You are on page 1of 11

pubs.acs.

org/biochemistry Article

A Mechanistic Basis for Phosphoethanolamine Modification of the


Cellulose Biofilm Matrix in Escherichia coli
Alexander C. Anderson, Alysha J. N. Burnett, Shirley Constable, Lana Hiscock, Kenneth E. Maly,
and Joel T. Weadge*
Cite This: Biochemistry 2021, 60, 3659−3669 Read Online

ACCESS Metrics & More Article Recommendations *


sı Supporting Information
See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.

ABSTRACT: Biofilms are communities of self-enmeshed bacteria


Downloaded via UNIV ESTADUAL PAULISTA on January 22, 2022 at 18:20:22 (UTC).

in a matrix of exopolysaccharides. The widely distributed human


pathogen and commensal Escherichia coli produces a biofilm matrix
composed of phosphoethanolamine (pEtN)-modified cellulose and
amyloid protein fibers, termed curli. The addition of pEtN to the
cellulose exopolysaccharide is accomplished by the action of the
pEtN transferase, BcsG, and is essential for the overall integrity of
the biofilm. Here, using the synthetic co-substrates p-nitrophenyl
phosphoethanolamine and β-D-cellopentaose, we demonstrate
using an in vitro pEtN transferase assay that full activity of the
pEtN transferase domain of BcsG from E. coli (EcBcsGΔN) requires
Zn2+ binding, a catalytic nucleophile/acid-base arrangement
(Ser278/Cys243/His396), disulfide bond formation, and other
newly uncovered essential residues. We further confirm that EcBcsGΔN catalysis proceeds by a ping-pong bisubstrate−biproduct
reaction mechanism and displays inefficient kinetic behavior (kcat/KM = 1.81 × 10−4 ± 2.81 × 10−5 M−1 s−1), which is typical of
exopolysaccharide-modifying enzymes in bacteria. Thus, the results presented, especially with respect to donor binding (as reflected
by KM), have importantly broadened our understanding of the substrate profile and catalytic mechanism of this class of enzymes,
which may aid in the development of inhibitors targeting BcsG or other characterized members of the pEtN transferase family,
including the intrinsic and mobile colistin resistance factors.

iofilms are dynamic communities of microorganisms fixed


B in a synthesized matrix of secreted polysaccharide
materials.1 This biofilm matrix serves as the interface between
beginning to be understood.13 For example, the biofilm of
Pseudomonas f luorescens SBW25 was observed to contain an
acetylated form of cellulose,8 and more recently, the discovery
its constituents and their outside environment, but also plays of phosphoethanolamine (pEtN) cellulose was reported in E.
many other roles to the benefit of the enclosed microorganisms coli and Salmonella enterica (Figure 1).14 Cellulose has also
that may justify the large resource cost of producing copious been recently proposed in the Gram-positive human pathogen
quantities of this extracellular matrix material.2 For example, Clostridioides dif f icile,12 and an operon for the production of
biofilms have been observed to serve in resource capture,3 Pel exopolysaccharide was reported in Bacillus cereus,15 thereby
accelerate cell growth,4 and improve tolerance to stressors or expanding the repertoire of bacteria known to produce these
disinfectants, such as shearing forces,2 desiccation,2 antimicro- exopolysaccharides beyond Gram-negative organisms.
bial compounds,2,5 extreme temperature,2,5,6 or sanitizing At present, all Gram-negative bacteria known to produce
agents2,3,5 for the enclosed microorganisms. The persistent cellulose as a component of their biofilm possess an operon
colonization of surfaces by the widely distributed human and encoding the essential and conserved protein complex
animal pathogens Escherichia coli and Salmonella spp. has been responsible for cellulose biosynthesis and export.16 Structural
linked to the production of biofilms composed of the and functional studies, informed by research on other
polysaccharide cellulose and amyloid protein fibers termed polysaccharide secretion systems, have provided evidence for
curli.7 In addition to these species, biofilms containing
cellulose have been reported in numerous other species,
including other pathogens of plants and humans, like Received: September 10, 2021
Agrobacterium tumefaciens, Cronobacter sakazakii, Acetobacter Revised: October 22, 2021
xylinum, Clostridium (previously Sarcina) spp., Rhizobium spp., Published: November 11, 2021
and some Pseudomonads.8−12 Although the biofilms of these
bacteria were long thought to contain a similar linear β-(1−4)-
glucan, the diversity of these biofilm polysaccharides is
© 2021 The Authors. Published by
American Chemical Society https://doi.org/10.1021/acs.biochem.1c00605
3659 Biochemistry 2021, 60, 3659−3669
Biochemistry pubs.acs.org/biochemistry Article

ide.14,25,27−29 The BcsE protein was shown to comprise both a


degenerate receiver domain and a GGDEF domain responsible
for sensing c-di-GMP.28 BcsE has also recently been proposed
to be responsible for the recruitment of BcsQ and BcsR to the
membrane during early Bcs macrocomplex assembly.29 Other
recent efforts with cryo-electron microscopy have resolved the
native Bcs macrocomplex in several states and demonstrate its
piecewise assembly.30,31 In light of the discovery of pEtN
cellulose, bcsG was also shown to be necessary for the
architecture and assembly of the biofilm matrix.14 BcsG was
further proposed to be the pEtN transferase responsible for the
postsynthetic modification of cellulose in the periplasm of
organisms producing a pEtN cellulose polysaccharide.14,27 We
previously reported the structure and activity of BcsG from E.
coli (EcBcsG), demonstrating that it acts as a pEtN transferase
with substrate specificity for cellulose.25 Additionally, the
structure and function of BcsG from Salmonella typhimurium
(StBcsG) was independently reported, and the results support
the conclusion that EcBcsG and StBcsG are functionally
equivalent pEtN transferases acting on cellulose polysacchar-
ides.27
Both EcBcsG and StBcsG studies reported the functional
complementation of bcsG using a library of BcsG amino acid
variants.25,27 These combined results demonstrated the
essential catalytic residues of BcsG, which are partially
Figure 1. Schematic of the reaction catalyzed by BcsG.
conserved among other characterized pEtN transferases,
Phosphoethanolamine is added to approximately 50% of glucose
units of bacterial cellulose, exclusively at the C6 position with an including the intrinsic and mobile colistin resistance factors.
unknown pattern across the polymer.14,25 Disruption of these essential residues in EcBcsG resulted in a
fragile pellicle biofilm phenotype in E. coli AR3110 that was
the roles of each gene product in this operon, often annotated indistinguishable from a strain bearing a chromosomal deletion
bcsABCZ.17 Biosynthesis of cellulose occurs by the action of of bcsG.25 However, the enzymatic consequences of amino acid
the processive glycosyltransferase BcsA at the cytoplasmic face substitutions that resulted in these phenotypes remain
of the inner membrane, using UDP-glucose as a substrate.18 unexplored. Using an in vitro pEtN transferase assay, we
Regulation of cellulose biosynthesis occurs through the binding show here that Zn2+ binding, disulfide bond formation, and the
of the second messenger molecule cyclic dimeric guanosine catalytic nucleophile/acid-base arrangement (Ser278/Cys243/
monophosphate (c-di-GMP) to the PilZ domain of BcsA, His396) are essential for full transferase activity of the
allosterically regulating the activity of the glycosyltransferase recombinant catalytic domain of EcBcsG (herein EcBcsGΔN).
domain.19 Translocation of the growing chain from the Furthermore, we expand the donor substrate profile to confirm
cytoplasm to the periplasm occurs co-synthetically, through a EcBcsGΔN is specific for pEtN transfer and report the
pore in BcsA, and is aided by a carbohydrate-binding domain Michaelis−Menten kinetics of EcBcsG using the co-substrate
found in the periplasmic protein BcsB.18,20 Similar to structural analogues p-nitrophenyl phosphoethanolamine (p-NPPE) and
and functional studies of other polysaccharide systems, cello-oligosaccharides. We also demonstrate that the kinetic
cellulose export across the outer membrane was recently behavior of EcBcsG further supports a ping-pong bisubstrate−
shown to require an outer membrane β-barrel and periplasmic biproduct (bibi) catalytic mechanism and expand the active
tetratricopeptide repeat (TPR)-containing protein, encoded by site to include new residues proposed by examination of the
BcsC.21−24 Finally, a periplasmic polysaccharide lyase or structural model to be involved in catalysis.
glycoside hydrolase is found in many known polysaccharide
biosynthesis operons.22 With respect to microbial cellulose,
structural and functional studies of BcsZ from E. coli and CcsZ
■ EXPERIMENTAL PROCEDURES
Materials. Culture media, isopropyl β-D-thiogalactopyrano-
from C. dif f icile demonstrated that these terminal enzymes side (IPTG), and kanamycin sulfate were purchased from
were from glycoside hydrolase families 8 and 5, respectively, BioBasic (Markham, ON). Cello-oligomers were purchased
and both possess endo-β-(1,4)-glucanase activity for mod- from Megazyme (Dublin, Ireland). Nickel-nitrilotriacetic acid
ification of the polymer.12,26 (Ni-NTA) agarose was purchased from Qiagen (Valencia,
Previously, the role of accessory operons or genes within CA). Unless otherwise specified, all other materials were
these operons remained unknown until the discovery of purchased from Sigma-Aldrich Canada Ltd. (Oakville, ON).
modified polysaccharide materials within biofilms.17 For Cloning, Expression, and Purification of EcBcsGΔN.
example, in cellulose-producing organisms, the role of a type The catalytic domain of the bcsG gene was cloned into the
II operon, containing bcsEFG, was unknown, and its gene pET28a expression vector, as described previously.25 Briefly,
products showed little similarity to other characterized proteins bcsGΔN was amplified from E. coli K12 chromosomal DNA and
based on sequence.17 These genes were annotated as necessary cloned into the pET28a expression vector using NdeI and XhoI
for maximal cellulose production,17 but more recently, restriction sites. The polymerase chain reaction (PCR)
structural and functional characterization has shed light on products were digested by the appropriate enzymes for 1 h,
their true roles in processing of the cellulose polysacchar- followed by purification using a GeneJET PCR purification kit
3660 https://doi.org/10.1021/acs.biochem.1c00605
Biochemistry 2021, 60, 3659−3669
Biochemistry pubs.acs.org/biochemistry Article

(Thermo Fisher Scientific). The purified PCR product was resulting solutions were loaded onto an Ä KTA Start fast-
ligated into an NdeI/XhoI doubly digested vector using T4 protein liquid chromatography instrument (Cytiva) equipped
DNA ligase (Thermo Fisher Scientific) at 22 °C for 4 h. The with a HiTrap Q FF cartridge (1 mL, Cytiva). The solvents
ligation mixture was used to transform E. coli TOP10 (Thermo were (A) 50 mM Tris (pH 8.0) and (B) 50 mM Tris and (pH
Fisher Scientific), and pET28a-bcsGΔN was isolated from 16− 8.0) and 1 M NaCl. The protein was bound by passing sample
18 h grown cultures using an EZ-10 plasmid prep kit (Bio five times over the column at a flow rate of 1 mL/min in 100%
Basic). The resulting recombinant BcsG C-terminal catalytic A. The column was washed for 20 min with 100% A, followed
domain, EcBcsGΔN, was comprised of residues Ala164−Gln559 by elution in a linear gradient from 0% to 100% B over 30 min.
from UniProt entry P37659 and included the addition of an N- Peaks of >50 mAU at 280 nm were collected using a Frac30
terminal hexahistidine tag conferred by the pET28a expression fraction collector (Cytiva). All purification buffers were
plasmid. supplemented with 5 mM MgCl2, as we observed this to
Site-Directed Mutagenesis. Site-directed mutagenesis improve the stability of recombinant EcBcsGΔN and deriva-
was performed using the PCR primer method with pET28a- tives. Recombinant protein eluted from purification columns
bcsGΔN as a template. Amino acid variants of Arg458 and Ser244 was pooled and concentrated to between 30 and 45 mg mL−1
were generated from site-directed mutagenesis of the wild-type in a 30 kDa molecular weight cutoff centrifugal filter unit
pET28a-bcsGΔN plasmid as a service provided by BioBasic. The (Cytiva) and stored at 4 or −20 °C for short- or long-term
primers used to generate the alanine EcBcsGΔN amino acid storage, respectively.
variants of Cys243, Tyr277, Ser278, His396, Glu442, and His443 are Measurement of the Enzymatic Rate. The enzymatic
available in Table S1. Following amplification of amino acid rate of EcBcsGΔN and derivatives was measured with the
variant plasmids, DpnI (1 unit) was added to reaction mixtures chromogenic assay using the mock co-substrates p-nitrophenyl
to digest methylated template DNA. The reaction mixture was phosphoethanolamine (p-NPPE), p-nitrophenyl phosphopro-
transformed into E. coli TOP10, and the resulting clones were panolamine (p-NPPP), p-nitrophenyl phosphate (p-NPP)
sequenced to confirm the presence of the desired mutations (Sigma), and cellooligosaccharides, as described previously.25
(The Centre for Applied Genomics). The synthesis of p-NPPE was performed as described
Expression and Purification of EcBcsGΔN and Deriv- previously.25 The detailed synthesis of p-NPPP is described
atives. pET28a-bcsGΔN and derivative plasmids thereof were in the Supporting Information. The esterase activity of
transformed into E. coli BL21 (DE3), and cells were grown in 1 EcBcsGΔN and amino acid variants thereof was measured
L volumes of rich medium (32 g of tryptone, 20 g of yeast using 7 mM p-NPPE and 50 μM enzyme. The substrates p-
extract, and 10 g of NaCl) containing kanamycin (50 μg NPPP and p-NPP were tested at 2 mM with and without 3
mL−1). Cultures were incubated at 37 °C while being shaken, mM cellopentaose. Transferase activity was measured under
until they reached an optical density at 600 nm of 0.6−0.8. the same conditions, but only in the presence of
Expression of EcBcsGΔN and amino acid variants thereof was cellooligosaccharides, as described previously.25 All reactions
induced by the addition of IPTG (1 mM), and growth was were carried out in 50 mM HEPES buffer (pH 7.5) containing
allowed to continue for 16−20 h at 23 °C. The cells were 50 mM NaCl at 37 °C using 100 μL volumes. Concentrations
collected by centrifugation (4300 × g for 15 min at 4 °C) and of 1 mM ethylenediamine tetraacetic acid (EDTA) or
stored at −20 °C until use. Cell pellets were resuspended in dithiothreitol (DTT) were exposed to EcBcsGΔN by buffer
lysis buffer [50 mM Tris (pH 8.0) and 300 mM NaCl] and exchange for no less than 24 h prior to the assay. Reactions
supplemented with 0.5 mg of RNase A (BioBasic), 0.25 mg of were monitored in a Cytation5 imaging plate reader (Biotek)
DNase I (BioBasic), and for some variants one EDTA-free at 37 °C for 30 min at 414 nm. The specific activity and
mini protease inhibitor tablet (Roche). The resulting cell velocity of the reactions were calculated from absorbance data
suspensions were lysed using a cell disruptor (Constant using a standard curve prepared with p-nitrophenol analytical
Systems) operating at 17000 psi (117211 kPa) and 4 °C. The reference material (Sigma, catalog no. 241326). All rates are
lysate was separated by centrifugation (28000 × g for 45 min at reported as the mean ± the standard deviation (SD) of at least
4 °C), and hexahistidine-tagged EcBcsGΔN (and amino acid three replicates. Figure preparation and statistical analysis were
variants thereof) were purified from the soluble fraction using performed using GraphPad Prism version 9.0.1.
nickel affinity chromatography. Ni-NTA resin (Thermo Fisher Steady-State Kinetics. Kinetic assays were performed
Scientific) pre-equilibrated in lysis buffer (2 mL) was using the same assay design described above but with p-NPPE
suspended in the cell lysate and incubated for 1 h at 4 °C to concentrations as specified that ranged from 0 to 14.5 mM.
facilitate binding. This solution was filtered by being passed Kinetics with the cellooligosaccharide acceptor were also
over a chromatography column and washing of the resin in at conducted with concentrations ranging from 0 to 6 mM with a
least three 10 mL volumes of lysis buffer, followed by three 10 fixed p-NPPE concentration of 7 mM. Michaelis−Menten
mL volumes of lysis buffer with the addition of 25 mM parameters (kcat and KM) were determined by nonlinear
imidazole. Elution was achieved by addition of 10 mL of regression analysis of plots of initial velocity (nanomoles per
elution buffer (lysis buffer with the addition of 250 mM second) as a function of p-NPPE concentration. All model
imidazole) to the chromatography resin. Buffer exchange was fitting, analysis, and figure preparation were performed using
performed by a two-step dialysis against 50 mM Tris (pH 8.0) GraphPad Prism version 9.0.1.
and 150 mM NaCl using two 2 L volumes for at least 1 h per Mass Spectrometry. Liquid chromatography−mass spec-
volume at 4 °C. Secondary purification was performed by trometry analyses were performed at the Mass Spectrometry
anion exchange chromatography when deemed necessary as Facility of the Advanced Analysis Centre (University of
assessed by sodium dodecyl sulfate−polyacrylamide gel Guelph, Guelph, ON) on an Agilent 1200 HPLC liquid
electrophoresis. For secondary purification, proteins were chromatograph interfaced with an Agilent UHD 6530 Q-TOF
buffer exchanged by dialysis against 50 mM Tris (pH 8.0) mass spectrometer. A C18 column (Agilent Extend-C18, 50
using two 2 L volumes for at least 1 h per volume at 4 °C. The mm × 2.1 mm, 1.8 μm) was used for chromatographic
3661 https://doi.org/10.1021/acs.biochem.1c00605
Biochemistry 2021, 60, 3659−3669
Biochemistry pubs.acs.org/biochemistry Article

Figure 2. Amino acid replacements of essential catalytic residues abrogate EcBcsGΔN activity. (A) The structure of the EcBcsGΔN active site (PDB
entry 6PD0) is depicted here and was previously mapped by functional complementation in vivo.25 (B) The measured rates of esterase (i.e.,
carbohydrate acceptor-free conditions) and transferase (i.e., in the presence of a cellopentaose acceptor) activities of EcBcsGΔN amino acid variants
within the catalytic site demonstrate these residues are essential in vitro with the exception of Y277. Asterisks denote statistical significance [Tukey’s
multiple comparisons, q(33) > 8.40 and p < 0.0001 for all comparisons].

separation. The following solvents were used for separation: in this study (C243A, S244A, Y277F, S278A, H396A, E442A,
water with 0.1% (v/v) formic acid (A) and acetonitrile with H443A, R458A, and R458H) were generated using site-
0.1% (v/v) formic acid (B). The initial mobile phase directed mutagenesis of pET28a-bcsGΔN. Each of these
conditions were 10% B, hold for 1 min, and then increase to recombinant proteins were expressed in E. coli BL21
100% B over 29 min. transformants with observed yields similar to that of wild-

■ RESULTS AND DISCUSSION


Amino Acid Replacements. To assess the catalytic
type EcBcsGΔN. These variant proteins were purified to
apparent homogeneity using the protocol described previ-
ously,25 with additional washing steps as deemed necessary for
consequences of amino acid substitutions within the EcBcsG purification to homogeneity. Care was taken to use new
active site, we constructed a library of amino acid EcBcsGΔN chromatography media for the purification of each respective
variants with alanine substitutions of active site residues enzyme to prevent enzyme carryover and cross-contamination
observed to be essential for catalytic activity in vivo (Cys243, of assay data with multiple enzyme forms.
Tyr277, Ser278, His396, Glu442, and His443). These residues As expected, replacement of the nucleophilic Ser278 (S278A)
represent the structural equivalent to the active site of known resulted in a loss of any detectable activity. The observed rates
pEtN transferases. Interestingly, only His443 and His396 are of both esterase (i.e., bulk solvent serves as the pEtN acceptor)
strictly conserved, although our previous work demonstrated and transferase activity (i.e., in the presence of a defined
that the remainder are catalytically essential in vivo and carbohydrate acceptor) of the S278A replacement were not
probably represent the functional equivalents to the conserved different from an enzyme-free control (Figure 2; t tests, t(4) =
residues of other pEtN transferases.25 However, the con- 0.3586 and p = 0.7830 for esterase; t(4) = 0.4935 and p =
sequences of these amino acid replacements remained untested 0.6475 for transferase). This lack of activity supports the role
on an enzymatic level, so herein we explored these residues, of this nucleophilic serine in the catalytic mechanism. His396 is
along with two new residues we suspected may be catalytically highly conserved and has also been shown to be catalytically
important (Ser244 and Arg458) based on inspection of the BcsG important in other pEtN transferases.32 Replacement of this
structure and that of its homologues. All of the constructs used histidine in BcsG (H396A) also led to reductions in activity
3662 https://doi.org/10.1021/acs.biochem.1c00605
Biochemistry 2021, 60, 3659−3669
Biochemistry pubs.acs.org/biochemistry Article

with a residual esterase activity of 16.6% and a residual the interface of the catalytic and membrane domains.36
transferase activity of 7.0% (Figure 2), which is consistent with Consequently, we rationalized that this disparity in our data
results previously noted for this mutation in the in vivo biofilm may indicate that the role of the Tyr277 hydroxyl may also
assays, where alanine substitution of this residue resulted in the include the stability of the local folding and interaction of the
fragile pellicle phenotype.25 EcBcsG C-terminal catalytic domain relative to the N-terminal
Substantial losses of activity were also observed with alanine transmembrane region and, thus, allow EcBcsG to sample
replacement of the metal-binding triad residues of the enzyme productive conformations during the catalytic cycle. This
(Figure 2). The Glu442 replacement (E442A) resulted in an hypothesis would account for the importance of Tyr277 in
esterase activity indistinguishable from an enzyme-free control EcBcsG activity observed in vivo, but was not needed for
(e.g., 1.96 ± 0.07 nmol min−1 mg−1), and only 2.3% residual EcBcsGΔN in vitro activity.25
transferase activity was detected in the presence of β-D- The MCR-1 pEtN transferase has been resolved by X-ray
cellopentaose. Replacement of Cys243 with alanine (C243A) crystallography with both one Zn2+ ion and two Zn2+ ions at
also resulted in low residual specific activities of 4.7% the active site32 (PDB entries 5LRN and 5LRM, respectively).
(esterase) and 2.4% (transferase). The His443 substitution Although the Zn2+ ion resolved in EcBcsGΔN is found at the
(H443A) also demonstrated reduced specific activities, conserved location found in all pEtN transferases, it remains
although to a lesser degree, representing 38.5% residual unclear if the second Zn2+ site reported for MCR-1 is
esterase and 10.4% residual transferase activity. The retention catalytically important and whether this site is found in other
of some activity by His443 may be explained by the observation pEtN transferases. Two His residues in MCR-1 coordinate this
of Ser278 also serving as a ligand for Zn2+ binding in some of second Zn2+ ion and are equivalent to His396 and Arg458 in
the structures when His443 is not. Completion of the Zn2+ EcBcsG. To assess if Arg458 plays a role in secondary Zn2+
coordination sphere in at least one of the structural models binding or catalysis, we measured the enzymatic rates of
shows the Zn2+ also bound to a water molecule, which is alanine (R458A) and histidine (R458H) replacements (Figure
consistent with the Zn2+ playing a catalytic and not a structural 2). We found that the R458A variant displayed 25.6% esterase
role.33 Combined, these observations and the low residual and no detectable transferase activity, while the R458H variant
specific activities are consistent with the essential Zn2+ binding displayed 11.4% esterase and also did not display detectable
roles of these residues that had previously been noted in the transferase activity compared to that of the wild type. As the
structural model and assayed in vivo.25 These findings further histidine imidazole group would be expected to functionally
support the observation that Zn2+ binding is indispensable for complement the arginine guanidino group in metal binding,
the catalytic activity of BcsG in vivo, suggesting the residues we these results suggest that Arg458 does not participate in
replaced here are truly equivalent to the catalytic residues of secondary metal binding. Instead, because Arg458 contributes a
other polymyxin resistance factors.25 guanidino group at a similar distance and across the face of the
Previous mass spectrometry and structural data we collected active site from His396 (i.e., 5 Å above the face of the Zn2+ ion),
suggested that an additional residue may contribute to the this configuration is consistent with the possibility that Arg458
catalytic fold and belonged to the loop that presents Cys243 to stabilizes the charged pEtN−enzyme intermediate formed
the active site.25 Although we did not test a full-length BcsG during the first mechanistic step that results in reduced esterase
amino acid replacement in vivo, we propose that the identity of activity in both mutants. Because replacement of Arg458 with
this residue is Ser244 (Figure 2). The side chain hydroxyl of either Ala or His also abrogated transferase activity, this
Ser244 is oriented toward the Zn2+ at an atomic distance of 4.5 residue likely also plays an essential role in accommodation of
Å, and this residue is the only one that could plausibly serve an the acceptor co-substrate during the second mechanistic step.
essential function given both our structural and mass An in situ assay for biofilm phenotype in S. typhimurium using
spectrometry data sets. In support of our hypothesis, we BcsG variants suggested that both His and Ala replacements of
found that replacement of Ser244 with alanine (S244A) resulted the equivalent to Arg458 resulted in reduced cellulose
in 7.2% residual esterase and 5.8% residual transferase activity, production by S. typhimurium,27 corroborating the role of
demonstrating that Ser244 is indeed part of the active site and is this conserved Arg residue in substrate binding and catalysis
required for full activity. rather than Zn2+ binding.
To our surprise, replacement of Tyr277 with Phe (Y277F) Molecular Determinants. To assess if the losses of
produced an enzyme variant with greater esterase activity and activity we observed for substitutions of the Zn2+-binding
84% residual transferase activity compared to wild-type residues were in fact due to the loss of Zn2+ in the EcBcsGΔN
EcBcsGΔN (Figure 2). This observation did not support our active site or instead due to structural changes in the enzyme,
earlier proposal that Tyr277 may function in cellulose we supplied the metal chelating agent ethylenediaminetetra-
accommodation in the proximity of the Zn2+ ion at the active acetic acid (EDTA) to EcBcsGΔN prior to assaying. We
center.25 An increase in solvent accessibility to the active site observed that treatment with 1 mM EDTA significantly
may account for the improved rate of esterase activity and an reduced the specific esterase activity of EcBcsGΔN to 10.6% of
increased level of competition for transfer to oligosaccharides the untreated activity [Figure 3; Tukey’s test, q(14) = 12.12,
that is reflected by the lower transferase activity. This and p < 0.0001]. A matched enzyme-free control containing 1
hypothesis is further supported by the fact that an equivalent mM EDTA displayed no difference in the rate of p-NPPE
tyrosine residue is not observed in available structures of other turnover, suggesting the effect seen was due to loss of Zn2+ in
pEtN transferases,32,34,35 nor was a tyrosine residue observed the active site. This significant loss of esterase activity agreed
to participate in contact with the lipid-binding site in the with the large reduction in esterase activity observed for the
structure of the full-length pEtN transferase EptA from alanine replacements of Cys243, Glu442, and His443. To our
Neisseria meningitidis [NmEptA, Protein Data Bank (PDB) surprise, however, the loss of transferase activity observed for
entry 5FGN].36 Instead, the residue equivalent to Tyr277 in EDTA treatment was markedly smaller than that observed for
NmEptA, Ser279, appears to shape the lipid-binding pocket at single-amino acid substitutions of the Zn2+-binding residues
3663 https://doi.org/10.1021/acs.biochem.1c00605
Biochemistry 2021, 60, 3659−3669
Biochemistry pubs.acs.org/biochemistry Article

that replacement of the Cys pair with Ala results in a loss of


catalytic activity in vivo.25 Surprisingly, this disulfide bond in
BcsG is not at the conserved location but serves to fix the same
helical element presenting the nucleophilic Ser278. We were
unable to isolate a C290A/C306A variant of EcBcsGΔN, further
suggesting that the disulfide bond is a critical structural feature
of BcsG. To assess the importance of disulfide bond formation
to enzymatic activity in vitro, we introduced the reducing agent
dithiothreitol (DTT) to EcBcsGΔN prior to measurement of
the enzymatic rate. An observed decrease in enzymatic activity,
representing 31.9% of residual esterase activity, was observed
(Figure 3). Surprisingly, however, the measured residual
transferase activity was 72.8% of that of wild-type EcBcsGΔN,
while a matched enzyme-free control showed no differences in
the rate of p-NPPE hydrolysis. These findings are in apparent
disagreement with the loss of function of the disulfide-deficient
EcBcsG variant we reported in vivo. These data may in part be
rationalized by the specific requirements of exopolysaccharide
modification for biofilm formation. For example, the biofilm
Figure 3. Zn2+ binding and disulfide bond formation are essential
polysaccharide poly-β-(1,6)-N-acetyl-D-glucosamine (PNAG)
catalytic features of EcBcsGΔN. Treatment of EcBcsGΔN with the requires only partial deacetylation by the enzyme PgaB or its
metal chelating agent EDTA or the reducing agent DTT results in orthologues for successful biofilm formation in various
impaired enzymatic activities, consistent with impaired biofilm- bacteria.39−41 PNAG found in biofilms has been observed to
forming phenotypes observed in vivo. Asterisks denote significant have approximately 15−20% of the N-acetyl-D-glucosamine
differences from respective controls [Tukey’s multiple comparisons, saccharide units deacetylated, with either more or less
q(14) > 6.5 and p < 0.004 for all comparisons]. extensive deacetylation causing disruption in biofilm for-
mation, probably due to the loss of a distributed cationic
(i.e., 43% residual activity of the wild type for the EDTA- charge on the polymer.40,42,43 Similarly, a loss of EcBcsGΔN
treated form, compared to 2.4%, 2.3%, and 10.4% residual activity of approximately 30% measured in the presence of
transferase activity for alanine replacements of Cys243, Glu442, DTT might be considered trivial for the biological function of
and His443, respectively). Our data suggest either that BcsG has some enzymes; however, this loss may not be enough to
an exceptional affinity for Zn2+, given that 1 mM EDTA is a maintain the 50% level of pEtN modification of D-glucose
standard concentration used to assess the metal dependency of saccharide units that has been observed for pEtN cellulose
enzymes, or that in addition to Zn2+ binding, Cys243, Glu442, biofilms isolated from E. coli or S. enterica.14 Thus, even subtle
and His443 are also catalytically important residues, particularly reductions in the enzymatic rate, such as those observed here,
during the latter step of catalysis when pEtN is transferred to could still plausibly alter the degree of pEtN substitution so
cellulose. In support of this theory, any general mechanism that it is not matched properly to the rate of cellulose
involving Ser278 as the catalytic nucleophile would also require synthesis, thereby disrupting the otherwise uniform charge and
a general acid−base residue to abstract the serine hydroxyl chemistry required for proper biofilm architecture and
proton during the formation of the covalent enzyme assembly in vivo.
intermediate and then subsequently replace it following Donor Co-substrate Preference. Although it has been
transfer of the pEtN group to its ultimate acceptor. Our established that the natural co-substrates are phosphatidyle-
results also mirror the findings that treatment of MCR-1- thanolamine14,27 and bacterial cellulose,14 we demonstrated
expressing cells with 1 mM EDTA reduces but does not that BcsG was not capable of transfer to equivalent linear β-
abolish the colistin MIC to MCR-1 negative levels. Similarly, 1,4-aminosugar polymers (i.e., chitin).25 However, the
replacement of the residue equivalent to Cys243 in MCR-1 sufficiency of BcsG to use alternative phosphatidylethanol-
(Glu246) reduces the colistin MIC to the same extent as the amine mimetics as phospho-donors remains unexplored, which
vector control, which is beyond what is observed for EDTA may be of interest for materials science and development. To
treatment. In the literature, there are examples of cysteine investigate the phospho-donor preference of EcBcsG, we
residues acting as a catalytic base when proximal to a assayed the enzyme in vitro using the commercially available
histidine,37 but this would be unprecedented among pEtN substrate analogue p-nitrophenyl phosphate (p-NPP). Addi-
family members. Indeed, cysteine is an objectively poorer tionally, using a synthetic approach similar to that of p-
acid−base catalyst, and the Glu442 residue seems intuitively NPPE,25 the related compound p-nitrophenyl phosphopropa-
better suited for this function. However, given that alanine nolamine (p-NPPP) was synthesized in two steps (see the
replacements of both Glu442 and Cys243 are significantly less Supporting Information). BcsG demonstrated detectable
active than either a single replacement of His443 or EDTA- esterase activity on both the minimal substrate p-NPP and
treated enzyme, these data suggest that these two amino acids the extended substrate p-NPPP, although to a lesser extent
and/or His396 (as noted below) are the leading candidates to than the preferred p-NPPE (Figure S1). We repeated these
participate as acid−base catalytic residues. experiments in the presence of cellopentaose as an acceptor
A disulfide bond that fixes the helical element presenting the substrate and observed no significant increase in the rate of
catalytic nucleophile is a conserved and essential feature of turnover of p-NPP or p-NPPP that would suggest catalytic
pEtN transferases.25,38 We previously showed that BcsG transfer of the ester-linked phosphate or phosphopropanol-
contains a disulfide bond between Cys290 and Cys306, and amine. Analysis of the enzymatic products by LC-MS
3664 https://doi.org/10.1021/acs.biochem.1c00605
Biochemistry 2021, 60, 3659−3669
Biochemistry pubs.acs.org/biochemistry Article

corroborated these results, as no evidence of phospho- or active site residues can be implied from the crystal structure
phosphopropanolamine cellulose was detected, thereby in- and from trapping of the covalent enzyme intermediate, the
dicating that BcsG is exclusively capable of transferring the catalytic importance of Ser244 and His396 was not obviated from
pEtN functional group under these conditions but also further prior data, notably for His396 that may plausibly serve as the
confirming it as the true biological substrate. complementary base for the reaction that has not otherwise
Steady-State Esterase Kinetics. To investigate the been identified.25,27 Although our specific activity data (noted
kinetic parameters of EcBcsGΔN, we measured the steady- above) suggest that these variants of EcBcsGΔN display
state esterase rate and calculated the specific activity at varied impaired ability to transfer pEtN to cellulosic substrates, it
concentrations of p-NPPE (0−14.5 mM) without the addition remained unclear if these residues are exclusively involved in
of an acceptor co-substrate (Figure 4A). The resulting data the latter half of the mechanism whereby pEtN is transferred to
cellulose, or if their roles might extend to cleavage of the pEtN
substrate. Therefore, we assessed the steady-state kinetics of
the S244A and H396A EcBcsGΔN variants in the presence of 3
mM cellopentaose.
The alanine replacement of Ser244 demonstrated a decrease
in turnover number (kcat) and a greatly reduced affinity for p-
NPPE (KM increases >3-fold) in the presence of acceptor,
suggesting this residue plays a role in both steps of the
proposed mechanism (Figure 4A and Table 1). While other
biochemically characterized pEtN transferases possess a
conserved Thr residue that is equivalent to Ser244 in BcsG,
none of the previous work has kinetically explored the role of
this residue in catalysis. Thus, the results presented herein,
especially with respect to donor binding (as reflected by KM),
have importantly broadened our understanding of the catalytic
mechanism of this class of enzymes, which may aid in the
development of inhibitors. Other biochemically characterized
pEtN transferases, including MCR-1, NmEptA, and CjEptC,
each possess a conserved histidine equivalent to the BcsG
His396 residue.32,34,35 Although this His residue has been
shown to be essential for at least MCR-1,32 it has also been
proposed to coordinate a second Zn2+ ion at the active site,
which is apparently required for catalysis (Figure 5C).32,44 In
addition to our specific activity results (Figure 2), our kinetic
analysis of His396 in EcBcsGΔN confirmed the importance of
this residue in catalysis and expanded our knowledge of its role
in the reaction mechanism. Specifically, H396A demonstrated
a lower turnover number (kcat decreases 2.6-fold) and a modest
reduction in affinity for p-NPPE in the presence of
Figure 4. Steady-state kinetics of EcBcsGΔN. (A) Steady-state esterase cellopentaose, thereby indicating that this residue is important
(WT) and transferase (mutant) kinetics using p-NPPE as the
phosphoethanolamine donor and 3 mM cellopentaose as the acceptor
in both steps of the proposed mechanism (Figure 4A and
co-substrate. (B) Steady-state transferase kinetics using 7 mM p- Table 1). However, our group and others27 were not able to
NPPE as the donor and cellooligosaccharides with DP values of 4−6 support the existence of a second Zn2+ ion in the active site of
as the acceptor co-substrates. BcsG, suggesting the role of His396 is truly in catalysis rather
than secondary metal binding among the pEtN transferases. In
were fit in agreement with the Michaelis−Menten model, with the structural models of BcsG, His396 is positioned above the
an R2 value of 0.9624 (Table 1). We calculated from our data zinc ion (approximately 5 Å) and in some models is within
an apparent KM of 2.40 ± 0.28 mM, a maximal rate (Vmax) of hydrogen bonding distance of the main chain carbonyl of
2.20 × 10−2 nmol s−1, and a kcat of 4.34 × 10−7 ± 1.30 × 10−8 Asp397, as well as a water molecule when present (Figure 5A).
s−1. The derived catalytic efficiency kcat/KM was 1.81 × 10−4 ± These results are consistent with the distances noted for the
2.81 × 10−5 M−1 s−1. equivalents to His396 and Asp397 in ICRMc (His429 and Gly430,
Ser244 and His396 Participate in Both Mechanistic respectively). However, in ICRMc, the water molecule is instead
Steps but Not in Zn2+ Binding. While the roles of the other occupied by the phosphoryl group of the bound phosphoe-

Table 1. Measured Michaelis−Menten Parameters for EcBcsG and Its Variants


parameter WT S244Aa H396Aa
kcat (s−1) 4.34 × 10−7 ± 1.30 × 10−8 2.98 × 10−7 ± 2.45 × 10−8 1.64 × 10−7 ± 9.58 × 10−9
KM (mM) 2.40 ± 0.28 7.85 ± 1.30 3.20 ± 0.56
Vmax (nmol s−1) 0.022 0.015 0.008
kcat/KM (M−1 s−1) 1.81 × 10−4 ± 2.81 × 10−5 3.80 × 10−5 ± 9.42 × 10−6 5.13 × 10−5 ± 8.98 × 10−6

a
With 3 mM cellopenatose co-substrate.

3665 https://doi.org/10.1021/acs.biochem.1c00605
Biochemistry 2021, 60, 3659−3669
Biochemistry pubs.acs.org/biochemistry Article

Figure 5. Conserved His396 functions in catalysis and not secondary metal binding. The active sites of (A) BcsG (PDB entry 6PD0), (B) IcrMc
(PDB entry 6BND), and (C) MCR-1 (PDB entry 5LRM) display a conserved His residue in the proximity of the active site. Although involved in
binding a second Zn2+ ion in some structures of MCR-1, this His residue is within H-bonding distance of the adjacent main chain carbonyl in each
of the three structures and likely functions in proton relay, supported by kinetic studies of the BcsG H396A variant.

Table 2. Measured Michaelis−Menten Parameters for EcBcsG Acceptor Co-substrates


parameter cellotetraose cellopentaose cellohexaose
kcat (s−1) 2.00 × 10−6 ± 1.42 × 10−7 3.80 × 10−6 ± 2.99 × 10−7 3.27 × 10−6 ± 1.97 × 10−7
KM (mM) 3.20 ± 0.47 3.08 ± 0.53 1.76 ± 0.21
Vmax (nmol s−1) 0.10 0.19 0.16
kcat/KM (M−1 s−1) (6.24 ± 1.41) × 10−4 1.23 × 10−3 ± 2.96 × 10−4 1.86 × 10−3 ± 3.18 × 10−4

thanolamine (PDB entry 6BND), which is further coordinated cellohexaose, we measured an order of magnitude increase in
from the opposite side by the zinc ion (Figure 5B). These catalytic efficiency in the presence of the acceptor (i.e., 1.86 ×
findings suggest that His396 is also positioned to be part of a 10−3 ± 3.18 × 10−4 compared to 1.81 × 10−4 ± 2.81 × 10−5
similar proton relay that is important for catalysis in BcsG. M−1 s−1 without an acceptor), suggesting that the enzyme is far
While further work to explore this role is warranted, the more active as a transferase than as an esterase, at least with the
essential nature of His396 in the previously proposed ping-pong artificial substrate p-NPPE.
bibi mechanism25 has been further evidenced by our kinetic Measurement of the kinetic parameters of other pEtN
and structural results. transferases has not been reported at present, likely due to the
Steady-State Transferase Kinetics. We previously absence of commercially available substrate analogues with
reported that introducing high relative concentrations of which to do so. Accordingly, a comparison between the kinetic
cellooligosaccharides with degrees of polymerization (DP) parameters of EcBcsGΔN measured here and those of other
from 4 to 6 significantly increased the specific activity of pEtN transferases is not possible. However, the measured kcat
EcBcsGΔN in a length-dependent manner.25 Under these and KM values we report are poor in comparison to those of
conditions, pEtN-modified cellooligosaccharides accumulated other biochemically characterized enzymes of virtually any
over time, demonstrating EcBcsGΔN is capable of transferase identity; a majority of enzymes have reported kcat (s−1) values
activity in vitro. However, the substrate preference of of >1 and catalytic efficiencies of at least 1.0 × 103 M−1 s−1.45
EcBcsGΔN using these cellooligosaccharides remains to be However, when examined against characterized enzymes that
seen. To that end, we performed steady-state kinetics using modify other bacterial exopolysaccharides, the BcsG values are
fixed and saturating concentrations of p-NPPE and varied the more typical. For example, the PNAG de-N-acetylases PgaB
concentration of cellotetraose, cellopentaose, and cellohexaose from E. coli and IcaB from Staphylococcus epidermidis have low
to understand EcBcsGΔN acceptor preference (Table 2). reported kcat (PgaB, 0.0013 s−1; IcaB, 0.0007 s−1) and kcat/KM
As our previous results suggested, EcBcsGΔN displays an values (PgaB, 0.26 M−1 s−1; IcaB, 0.03 M−1 s−1), using a similar
increasing affinity for cellooligosaccharides with an increasing assay design with a synthetic fluorogenic substrate 4-
DP, although due to limited solubility, cellohexaose represents methylumbelliferyl acetate.43,46 In addition to PNAG, the
the highest DP that can be assayed using our experimental alginate epimerase AlgG from Pseudomonas aeruginosa (kcat/KM
design (Figure 4B). We measured apparent KM values of 3.20 ∼ 0.02 M−1 s−1) is also consistent with EcBcsGΔN.47 It is worth
± 0.47, 3.08 ± 0.53, and 1.76 ± 0.21 mM for noting that the analyses mentioned above were all performed
celloligosaccharides with DP values of 4, 5, and 6, respectively with artificial co-substrates that are not present in a cellular
(Figure 4B and Table 2). A similar trend in calculated kcat context, which may also explain in part the catalytic
values was observed with values of 2.00 × 10−6 ± 1.41 × 10−7, inefficiencies of these enzymes to act upon them. Regardless,
3.80 × 10−6 ± 3.00 × 10−7, and 3.27 × 10−6 ± 1.97 × 10−7 s−1, the generally low turnover numbers found in exopolysacchar-
respectively, and the resulting catalytic efficiencies (kcat/KM) ide-modifying enzymes have been rationalized previously by
were then calculated to be (6.24 ± 1.41) × 10−4, 1.23 × 10−3 the partial degree of modification achieved in the mature
± 2.96 × 10−4, and 1.86 × 10−3 ± 3.18 × 10−4 M−1 s−1 for DP polymer,39,43 which we mentioned above. Accordingly, only
values of 4, 5, and 6, respectively (Table 2). In the case of 50% of the saccharide units are C6-pEtN substituted in the
3666 https://doi.org/10.1021/acs.biochem.1c00605
Biochemistry 2021, 60, 3659−3669
Biochemistry pubs.acs.org/biochemistry Article

Figure 6. Proposed mechanism of BcsG as a pEtN transferase. In the first mechanistic step, the hydroxyl proton is abstracted from Ser278 by His443
and Glu442 (2). Protonation and collapse of the resulting oxyanion intermediate are plausibly achieved by His396 and Cys243, resulting in the
covalent enzyme intermediate and release of the diacylglycerol co-product (3). In the second mechanistic step, the C6 hydroxyl of cellulose is
deprotonated, possibly by Cys243, and attacks the phosphoryl-Ser278 intermediate. The resulting Ser278 oxyanion is protonated, likely by the adjacent
His443, followed by release of the pEtN cellulose co-product (4). Our enzymatic data also suggest that Ser244 and Arg458 serve as important polar
contacts during both mechanistic steps, especially Arg458 in the second mechanistic step.

mature pEtN cellulose polymer,14 which is in better agreement our data suggest a shared mono-Zn2+ mechanism for the pEtN
with the low EcBcsGΔN turnover number and catalytic transferase family and that BcsG possesses an active site similar
efficiency. to, but distinct from, that of the colistin resistance enzymes,
The rate of cellulose synthesis by BcsAB has been measured, thereby pointing to new roles for the family.
and surprisingly, Omadjela and co-workers reported a rate for On the basis of our current data, we propose a catalytic
BcsAB that is orders of magnitude greater than that of BcsG.18
mechanism for transfer of phosphoethanolamine to cellulose
However, they similarly acknowledge for their in vitro assay, as
we did above for BcsG, that these assays can fail to capture by BcsG that involves activation of Ser278 by abstraction of the
nuances of the biological complexity of the Bcs macrocomplex hydroxyl proton by His443 and Glu442 (Figure 6). Following a
and that the rate of cellulose synthesis that they report for nucleophilic attack of the phosphate, the loss of the
BcsAB is probably much lower in vivo. Direct rate comparisons diacylglycerol product is probably achieved by protonation of
between these disparately designed assays should also be the oxyanion intermediate, stabilized by the Zn2+ ion, possibly
interpreted with caution. However, it should be noted that the through Cys243 and His396. Although a catalytic Cys base is
native E. coli Bcs macrocomplex has been resolved with two unprecedented in the pEtN family, the Glu442/His443 acid−
copies of BcsG associated with a single BcsA catalytic base pair would be too distant to serve this role, separated by
subunit.23 Taken together, these observations may partly at least 5 Å. Our data presented herein support a catalytic role
reconcile the differences in rates between BcsG and BcsA in
for Cys243, and its presence in place of the typical glutamic acid
such a way as to generate the level of pEtN modification
required for biofilm formation in vivo. may be rationalized by the lower overall catalytic efficiency


required of BcsG in general. The essential substrate
CONCLUDING REMARKS recognition and catalytic features observed thus provide a
We have demonstrated that full EcBcsGΔN activity in vitro is molecular level understanding of BcsG and will facilitate the
dependent upon Zn 2+ binding, the putative catalytic design of biofilm inhibitors targeting this enzyme.


nucleophile and acid−base arrangement (Ser278/Cys243/
His396), the previously unidentified active site residues Arg458 ASSOCIATED CONTENT
and Ser244, and proper formation of a disulfide bond between
Cys290 and Cys306. A kinetic analysis of EcBcsGΔN demon- *
sı Supporting Information

strated that it is specific for the transfer of phosphoethanol- The Supporting Information is available free of charge at
amine over other donor substrates tested. Our data also https://pubs.acs.org/doi/10.1021/acs.biochem.1c00605.
showed that EcBcsG shares equivalent catalytic residues with
A complete list of plasmids, primers, and strains used in
the characterized colistin resistance enzymes, although our
this study, the synthesis of p-NPPP, and representative
model does not support a two-Zn2+ model for catalysis that 1
was suggested for one other group member, MCR-1. Instead, H, 13C, and 31P NMR shifts for p-NPPP (PDF)

3667 https://doi.org/10.1021/acs.biochem.1c00605
Biochemistry 2021, 60, 3659−3669
Biochemistry


pubs.acs.org/biochemistry Article

AUTHOR INFORMATION (2) Flemming, H.-C.; Wingender, J.; Szewzyk, U.; Steinberg, P.;
Rice, S. A.; Kjelleberg, S. Biofilms: An Emergent Form of Bacterial
Corresponding Author Life. Nat. Rev. Microbiol. 2016, 14 (9), 563−575.
Joel T. Weadge − Department of Biology, Wilfrid Laurier (3) Marsh, P. Dental Plaque as a Microbial Biofilm Dental Plaque −
University, Waterloo, ON N2L3C5, Canada; orcid.org/ Existing Perspective. Caries Res. 2004, 38, 204−211.
0000-0002-8020-0015; Phone: 519-884-0710, ext. 2161; (4) Flickinger, S. T.; Copeland, M. F.; Downes, E. M.; Braasch, A.
Email: jweadge@wlu.ca T.; Tuson, H. H.; Eun, Y.-J.; Weibel, D. B. Quorum Sensing between
Pseudomonas Aeruginosa Biofilms Accelerates Cell Growth. J. Am.
Authors Chem. Soc. 2011, 133 (15), 5966−5975.
Alexander C. Anderson − Department of Biology, Wilfrid (5) Fux, C. A.; Costerton, J. W.; Stewart, P. S.; Stoodley, P. Survival
Laurier University, Waterloo, ON N2L3C5, Canada; Strategies of Infectious Biofilms. Trends Microbiol. 2005, 13 (1), 34−
Present Address: A.C.A.: Department of Molecular and 40.
Cellular Biology, University of Guelph, Guelph, ON (6) Römling, U.; Bokranz, W.; Rabsch, W.; Zogaj, X.; Nimtz, M.;
N1G2W1, Canada; orcid.org/0000-0002-1870-2903 Tschäpe, H. Occurrence and Regulation of the Multicellular
Alysha J. N. Burnett − Department of Biology, Wilfrid Laurier Morphotype in Salmonella Serovars Important in Human Disease.
University, Waterloo, ON N2L3C5, Canada Int. J. Med. Microbiol. 2003, 293 (4), 273−285.
(7) Saldaña, Z.; Xicohtencatl-Cortes, J.; Avelino, F.; Phillips, A. D.;
Shirley Constable − Department of Biology, Wilfrid Laurier
Kaper, J. B.; Puente, J. L.; Girón, J. A. Synergistic Role of Curli and
University, Waterloo, ON N2L3C5, Canada Cellulose in Cell Adherence and Biofilm Formation of Attaching and
Lana Hiscock − Department of Biology and Department of Effacing Escherichia Coli and Identification of Fis as a Negative
Chemistry & Biochemistry, Wilfrid Laurier University, Regulator of Curli. Environ. Microbiol. 2009, 11 (4), 992−1006.
Waterloo, ON N2L3C5, Canada (8) Spiers, A. J.; Bohannon, J.; Gehrig, S. M.; Rainey, P. B. Biofilm
Kenneth E. Maly − Department of Chemistry & Biochemistry, Formation at the Air-Liquid Interface by the Pseudomonas
Wilfrid Laurier University, Waterloo, ON N2L3C5, Fluorescens SBW25 Wrinkly Spreader Requires an Acetylated Form
Canada; orcid.org/0000-0002-3695-4995 of Cellulose. Mol. Microbiol. 2003, 50 (1), 15−27.
Complete contact information is available at: (9) Hu, L.; Grim, C. J.; Franco, A. A.; Jarvis, K. G.; Sathyamoorthy,
V.; Kothary, M. H.; McCardell, B. A.; Tall, B. D. Analysis of the
https://pubs.acs.org/10.1021/acs.biochem.1c00605
Cellulose Synthase Operon Genes, BcsA, BcsB, and BcsC in
Cronobacter Species: Prevalence among Species and Their Roles in
Author Contributions Biofilm Formation and Cell-Cell Aggregation. Food Microbiol. 2015,
J.T.W. and A.C.A. conceived the research and acquired 52, 97−105.
funding. A.C.A. and A.J.N.B. designed and carried out the (10) Ross, P.; Mayer, R.; Benziman, M. Cellulose Biosynthesis and
enzymology experiments and prepared the manuscript. L.H. Function in Bacteria. Microbiol. Rev. 1991, 55 (1), 35−58.
and K.E.M. carried out the synthesis and validation of p-NPPE (11) Canale-Parola, E.; Borasky, R.; Wolfe, R. S. Studies on Sarcina
and p-NPPP. S.C. carried out some of the mutant-based Ventriculi III. Localization of Cellulose. J. Bacteriol. 1961, 81 (2),
enzymology. J.T.W., A.C.A., A.J.N.B., S.C., L.H., and K.E.M. 311−318.
were involved in manuscript review and editing. All authors (12) Scott, W.; Lowrance, B.; Anderson, A. C.; Weadge, J. T.
have given approval to the final version of the manuscript. Identification of the Clostridial Cellulose Synthase and Character-
ization of the Cognate Glycosyl Hydrolase, CcsZ. PLoS One 2020, 15
Funding (12), e0242686.
The authors acknowledge the support of the Natural Sciences (13) Whitfield, G. B.; Marmont, L. S.; Howell, P. L. Enzymatic
and Engineering Research Council of Canada (NSERC) in the Modifications of Exopolysaccharides Enhance Bacterial Persistence.
form of a grant to J.T.W. (229971) and a graduate scholarship Front. Microbiol. 2015, 6, 471.
(PGS-D) to A.C.A. A.C.A. and A.J.N.B. were also supported by (14) Thongsomboon, W.; Serra, D. O.; Possling, A.;
Ontario Graduate Scholarships via Wilfrid Laurier University. Hadjineophytou, C.; Hengge, R.; Cegelski, L. Phosphoethanolamine
Cellulose: A Naturally Produced Chemically Modified Cellulose.
Notes
Science (Washington, DC, U. S.) 2018, 359 (6373), 334−338.
The authors declare no competing financial interest.


(15) Whitfield, G. B.; Marmont, L. S.; Bundalovic-Torma, C.; Razvi,
E.; Roach, E. J.; Khursigara, C. M.; Parkinson, J.; Howell, P. L.
ACKNOWLEDGMENTS Discovery and Characterization of a Gram- Positive Pel Poly-
The authors thank D. Brewer and A. Charchoglyan at the saccharide Biosynthetic Gene Cluster. PLoS Pathog. 2020, 16 (4),
University of Guelph for expert technical assistance with mass e1008281.
spectrometry experiments. (16) Römling, U. Molecular Biology of Cellulose Production in


Bacteria. Res. Microbiol. 2002, 153 (4), 205−212.
(17) Römling, U.; Galperin, M. Y. Bacterial Cellulose Biosynthesis:
ABBREVIATIONS Diversity of Operons, Subunits, Products, and Functions. Trends
pEtN, phosphoethanolamine; c-di-GMP, cyclic dimeric Microbiol. 2015, 23 (9), 545−557.
guanosine monophosphate; TPR, tetratricopeptide repeat; p- (18) Omadjela, O.; Narahari, A.; Strumillo, J.; Mélida, H.; Mazur,
NPPE, p-nitrophenyl phosphoethanolamine; p-NPPP, p-nitro- O.; Bulone, V.; Zimmer, J. BcsA and BcsB Form the Catalytically
phenyl phosphopropanolamine; p-NPP, p-nitrophenyl phos- Active Core of Bacterial Cellulose Synthase Sufficient for in Vitro
phate; Ni-NTA, nickel-nitrilotriacetic acid; IPTG, isopropyl β- Cellulose Synthesis. Proc. Natl. Acad. Sci. U. S. A. 2013, 110 (44),
D-thiogalactopyranoside; EDTA, ethylenediaminetetraacetic
17856−17861.
(19) Morgan, J. L. W.; McNamara, J. T.; Zimmer, J. Mechanism of
acid; DTT, dithiothreitol; DP, degree of polymerization.


Activation of Bacterial Cellulose Synthase by Cyclic Di-GMP. Nat.
Struct. Mol. Biol. 2014, 21 (5), 489−496.
REFERENCES (20) Morgan, J. L. W.; Strumillo, J.; Zimmer, J. Crystallographic
(1) Mah, T.-F. C.; O’Toole, G. A. Mechanisms of Biofilm Resistance Snapshot of Cellulose Synthesis and Membrane Translocation. Nature
to Antimicrobial Agents. Trends Microbiol. 2001, 9 (1), 34−39. 2013, 493 (7431), 181−186.

3668 https://doi.org/10.1021/acs.biochem.1c00605
Biochemistry 2021, 60, 3659−3669
Biochemistry pubs.acs.org/biochemistry Article

(21) Whitfield, C.; Mainprize, I. L. TPR Motifs: Hallmarks of a New into Polymyxin Resistance Mediated by EptC Originating from
Polysaccharide Export Scaffold. Structure 2010, 18 (2), 151−153. Escherichia Coli. FEBS J. 2019, 286 (4), 750−764.
(22) Low, K. E.; Howell, P. L. Gram-Negative Synthase-Dependent (39) Arciola, C. R.; Campoccia, D.; Ravaioli, S.; Montanaro, L.
Exopolysaccharide Biosynthetic Machines. Curr. Opin. Struct. Biol. Polysaccharide Intercellular Adhesin in Biofilm: Structural and
2018, 53, 32−44. Regulatory Aspects. Front. Cell. Infect. Microbiol. 2015, 5, 7.
(23) Acheson, J. F.; Derewenda, Z. S.; Zimmer, J. Architecture of the (40) Little, D. J.; Milek, S.; Bamford, N. C.; Ganguly, T.;
Cellulose Synthase Outer Membrane Channel and Its Association Difrancesco, B. R.; Nitz, M.; Deora, R.; Howell, P. L. The Protein
with the Periplasmic TPR Domain. Structure 2019, 27 (12), 1855− BpsB Is a Poly-β-1,6-N-Acetyl-D-Glucosamine Deacetylase Required
1861.e3. for Biofilm Formation in Bordetella Bronchiseptica. J. Biol. Chem.
(24) Nojima, S.; Fujishima, A.; Kato, K.; Ohuchi, K.; Shimizu, N.; 2015, 290 (37), 22827−22840.
Yonezawa, K.; Tajima, K.; Yao, M. Crystal Structure of the Flexible (41) Little, D. J.; Poloczek, J.; Whitney, J. C.; Robinson, H.; Nitz,
Tandem Repeat Domain of Bacterial Cellulose Synthesis Subunit C. M.; Howell, P. L. The Structure- and Metal-Dependent Activity of
Sci. Rep. 2017, 7 (1), 13018. Escherichia Coli PgaB Provides Insight into the Partial de-N-
(25) Anderson, A. C.; Burnett, A. J. N.; Hiscock, L.; Maly, K. E.; Acetylation of Poly-β-1,6-N-Acetyl- D-Glucosamine. J. Biol. Chem.
Weadge, J. T. The Escherichia Coli Cellulose Synthase Subunit G 2012, 287 (37), 31126−31137.
(BcsG) Is a Zn2+-Dependent Phosphoethanolamine Transferase. J. (42) Little, D. J.; Li, G.; Ing, C.; DiFrancesco, B. R.; Bamford, N. C.;
Biol. Chem. 2020, 295 (18), 6225−6235. Robinson, H.; Nitz, M.; Pomes, R.; Howell, P. L. Modification and
(26) Mazur, O.; Zimmer, J. Apo- and Cellopentaose-Bound Periplasmic Translocation of the Biofilm Exopolysaccharide Poly-
Structures of the Bacterial Cellulose Synthase Subunit BcsZ. J. Biol. −1,6-N-Acetyl-D-Glucosamine. Proc. Natl. Acad. Sci. U. S. A. 2014,
Chem. 2011, 286 (20), 17601−17606. 111 (30), 11013−11018.
(27) Sun, L.; Vella, P.; Schnell, R.; Polyakova, A.; Bourenkov, G.; Li, (43) Pokrovskaya, V.; Poloczek, J.; Little, D. J.; Griffiths, H.; Howell,
F.; Cimdins, A.; Schneider, T. R.; Lindqvist, Y.; Galperin, M. Y.; P. L.; Nitz, M. Functional Characterization of Staphylococcus
Schneider, G.; Römling, U. Structural and Functional Character- Epidermidis IcaB, a De-N-Acetylase Important for Biofilm Formation.
ization of the BcsG Subunit of the Cellulose Synthase in Salmonella Biochemistry 2013, 52 (32), 5463−5471.
Typhimurium. J. Mol. Biol. 2018, 430 (18), 3170−3189. (44) Suardíaz, R.; Lythell, E.; Hinchliffe, P.; Van Der Kamp, M.;
(28) Fang, X.; Ahmad, I.; Blanka, A.; Schottkowski, M.; Cimdins, A.; Spencer, J.; Fey, N.; Mulholland, A. J. Catalytic Mechanism of the
Galperin, M. Y.; Römling, U.; Gomelsky, M. GIL, a New C-di-GMP- Colistin Resistance Protein MCR-1. Org. Biomol. Chem. 2021, 19,
binding Protein Domain Involved in Regulation of Cellulose 3813−3819.
Synthesis in Enterobacteria. Mol. Microbiol. 2014, 93 (3), 439−452. (45) Bar-Even, A.; Noor, E.; Savir, Y.; Liebermeister, W.; Davidi, D.;
(29) Zouhir, S.; Abidi, W.; Caleechurn, M.; Krasteva, P. V. Structure Tawfik, D. S.; Milo, R. The Moderately Efficient Enzyme:
and Multitasking of the C-Di-GMP-Sensing Cellulose Secretion Evolutionary and Physicochemical Trends Shaping Enzyme Param-
Regulator BcsE. mBio 2020, 11, 4. eters. Biochemistry 2011, 50 (21), 4402−4410.
(30) Acheson, J. F.; Ho, R.; Goularte, N. F.; Cegelski, L.; Zimmer, J. (46) Chibba, A.; Poloczek, J.; Little, D. J.; Howell, P. L.; Nitz, M.
Molecular Organization of the E. Coli Cellulose Synthase Macro- Synthesis and Evaluation of Inhibitors of E. Coli PgaB, a
complex. Nat. Struct. Mol. Biol. 2021, 28, 310. Polysaccharide de-N-Acetylase Involved in Biofilm Formation. Org.
(31) Abidi, W.; Zouhir, S.; Caleechurn, M.; Roche, S.; Krasteva, P. Biomol. Chem. 2012, 10 (35), 7103−7107.
V. Architecture and Regulation of an Enterobacterial Cellulose (47) Jerga, A.; Raychaudhuri, A.; Tipton, P. A. Pseudomonas
Secretion System. Sci. Adv. 2021, 7 (5), 1−16. Aeruginosa C5-Mannuronan Epimerase: Steady-State Kinetics and
(32) Hinchliffe, P.; Yang, Q. E.; Portal, E.; Young, T.; Li, H.; Tooke, Characterization of the Product. Biochemistry 2006, 45 (2), 552−560.
C. L.; Carvalho, M. J.; Paterson, N. G.; Brem, J.; Niumsup, P. R.;
Tansawai, U.; Lei, L.; Li, M.; Shen, Z.; Wang, Y.; Schofield, C. J.;
Mulholland, A. J.; Shen, J.; Fey, N.; Walsh, T. R.; Spencer, J. Insights
into the Mechanistic Basis of Plasmid-Mediated Colistin Resistance
from Crystal Structures of the Catalytic Domain of MCR-1. Sci. Rep.
2017, 7, 39392.
(33) McCall, K. A.; Huang, C.; Fierke, C. A. Function and
Mechanism of Zinc Metalloenzymes. J. Nutr. 2000, 130 (5), 1437S−
1446S.
(34) Fage, C. D.; Brown, D. B.; Boll, J. M.; Keatinge-Clay, A. T.;
Trent, M. S. Crystallographic Study of the Phosphoethanolamine
Transferase EptC Required for Polymyxin Resistance and Motility in
Campylobacter Jejuni. Acta Crystallogr., Sect. D: Biol. Crystallogr. 2014,
70 (10), 2730−2739.
(35) Wanty, C.; Anandan, A.; Piek, S.; Walshe, J.; Ganguly, J.;
Carlson, R. W.; Stubbs, K. A.; Kahler, C. M.; Vrielink, A. The
Structure of the Neisserial Lipooligosaccharide Phosphoethanolamine
Transferase A (LptA) Required for Resistance to Polymyxin. J. Mol.
Biol. 2013, 425 (18), 3389−3402.
(36) Anandan, A.; Evans, G. L.; Condic-Jurkic, K.; O’Mara, M. L.;
John, C. M.; Phillips, N. J.; Jarvis, G. A.; Wills, S. S.; Stubbs, K. A.;
Moraes, I.; Kahler, C. M.; Vrielink, A. Structure of a Lipid A
Phosphoethanolamine Transferase Suggests How Conformational
Changes Govern Substrate Binding. Proc. Natl. Acad. Sci. U. S. A.
2017, 114 (9), 2218−2223.
(37) Moynihan, M. M.; Murkin, A. S. Cysteine Is the General Base
That Serves in Catalysis by Isocitrate Lyase and in Mechanism-Based
Inhibition by 3-Nitropropionate. Biochemistry 2014, 53 (1), 178−187.
(38) Zhao, Y.; Meng, Q.; Lai, Y.; Wang, L.; Zhou, D.; Dou, C.; Gu,
Y.; Nie, C.; Wei, Y.; Cheng, W. Structural and Mechanistic Insights

3669 https://doi.org/10.1021/acs.biochem.1c00605
Biochemistry 2021, 60, 3659−3669

You might also like