1
Genome-wide peptidoglycan profiling of Vibrio cholerae.
1
Sara B. Hernandez
1, a #
, Laura Alvarez
1, #
, Barbara Ritzl-Rinkenberger
1, #
, Bastian
2
Schiffthaler
2
, Alonso R. Serrano
2, b
and Felipe Cava
1, *
.
3
4
# Contributed equally
5
6
Affiliations:
7
1
The laboratory for Molecular Infection Medicine Sweden (MIMS), Department of
8
Molecular Biology, Umeå University, Umeå, Sweden.
9
a
Current address: Instituto de Bioquímica Vegetal y Fotosíntesis, Consejo Superior
10
de Investigaciones Científicas and Universidad de Sevilla, Seville, Spain.
11
2
Umeå Plant Science Centre, Department of Plant Physiology, Umeå University,
12
Umeå, Sweden.
13
b
Current address: Codon Consulting AB, Stockholm, Sweden.
14
15
Lead contact: Felipe Cava
16
Running title: Genome-wide peptidoglycan profiling
17
Keywords: cell wall, peptidoglycan, high throughput, Vibrio cholerae, penicillin
18
binding proteins.
19
.CC-BY-NC-ND 4.0 International licensemade available under a
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is
The copyright holder for this preprintthis version posted August 26, 2022. ; https://doi.org/10.1101/2022.08.25.505259doi: bioRxiv preprint
2
SUMMARY
20
Most bacteria cells are protected by a peptidoglycan cell wall. Defining the chemical
21
structure of the peptidoglycan has been instrumental to characterize cell wall
22
associated proteins and to illuminate the mode of action of cell wall-acting antibiotics.
23
However, a major roadblock for a comprehensive understanding of peptidoglycan
24
homeostasis has been the lack of methods to conduct large-scale, systematic
25
studies. Here we have developed and applied an innovative high throughput
26
peptidoglycan analytical pipeline to analyze the entire non-essential, arrayed mutant
27
library of Vibrio cholerae. The unprecedented breadth of these analyses revealed
28
that peptidoglycan homeostasis is preserved by a large percentage of the genome
29
organized in complex networks that functionally link peptidoglycan features with
30
genetic determinants. As an example, we discovered a novel bifunctional penicillin-
31
binding protein in V. cholerae. Collectively, genome-wide peptidoglycan profiling
32
provides a fast, easy, and unbiased method for systematic identification of the
33
genetic determinants of peptidoglycan synthesis and remodeling.
34
.CC-BY-NC-ND 4.0 International licensemade available under a
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is
The copyright holder for this preprintthis version posted August 26, 2022. ; https://doi.org/10.1101/2022.08.25.505259doi: bioRxiv preprint
3
INTRODUCTION
35
Most bacteria present a strong, yet elastic, cell wall surrounding the cytosolic
36
membrane that maintains cell turgor, determines the bacterial shape and serves as
37
scaffold for the attachment of other cellular components (Dufresne and Paradis-
38
Bleau, 2015). The main component of the bacterial cell wall is the peptidoglycan
39
(PG), a polymer of glycan chains of the alternating N-acetyl-glucosamine (NAG) and
40
N-acetyl-muramic acid (NAM) disaccharide that are crosslinked through stem
41
peptides. PG composition and structure can vary between bacterial species and also
42
in response to environmental changes (i.e., growth conditions) (Vollmer et al., 2008).
43
PG is a unique and essential component in bacteria, and hence the enzymes that
44
synthesize and remodel this polymer are commonly used as antibacterial targets.
45
Consequently, extensive research has been performed during decades to provide a
46
more comprehensive understanding of the genetic and environmental factors that
47
govern PG homeostasis (Dorr, 2021; Hernandez and Cava, 2021; Kumar et al.,
48
2022; Porfirio et al., 2019; Rohs and Bernhardt, 2021). Unfortunately, conventional
49
methods for the fine analysis of the chemical structure of the PG are tedious, time-
50
consuming and require a significant culture size as starting material, making them
51
largely incompatible with large‐scale, systematic screenings.
52
Current knowledge about the bacterial PG chemical structure has been mainly
53
obtained by analytical methods proposed by Glauner et al. more than 30 years ago
54
(Glauner et al., 1988), which involves: purification of the mature PG sacculus,
55
removal of proteins covalently linked to the PG, digestion with muramidase (i.e.,
56
lysozyme) to generate soluble disaccharide peptides (muropeptides) and separation
57
of the muropeptides by liquid chromatography (LC) (Alvarez et al., 2016; Desmarais
58
.CC-BY-NC-ND 4.0 International licensemade available under a
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is
The copyright holder for this preprintthis version posted August 26, 2022. ; https://doi.org/10.1101/2022.08.25.505259doi: bioRxiv preprint
4
et al., 2013). These analyses have significantly improved during the last decade
59
thanks to the use of a more advanced ultra-performance liquid chromatography
60
(UPLC) technology replacing high pressure liquid chromatography (HPLC) systems,
61
and the use of in-line LC-mass spectrometry (MS) systems for PG analysis (Alvarez
62
et al., 2016; Bern et al., 2017; Desmarais et al., 2013; Kuhner et al., 2014; Patel et
63
al., 2021). Additionally, the PG isolation protocol has also been simplified and
64
accelerated by replacing the use of the ultra-centrifuge by bench centrifuges, and the
65
reduction or removal of detergents used during sacculi isolation (Alvarez et al., 2016;
66
Kuhner et al., 2014). Yet, current protocols are still far from being truly high
67
throughput to be used in large-scale studies.
68
Here, we have developed a high throughput (HT) method for PG isolation and LC-
69
based analysis appropriate for hundreds to thousands of samples of both Gram-
70
negative and Gram-positive bacteria. This approach has empowered an
71
unprecedented breadth and detail in PG homeostasis genetic determination and PG
72
functional networks. As a proof of concept, we have screened the muropeptide
73
profiles of the non-redundant mutant library of the Gram-negative pathogen Vibrio
74
cholerae. V. cholerae genome-wide PG profiling revealed a much broader
75
interrelation between PG homeostasis and other cellular processes/pathways than
76
anticipated. Paradoxically, multiple mutants in cell wall genes did not show significant
77
changes in the chemical structure of the PG thus underscoring the existence of
78
functional redundancies and/or conditional determinants. Correlative analyses have
79
allowed to globally visualize the degree of interdependence between PG features as
80
well as potential compensatory mechanisms.
81
In-depth scrutiny of the V. cholerae genome-wide PG profiling uncovered vc1321.
82
Despite its annotation as a hypothetical protein and the apparent lack of phenotype,
83
.CC-BY-NC-ND 4.0 International licensemade available under a
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is
The copyright holder for this preprintthis version posted August 26, 2022. ; https://doi.org/10.1101/2022.08.25.505259doi: bioRxiv preprint
5
our analysis revealed that VC1321 is a novel high molecular weight penicillin-binding
84
protein (PBP1V, for Penicillin Binding Protein 1 of Vibrio) that contributes to V.
85
cholerae PG amount and crosslink. Collectively, these results highlight the capability
86
of our innovative HT PG profiling method to perform genome-wide PG analysis to
87
overcome phenotypic and genomic annotation hurdles towards unveiling the full
88
repertoire of cell wall determinants in bacteria.
89
.CC-BY-NC-ND 4.0 International licensemade available under a
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is
The copyright holder for this preprintthis version posted August 26, 2022. ; https://doi.org/10.1101/2022.08.25.505259doi: bioRxiv preprint
6
RESULTS
90
Development of a high throughput method for peptidoglycan isolation
91
The most commonly used PG isolation protocols were developed in the 60’s
92
(Kolenbrander and Ensign, 1968; Porfirio et al., 2019). These methods capitalize on
93
the separation of SDS-insoluble PG (i.e., murein) from the rest of the cell
94
components and require an individual processing of the samples incompatible with
95
large-scale analyses (Fig. 1a,b). First, the bacterial suspension is added dropwise
96
into boiling SDS and stirred for at least one hour to guarantee the complete lysis.
97
Prior to the enzymatic treatment that renders the soluble muropeptides from the
98
macromolecular murein, the detergent is washed away by repeated
99
ultracentrifugation. To overcome these roadblocks, we explored alternative methods
100
compatible with HT sample processing (Supplementary Fig. 1a). To this end, we
101
boiled bacterial cells with detergent using different heat transmission systems: water
102
bath, in a heat block (dry) or autoclave. Autoclaving performed the best compared to
103
the other methods, yielding the highest PG amount from the same starting material.
104
Since autoclaving can be easily scaled up to 96-well plates, we selected this
105
approach for the first step.
106
Removing the SDS by series of ultracentrifugation steps is time consuming. Further,
107
since a considerable amount of the sample is lost it requires sufficient cells to
108
produce good quality PG analysis. Traditionally, cultures ranging from 10-100 ml (ca.
109
10
11
cells for E. coli) are typically used as starting material. This is an obvious
110
bottleneck that limits not just the scale and throughput of the studies, but also the
111
type of samples that can be successfully processed by this protocol, i.e., low-
112
concentrated environmental or in vivo samples. As purified sacculus retains the
113
.CC-BY-NC-ND 4.0 International licensemade available under a
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is
The copyright holder for this preprintthis version posted August 26, 2022. ; https://doi.org/10.1101/2022.08.25.505259doi: bioRxiv preprint
7
shape of the cell (Formanek and Formanek, 1970) we reasoned that filtration
114
methods used for bacteria could also work for sacculi isolation. Using 0.2 μm pore-
115
size hydrophilic polypropylene (GHP) filters we washed away the SDS from sacculi
116
of different bacteria species (Supplementary Fig. 1b) without significant sample
117
loss, thereby allowing the analysis of smaller starting cultures. In fact, SDS-sacculi
118
from only 500 μl of a saturated bacterial culture (⁓10
9
total cells) was successfully
119
processed.
120
Since the sacculi are retained by the filters we reasoned that the subsequent
121
enzymatic digestion steps could be performed in-a-pot. To this end, after protease
122
treatment, we discarded the flow-through and the retained sacculi were subjected to
123
digestion with muramidase to finally elute the soluble muropeptides. Additional
124
modifications of the original protocol were implemented to further cut down the total
125
processing time: i) the final SDS concentration was lowered (from 5 to 1.5% (w/v)) to
126
reduce the number of washes required, ii) the time of the enzymatic reactions was
127
adjusted to ensure their best performance in the shortest time iii) and the reaction
128
buffer was substituted by water since the muramidase performs equally in both
129
solutions (Supplementary Fig. 1c,d). Finally, we also adapted for automated liquid
130
handling the sodium borohydride treatment that reduces NAM residues to NA-
131
muraminitol to avoid double peaks due to α- and β- anomeric configurations. All in
132
all, these modifications resulted in the development of an isolation procedure
133
compatible with HT methods using multiwell plates and automated pipetting systems
134
to process simultaneously large number of samples more than ten times faster than
135
the traditional methods (Fig. 1b,c). This novel filter-based PG isolation protocol
136
together with the higher speed and sensitivity of UPLC systems (Alvarez et al., 2020;
137
Alvarez et al., 2016) enables the production of large chromatographic datasets for a
138
.CC-BY-NC-ND 4.0 International licensemade available under a
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is
The copyright holder for this preprintthis version posted August 26, 2022. ; https://doi.org/10.1101/2022.08.25.505259doi: bioRxiv preprint
8
great range of bacterial species, including both Gram-negative and Gram-positive
139
bacteria (Fig. 1d), that can be further analyzed using a variety of bioinformatic tools
140
(Kumar et al., 2017).
141
142
Validation of the high throughput muropeptide profiling
143
As proof of concept, we used the model organism V. cholerae. The muropeptides of
144
the wildtype strain were first identified by UPLC-MS (Supplementary Fig. 2a,b). PG
145
analysis of the wildtype strain and over 900 samples grown in duplicates resulted in
146
highly correlated relative abundances for each muropeptide (Pearson correlation
147
score 0.994) (Fig. 2a) and demonstrated the robustness and reproducibility of the
148
method. To assess the potential of the HT PG isolation method in discovering novel
149
genetic determinants involved in PG homeostasis, we first analyzed as controls well-
150
characterized V. cholerae mutants in cell wall biosynthetic genes: the bifunctional
151
PBP1A (Dorr et al., 2014b), the LD-transpeptidase LdtA (Cava et al., 2011) and the
152
DD-endopeptidase DacA1 (Moll et al., 2015), all of them known to present specific
153
structural changes in their PG. To facilitate comparative analysis between samples,
154
the muropeptide profiles (Fig. 2b) were transformed into numeric data. The relative
155
abundance for each muropeptide was calculated by integrating the area under
156
individual peaks divided by the total area of the whole chromatogram. The Log2 fold
157
change (Log2FC) was calculated for every muropeptide using the wildtype values as
158
control (Supplementary Fig. 2c). Replicas of each strain properly clustered together
159
and separated from the rest in a principal component analysis (PCA) (Fig. 2c).
160
Expected differences between the wildtype PG and that of the three mutant strains
161
were clearly highlighted: the PBP1A
-
mutant showed a defect in PG amount (Fig.
162
2d); LdtA
-
a decrease of LD-crosslinked muropeptides, and the DacA1
-
mutant an
163
.CC-BY-NC-ND 4.0 International licensemade available under a
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is
The copyright holder for this preprintthis version posted August 26, 2022. ; https://doi.org/10.1101/2022.08.25.505259doi: bioRxiv preprint
9
increase in pentapeptide-containing muropeptides (Fig. 2d-f). Variations in the PG
164
profile of the LdtA- and DacA1- mutants, correspond to muropeptide changes of ca.
165
2% of the total area demonstrating that HT PG profiling is reliable to uncover subtle
166
PG variations.
167
168
Genome-wide peptidoglycan profiling of V. cholerae
169
To test the potential of the HT PG isolation method in large-scale PG screenings we
170
processed the whole transposon-mutant library of V. cholerae (Cameron et al., 2008)
171
(Fig. 3a). After discarding inconclusive samples (e.g., mutants that fail to grow), we
172
analysed the PG profile of 3,030 mutants (96% of the library) which altogether
173
generated a comprehensive quantitative dataset for each muropeptide and the main
174
PG-features (Extended Data 1).
175
The comparison of the standard deviation (SD) of the relative abundance of each
176
muropeptide in the library dataset with the variation between replicas demonstrated
177
that, except for peaks of very low abundance (i.e., D43 or T344), technical variability
178
is lower than that between biological samples (i.e., mutants), thus supporting the
179
robustness of our method to identify mutants with altered PG (Supplementary Fig.
180
3a). Additionally, we studied the variability of each muropeptide with respect to their
181
relative abundance (Supplementary Fig. 3b). Most of the muropeptides adjusted to
182
a linear regression model, except for the M2 and D34 peaks, which show a higher
183
variation degree, indicating that homeostasis of these muropeptides in V. cholerae is
184
safeguarded by a high number of genetic determinants.
185
Relative abundance is not a reliable metric for sample discrimination (Fig. 3b and
186
Supplementary Fig. 3c-d). The scree plot and biplot show that the most abundant
187
.CC-BY-NC-ND 4.0 International licensemade available under a
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is
The copyright holder for this preprintthis version posted August 26, 2022. ; https://doi.org/10.1101/2022.08.25.505259doi: bioRxiv preprint
10
muropeptides, M4, D44 and D44
Anh
, are the major contributors to the sample
188
variability while fluctuations in less abundant peaks have a more limited impact. On
189
the contrary, Log2FC provided a deeper statistical analysis where the contribution of
190
all peaks had similar weight to define sample’s variability (Supplementary Fig. 3e).
191
Hierarchical clustering of the Log2FC values was used to identify sets of genes with
192
similar PG profiles that may belong to the same biological pathway (Fig. 3c). We
193
focused on the main PG features: PG amount, type and degree of crosslink, and
194
length of the PG chains. As expected, mutants in genes organized in the same
195
operon (based on multiple RNA-seq datasets) clustered together based on their PG
196
profile (Fig. 3d-f and Extended Data 2). As an example, mutation of the loci vc0975
197
and vc0976 results in a significant decrease in the relative amount of PG and an
198
increase in LD-crosslink (Fig. 3d). These genes encode two conserved predicted
199
proteases (Teoh et al., 2015) whose function has not been described yet. Similar
200
changes in the PG were observed for the vc0377 and vc0378 mutants (Fig. 3e),
201
encoding a CheX-like protein proposed to be involved in motility regulation (Altinoglu
202
et al., 2022) and the zinc uptake regulator Zur (Sheng et al., 2015) respectively,
203
which influences PG endopeptidase activity in V. cholerae (Murphy et al., 2019).
204
Conversely, mutation of vc0694 or vc0693 presented a significant increase of LD-
205
crosslink (Fig. 3f). These encode an uncharacterized two component system that
206
has been previously related to intestinal colonization in infant mice (Cheng et al.,
207
2015).
208
The Log2FC-based PCA for all 20 V. cholerae muropeptides revealed outliers
209
representing those mutants with the most defective PGs at a general level (Fig. 3g).
210
Even when the PCA is a multiparametric spatial classification of the samples, these
211
outliers grouped in 6 major clusters (Fig. 3g-h and Extended Data 3) that shared
212
.CC-BY-NC-ND 4.0 International licensemade available under a
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is
The copyright holder for this preprintthis version posted August 26, 2022. ; https://doi.org/10.1101/2022.08.25.505259doi: bioRxiv preprint
11
variability in specific muropeptides (e.g., M2, M4
M
and D45 muropeptides). The 50
213
outlier candidates presenting the highest PCA scores included genes of a wide
214
variety of COG (SI3) suggesting that multiple cellular pathways preserve bacterial
215
PG homeostasis. As an example, we identified a mutant in vc0245 (Fig. 3h), a gene
216
of unknown function but predicted as an LPS O-antigen polymerase (Manning et al.,
217
1995). This result is in agreement with recent reports pointing the cooperative
218
crosstalk between PG and LPS to maintain envelope integrity (Boll et al., 2016;
219
Goodall et al., 2021; Simpson et al., 2021).
220
221
Genome-wide peptidoglycan profiling reveals correlation between different
222
peptidoglycan features in V. cholerae.
223
Multivariate analysis (i.e., PCA) has low resolution to identify genetic determinants of
224
PG homeostasis because while it identifies major general alterations of cell wall’s
225
chemical structure, it fails to discriminate more subtle changes. To solve this
226
bottleneck, we performed univariate data analyses of single PG-features and
227
observed that mutants in known PG-related loci appeared as outliers in the
228
corresponding analyses (Fig. 4a). For example, mutation of genes involved in PG
229
synthesis are typically characterized by a defective (sedimentable) PG amount.
230
Accordingly, both the mutant in the PG synthase PBP1A (encoded by mrcA), and its
231
cognate regulator CsiV (Dorr et al., 2014a) show up as outliers with low PG amount.
232
Further, the mutant in vc1718, V. cholerae homolog of E. coli’s envelope biogenesis
233
factor ElyC also displayed low PG amount, in agreement with a previous report
234
(Paradis-Bleau et al., 2014).
235
.CC-BY-NC-ND 4.0 International licensemade available under a
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is
The copyright holder for this preprintthis version posted August 26, 2022. ; https://doi.org/10.1101/2022.08.25.505259doi: bioRxiv preprint
12
Additional examples can be found in relation to PG chain length. PG strands
236
terminate in anhydro NAM (Anh) caps generated by the action of lytic
237
transglycosylases (Irazoki et al., 2019). Therefore, quantification of the
238
anhydromuropeptide levels is a proxy to calculate average chain length: high
239
anhydromuropeptide level correlates with shorter PG chains and is indicative of high
240
autolytic activities, while low anhydromuropeptide content is indicative of longer
241
chains typically associated with defects in PG turnover pathways. As expected, most
242
of the mutants in genes encoding lytic transglycosylases presented longer PG chains
243
(Fig. 4a). These results are in agreement with recent studies that demonstrated non-
244
redundant roles for some of these lytic transglycosylases in V. cholerae (Weaver et
245
al., 2022; Weaver et al., 2019).
246
The type and degree of the PG crosslinking is another critical aspect of the
247
functionality of the cell wall. V. cholerae uses two types of crosslinks: the DD and LD
248
type. DD-crosslinks (AKA 4-3 type) are catalyzed by the DD-transpeptidase activity
249
of certain PBPs such as the bifunctional high molecular weight class A PBPs or the
250
monofunctional class B PBPs. Alternatively, the LD-crosslinks (so called 3-3 type)
251
depend on the activity of LD-transpeptidases (Aliashkevich and Cava, 2021). As for
252
most bacteria, V. cholerae primary crosslinking type is DD. This is clearly illustrated
253
in Fig. 4b by the direct correlation between DD-crosslink and total crosslink values
254
across mutants. Consistently, mutations in mrcA (the bifunctional PBP1A) or its
255
activators lpoA or csiV resulted in both decreased PG levels and crosslinkage. As
256
mutually inclusive features, PG amount and DD-crosslinking share most of their
257
outliers. However, there are also a number of exceptions (e.g., elyC mutant in PG
258
amount but not in crosslinking) thereby suggesting the existence of highly complex
259
functional networks governing PG homeostasis.
260
.CC-BY-NC-ND 4.0 International licensemade available under a
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is
The copyright holder for this preprintthis version posted August 26, 2022. ; https://doi.org/10.1101/2022.08.25.505259doi: bioRxiv preprint
13
In relation with LD-crosslinkage, our data show that the mutant lacking the LD-
261
transpeptidase A (LdtA) and its regulator RpoS are deficient in this crosslinkage type
262
(Cava et al., 2011). Also consistent with recent reports (Hernandez et al., 2020), the
263
mutant in the recycling LD-carboxypeptidase LdcV shows elevated LD-crosslink.
264
Interestingly, a mutant in vc1834, which encodes the homolog of E. coli’s coordinator
265
of PG synthesis and outer membrane constriction CpoB (Gray et al., 2015), also
266
showed high levels of LD-crosslink. In E. coli CpoB is associated to PBP1B, its main
267
PG-synthetase, but in V. cholerae PBP1A plays a more prominent role in cell wall
268
synthesis than PBP1B and deletion of PBP1B in V. cholerae has not produced any
269
PG phenotype (Dorr et al., 2014b). In V. cholerae the two bifunctional PBPs PBP1A
270
and PBP1B have been shown to play distinct roles than in E. coli, the PG-phenotype
271
observed here for the CpoB homolog mutant also suggests a different role for this
272
PG-synthesis coordinator in V. cholerae, compared to E. coli.
273
Finally, our genome-wide PG analysis also revealed a clear inverse correlation
274
between chain length and total crosslink (Fig. 4b). This is consistent with previous
275
observations (Quintela et al., 1995) supporting that a higher number of peptide-
276
crosslinks is required to maintain the mesh integrity of sacculi consisting of shorter
277
glycan chains in average. Collectively these data show that univariate analysis
278
complements the discovery potential of the genome-wide PG profiling.
279
The normal distribution of the data (Fig. 4b) classified as outliers those samples that
280
deviate 1, 2 or 3 SD for at least one muropeptide or PG feature. These data revealed
281
an unexpectedly high genomic contribution to PG homeostasis (Supplementary Fig.
282
3g) that spans beyond traditionally implicated cell wall enzymes and regulators.
283
Interestingly, a Venn diagram representing the outliers that deviate more than 1SD
284
revealed that more than half of the identified mutants (57.4%) displays changes in
285
.CC-BY-NC-ND 4.0 International licensemade available under a
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is
The copyright holder for this preprintthis version posted August 26, 2022. ; https://doi.org/10.1101/2022.08.25.505259doi: bioRxiv preprint
14
more than one PG feature. In this line, the aforementioned contribution of DD-
286
crosslink to the total crosslink is supported by the high number of mutants with a
287
phenotype for both features (159), compared to the features alone (27 and 66
288
respectively). Collectively, genome-wide PG analysis permitted to appreciate a high
289
degree of genomic contribution and cell wall structural interdependence.
290
291
Identification of a new high molecular weight penicillin binding protein in V.
292
cholerae
293
To identify biological processes related with the change of the different PG features,
294
we performed gene-ontology (GO) enrichment analyses using the sets of mutants
295
with significantly altered PG features (i.e., low or high PG amount, anhydro
296
muropeptides, and different kind of crosslink) (Supplementary Fig. 4 and Extended
297
Data 4). We focused on the samples with lower PG amount, since determinants of
298
PG synthesis are a common target for antibiotics. The biological processes
299
“peptidoglycan metabolism” and “peptidoglycan-based cell wall biogenesis” appeared
300
significantly enriched (p-val <0.05) (Fig. 5a). In addition to genes encoding known
301
PG-related proteins (e.g., the synthetases PBP1A and PBP1B, or the PG
302
degradative enzymes MltA, MltF and DacA2) (Alvarez et al., 2021), we found
303
vc1321, which was annotated as a hypothetical protein in the KEGG database (Fig.
304
5b). We constructed the deletion mutant Δvc1321 and its complemented strain
305
derivative. Analysis of the PG confirmed that the disruption of this gene not only
306
impairs the PG amount, but also decreases the amount of total crosslink (Fig.
307
5c,d). We confirmed that VC1321 can be specifically labelled by the fluorescent
308
penicillin Bocillin-FL, strongly suggesting that this protein is a new type of class A
309
PBP and so we named it as PBP1V (for Penicillin Binding Protein 1 of Vibrio) (Fig.
310
.CC-BY-NC-ND 4.0 International licensemade available under a
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is
The copyright holder for this preprintthis version posted August 26, 2022. ; https://doi.org/10.1101/2022.08.25.505259doi: bioRxiv preprint
15
5e). Sequence similarity and identity was higher between PBP1A and PBP1B than
311
between these PBPs and PBP1V (Fig. 5f). In silico analysis predicted that PBP1V
312
contains an N-terminal transglycosylase domain (TG) and a putative C-terminal
313
transpeptidase (TP) domain that is split by a region with no homology to previously
314
characterized class A PBPs (Fig. 5f). Based on sequence alignments with other high
315
molecular weight PBPs (Sauvage et al., 2008) (Supplementary Fig. 5), we selected
316
and mutated three residues: the E172 as the putative TG active site; and S504 and
317
S610 as the potential TP active serines in the SxxK and SxN conserved motifs,
318
respectively (Fig. 5f). Mutation of S504 but not of S610 impaired PBP1V binding of
319
Bocillin-FL suggesting that S504 is the TP catalytic serine. Interestingly mutation of
320
E172 also prevented Bocillin-FL binding suggesting a regulatory role of the TG
321
domain in PBP1V transpeptidation activity (Fig. 5g).
322
As PBP1V’s molecular weight was larger than the theoretical value, we hypothesized
323
it could be present in a dimeric form. We noticed that the sequence of PBP1V has
324
two cysteine residues that could be involved in the formation of disulfide bridges
325
between subunits (C139 and C1013). Effectively, V. cholerae wildtype membrane
326
fraction treated with the reducing agent DTT resulted in a band shift which coincided
327
with the monomeric size (116 KDa). This result was further confirmed by mutating
328
any of these cysteines (Fig. 5h). PG analysis of these mutants demonstrated that the
329
oligomeric state of PVP1V does not affect its role in PG synthesis under the condition
330
tested (Supplementary Fig. 6).
331
Interestingly, unlike the PBP1A mutant, the absence of PBP1V did not impair proper
332
morphology or growth under tested conditions (Fig. 5i) reinforcing the idea that
333
genome-wide PG profiling is a powerful tool to identify novel genetic determinants on
334
PG homeostasis, some of which might lack functional annotation or play relevant
335
.CC-BY-NC-ND 4.0 International licensemade available under a
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is
The copyright holder for this preprintthis version posted August 26, 2022. ; https://doi.org/10.1101/2022.08.25.505259doi: bioRxiv preprint
16
biological roles in specific conditions. Finally, while PBP1V is only conserved in V.
336
cholerae and its close relative V. mimicus within the genus of Vibrio, orthologs of this
337
PBP are also encoded by other species frequently associated with marine lifestyles
338
(Fig. 5j).
339
.CC-BY-NC-ND 4.0 International licensemade available under a
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is
The copyright holder for this preprintthis version posted August 26, 2022. ; https://doi.org/10.1101/2022.08.25.505259doi: bioRxiv preprint
17
DISCUSSION
340
We have developed a filtration-based method for PG isolation adapted to 96-well
341
plates and automated pipetting systems that enables HT PG biochemical analyses.
342
We have proved that the new protocol is applicable to different bacterial species and
343
provides PG profiles comparable to those obtained using the traditional procedure.
344
The use of filters drastically reduces the initial number of bacteria required for PG
345
isolation by ca. 2 orders of magnitude compared to the minimum amount required by
346
protocols using ultracentrifugation. Therefore, in addition to enabling larger-scale
347
analyses, this method opens up the possibility to characterize the PG chemical
348
structure of low-PG challenging samples such as intracellular non-proliferating
349
bacteria. Moreover, as filtration prevents PG loss, the limiting factor with this method
350
is no longer the sample size (amount of bacteria) but rather the detection limit of the
351
analytical system. In this regard, the accuracy and sensitivity of current MS has been
352
improving during the last decades and can facilitate the muropeptide characterization
353
of challenging low-abundant PG samples.
354
Developing a HT PG isolation method has been instrumental to analyse the cell wall
355
of the non-redundant mutant library of V. cholerae, i.e., the first genome-wide PG
356
profiling of a bacterium. Analysis of the data confirmed the role of known PG-related
357
genes and established new links between previously unrelated genes with cell wall
358
homeostasis, including genes encoding for proteins with previously described or
359
unknown functions. The unprecedented width of the omic-scale PG profiling has
360
made possible to witness a remarkably high genomic contribution to PG
361
homeostasis. For instance, almost 98% of the mutants of the V. cholerae transposon
362
library present a statistically significant alteration of their PG profile (deviate at least
363
1SD from the average), and more than the 50% of the mutants presented a variation
364
.CC-BY-NC-ND 4.0 International licensemade available under a
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is
The copyright holder for this preprintthis version posted August 26, 2022. ; https://doi.org/10.1101/2022.08.25.505259doi: bioRxiv preprint
18
higher than 2SD of the average of the whole library for at least one muropeptide or a
365
PG-feature) that included loci from all COG categories. These results are in
366
agreement with studies that linked the cell wall with seemingly unrelated
367
pathways/processes (e.g.,(Bancroft et al., 2020; Campbell et al., 2021; Masser et al.,
368
2021)). Conversely, certain mutants in cell wall-annotated genes did not show an
369
altered PG profile suggesting the existence of functional redundancies or condition-
370
specific activities. Future research should uncover the functional relationships within
371
these complex genetic networks and what regulatory rewiring occurs in response to
372
environmental changes.
373
Additionally, we observed that variations in PG features and muropeptides are in
374
general mutually inclusive: i.e., a significant proportion of the mutants present
375
changes in more than one PG feature. This result is consistent with the idea of PG
376
structure being safeguarded by compensatory mechanisms that preserve the cell
377
wall integrity. This balance can be explained following a “substrate-product” logic as
378
it is for the inverse relationship between monomers and dimers or between
379
pentapeptides and tetrapeptides, but often seems to be the result of more complex
380
regulatory associations (e.g., crosslink vs chain length) implicating multiple variables.
381
A main strength of this global study is its potential to identify novel cell wall genetic
382
determinants even when lacking a growth phenotype under the conditions studied.
383
As an example, we discovered PBP1V, a third high molecular weight PBP in V.
384
cholerae which differs from the previously characterized PBP1A and PBP1B of this
385
bacterium and is also not homolog to PBP1C proteins found in other bacteria.
386
Deletion of pbp1V did not produce any growth defect in our analyses, but it
387
decreased the total amount and the degree of crosslink of the PG. Although only the
388
TG domain was predicted for this protein, sequence alignment with other high
389
.CC-BY-NC-ND 4.0 International licensemade available under a
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is
The copyright holder for this preprintthis version posted August 26, 2022. ; https://doi.org/10.1101/2022.08.25.505259doi: bioRxiv preprint
19
molecular weight PBPs suggested a potential TP domain consisting of the canonical
390
active site sequences separated by a 186 amino acids domain that was not
391
previously described in any bi-functional PBPs so far studied. Mutation of the S504
392
residue of PBP1V that forms part of a conserved S*xxK TP motif, prevented binding
393
of PBP1V to Bocillin FL, strongly suggesting that this is the catalytic serine of the TP
394
domain. Surprisingly, mutation of E172, which is not found into a conserved
395
EDxxFxxHxG sequence typical of TG domains, also affected binding of Bocillin FL,
396
suggesting that the TG activity might also influence the TP activity of PBP1V’s
397
transpeptidation, similar as previously reported (Zijderveld et al., 1991). PBP1V was
398
found to form dimers by intermolecular disulfide bonds between cysteines C139 and
399
C1013. Dimerization of high molecular weight PBPs by disulfide bonds has been
400
previously reported as a mechanism of redox control of their activity (Bukowska-
401
Faniband and Hederstedt, 2017). However, the monomeric PBP1V retains its TP
402
activity suggesting that dimerization might rather affect other aspects such as
403
localization or protein interactions.
404
The HT method presented here for PG isolation and genome-wide PG profiling
405
represents a promising tool for discovering new genetic determinants of PG
406
biosynthesis and remodeling (i.e., PBP1V) and their genetic networks, the functional
407
relations between the PG features and the structural constraints that define a
408
“healthy” PG sacculus. By profiling mutant libraries of diverse bacterial species
409
(including some of the most important bacterial pathogens), the publicly available
410
experimental platforms and databases generated will have an even wider impact due
411
to their utility to uncover general and species-specific mechanisms underpinning PG
412
homeostasis. Additionally, by using diverse growth conditions we can learn how
413
bacteria adapt their cell wall to relevant environments (e.g. the host) and which
414
.CC-BY-NC-ND 4.0 International licensemade available under a
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is
The copyright holder for this preprintthis version posted August 26, 2022. ; https://doi.org/10.1101/2022.08.25.505259doi: bioRxiv preprint
20
constraints cannot be surpassed in the PG structure to preserve viability and shape.
415
Interlinking this PG-biochemical information with phenotypic analyses of the same
416
mutant libraries (Nichols et al., 2011) will permit to create highly informative PG
417
profile-phenotype-gene clusters to better understand the biological consequences of
418
PG synthesis and remodeling. Finally, since the cell wall is one of the major “Achilles
419
heels” of bacteria, this systematic survey of new players in the synthesis and
420
regulation of the PG structure will be particularly relevant on antibiotic resistant
421
pathogenic bacteria and infection-mimicking conditions as it has a great potential to
422
promote the discovery of new pathways that can be targeted in antimicrobial
423
therapies.
424
425
.CC-BY-NC-ND 4.0 International licensemade available under a
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is
The copyright holder for this preprintthis version posted August 26, 2022. ; https://doi.org/10.1101/2022.08.25.505259doi: bioRxiv preprint
21
ACKNOWLEDGEMENTS
426
Research in the Cava lab is supported by The Swedish Research Council (VR), The
427
Knut and Alice Wallenberg Foundation (KAW), The Laboratory of Molecular Infection
428
Medicine Sweden (MIMS) and The Kempe Foundation. We thank John J. Mekalanos
429
for providing the V. cholerae transposon mutant library and the Cava Lab and Miguel
430
A. de Pedro for their helpful discussions.
431
432
AUTHOR CONTRIBUTION
433
SBH, LAM and FC designed the experiments. SBH, LAM and BRR performed
434
research. SBH, LAM, BRR, BS and AS analysed data. SBH, LAM and FC wrote the
435
paper with input from the rest of the authors.
436
437
COMPETING INTEREST
438
None
439
440
.CC-BY-NC-ND 4.0 International licensemade available under a
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is
The copyright holder for this preprintthis version posted August 26, 2022. ; https://doi.org/10.1101/2022.08.25.505259doi: bioRxiv preprint
22
FIGURE LEGENDS
441
442
Fig. 1. Development of a high throughput method for peptidoglycan sample
443
preparation and analysis.
444
(a) Schematic of the comparison between the traditional and the novel high
445
throughput (HT) method for PG isolation and examples of the potential types of
446
analyses that can be carried out with large-scale muropeptide profile datasets. (b)
447
Comparative pipeline for the HT and the traditional PG isolation method. (c)
448
Comparative timescales for the isolation of PG from 96 samples following the
449
traditional or the HT sample preparation method. (d) Representative PG profiles
450
obtained for diverse bacteria using the HT PG isolation method. See also
451
Supplementary Fig. 1.
452
453
454
Fig. 2. Validation of the high throughput method for peptidoglycan sample
455
preparation.
456
(a) Pairwise scatter plot comparing the relative abundance of every muropeptide
457
between replicates of 900 samples. Mean Pearson correlation coefficient is shown.
458
(b-f) Validation of the HT method for detection of minor changes in the PG structure
459
by analysis of mutants known to have a phenotype in their muropeptide profile:
460
Pbp1A (PG synthetase), LdtA (LD-transpeptidase) and DacA1 (DD-
461
carboxypeptidase); (b) representative chromatogram of each strain; (c) PCA using
462
the Log2FC of peak areas of all the muropeptides of each sample; (d) analysis of PG
463
features; (e) heatmap presenting the relative peak area of each muropeptide
464
numbered in panel b; (f) heatmap of the Log2FC values (relative to the wildtype)
465
calculated for each peak numbered in b. In e and f, peaks expected to change in
466
LdtA
-
and DacA1
-
mutants are highlighted in green and blue respectively. Error bars
467
in d correspond to the standard deviation of triplicates. See also Supplementary
468
Fig. 2.
469
470
471
Fig. 3. Peptidoglycan profiling of the V. cholerae non-redundant mutant library.
472
(a) Total number of mutant strains available, processed and analyzed, and their
473
classification into main COG groups. (b) Muropeptide abundance and variation (SD)
474
.CC-BY-NC-ND 4.0 International licensemade available under a
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is
The copyright holder for this preprintthis version posted August 26, 2022. ; https://doi.org/10.1101/2022.08.25.505259doi: bioRxiv preprint
23
across the whole dataset. (c) Hierarchical clustering and heat-map representing the
475
Log2FC calculated for different PG-features and muropeptides. (d-f) Analysis of the
476
PG profile of genes belonging to the same transcriptional unit: PG features, heatmap
477
and clustering of the PG profiles of indicated mutants and the expression data of
478
each gene observed in 307 different RNAseq experiments are shown. (g) PCA
479
obtained for the mutants of the library based on the Log2FC values observed for the
480
20 identified muropeptides. Outliers are colour-coded based on their PG-similarities
481
clustering (see h). (h) Hierarchical clustering of the 50 outliers selected in g. See
482
also Supplementary Fig. 3 and Extended Data 1-3.
483
484
485
Fig. 4. Analysis of the main peptidoglycan features of the V. cholerae mutant
486
library.
487
(a) Violin plots showing the distribution of samples for the indicated PG-features in
488
the 3030 analyzed mutants. Mutants expected to present a distinctive feature are
489
labelled in each group. (b) Correlation scatterplot matrix for the different PG features
490
in the V. cholerae mutant library. Histograms showing the distribution of samples for
491
every PG feature are represented in the diagonal.
492
493
494
Fig. 5. vc1321 encodes a new high molecular weight PBP in V. cholerae.
495
(a) Over-represented biological processes associated to those mutants presenting low
496
relative amount of PG (Log2FC <-0.5). The most representative biological processes with a
497
fold enrichment higher than 2 are represented. Dot size indicates the number of loci/category
498
and colour indicates the significance of the enrichment (-Log10(P-values)). (b) Plot showing
499
the Log2FC of the PG amount for the complete dataset. Genes belonging to the
500
”peptidoglycan metabolism” category in a are labelled. (c) Representative PG
501
chromatograms of the wildtype, the Δvc1321 mutant strain and the Δvc1321 strain
502
overexpressing vc1321 from the arabinose-inducible promoter of the pBAD plasmid. (d)
503
Quantification of the relative PG amount and DD-crosslink of the wildtype, Δvc1321 mutant
504
strain, and the Δvc1321 carrying the pBAD empty plasmid as control or overexpressing
505
vc1321 from the arabinose-inducible promoter. (e) Bocillin gel of indicated strains showing
506
the band corresponding to the product of vc1321 (named PBP1 of V. cholerae, PBP1V) (f)
507
Schematic representation of the transglycosylase (TG) and transpeptidase (TP) domains
508
predicted in the sequence of VC1321. (g-h) Bocillin gels of vc1321 point mutants in different
509
.CC-BY-NC-ND 4.0 International licensemade available under a
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is
The copyright holder for this preprintthis version posted August 26, 2022. ; https://doi.org/10.1101/2022.08.25.505259doi: bioRxiv preprint
24
putative catalytic residues (h) and in candidate cysteines involved in the dimerization of the
510
protein. (i) Deletion of pbp1V does not result in growth or morphological defect. Scale bar = 2
511
µm. (j) PBP1V homologs across different species. See also Supplementary Figs. 4-6 and
512
Extended Data 4.
513
.CC-BY-NC-ND 4.0 International licensemade available under a
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is
The copyright holder for this preprintthis version posted August 26, 2022. ; https://doi.org/10.1101/2022.08.25.505259doi: bioRxiv preprint
25
SUPPLEMENTARY FIGURES TITLES AND LEGENDS
514
515
Supplementary Fig. 1. High throughput method optimization, Related to Fig. 1.
516
(a) Comparison of the efficiency of PG isolation methods. Samples corresponding to different
517
cell numbers were collected, supplemented with SDS 1.5% (w/v) final concentration, and
518
solubilized by boiling, autoclaving or using a heat-block during different times. Samples were
519
processed and analyzed by UPLC. Yield is quantified as maximum intensity of absorbance at
520
204 nm for the M4 muropeptide, in arbitrary units (AU). (b) Phase contrast microscopy of
521
different bacteria and their purified sacculi processed using the traditional or the HT
522
purification method. Scale bar = 2 µm. (c) Relative muramidase activity at different time
523
points. (d) Efficiency of the muramidase digestion in buffer or water on sacculi from two
524
different concentrated starting cultures at two time points. Yield is quantified as maximum
525
intensity of absorbance at 204 nm for the M4 muropeptide, in arbitrary units (AU).
526
527
Supplementary Fig. 2. Muropeptide profile of V. cholerae’s peptidoglycan obtained
528
using the high throughput sample preparation method and analysis of data, Related to
529
Figs. 2-3.
530
(a) Representative chromatogram obtained for V. cholerae wildtype strain using the HT
531
sample preparation method. Peaks of interest are listed. (b) Table of identified muropeptides
532
in a. Identity was confirmed by MS/MS analysis, theoretical and observed neutral mass in Da
533
are indicated. GlcNAc: N-acetyl glucosamine; MurNAc: N-acetyl muramic acid; (1-6anhydro)
534
MurNAc: 1-6 anhydro N-acetyl muramic acid, terminal muropeptide; L-Ala: L-alanine; D-Glu:
535
D-glutamic acid; DAP: meso-diaminopimelic acid; D-Ala: D-alanine; D-Met: D-methionine;
536
Gly: glycine; Lpp: Braun’s lipoprotein. (c) Pipeline for the data transformation and analysis: i)
537
abundance was calculated by integration of the area of the peaks, ii) relative area of each
538
muropeptide was calculated by dividing the peak area by the total area of the chromatogram,
539
iii) finally, fold change (FC) relative to the control sample was calculated and Log2FC was
540
used for representation in heat maps.
541
542
Supplementary Fig. 3. Analysis of the peptidoglycan profiling of the V. cholerae non-
543
redundant mutant library, Related to Fig. 3.
544
(a) Biological versus technical variability: comparison of the standard deviation of the relative
545
abundance of each muropeptide across all samples in the V. cholerae mutant library versus
546
the average standard deviation across replicas of 300 samples. (b) Correlation between
547
each muropeptide relative abundance and its standard deviation across the library dataset.
548
(c) Scree plot of the main components used for the PCA using relative abundance of
549
.CC-BY-NC-ND 4.0 International licensemade available under a
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is
The copyright holder for this preprintthis version posted August 26, 2022. ; https://doi.org/10.1101/2022.08.25.505259doi: bioRxiv preprint
26
muropeptides and Log2FC calculated values. (d) Biplot showing the contribution of the
550
different variables (muropeptides) to the data variability in the dataset of relative amounts. (e)
551
Biplot showing the contribution of the different variables (muropeptides) to the data variability
552
in the Log2FC dataset. (f) Violin plots showing the distribution of the samples for each
553
muropeptide. Log2FC data was used. Median (continued lines) and quartiles (dashed lines)
554
are represented. (g) Table representing the number and percentage of mutants with a
555
significant phenotype for at least one muropeptide or a PG feature. A significant phenotype is
556
considered when the sample is further than 1, 2 or 3 standard deviations (SD) from the
557
average. (h) Venn diagram showing the number of outliers with a phenotype (using 1SD as
558
threshold) in the main PG features.
559
560
Supplementary Fig. 4. Enrichment analysis of mutants presenting differential PG
561
features, Related to Fig. 5.
562
Over-represented biological processes of mutant candidates presenting low (left, red dots)
563
and high (right, blue dots) relative amounts of: (a) chain length (Log2FC <-0.2 and >0.18);
564
(b) total crosslink (Log2FC <-0.15 and >0.15); (c) DD-crosslink (Log2FC <-0.2 and >0.2);
565
and (d) LD-crosslink (Log2FC <-0.7 and >0.7). The most representative biological processes
566
with a fold enrichment higher than 2 are represented. The dot size in the balloon plots
567
indicates the number of loci included in each category and the colour the significance of the
568
enrichment (-Log10(p-values)).
569
570
Supplementary Fig. 5. Multiple sequence alignment of PBP1V with other high
571
molecular weight penicillin-binding proteins, Related to Fig. 5.
572
Sequence alignment of V. cholerae PBP1V (Vc-PBP1V, Uniprot: Q9KSD7), V. cholerae
573
PBP1A (Vc-PBP1A, Uniprot: Q9KNU5), V. cholerae PBP1B (Vc-PBP1B, Uniprot: Q9KUC0),
574
E. coli PBP1A (Ec-PBP1A, Uniprot: P02918), E. coli PBP1B (Ec-PBP1B, Uniprot: P02919)
575
and E. coli PBP1C (Ec-PBP1C, Uniprot: A0A093EN65), performed with T-COFFEE
576
Expresso. Background residue colour indicates degree of conservation. TG domain is
577
highlighted in yellow, TP domain is highlighted in light blue, PBP1C-binding domain is
578
highlighted in green. PBPs conserved motifs are indicated in the red boxes. Vc-PBP1V new
579
domain in red letters.
580
Supplementary Fig. 6. Peptidoglycan analysis of PBP1V cysteine point mutants,
581
Related to Fig. 5.
582
Quantification of the relative PG amount and DD-crosslink of the wildtype, Δpbp1V mutant
583
strain, and indicated cysteine mutants. (ns: not significant).
584
.CC-BY-NC-ND 4.0 International licensemade available under a
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is
The copyright holder for this preprintthis version posted August 26, 2022. ; https://doi.org/10.1101/2022.08.25.505259doi: bioRxiv preprint
27
EXTENDED DATA FILES AND LEGENDS
585
586
Extended Data 1. PG dataset from the V. cholerae Tn-library, Related to Fig. 3.
587
Extended Data 2. SRA numbers used for the study of gene expression analysis, Related to
588
Fig. 3 d-f.
589
Extended Data 3. Data 50 selected PCA outliers, Related to Fig. 3 g-h.
590
Extended Data 4. GO-enrichments, Related to Fig. 5.
591
Extended Data 5. Bacterial strains list, Related to Methods.
592
Extended Data 6. List of primers, Related to Methods.
593
594
595
.CC-BY-NC-ND 4.0 International licensemade available under a
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is
The copyright holder for this preprintthis version posted August 26, 2022. ; https://doi.org/10.1101/2022.08.25.505259doi: bioRxiv preprint
28
METHODS
596
- Lead Contact and Materials Availability
597
• All the data obtained from the genome-wide HTPGP of V. cholerae non-redundant
598
mutant collection is available in the Extended Data files.
599
• Further information and requests for resources and data should be directed and will
600
be fulfilled by the Lead Contact, Felipe Cava ([email protected]).
601
602
- Bacterial strains, media, and antibiotics
603
All the bacterial strains used in this study are listed in Extended Data 5. Lysogeny
604
broth (LB) containing 10 or 5 mg/ml of NaCl (for V. cholerae or the other bacterial
605
species respectively) was used as the standard growth medium. Agar 1.5% (w/v)
606
was used in solid plates. Unless otherwise specified, antibiotics were used at the
607
following concentrations: streptomycin 200 μg/ml, kanamycin 50 μg/ml.
608
Plasmids were constructed by standard DNA cloning techniques. V. cholerae mutant
609
strains were constructed using the primers listed in Extended Data 6, and standard
610
allele exchange techniques with derivatives of the suicide plasmid pCVD442 as
611
described previously (Donnenberg and Kaper, 1991). Plasmid used for
612
complementation are derivatives of pBAD18 (Guzman et al., 1995), where
613
expression is controlled by the P
BAD
promoter. The fidelity of the DNA regions
614
generated by PCR amplification was confirmed by DNA sequencing.
615
616
- Growth conditions
617
Bacterial species were grown in LB at 30 or 37 ºC for sacculi isolation. The culture
618
volume used for high throughput sacculi preparation ranged between 0.5 and 2 ml
619
and was adjusted to obtain the best PG-profile (highest absorbance, UPLC signal)
620
.CC-BY-NC-ND 4.0 International licensemade available under a
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is
The copyright holder for this preprintthis version posted August 26, 2022. ; https://doi.org/10.1101/2022.08.25.505259doi: bioRxiv preprint
29
without saturating the filters during the purification. For the analysis of the V.
621
cholerae mutant library, cells from glycerol stocks were first inoculated in 96-well
622
plates containing 200 μl of LB medium, grown overnight at 37 °C and further diluted
623
1000‐fold into 500 μl of fresh medium in 96-well deep-well plates, where they were
624
finally grown at 37 °C with shaking overnight. For induction of the P
BAD
promoter,
625
0.2% (w/v) of arabinose (final concentration) was added to bacterial cultures.
626
627
- High Throughput (HT) sacculi preparation and PG isolation and digestion
628
For optimization of the method, different sacculi isolation protocols were tested,
629
shown in Supplementary Fig. 1. Bacterial cultures grown overnight in 96-well deep-
630
well plates were pelleted in the same culture plates at 5,250 x g for 30 min. The
631
supernatant was discarded, and the pellets were resuspended with 50 μl of fresh
632
medium. After addition of 25 μl of 5% SDS, samples were tightly sealed using
633
silicone mats and tape, and autoclaved (15 min at 120 °C 1atm). Samples were then
634
transferred to 96-well-filter plates (AcroPrep Advance 96-well 0.2 μm GHP
635
membrane filter plates, 2 ml volume, PALL). SDS was removed by washing twice
636
with 1 ml Milli-Q water (5,250 x g for 15 min). SDS-free sacculi samples were treated
637
with 200 μl proteinase K (40 μg/ml in 100 mM TrisHCl pH 8.0, 30 min at 37°C) for
638
removal of the Braun's lipoprotein. After an additional 1 ml wash with water, samples
639
were digested with 50 μl muramidase (100 μg/ml in water, overnight at 37°C).
640
Enzymatic reactions were performed directly on the filter, sealing the plates with
641
aluminium foil and using a humidity chamber to reduce evaporation.
642
After muramidase digestion, soluble muropeptides were eluted into a new 96-well
643
deep-well plate by centrifuging the filter plate at 5,250 x g for 5 min. Finally, eluted
644
muropeptides were reduced using 10 μl borate buffer (0.5 M pH 9.0) and 25 μl
645
.CC-BY-NC-ND 4.0 International licensemade available under a
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is
The copyright holder for this preprintthis version posted August 26, 2022. ; https://doi.org/10.1101/2022.08.25.505259doi: bioRxiv preprint
30
NaBH
4
(10 mg/ml) for 30 min at room temperature. The pH of the samples was
646
adjusted to 3.5 using 8 μl orthophosphoric acid 25% (v/v) and the samples were
647
stored frozen at -20 °C until analysis by UPLC.
648
649
- Data acquisition by liquid chromatography
650
PG samples were analyzed by liquid chromatography, as previously described
651
(Alvarez et al., 2020). Briefly, muropeptides were separated using an UPLC system
652
(Waters) equipped with a trapping cartridge precolumn (SecurityGuard ULTRA
653
Cartridge UHPLC C18 2.1 mm, Phenomenex) and an analytical column (Waters
654
ACQUITY UPLC BEH C18, 130Å, 1.7 µm, 2.1 mm×150 mm), using as solvents 0.1%
655
formic acid in Milli-Q water (buffer A) and 0.1% formic acid (v/v) in acetonitrile (buffer
656
B) in a 14 min run. Muropeptides were detected by measuring the absorbance at 204
657
nm. Identity of the peaks was first assigned by LC coupled to a QTOF-MS instrument
658
and then by comparison of their retention time. The QTOF–MS instrument was
659
operated in positive ionization mode. Detection of muropeptides was performed by
660
MS
E
to allow for the acquisition of precursor and product ion data simultaneously,
661
using the following parameters: capillary voltage at 3.0 kV, source temperature to
662
120 ºC, desolvation temperature to 350 ºC, sample cone voltage to 40 V, cone gas
663
flow 100 l/h, desolvation gas flow 500 l/h and collision energy (CE): low CE: 6 eV and
664
high CE ramp: 15-40 eV. Mass spectra were acquired at a speed of 0.25 s/scan. The
665
scan was in a range of 100–2000 m/z. Data acquisition and processing was
666
performed using UNIFI software package (Waters Corp.).
667
For the study of the PG chemical variability between samples, collected UPLC
668
chromatographic data were analyzed using the PG-metrics pipeline (Kumar et al.,
669
2017). In brief, raw data were pre-processed by trimming out irrelevant segments
670
.CC-BY-NC-ND 4.0 International licensemade available under a
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is
The copyright holder for this preprintthis version posted August 26, 2022. ; https://doi.org/10.1101/2022.08.25.505259doi: bioRxiv preprint
31
and by baseline correction subtracting a synthetically created baseline. Trimmed and
671
baseline corrected datasets were then aligned by segments using the COW
672
algorithm by selecting appropriate reference samples and combinations of the
673
segment length and slack parameters. The peak area of aligned chromatograms was
674
calculated using a MATLAB routine based on the trapezoidal integration approach
675
wherein the boundaries of the integration were manually selected.
676
The relative area of each muropeptide was calculated dividing its peak area by the
677
total area of the chromatogram (Supplementary Fig. 2b). The different PG features
678
were calculated as described previously (Alvarez et al., 2020; Vigouroux et al.,
679
2020). When relative peak areas of each mutant were used, the most abundant
680
muropeptides presented the greatest impact on the results of PCA analyses
681
(Supplementary Fig. 3a-c), thus the log2 fold change (Log2FC) was calculated for
682
more comprehensive analysis of the data. Log2FCs were calculated by dividing each
683
value (peak area or PG-feature) by the one of the control sample when available or
684
by the mean value of all samples from the same sample set (i.e., 96-well plates)
685
(Supplementary Fig. 2c).
686
687
- Data analysis and representation
688
Prism 8.0 (GraphPad Software) was used to plot and analyze numerical data.
689
Statistical tests in Prism calculated significance of measurements as reported in the
690
corresponding figure legends.
691
Statistical analyses (correlation, standard deviation, PCA, biplots, scree plots) were
692
carried out using R version 4.1.2.
693
.CC-BY-NC-ND 4.0 International licensemade available under a
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is
The copyright holder for this preprintthis version posted August 26, 2022. ; https://doi.org/10.1101/2022.08.25.505259doi: bioRxiv preprint
32
Venn diagrams were created selecting as outliers those mutants with a PG
694
phenotype (more than 1, 2 or 3 standard deviations) in at least one muropeptide or
695
PG feature.
696
Heatmaps and hierarchical clustering analyses (agglomeration method: ward;
697
distance metric: euclidean) were performed using the NG-CHM Heat Map Builder,
698
version 2.20.2 (Ryan et al., 2019).
699
700
- Gene expression analysis
701
In order to calculate gene expression estimates, 307 RNA-Sequencing data
702
annotated to V. cholerae and performed on an Illumina sequencing machine were
703
downloaded from the NCBI’s sequence read archive (SRA) (Leinonen et al., 2011)
704
(See Extended Data 2). We then used salmon/v1.3.0 (Patro et al., 2017) with
705
parameters --libType A --seqBias and --posBias to quantify expression against a
706
decoy-aware index generated from the GCF_000829215.1_ASM82921v1 reference
707
(Okada et al., 2014) as described in Srivastava et al. (Srivastava et al., 2020). Post
708
quantification, we filtered experiments where less than 50% of reads could be
709
matched to a reference gene and imported the estimated counts into R/v4.0 using
710
the tximport package (Soneson et al., 2015). After importing, we used the
711
varianceStabilizingTransformation procedure implemented in the DESeq2 (Love et
712
al., 2014) R package to generate a blind, homoscedastic, and pseudo-log2 count set.
713
714
- Functional enrichment analysis
715
The Protein ANalysis THrough Evolutionary Relationships (PANTHER) Classification
716
System and the PANTHER Statistical Overrepresentation Test (released 20200407)
717
were used to categorize the selected datasets by Gene Ontology (GO) and to
718
.CC-BY-NC-ND 4.0 International licensemade available under a
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is
The copyright holder for this preprintthis version posted August 26, 2022. ; https://doi.org/10.1101/2022.08.25.505259doi: bioRxiv preprint
33
determine the overrepresented GO terms (Mi et al., 2019a; Mi et al., 2019b). For this,
719
selected datasets were searched using the “GO biological process complete”
720
annotation classification available in PANTHER to find functional classes statistically
721
overrepresented when compared to the reference list containing the 3030 proteins
722
corresponding to the Tn-mutants analysed by HT-PG profiling. GO terms were
723
considered significantly overrepresented only when the P-value (Fisher’s exact test)
724
was lower than 0.05. The resulting list for each dataset was summarized (similar
725
terms including the same list of genes were removed to avoid redundancy) and
726
visually represented as described by Bonnot et al. (Bonnot et al., 2019).
727
728
- Microscopy imaging
729
Bacteria were immobilized on LB pads containing 1% agarose (w/v). Phase contrast
730
microscopy was performed using a Zeiss Axio Imager .Z2 microscope (Zeiss,
731
Germany) equipped with a Plan-Apochromat 63X phase contrast objective lens and
732
an ORCA-Flash 4.0 LT digital CMOS camera (Hamamatsu Photonics, Japan), using
733
the Zeiss Zen Blue software. Image analysis and processing were performed using
734
ImageJ software (Rasband, W.S., ImageJ, U. S. National Institutes of Health,
735
Bethesda, Maryland, USA, https://imagej.nih.gov/ij/, 1997-2018).
736
737
- Membrane preparation and detection of penicillin-binding proteins
738
For the preparation of membranes of V. cholerae, 100 ml cell cultures were grown
739
over night and harvested by centrifugation at 15,000 x g for 15 min at 4 °C (Beckman
740
JLA-16.250 rotor, Beckman Avanti J-25, Beckman Coulter). Pellets were washed
741
with 20 mM potassium phosphate (pH 7.5) and 140 mM sodium chloride buffer.
742
Following another centrifugation for 30 min at 360 x g at 4 °C, cells were disrupted
743
twice using a French press (Pressure Cell Homogenizer FPG12800, Stansted Fluid
744
.CC-BY-NC-ND 4.0 International licensemade available under a
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is
The copyright holder for this preprintthis version posted August 26, 2022. ; https://doi.org/10.1101/2022.08.25.505259doi: bioRxiv preprint
34
Power Ltd). The resulting cell lysates were subjected to centrifugation at 360 x g for
745
10 min at 4 °C. The supernatant fractions were collected and centrifuged at 75,600 x
746
g for 90 min at 4 °C (Beckman JA25.50 rotor, Beckman Avanati J-25, Beckman
747
Coulter). The resulting pellets were washed again and resuspended in 500 µl
748
potassium phosphate buffer. The concentration of the membrane preparations was
749
measured using Bradford reagent (Bio-Rad) with bovine serum albumin as standard
750
(Sigma).
751
For the detection of the PBPs, membranes were incubated for 30 min at 37 °C with a
752
fluorescent labelling agent, Bocillin FL Penicillin (Invitrogen, Thermo Fisher Scientific)
753
at a final concentration of 0.05 mM. Finally, the samples were denatured with 4x
754
Laemmli sample buffer (Bio-Rad) at 95 °C for 5 min, followed by centrifugation to
755
remove membrane debris. 20 µl of the reaction mixture were subjected to SDS-
756
PAGE analysis on an 8% polyacrylamide gel. Penicillin-binding proteins were
757
visualised with a laser scanner, Typhoon FLA 9500 (GE Healthcare Life Sciences) at
758
473 nm with a 530DF20 emission filter.
759
760
- Multiple sequence alignments
761
Sequence from V. cholerae PBP1V (Vc-PBP1V, Uniprot: Q9KSD7), V. cholerae
762
PBP1A (Vc-PBP1A, Uniprot: Q9KNU5), V. cholerae PBP1B (Vc-PBP1B, Uniprot:
763
Q9KUC0), E. coli PBP1A (Ec-PBP1A, Uniprot: P02918), E. coli PBP1B (Ec-PBP1B,
764
Uniprot: P02919) and E. coli PBP1C (Ec-PBP1C, Uniprot: A0A093EN65) were
765
aligned using T-COFFEE Expresso Version 11.00 (Notredame et al., 2000) and
766
visualized with JalView (Waterhouse et al., 2009).
767
We run BLAST to study the homology between V. cholerae HMW-PBPs (Altschul et
768
al., 1997).
769
.CC-BY-NC-ND 4.0 International licensemade available under a
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is
The copyright holder for this preprintthis version posted August 26, 2022. ; https://doi.org/10.1101/2022.08.25.505259doi: bioRxiv preprint
35
770
- Protein structure prediction
771
PBP1V protein 3D structural model was built with Alphafold, using both monomer
772
and multimer modes (Evans et al., 2022; Jumper et al., 2021). Only the best model
773
among the five best given by default was examined in detail and represented in the
774
figures. Molecular graphics and analyses were performed with ChimeraX (Goddard
775
et al., 2018; Pettersen et al., 2021).
776
777
.CC-BY-NC-ND 4.0 International licensemade available under a
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is
The copyright holder for this preprintthis version posted August 26, 2022. ; https://doi.org/10.1101/2022.08.25.505259doi: bioRxiv preprint
36
REFERENCES
778
779
Aliashkevich, A., and Cava, F. (2021). LD-transpeptidases: the great unknown
780
among the peptidoglycan cross-linkers. The FEBS journal.
781
Altinoglu, I., Abriat, G., Carreaux, A., Torres-Sanchez, L., Poidevin, M., Krasteva,
782
P.V., and Yamaichi, Y. (2022). Analysis of HubP-dependent cell pole protein
783
targeting in Vibrio cholerae uncovers novel motility regulators. PLoS Genet 18,
784
e1009991.
785
Altschul, S.F., Madden, T.L., Schäffer, A.A., Zhang, J., Zhang, Z., Miller, W., and
786
Lipman, D.J. (1997). Gapped BLAST and PSI-BLAST: a new generation of protein
787
database search programs. Nucleic Acids Res 25, 3389-3402.
788
Alvarez, L., Cordier, B., van Teeffelen, S., and Cava, F. (2020). Analysis of Gram-
789
negative bacteria peptidoglycan by Ultra-Performance Liquid Chromatography. Bio-
790
protocol 10, e3780.
791
Alvarez, L., Hernandez, S.B., and Cava, F. (2021). Cell Wall Biology of Vibrio
792
cholerae. Annual review of microbiology 75, 151-174.
793
Alvarez, L., Hernandez, S.B., de Pedro, M.A., and Cava, F. (2016). Ultra-sensitive,
794
high-resolution liquid chromatography methods for the high-throughput quantitative
795
analysis of bacterial cell wall chemistry and structure. Methods in molecular biology
796
1440, 11-27.
797
Bancroft, P.J., Turapov, O., Jagatia, H., Arnvig, K.B., Mukamolova, G.V., and Green,
798
J. (2020). Coupling of peptidoglycan synthesis to central metabolism in
799
Mycobacteria: post-transcriptional control of CwlM by aconitase. Cell Rep 32,
800
108209.
801
Bern, M., Beniston, R., and Mesnage, S. (2017). Towards an automated analysis of
802
bacterial peptidoglycan structure. Anal Bioanal Chem 409, 551-560.
803
Boll, J.M., Crofts, A.A., Peters, K., Cattoir, V., Vollmer, W., Davies, B.W., and Trent,
804
M.S. (2016). A penicillin-binding protein inhibits selection of colistin-resistant,
805
lipooligosaccharide-deficient Acinetobacter baumannii. Proc Natl Acad Sci U S A
806
113, E6228-e6237.
807
Bonnot, T., Gillard, M.B., and Nagel, D.H. (2019). A simple protocol for informative
808
visualization of enriched gene ontology terms. Bio-protocol 9, e3429.
809
Bukowska-Faniband, E., and Hederstedt, L. (2017). Transpeptidase activity of
810
penicillin-binding protein SpoVD in peptidoglycan synthesis conditionally depends on
811
the disulfide reductase StoA. Molecular microbiology 105, 98-114.
812
Cameron, D.E., Urbach, J.M., and Mekalanos, J.J. (2008). A defined transposon
813
mutant library and its use in identifying motility genes in Vibrio cholerae. Proc Natl
814
Acad Sci U S A 105, 8736-8741.
815
.CC-BY-NC-ND 4.0 International licensemade available under a
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is
The copyright holder for this preprintthis version posted August 26, 2022. ; https://doi.org/10.1101/2022.08.25.505259doi: bioRxiv preprint
37
Campbell, C., Fingleton, C., Zeden, M.S., Bueno, E., Gallagher, L.A., Shinde, D.,
816
Ahn, J., Olson, H.M., Fillmore, T.L., Adkins, J.N., et al. (2021). Accumulation of
817
succinyl coenzyme A perturbs the Methicillin-Resistant Staphylococcus aureus
818
(MRSA) succinylome and is associated with increased susceptibility to beta-lactam
819
antibiotics. mBio 12, e0053021.
820
Cava, F., de Pedro, M.A., Lam, H., Davis, B.M., and Waldor, M.K. (2011). Distinct
821
pathways for modification of the bacterial cell wall by non-canonical D-amino acids.
822
EMBO J 30, 3442-3453.
823
Cheng, A.T., Ottemann, K.M., and Yildiz, F.H. (2015). Vibrio cholerae response
824
regulator VxrB controls colonization and regulates the Type VI Secretion System.
825
PLoS Pathog 11, e1004933.
826
Desmarais, S.M., De Pedro, M.A., Cava, F., and Huang, K.C. (2013). Peptidoglycan
827
at its peaks: how chromatographic analyses can reveal bacterial cell wall structure
828
and assembly. Molecular microbiology 89, 1-13.
829
Donnenberg, M.S., and Kaper, J.B. (1991). Construction of an eae deletion mutant of
830
enteropathogenic Escherichia coli by using a positive-selection suicide vector.
831
Infection and immunity 59, 4310-4317.
832
Dorr, T. (2021). Understanding tolerance to cell wall-active antibiotics. Ann N Y Acad
833
Sci 1496, 35-58.
834
Dorr, T., Lam, H., Alvarez, L., Cava, F., Davis, B.M., and Waldor, M.K. (2014a). A
835
novel peptidoglycan binding protein crucial for PBP1A-mediated cell wall biogenesis
836
in Vibrio cholerae. PLoS Genet 10, e1004433.
837
Dorr, T., Moll, A., Chao, M.C., Cava, F., Lam, H., Davis, B.M., and Waldor, M.K.
838
(2014b). Differential requirement for PBP1a and PBP1b in in vivo and in vitro fitness
839
of Vibrio cholerae. Infect Immun 82, 2115-2124.
840
Dufresne, K., and Paradis-Bleau, C. (2015). Biology and assembly of the bacterial
841
envelope. Advances in experimental medicine and biology 883, 41-76.
842
Evans, R., O’Neill, M., Pritzel, A., Antropova, N., Senior, A., Green, T., Žídek, A.,
843
Bates, R., Blackwell, S., Yim, J., et al. (2022). Protein complex prediction with
844
AlphaFold-Multimer. 2021.2010.2004.463034.
845
Formanek, H., and Formanek, S. (1970). Specific staining for electron microscopy of
846
murein sacculi of bacterial cell walls. European journal of biochemistry 17, 78-84.
847
Glauner, B., Holtje, J.V., and Schwarz, U. (1988). The composition of the murein of
848
Escherichia coli. The Journal of biological chemistry 263, 10088-10095.
849
Goddard, T.D., Huang, C.C., Meng, E.C., Pettersen, E.F., Couch, G.S., Morris, J.H.,
850
and Ferrin, T.E. (2018). UCSF ChimeraX: Meeting modern challenges in
851
visualization and analysis. Protein science : a publication of the Protein Society 27,
852
14-25.
853
.CC-BY-NC-ND 4.0 International licensemade available under a
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is
The copyright holder for this preprintthis version posted August 26, 2022. ; https://doi.org/10.1101/2022.08.25.505259doi: bioRxiv preprint
38
Goodall, E.C.A., Isom, G.L., Rooke, J.L., Icke, C., Pullela, K., Zhang, B., Rae, J.,
854
Tan, W.B., Winkle, M., Delhaye, A., et al. (2021). YhcB coordinates peptidoglycan
855
and LPS biogenesis with phospholipid synthesis during <em>Escherichia coli</em>
856
cell growth. 2021.2004.2016.440158.
857
Gray, A.N., Egan, A.J., Van't Veer, I.L., Verheul, J., Colavin, A., Koumoutsi, A.,
858
Biboy, J., Altelaar, A.F., Damen, M.J., Huang, K.C., et al. (2015). Coordination of
859
peptidoglycan synthesis and outer membrane constriction during Escherichia coli cell
860
division. Elife 4.
861
Guzman, L.M., Belin, D., Carson, M.J., and Beckwith, J. (1995). Tight regulation,
862
modulation, and high-level expression by vectors containing the arabinose PBAD
863
promoter. Journal of bacteriology 177, 4121-4130.
864
Hernandez, S.B., and Cava, F. (2021). New approaches and techniques for bacterial
865
cell wall analysis. Curr Opin Microbiol 60, 88-95.
866
Hernandez, S.B., Dorr, T., Waldor, M.K., and Cava, F. (2020). Modulation of
867
peptidoglycan synthesis by recycled cell wall tetrapeptides. Cell Rep 31, 107578.
868
Irazoki, O., Hernandez, S.B., and Cava, F. (2019). Peptidoglycan Muropeptides:
869
Release, Perception, and Functions as Signaling Molecules. Front Microbiol 10, 500.
870
Jumper, J., Evans, R., Pritzel, A., Green, T., Figurnov, M., Ronneberger, O.,
871
Tunyasuvunakool, K., Bates, R., Žídek, A., Potapenko, A., et al. (2021). Highly
872
accurate protein structure prediction with AlphaFold. Nature 596, 583-589.
873
Kolenbrander, P.E., and Ensign, J.C. (1968). Isolation and chemical structure of the
874
peptidoglycan of Spirillum serpens cell walls. Journal of bacteriology 95, 201-210.
875
Kuhner, D., Stahl, M., Demircioglu, D.D., and Bertsche, U. (2014). From cells to
876
muropeptide structures in 24 h: peptidoglycan mapping by UPLC-MS. Sci Rep 4,
877
7494.
878
Kumar, K., Espaillat, A., and Cava, F. (2017). PG-Metrics: A chemometric-based
879
approach for classifying bacterial peptidoglycan data sets and uncovering their
880
subjacent chemical variability. PLoS One 12, e0186197.
881
Kumar, S., Mollo, A., Kahne, D., and Ruiz, N. (2022). The bacterial cell wall: From
882
lipid II flipping to polymerization. Chem Rev 122, 8884-8910.
883
Leinonen, R., Sugawara, H., Shumway, M., and International Nucleotide Sequence
884
Database, C. (2011). The sequence read archive. Nucleic Acids Res 39, D19-21.
885
Love, M.I., Huber, W., and Anders, S. (2014). Moderated estimation of fold change
886
and dispersion for RNA-seq data with DESeq2. Genome Biol 15, 550.
887
Manning, P.A., Stroeher, U.H., Karageorgos, L.E., and Morona, R. (1995). Putative
888
O-antigen transport genes within the rfb region of Vibrio cholerae O1 are
889
homologous to those for capsule transport. Gene 158, 1-7.
890
.CC-BY-NC-ND 4.0 International licensemade available under a
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is
The copyright holder for this preprintthis version posted August 26, 2022. ; https://doi.org/10.1101/2022.08.25.505259doi: bioRxiv preprint
39
Masser, E.A., Burby, P.E., Hawkins, W.D., Gustafson, B.R., Lenhart, J.S., and
891
Simmons, L.A. (2021). DNA damage checkpoint activation affects peptidoglycan
892
synthesis and late divisome components in Bacillus subtilis. Molecular microbiology
893
116, 707-722.
894
Mi, H., Muruganujan, A., Ebert, D., Huang, X., and Thomas, P.D. (2019a). PANTHER
895
version 14: more genomes, a new PANTHER GO-slim and improvements in
896
enrichment analysis tools. Nucleic Acids Res 47, D419-D426.
897
Mi, H., Muruganujan, A., Huang, X., Ebert, D., Mills, C., Guo, X., and Thomas, P.D.
898
(2019b). Protocol Update for large-scale genome and gene function analysis with the
899
PANTHER classification system (v.14.0). Nat Protoc 14, 703-721.
900
Moll, A., Dorr, T., Alvarez, L., Davis, B.M., Cava, F., and Waldor, M.K. (2015). A D,
901
D-carboxypeptidase is required for Vibrio cholerae halotolerance. Environ Microbiol
902
17, 527-540.
903
Murphy, S.G., Alvarez, L., Adams, M.C., Liu, S., Chappie, J.S., Cava, F., and Dörr, T.
904
(2019). Endopeptidase regulation as a novel function of the Zur-dependent Zinc
905
starvation response. mBio 10.
906
Nichols, R.J., Sen, S., Choo, Y.J., Beltrao, P., Zietek, M., Chaba, R., Lee, S.,
907
Kazmierczak, K.M., Lee, K.J., Wong, A., et al. (2011). Phenotypic landscape of a
908
bacterial cell. Cell 144, 143-156.
909
Notredame, C., Higgins, D.G., and Heringa, J. (2000). T-Coffee: A novel method for
910
fast and accurate multiple sequence alignment. Journal of molecular biology 302,
911
205-217.
912
Okada, K., Na-Ubol, M., Natakuathung, W., Roobthaisong, A., Maruyama, F.,
913
Nakagawa, I., Chantaroj, S., and Hamada, S. (2014). Comparative genomic
914
characterization of a Thailand–Myanmar isolate, MS6, of Vibrio cholerae O1 El Tor,
915
which is phylogenetically related to a “US gulf coast” clone. PLOS ONE 9, e98120.
916
Paradis-Bleau, C., Kritikos, G., Orlova, K., Typas, A., and Bernhardt, T.G. (2014). A
917
genome-wide screen for bacterial envelope biogenesis mutants identifies a novel
918
factor involved in cell wall precursor metabolism. PLoS Genet 10, e1004056.
919
Patel, A.V., Turner, R.D., Rifflet, A., Acosta-Martin, A.E., Nichols, A., Awad, M.M.,
920
Lyras, D., Gomperts Boneca, I., Bern, M., Collins, M.O., et al. (2021). PGFinder, a
921
novel analysis pipeline for the consistent, reproducible, and high-resolution structural
922
analysis of bacterial peptidoglycans. Elife 10.
923
Patro, R., Duggal, G., Love, M.I., Irizarry, R.A., and Kingsford, C. (2017). Salmon
924
provides fast and bias-aware quantification of transcript expression. Nat Methods 14,
925
417-419.
926
Pettersen, E.F., Goddard, T.D., Huang, C.C., Meng, E.C., Couch, G.S., Croll, T.I.,
927
Morris, J.H., and Ferrin, T.E. (2021). UCSF ChimeraX: Structure visualization for
928
researchers, educators, and developers. Protein science : a publication of the Protein
929
Society 30, 70-82.
930
.CC-BY-NC-ND 4.0 International licensemade available under a
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is
The copyright holder for this preprintthis version posted August 26, 2022. ; https://doi.org/10.1101/2022.08.25.505259doi: bioRxiv preprint
40
Porfirio, S., Carlson, R.W., and Azadi, P. (2019). Elucidating peptidoglycan structure:
931
An analytical toolset. Trends in microbiology 27, 607-622.
932
Quintela, J.C., Caparros, M., and de Pedro, M.A. (1995). Variability of peptidoglycan
933
structural parameters in gram-negative bacteria. FEMS Microbiol Lett 125, 95-100.
934
Rohs, P.D.A., and Bernhardt, T.G. (2021). Growth and Division of the Peptidoglycan
935
Matrix. Annual review of microbiology 75, 315-336.
936
Ryan, M.C., Stucky, M., Wakefield, C., Melott, J.M., Akbani, R., Weinstein, J.N., and
937
Broom, B.M. (2019). Interactive Clustered Heat Map Builder: An easy web-based tool
938
for creating sophisticated clustered heat maps. F1000Res 8.
939
Sauvage, E., Kerff, F., Terrak, M., Ayala, J.A., and Charlier, P. (2008). The penicillin-
940
binding proteins: structure and role in peptidoglycan biosynthesis. FEMS Microbiol
941
Rev 32, 234-258.
942
Sheng, Y., Fan, F., Jensen, O., Zhong, Z., Kan, B., Wang, H., and Zhu, J. (2015).
943
Dual Zinc transporter systems in Vibrio cholerae promote competitive advantages
944
over gut microbiome. Infect Immun 83, 3902-3908.
945
Simpson, B.W., Nieckarz, M., Pinedo, V., McLean, A.B., Cava, F., and Trent, M.S.
946
(2021). Acinetobacter baumannii can survive with an outer membrane lacking
947
lipooligosaccharide due to structural support from elongasome peptidoglycan
948
synthesis. mBio 12, e0309921.
949
Soneson, C., Love, M.I., and Robinson, M.D. (2015). Differential analyses for RNA-
950
seq: transcript-level estimates improve gene-level inferences. F1000Res 4, 1521.
951
Srivastava, A., Malik, L., Sarkar, H., Zakeri, M., Almodaresi, F., Soneson, C., Love,
952
M.I., Kingsford, C., and Patro, R. (2020). Alignment and mapping methodology
953
influence transcript abundance estimation. Genome Biol 21, 239.
954
Teoh, W.P., Matson, J.S., and DiRita, V.J. (2015). Regulated intramembrane
955
proteolysis of the virulence activator TcpP in Vibrio cholerae is initiated by the tail-
956
specific protease (Tsp). Molecular microbiology 97, 822-831.
957
Vigouroux, A., Cordier, B., Aristov, A., Alvarez, L., Ozbaykal, G., Chaze, T.,
958
Oldewurtel, E.R., Matondo, M., Cava, F., Bikard, D., et al. (2020). Class-A penicillin
959
binding proteins do not contribute to cell shape but repair cell-wall defects. Elife 9.
960
Vollmer, W., Blanot, D., and de Pedro, M.A. (2008). Peptidoglycan structure and
961
architecture. FEMS Microbiol Rev 32, 149-167.
962
Waterhouse, A.M., Procter, J.B., Martin, D.M., Clamp, M., and Barton, G.J. (2009).
963
Jalview Version 2--a multiple sequence alignment editor and analysis workbench.
964
Bioinformatics 25, 1189-1191.
965
Weaver, A.I., Alvarez, L., Rosch, K.M., Ahmed, A., Wang, G.S., van Nieuwenhze,
966
M.S., Cava, F., and Dorr, T. (2022). Lytic transglycosylases mitigate periplasmic
967
crowding by degrading soluble cell wall turnover products. Elife 11.
968
.CC-BY-NC-ND 4.0 International licensemade available under a
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is
The copyright holder for this preprintthis version posted August 26, 2022. ; https://doi.org/10.1101/2022.08.25.505259doi: bioRxiv preprint
41
Weaver, A.I., Jimenez-Ruiz, V., Tallavajhala, S.R., Ransegnola, B.P., Wong, K.Q.,
969
and Dorr, T. (2019). Lytic transglycosylases RlpA and MltC assist in Vibrio cholerae
970
daughter cell separation. Molecular microbiology 112, 1100-1115.
971
Zijderveld, C.A., Aarsman, M.E., den Blaauwen, T., and Nanninga, N. (1991).
972
Penicillin-binding protein 1B of Escherichia coli exists in dimeric forms. Journal of
973
bacteriology 173, 5740-5746.
974
975
.CC-BY-NC-ND 4.0 International licensemade available under a
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is
The copyright holder for this preprintthis version posted August 26, 2022. ; https://doi.org/10.1101/2022.08.25.505259doi: bioRxiv preprint
Fig. 1. Development of a high throughput method for peptidoglycan sample preparation and analysis. (a)
Schematic of the comparison between the traditional and the novel high throughput (HT) method for PG isolation
and examples of the potential types of analyses that can be carried out with large-scale muropeptide profile
datasets. (b) Comparative pipeline for the HT and the traditional PG isolation method. (c) Comparative timescales
for the isolation of PG from 96 samples following the traditional or the HT sample preparation method. (d)
Representative PG profiles obtained for diverse bacteria using the HT PG isolation method.
See also Supplementary Fig. 1.
.CC-BY-NC-ND 4.0 International licensemade available under a
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is
The copyright holder for this preprintthis version posted August 26, 2022. ; https://doi.org/10.1101/2022.08.25.505259doi: bioRxiv preprint
Fig. 2. Validation of the high throughput method for peptidoglycan sample preparation.
(a) Pairwise scatter plot comparing the relative abundance of every muropeptide between replicates of 900 samples.
Mean Pearson correlation coefficient is shown. (b-f) Validation of the HT method for detection of minor changes in the
PG structure by analysis of mutants known to have a phenotype in their muropeptide profile: Pbp1A (PG synthetase),
LdtA (LD-transpeptidase) and DacA1 (DD-carboxypeptidase); (b) representative chromatogram of each strain; (c) PCA
using the Log2FC of peak areas of all the muropeptides of each sample; (d) analysis of PG features; (e) heatmap
presenting the relative peak area of each muropeptide numbered in panel b; (f) heatmap of the Log2FC values (relative
to the wildtype) calculated for each peak numbered in b. In e and f, peaks expected to change in LdtA
-
and DacA1
-
mutants are highlighted in green and blue respectively. Error bars in d correspond to the standard deviation of
triplicates.
See also Supplementary Fig. 2.
.CC-BY-NC-ND 4.0 International licensemade available under a
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is
The copyright holder for this preprintthis version posted August 26, 2022. ; https://doi.org/10.1101/2022.08.25.505259doi: bioRxiv preprint
Fig. 3. Peptidoglycan profiling of the V. cholerae non-redundant mutant library.
(a) Total number of mutant strains available, processed and analyzed, and their classification into main COG groups.
(b) Muropeptide abundance and variation (SD) across the whole dataset. (c) Hierarchical clustering and heat-map
representing the Log2FC calculated for different PG-features and muropeptides. (d-f) Analysis of the PG profile of
genes belonging to the same transcriptional unit: PG features, heatmap and clustering of the PG profiles of indicated
mutants and the expression data of each gene observed in 307 different RNAseq experiments are shown. (g) PCA
obtained for the mutants of the library based on the Log2FC values observed for the 20 identified muropeptides.
Outliers are colour-coded based on their PG-similarities clustering (see h). (h) Hierarchical clustering of the 50 outliers
selected in g.
See also Supplementary Fig. 3 and Extended Data 1-3.
.CC-BY-NC-ND 4.0 International licensemade available under a
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is
The copyright holder for this preprintthis version posted August 26, 2022. ; https://doi.org/10.1101/2022.08.25.505259doi: bioRxiv preprint
Fig. 4. Analysis of the main peptidoglycan features of the V. cholerae mutant library
(a) Violin plots showing the distribution of samples for the indicated PG-features in the 3030 analyzed mutants. Mutants
expected to present a distinctive feature are labelled in each group. (b) Correlation scatterplot matrix for the different
PG features in the V. cholerae mutant library. Histograms showing the distribution of samples for every PG feature are
represented in the diagonal.
.CC-BY-NC-ND 4.0 International licensemade available under a
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is
The copyright holder for this preprintthis version posted August 26, 2022. ; https://doi.org/10.1101/2022.08.25.505259doi: bioRxiv preprint
Fig. 5. vc1321 encodes a new high molecular weight PBP in V. cholerae
(a) Over-represented biological processes associated to those mutants presenting low relative amount of PG (Log2FC
<-0.5). The most representative biological processes with a fold enrichment higher than 2 are represented. Dot size
indicates the number of loci/category and colour indicates the significance of the enrichment (-Log10(P-values)). (b)
Plot showing the Log2FC of the PG amount for the complete dataset. Genes belonging to the ”peptidoglycan
metabolism” category in a are labelled. (c) Representative PG chromatograms of the wildtype, the Δvc1321 mutant
strain and the Δvc1321 strain overexpressing vc1321 from the arabinose-inducible promoter of the pBAD plasmid. (d)
Quantification of the relative PG amount and DD-crosslink of the wildtype, Δvc1321 mutant strain, and the Δvc1321
carrying the pBAD empty plasmid as control or overexpressing vc1321 from the arabinose-inducible promoter. (e)
Bocillin gel of indicated strains showing the band corresponding to the product of vc1321 (named PBP1 of V. cholerae,
PBP1V) (f) Schematic representation of the transglycosylase (TG) and transpeptidase (TP) domains predicted in the
sequence of VC1321. (g-h) Bocillin gels of vc1321 point mutants in different putative catalytic residues (h) and in
candidate cysteines involved in the dimerization of the protein. (i) Deletion of pbp1V does not result in growth or
morphological defect. Scale bar = 2 µm. (j) PBP1V homologs across different species.
See also Supplementary Fig. 4-6 and Extended Data 4.
.CC-BY-NC-ND 4.0 International licensemade available under a
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is
The copyright holder for this preprintthis version posted August 26, 2022. ; https://doi.org/10.1101/2022.08.25.505259doi: bioRxiv preprint
