water molecules are
represented as spheres. Potential
hydrogen bonds (within 3.2 Å)
are indicated by the dashed lines
of dTDP-3-amino-3,6-dideoxy-D-galactose no longer resides
within hydrogen bonding distance to the carbonyl group
of Phe 78. Note that the potassium ion observed in the first
model was not found in the structure with bound dTDP-
3-amino-3,6-dideoxy-D-galactose. The reason for this is
unclear but could be a function of slightly different crystal-
lization conditions or dTDP-ligand preparations.
As noted above, the apoenzyme structure was also
determined in this investigation. The model was refined
to an overall R-factor of 16.9% at 1.8 Å resolution. The
α-carbons for ligand bound structures superimpose onto
the apoenzyme model with a root-mean-square deviation
of 0.4 Å. Other than some minor changes in solvent posi-
tions, the three models are virtually identical. Clearly, the

HEISDORF ET AL.
2153
FIGURE 9 The structure of
the N-formyltransferase/dTDP-
3-amino-3,6-dideoxy-D-galactose
complex. The electron density
corresponding to the bound
ligand, contoured at 3σ, is
presented in stereo in (a). The
map was calculated as described
in the legend to Figure 4. Shown
in stereo in (b) is a superposition
of the N-formyltransferase/
dTDP-3-amino-3,6-dideoxy-D-
glucose and the N-
formyltransferase/dTDP-
3-amino-3,6-dideoxy-D-galactose
complexes highlighted in light
blue and violet, respectively,
near the active site vicinity
conformation of the N-formyltransferase is ideally poised
to accept the cofactor and the dTDP-sugar substrate into
the active site region with no required conformational
changes.
virulence factors.44 As another example, the O-antigen
gene cluster has been shown to have a major role in the
virulence of the Escherichia coli meningitis clone O45:K1:
H7.45 And for those who enjoy fishing, Aliivibrio sal-
monicida, the causative agent of hemorrhagic septicemia
of Atlantic salmon, rainbow trout, and Atlantic cod,
requires its O-antigen for virulence.46 Clearly, there is
mounting evidence in the literature that the O-antigen
is important for virulence, but the manner in which it
enhances pathogenicity is presently unclear. Factors such
as serum resistance, immunomodulation, and adhesion
may play critical roles.
On the basis of the data presented herein, we have
now verified that H. canadensis has the enzymes required
for the biosynthesis of dTDP-3-formamido-3,6-dideoxy-D-
glucose, and we suggest that this sugar will be found in
the O-antigen of its lipopolysaccharide or possibly in the
capsular polysaccharide.
3 | DISCUSSION
Bacterial O-antigens are thought to be important for bac-
terial survival and niche adaptation.38 It has been demon-
strated, for example, that Salmonella serogroups
containing deoxygenated sugars are more pathogenic
than those containing only common sugars.39,40 In addi-
tion, studies have shown that laboratory animals chal-
lenged with one serogroup of Salmonella displaying
unusual sugars do not gain immunity to other serogroups
with different deoxysugars.41,42 As another example, inac-
tivation of the first gene (wbtA) in the O-antigen cluster
in Francisella tularensis, the causative agent of rabbit
fever, results in severe attenuation of virulence.43 It has
also been demonstrated that the O-antigen is essential for
full virulence of Yersinia enterocolitica O:8 and, strik-
ingly, its absence affects the expression of other Yersinia
With respect to the ketoisomerase, a search utilizing
the PDBeFold server demonstrates that WlaRB from
C. jejuni is a close relative of the H. canadensis isomer-
ase.47 These two enzymes demonstrate a sequence iden-
tity and a similarity of 78 and 88%, respectively, and both

2154
HEISDORF ET AL.
3000
2500
2000
1500
1000
500
tetramers superimpose with a root-mean-square devia-
tion of ~1 Å. A subsequent gel filtration experiment on
WlaRB confirmed its tetrameric quaternary structure
with no evidence of dimer formation.
(a)
FdtA
In our past investigations, we assumed that all sugar
3,4-ketoisomerases functioned as dimers. We thus re-
evaluated the quaternary structures of WlaRA from
C. jejuni, FdtA from Aneurinibacillus thermoaerophilus, and
QdtA from Thermoanaerobacterium thermosaccharolyticum.
The first two enzymes produce dTDP-3-keto-6-deoxy-D-
galactose, whereas the last yields dTDP-3-keto-6-deoxy-D-
glucose. Shown in Figure 10a,b is the result from the gel
filtration experiments. Clearly, FdtA and WlaRA migrate as
dimers. Interestingly, QdtA migrates as an octamer. The
crystals utilized in the structural analysis of QdtA belonged
to the space group P4212 and contained a dimer in the
asymmetric unit. Visual inspection of the crystalline pack-
ing arrangement (PDB accession code 4O9G) shows that,
indeed, QdtA is an octamer with D4 symmetry. The protein
packed in the lattice such that local fourfold rotational axis
was coincident to a crystallographic fourfold.
The bacterial 3,4-ketoisomerases are an especially
intriguing class of enzymes in that they all function on
dTDP-4-keto-6-deoxy-D-glucose but some yield dTDP-
3-keto-6-deoxy-D-glucose whereas other produce dTDP-
3-keto-6-deoxy-D-galactose. The structural features that
determine the stereochemical outcome of the reaction
are extremely subtle and most likely extend away from
the immediate vicinity of the active site. All those
enzymes studied thus far that produce dTDP-3-keto-
6-deoxy-D-galactose are dimers. However, those that pro-
duce dTDP-3-keto-6-deoxy-D-glucose function as higher
order multi-subunit complexes. It may be that the quater-
nary structure of the ketoisomerase plays a more impor-
tant role in stereochemical outcome than originally
envisioned. This will be addressed in a subsequent
investigation.
cytochrome c
carbonic anhydrase
BSA
WlaRA
7
9
11
13
15
17
19
21
23
Retention volume (ml)
(b)
2500
2000
1500
1000
500
ketoisomerase
QdtA
cytochrome c
BSA
WlaRB
carbonic anhydrase
7
9
11
13
15
17
19
21
23
Retention volume (ml)
FIGURE 10 Quaternary structures of additional sugar
3,4-ketoisomerases studied in the laboratory. Shown in (a) is a
graph of retention times versus milliabsorbance units for the
3,4-ketoisomerases that produce dTDP-4-keto-6-deoxy-D-galactose
(calculated subunit MW of 16,040 and 17,290 for FdtA and WlaRA,
respectively). Shown in (b) is a graph of retention times versus
milliabsorbance units for the 3,4-ketoisomerases that produce
dTDP-4-keto-6-deoxy-D-glucose (calculated subunit MW of 16,070
and 15,730 for QdtA and WlaRb, respectively). The molecular
weight standards were bovine serum albumin (MW 66,000),
carbonic anhydrase (MW 29,000), and cytochrome c (MW 12,400).
Note that the commercially available bovine serum albumin
contains a contaminant, hence the double peak
Like the ketoisomerase, the H. canadensis amino-
transferase and N-formyltransferase are also structurally
similar to the corresponding enzymes in C. jejuni,
namely, WlaRG and WlaRD, respectively. Indeed, the
α-carbons between the two aminotransferases superim-
pose with a root-mean-square deviation of 0.7 Å whereas
those for the two N-formyltransferases correspond with a
root-mean-square deviation of 0.5 Å. It is interesting that
both of these Gram-negative pathogens have the biosyn-
thetic machinery necessary for both capsular polysaccha-
ride export and the production of N-linked glycosylated
proteins.
yield dTDP-3-formamido-3,6-dideoxy-D-glucose. Origi-
nally, we suggested that WlaRB functions as a dimer.30
Interestingly, the crystals utilized in the investigation
contained two dimers in the asymmetric unit. Upon fur-
ther consideration of the packing arrangement of these
dimers, it is clear that WlaRB also assumes a tetrameric
quaternary structure. Indeed, the α-carbons for the two
N-formylated sugars were first observed in 1985 on
the lipopolysaccharides of certain strains of Pseudomonas
aeruginosa but until recently were thought not to be
widespread.48 It is now known, however, that pathogens

HEISDORF ET AL.
2155
such as F. tularensis, Mycobacterium tuberculosis, Brucella
melitensis, Providencia alcalifaciens, Salmonella enterica
O60, Pantoea ananatis, Shewanella spp., and Pseudomo-
nas congelans all have genes encoding for N-formyl-
transferases.15,17,19–23
Are N-formylated sugars associated with increased
pathogenicity? The jury is still out, but intriguingly, loss
of activity of an N-formyltransferase in Brucella abortus
results in a bacterium with attenuated pathogenicity. It
will be of great interest to determine how widespread the
genes are that encode for N-formyltransferases in non-
pylori Helicobacter organisms given that there is evidence
to suggest that such bacteria may play a role in the initia-
tion of human inflammatory bowel disease.49
The N-terminally and C-terminally tagged amino-
transferases were dialyzed against 10 mM Tris–HCl
(pH 8.0) and 200 mM NaCl, and then concentrated to
21 mg/ml and 23 mg/ml, respectively, based on an
extinction coefficient of 0.84 (mg/ml)ꢁ1 cmꢁ1. A tag-free
version of the aminotransferase was generated by
removal of the polyhistidine tag via digestion with rTEV
protease. The rTEV protease and remaining tagged pro-
teins were removed by passage over Ni-NTA agarose, and
the tag-free protein was dialyzed against 10 mM Tris–
HCl (pH 8.0) and 200 mM NaCl, followed by concentra-
tion to 20 mg/ml.
The ketoisomerase was purified in the same manner
and concentrated to 18.5 mg/ml (N-terminally tagged),
19 mg/ml (C-terminally tagged), and 16 mg/ml (tag-free)
based on an extinction coefficient of 0.90 (mg/ml)ꢁ1 cmꢁ1
.
4 | MATERIALS AND METHODS
4.1 | Protein expression and purification
The N-formyltransferase was similarly purified, with
the exception that the final dialysis solution contained
10 mM Tris–HCl (pH 8.0) and 500 mM NaCl. Final con-
centrations were 16.3 mg/ml (C-terminally tagged) and
15.0 mg/ml (tag-free) based on an extinction coefficient
of 0.88 (mg/ml)ꢁ1 cmꢁ1. The N-terminally protein precip-
itated upon concentration.
The genes HCAN_0198, HCAN_0204, and HCAN_0200
(encoding for the ketoisomerase, the aminotransferase, and
the N-formyltransferase, respectively) from H. canadensis
MIT 98-5491 were synthesized by Integrated DNA Technol-
ogies. They were placed into either pET31(b+) plasmids
(Novagen) to generate proteins with non-cleavable
C-terminal polyhistidine tags (LEHHHHHH) or pET28t3g
plasmids that lead to proteins with N-terminal polyhistidine
tags possessing rTEV cleavage sites as previously
described.50 The plasmids were utilized to transform
Rosetta2 (DE3) E. coli cells (Novagen) for protein expres-
sion. Cultures were grown in lysogeny broth supplemented
with 100 mg/L ampicillin and 50 mg/L chloramphenicol
for the pET31(b+) constructs or 50 mg/L kanamycin and
50 mg/L chloramphenicol for the pET28t3g constructs. The
cultures were grown at 37ꢂC with shaking until an optical
density of 0.8 was reached at 600 nm. The flasks were
cooled in an ice bath, and the cells were induced with the
addition of 1 mM isopropyl β-D-1-thiogalactopyranoside
and allowed to express protein at 21ꢂC for 24 hr.
The cells were harvested by centrifugation and fro-
zen as pellets in liquid nitrogen. These pellets were sub-
sequently disrupted by sonication on ice in a lysis buffer
composed of 50 mM sodium phosphate, 20 mM imidaz-
ole, 10% glycerol, and 300 mM NaCl (pH 8.0). The
lysates were cleared by centrifugation, and the proteins
were purified at 4ꢂC utilizing Prometheus™ Ni-NTA
agarose (Prometheus Protein Biology Products)
according to the manufacturer's instructions. All buffers
were adjusted to pH 8.0 and contained 50 mM sodium
phosphate, 300 mM NaCl, and imidazole concentrations
of 20 mM for the wash buffer and 300 mM for the elu-
tion buffer.
4.2 | Crystallization
Crystals of the ketoisomerase (N-terminally tagged) were
grown via the hanging drop method of vapor diffusion at
room temperature from 7 to 13% poly(ethylene glycol) 8000,
1 M tetramethylammonium chloride, and 100 mM HEPES
(pH 7.5) in the presence of 5 mM dTDP. They belonged to
the trigonal space group P3121 with unit cell dimensions of
a = b = 92.9 Å and c = 85.6 Å. The asymmetric unit con-
tained two subunits. For X-ray data collection, the crystals
were transferred to a cryo-protectant solution composed of
20% poly(ethylene glycol) 8000, 200 mM NaCl, 1.25 M
tetramethylammonium chloride, 12% ethylene glycol, 5 mM
dTDP, and 100 mM HEPES (pH 7.5).
Crystals of C-terminally tagged aminotransferase
were grown via the hanging drop method of vapor diffu-
sion at room temperature from 15 to 18% poly(ethylene
glycol) 3350, 100 mM MgCl2, and 100 mM MES (pH 6.0)
in the presence of 1 mM PLP and 5 mM dTDP-3-amino-
3,6-dideoxy-D-glucose. They belonged to the orthorhom-
bic space group C2221 with unit cell dimensions of
a = 61.0 Å, b = 150.8 Å, and c = 91.1 Å. The asymmetric
unit contained one subunit. For X-ray data collection, the
crystals were transferred to a cryo-protectant solution
composed of 26% poly(ethylene glycol) 3350, 300 mM
NaCl, 100 mM MgCl2, 18% ethylene glycol, 1 mM PLP,
5 mM dTDP-3-amino-3,6-dideoxy-D-glucose, and 100 mM
MES (pH 6.0).

2156
HEISDORF ET AL.
Crystals
of
the
C-terminally
tagged
N-
4.5 | Synthesis of dTDP-4-keto-6-deoxy-D-
formyltransferase were grown via the hanging drop
method of vapor diffusion at room temperature from
11–14% poly(ethylene glycol) 8000, 2% DMSO, and
100 mM Homopipes (pH 5.0). They belonged to the
monoclinic space group C2 with unit cell dimensions
of a = 129.0 Å, b = 74.7 Å, c = 82.3 Å and β = 113.7ꢂ.
The asymmetric unit contained one dimer. For X-ray
data collection, the crystals were transferred to a cryo-
protectant solution composed of 20% poly(ethylene gly-
col) 8000, 600 mM NaCl, 18% ethylene glycol, and
100 mM Homopipes (pH 5.0).
glucose
A buffered solution of 50 mM dTDP-D-glucose was pre-
pared in 50 mM HEPPS (pH 8.0). To this solution, DesIV,
a dTDP-glucose-4,6-dehydratase from S. venezuelae57 was
added to a final concentration of 20 mg/ml. The reaction
was allowed to proceed overnight at room temperature.
After removal of the enzyme by filtration through a
30-kDa membrane, the conversion of dTDP-D-glucose to
the dTDP-4-keto-6-deoxy-D-glucose product was verified
by HPLC.
Ligand complexes were obtained by transferring the crys-
tals to soaking solutions containing 26% poly(ethylene gly-
col) 8000, 600 mM NaCl, 5 mM N5-formyl-THF, and 10 mM
dTDP-3-amino-3,6-dideoxy-D-glucose (or dTDP-3-amino-3,-
6-dideoxy-D-galactose), and 100 mM MES (pH 6.0). After
soaking for 24 hr, they were then transferred to a cryoprotec-
tant solution composed of 32% poly(ethylene glycol) 8000,
700 mM NaCl, 15% ethylene glycol, 5 mM N5-formyl-THF,
10 mM dTDP-3-amino-3,6-dideoxy-D-glucose (or dTDP-
3-amino-3,6-dideoxy-D-galactose), and 100 mM MES (pH 6.0).
4.6 | Synthesis of dTDP-3-amino-
3,6-dideoxy-D-glucose and dTDP-3-amino-
3,6-dideoxy-D-galactose
A typical 50 ml reaction contained 7 mM dTDP-D-glu-
cose, 0.5 mg/ml DesIV, 0.1 mg/ml of the HCAN_0198
encoded ketoisomerase, 2 mg/ml HCAN_0204 encoded
aminotransferase, 0.2 mM PLP, 50 mM sodium gluta-
mate, and 50 mM HEPPS buffered at pH 8.0. The reac-
tion was allowed to proceed overnight at room
temperature. Enzymes were then removed by filtration
through a 10,000 MW cutoff Amicon filter. After 15-fold
dilution with water, the reaction mixture was loaded
onto a 50 ml HiLoad 26/10 Q-Sepharose HP column.
Purification was accomplished using a 20-column vol-
ume gradient of 0–400 mM ammonium bicarbonate at
pH 8.5. Substitution of the HCAN_0198 encoded
ketoisomerase with FdtA (the 3,4-ketoisomerase from
A. thermoaerophilus)26 led to the formation of the dTDP-
3-amino-3,6-dideoxy-D-galactose, which was purified in
the same manner as that for dTDP-3-amino-3,6-dideoxy-
D-glucose.
4.3 | X-ray data collection and
processing
All X-ray data were collected using
a BRUKER
D8-VENTURE sealed tube system equipped with
HELIOS optics and a PHOTON II detector. The X-ray
data sets were processed with SAINT and scaled with
SADABS (Bruker AXS). Relevant X-ray data collection
statistics are listed in Table 2.
4.4 | Structure solution and model
refinement
The structure of the apo-form of the N-formyltransferase
was determined via molecular replacement with
PHASER52 using PDB entry 4LXQ as a search probe.14
Iterative cycles of model building with COOT53,54 and
refinement with REFMAC55 led to a final X-ray model
with an overall R-factor of 16.9%. Relevant refinement
statistics are provided in Table 2. The structures in com-
plex with N5-formyl-THF and the dTDP-sugars were
solved via difference Fourier methods using the apoen-
zyme structure as a starting model. Relevant refinement
statistics are provided in Table 2. The aminotransferase
structure was solved with PHASER (20) using PDB entry
5U1Z as a starting model.56 The ketoisomerase structure
was determined with PHASER (20) using PDB entry
5TPU as a starting model.30
4.7 | Determination of kinetic constants
Steady-state
kinetic
parameters
for
the
N-
formyltransferase were evaluated via a discontinuous
assay using an ÄKTA HPLC system. Reaction rates were
determined by calculating the amount of product (dTDP-
3-formamido-3,6-dideoxy-D-glucose or dTDP-3-formamido-
3,6-dideoxy-D-galactose) formed on the basis of the peak
area on an HPLC trace measured at 267 nm. The area was
correlated to concentration via a calibration curve created
with standard samples that had been treated in the same
manner as the reaction aliquots.
The 1.5 ml reactions contained 50 mM HEPPS (pH 8.0),
5 mM N10-formyl-THF, and 0.025 mg/ml (0.75 mM) N-
formyltransferase. The dTDP-3-amino-3,6-dideoxy-D-glucose

HEISDORF ET AL.
2157

2158
HEISDORF ET AL.
(or dTDP-3-amino-3,6-dideoxy-D-galactose) concentrations
were varied from 0.005 to 1.5 millimolar. Seven 200 ml ali-
quots were taken over a 4-min time period and then
quenched by the addition of 10 ml of 6 M HCl. Subse-
quently, 200 ml of carbon tetrachloride was added, the sam-
ples were vortexed and spun at 14000g for 2 min. Of note,
150 ml of the aqueous phase was taken for HPLC analysis.
The samples were diluted with 2 ml of water and loaded
onto a 1 ml ResQ column. The N-formylated sugar product
was quantified after elution with an eight-column volume
gradient from 0 to 400 M lithium chloride (pH 4.0, HCl) as
previously described.14
The data were fitted to the equation: v0 = (Vmax[S])/
(KM + [S]) using PRISM (GraphPad Software, Inc.). The
kcat values were calculated according to the equation:
kcat = Vmax/[ET]. Relative kinetic parameters can be
found in Table 1.
Steady-state kinetic parameters for the ketoisomerase
were determined using a continuous and coupled assay
with a Beckman Coulter DU-640 spectrophotometer.
Enzyme activities were measured by monitoring the
decrease in absorbance at 340 nm as NADPH is oxidize to
NADP+ by KijD10 as previously described.29 The reaction
mixtures contained 50 mM HEPES (pH 7.5), 1 mg/ml
KijD10, 0.2 mM NADPH, and dTDP-4-keto-6-deoxy-D-
glucose varying from 0.01 to 3.0 mM. The reactions were
initiated by the addition of the ketoisomerase to a final con-
centration of 0.55 mM. The data were fitted as described
above and the kinetic parameters are presented in Table 1.
Thoden: Conceptualization; data curation; formal analy-
sis; investigation; validation; visualization. Hazel
Holden: Conceptualization; formal analysis; project
administration; validation; visualization; writing - origi-
nal draft; writing-review & editing.
ORCID
Hazel M. Holden https://orcid.org/0000-0001-6214-
3638
REFERENCES
1. Marshall BJ, Warren JR. Unidentified curved bacilli in the
stomach of patients with gastritis and peptic ulceration. Lancet.
1984;1:1311–1315.
2. Goodwin CS, Armstrong JA, Chilvers T, et al. Transfer of Campylo-
bacter pylori and Campylobacter mustelae to Helicobacter gen. Nov.
as Helicobacter pylori comb. nov. and Helicobacter mustelae comb.
nov., respectively. Intl J System Evol Microbiol. 1989;39:397–405.
3. Ahmed N. 23 years of the discovery of Helicobacter pylori: Is
the debate over? Ann Clin Microbiol Antimicrob. 2005;4:17.
4. Suerbaum S, Michetti P. Helicobacter pylori infection. N Engl J
Med. 2002;347:1175–1186.
5. Cao DM, Lu QF, Li SB, et al. Comparative genomics of
H. pylori and non-Pylori Helicobacter species to identify new
regions associated with its pathogenicity and adaptability.
Biomed Res Int. 2016;2016:6106029.
6. Fox JG, Chien CC, Dewhirst FE, et al. Helicobacter canadensis
sp. nov. isolated from humans with diarrhea as an example of
an emerging pathogen. J Clin Microbiol. 2000;38:2546–2549.
7. Andersen LP. New Helicobacter species in humans. Dig Dis.
2001;19:112–115.
8. Solnick JV, Schauer DB. Emergence of diverse Helicobacter spe-
cies in the pathogenesis of gastric and enterohepatic diseases.
Clin Microbiol Rev. 2001;14:59–97.
9. Waldenstrom J, On SL, Ottvall R, Hasselquist D,
Harrington CS, Olsen B. Avian reservoirs and zoonotic poten-
tial of the emerging human pathogen Helicobacter canadensis.
Appl Environ Microbiol. 2003;69:7523–7526.
10. Goto K, Jiang W, Zheng Q, et al. Epidemiology of Helicobacter
infection in wild rodents in the Xinjiang-Uygur autonomous
region of China. Curr Microbiol. 2004;49:221–223.
11. Nebbia P, Tramuta C, Ortoffi M, Bert E, Cerruti Sola S,
Robino P. Identification of enteric Helicobacter in avian species.
Schweiz Arch Tierheilkd. 2007;149:403–407.
12. Van den Bulck K, Decostere A, Baele M, Marechal M,
Ducatelle R, Haesebrouck F. Low frequency of Helicobacter
species in the stomachs of experimental rabbits. Lab Anim.
2006;40:282–287.
13. Loman NJ, Snyder LA, Linton JD, Langdon RL, A. J.,
Weinstock GM, Wren BW, Pallen MJ. Genome sequence of the
emerging pathogen Helicobacter canadensis. J Bacteriol. 2009;
191:5566–5567.
4.8 | Analysis of quaternary structure
The quaternary structures of the various ketoisomerases
were analyzed via a Superdex 200 10/300 (GE Healthcare)
gel filtration column and an ÄKTA HPLC system. The
samples were loaded and run in 10 mM Tris–HCL
(pH 8.0) and 200 mM NaCl. The column was run at a
speed of 0.5 ml/min at ambient temperature.
ACKNOWLEDGMENTS
X-ray coordinates have been deposited in the Research
Collaboratory for Structural Bioinformatics, Rutgers Uni-
versity, New Brunswick, NJ (accession nos. 7N63, 7N67,
7N7A, 7N7B, and 7N7C). This work was supported by
NIH R35 GM134643 to H. M. H.
CONFLICT OF INTEREST
The authors have no competing financial interests.
14. Thoden JB, Goneau MF, Gilbert M, Holden HM. Structure of a
sugar N-formyltransferase from Campylobacter jejuni. Bio-
chemistry. 2013;52:6114–6126.
AUTHOR CONTRIBUTIONS
Colton Heisdorf: Data curation; investigation. William
Griffiths: Data curation; formal analysis. James
15. Zimmer AL, Thoden JB, Holden HM. Three-dimensional struc-
ture of a sugar N-formyltransferase from Francisella tularensis.
Protein Sci. 2014;23:273–283.

HEISDORF ET AL.
2159
16. Genthe NA, Thoden JB, Benning MM, Holden HM. Molecular
structure of an N-formyltransferase from Providencia
alcalifaciens O30. Protein Sci. 2015;24:976–986.
17. Woodford CR, Thoden JB, Holden HM. New role for the
ankyrin repeat revealed by a study of the N-formyltransferase
from Providencia alcalifaciens. Biochemistry. 2015;54:631–638.
18. Genthe NA, Thoden JB, Holden HM. Structure of the
Escherichia coli ArnA N-formyltransferase domain in complex
with N(5) -formyltetrahydrofolate and UDP-Ara4N. Protein
Sci. 2016;25:1555–1562.
specific polysaccharide of Hafnia alvei 1204 containing 3,6-dideoxy-
3-formamido-D-glucose. Carbohydr Res. 1995;273:187–195.
36. Kocharova NA, Maszewska A, Zatonsky GV, et al. Structure of
the O-polysaccharide of Providencia alcalifaciens O21 con-
taining 3-formamido-3,6-dideoxy-D-galactose. Carbohydr Res.
2003;338:1425–1430.
37. Burgie ES, Holden HM. Three-dimensional structure of DesVI
from Streptomyces venezuelae: A sugar N,N-dimethyltransferase
required for dTDP-desosamine biosynthesis. Biochemistry.
2008;47:3982–3988.
19. Dunsirn MM, Thoden JB, Gilbert M, Holden HM. Biochemical
investigation of Rv3404c from Mycobacterium tuberculosis. Bio-
chemistry. 2017;56:3818–3825.
20. Woodford CR, Thoden JB, Holden HM. Molecular architecture
of an N-formyltransferase from Salmonella enterica O60.
J Struct Biol. 2017;200:267–278.
21. Riegert AS, Chantigian DP, Thoden JB, Tipton PA, Holden HM.
Biochemical characterization of WbkC, an N-formyltransferase
from Brucella melitensis. Biochemistry. 2017;56:3657–3668.
22. Hofmeister DL, Thoden JB, Holden HM. Investigation of a
sugar N-formyltransferase from the plant pathogen Pantoea
ananatis. Protein Sci. 2019;28:707–716.
38. Liu B, Knirel YA, Feng L, et al. Structure and genetics of Shi-
gella O antigens. FEMS Microbiol Rev. 2008;32:627–653.
39. Luderitz O, Staub AM, Westphal O. Immunochemistry of O
and R antigens of Salmonella and related Enterobacteriaceae.
Bacteriol Rev. 1966;30:192–255.
40. Guo R, Li Z, Jiao Y, et al. O-polysaccharide is important for Sal-
monella pullorum survival in egg albumen, and virulence and col-
onization in chicken embryos. Avian Pathol. 2017;46:535–540.
41. Lindberg AA, Segall T, Weintraub A, Stocker BA. Antibody
response and protection against challenge in mice vaccinated
intraperitoneally with a live aroA O4-O9 hybrid Salmonella
Dublin strain. Infect Immun. 1993;61:1211–1221.
23. Girardi NM, Thoden JB, Holden HM. Misannotations of the
genes encoding sugar N-formyltransferases. Protein Sci. 2020;
29:930–940.
24. Holden HM, Thoden JB, Gilbert M. Enzymes required for the
biosynthesis of N-formylated sugars. Curr Opin Struct Biol.
2016;41:1–9.
25. Dunwell JM, Purvis A, Khuri S. Cupins: The most functionally
diverse protein superfamily? Phytochemistry. 2004;65:7–17.
26. Davis ML, Thoden JB, Holden HM. The x-ray structure of
dTDP-4-keto-6-deoxy-D-glucose-3,4-ketoisomerase. J Biol Chem.
2007;282:19227–19236.
27. Chantigian DP, Thoden JB, Holden HM. Structural and bio-
chemical characterization of a bifunctional ketoisomerase/N-
acetyltransferase from Shewanella denitrificans. Biochemistry.
2013;52:8374–8385.
42. Kingsley RA, Baumler AJ. Host adaptation and the emergence
of infectious disease: The Salmonella paradigm. Mol Microbiol.
2000;36:1006–1014.
43. Raynaud C, Meibom KL, Lety MA, et al. Role of the wbt locus
of Francisella tularensis in lipopolysaccharide O-antigen bio-
genesis and pathogenicity. Infect Immun. 2007;75:536–541.
44. Bengoechea JA, Najdenski H, Skurnik M. Lipopolysaccharide O
antigen status of Yersinia enterocolitica O:8 is essential for viru-
lence and absence of O antigen affects the expression of other
Yersinia virulence factors. Mol Microbiol. 2004;52:451–469.
45. Plainvert C, Bidet P, Peigne C, et al. A new O-antigen gene
cluster has a key role in the virulence of the Escherichia coli
meningitis clone O45:K1:H7. J Bacteriol. 2007;189:8528–8536.
46. Norstebo SF, Lotherington L, Landsverk M, Bjelland AM,
Sorum H. Aliivibrio salmonicida requires O-antigen for virulence in
Atlantic salmon (Salmo salar L.). Microb Pathog. 2018;124:322–331.
47. Krissinel E, Henrick K. Secondary-structure matching (SSM), a
new tool for fast protein structure alignment in three dimen-
sions. Acta Crystallogr. 2004;D60:2256–2268.
48. Knirel YA, Vinogradov EV, Shashkov AS, et al. Somatic antigens
of Pseudomonas aeruginosa. The structure of the O-specific poly-
saccharide chains of lipopolysaccharides of P. aeruginosa sero-
group O4 (Lanyi) and related serotype O6 (Habs) and
immunotype 1 (fisher). Eur J Biochem. 1985;150:541–550.
49. Hansen R, Thomson JM, Fox JG, El-Omar EM, Hold GL. Could
Helicobacter organisms cause inflammatory bowel disease?
FEMS Immunol Med Microbiol. 2011;61:1–14.
28. Thoden JB, Holden HM. The molecular architecture of QdtA, a
sugar
3,4-ketoisomerase
from
Thermoanaerobacterium
thermosaccharolyticum. Protein Sci. 2014;23:683–692.
29. Thoden JB, Vinogradov E, Gilbert M, Salinger AJ, Holden HM.
Bacterial sugar 3,4-ketoisomerases: Structural insight into
product stereochemistry. Biochemistry. 2015;54:4495–4506.
30. Li ZZ, Riegert AS, Goneau MF, et al. Characterization of the
dTDP-Fuc3N and dTDP-Qui3N biosynthetic pathways in Cam-
pylobacter jejuni 81116. Glycobiology. 2017;27:358–369.
31. DeLano WL. Unraveling hot spots in binding interfaces: Pro-
gress and challenges. Curr Opin Struct Biol. 2002;12:14–20.
32. Melancon CE 3rd, Hong L, White JA, Liu YN, Liu HW. Char-
acterization of TDP-4-keto-6-deoxy-D-glucose-3,4-ketoisomerase
from the D-mycaminose biosynthetic pathway of Streptomyces
fradiae: In vitro activity and substrate specificity studies. Bio-
chemistry. 2007;46:577–590.
50. Thoden JB, Holden HM. The molecular architecture of human N-
acetylgalactosamine kinase. J Biol Chem. 2005;280:32784–32791.
51. Laskowski RA, MacArthur MW, Moss DS, Thornton JM.
PROCHECK: A program to check the stereochemical quality of
protein structures. J Appl Cryst. 1993;26:283–291.
52. McCoy AJ, Grosse-Kunstleve RW, Adams PD, Winn MD,
Storoni LC, Read RJ. Phaser crystallographic software. J Appl
Cryst. 2007;40:658–674.
33. Jansonius JN. Structure, evolution and action of vitamin
B6-dependent enzymes. Curr Opin Struct Biol. 1998;8:759–769.
34. Toney MD. Aspartate aminotransferase: An old dog teaches
new tricks. Arch Biochem Biophys. 2014;544:119–127.
53. Emsley P, Cowtan K. Coot: Model-building tools for molecular
35. Katzenellenbogen E, Romanowska E, Kocharova NA,
graphics. Acta Crystallogr. 2004;D60:2126–2132.
Shashkov AS, Knirel YA, Kochetkov NK. Structure of the O-

2160
HEISDORF ET AL.
54. Emsley P, Lohkamp B, Scott WG, Cowtan K. Features and
development of coot. Acta Crystallogr. 2010;D66:486–501.
55. Murshudov GN, Vagin AA, Dodson EJ. Refinement of macro-
molecular structures by the maximum-likelihood method. Acta
Crystallogr. 1997;D53:240–255.
56. Dow GT, Gilbert M, Thoden JB, Holden HM. Structural investi-
gation on WlaRG from Campylobacter jejuni: A sugar amino-
transferase. Protein Sci. 2017;26:586–599.
How to cite this article: Heisdorf CJ,
Griffiths WA, Thoden JB, Holden HM.
Investigation of the enzymes required for the
biosynthesis of an unusual formylated sugar in the
emerging human pathogen Helicobacter
canadensis. Protein Science. 2021;30:2144–60.
https://doi.org/10.1002/pro.4169
57. Allard ST, Cleland WW, Holden HM. High resolution X-ray
structure of dTDP-glucose 4,6-dehydratase from Streptomyces
venezuelae. J Biol Chem. 2004;279:2211–2220.