Biochemistry and Biophysics Reports 25 (2021) 100921Availa

Biochemistry and Biophysics Reports 25 (2021) 100921
Available online 22 January 2021
2405-5808/© 2021 The Authors. Published by Elsevier B.V.
This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/).
A high methionine and low folate diet alters glucose
homeostasis and
gut microbiome
Chak Kwong Cheng a, b, Chenguang Wang a, b, c, Wenbin
Shang a, b, Chi Wai Lau a, b,
Jiang-Yun Luo a, b, Li Wang a, b, Yu Huang a, b, *
a School of Biomedical Sciences and Li Ka Shing Institute of
Health Science, The Chinese University of Hong Kong, Hong
Kong SAR, China
b Heart and Vascular Institute and Shenzhen Research Institute,
The Chinese University of Hong Kong, Hong Kong SAR, China
c Shenzhen Institute of Advanced Technology, Chinese
Academy of Sciences, Shenzhen, China
A R T I C L E I N F O
Keywords:
HMLF diet
Hyperhomocysteinemia
Glucose homeostasis
Gut microbiome
16S rRNA sequencing
Porphyromonadaceae

A B S T R A C T
Hyperhomocysteinemia (HHcy) is considered as a risk factor for
several complications, including cardiovascular
and neurological disorders. A high methionine low folate
(HMLF) diet chronically causes HHcy by accumulating
homocysteine in the systemic circulation. Elevated Hcy level is
also associated with the incidence of diabetes
mellitus. However, very few studies focus on the impact of
HMLF diet on glucose homeostasis, and that on gut
microbiome profile. HHcy was induced by feeding C57BL/6
mice a HMLF diet for 8 weeks. The HMLF diet
feeding resulted in a progressive body weight loss, and
development of slight glucose intolerance and insulin
resistance in HHcy mice. Notably, the HMLF diet alters the gut
microbiome profile and increases the relative
abundance of porphyromonadaceae family of bacteria in HHcy
mice. These findings provide new insights into the
roles of dysregulated glucose homeostasis and gut flora in the
pathogenesis of HHcy-related complications.
1. Introduction
Homocysteine (Hcy) is a non-proteinogenic sulfur-containing
ho-
mologue of cysteine. Hcy is synthesized through elimination of
a ter-
minal methyl group from methionine [1]. In normal
physiological
conditions, Hcy is converted to either methionine or cysteine in
the
presence of B-vitamins, which serve as critical co-factors for
normal
function of methionine synthase and cystathionine synthase [2].
Dietary

consumption of excessive methionine and dietary deficiency of
B-vita-
mins (pyridoxine (B6), folate (B9) and B12) cause Hcy
accumulation in
the systemic circulation [3]. A diet of high methionine and low
folate
(HMLF) content has been therefore used to induce hyper-
homocysteinemia (HHcy) in rodents [4]. An elevated level of
plasma
Hcy (200–300 μmol L− 1) is regarded as HHcy [5], a risk factor
for car-
diovascular complications (e.g. coronary heart disease,
hypertension)
and neurological disorders (e.g. the Alzheimer’s disease, the
Parkinson’s
disease) [6]. Furthermore, increased level of serum Hcy has
been re-
ported in patients with type 2 diabetes mellitus [7]. Previous
studies also
showed that an elevated Hcy level correlates with higher risk of
vascular
complications and mortality in type 2 diabetic patients [8].
Non-enzymatic factors, particularly insulin, can affect the
circulatory
level of Hcy. In brief, higher plasma level of insulin down-
regulates the
hepatic expression of cystathionine beta-synthase, a crucial
enzyme
involved in the biosynthesis of cysteine from Hcy [9], implying
that
hyperinsulinemia may result in elevated Hcy level. Most often,
hyper-
insulinemia results from insulin resistance. When the cells in
liver, fat

and muscles fail to respond to insulin due to insulin resistance,
the
pancreas compensatively secretes more insulin, a condition
called
hyperinsulinemia [10]. Insulin resistance was shown to be
associated
with HHcy in a clinical trial of 2214 individuals [11]. Another
study
suggested that high glucose level can facilitate Hcy clearance
by
enhancing remethylation of Hcy, and hence methionine
synthesis [12].
However, whether HHcy negatively impacts on glucose
metabolism and
insulin sensitivity remains inconclusive and contradictory. A
previous
clinical study reported a correlation between HHcy and insulin
resis-
tance in patients with hypothyroidism [13], whereas another
Mendelian
Abbreviations: Hcy, homocysteine; HDL, high-density
lipoprotein; HHcy, hyperhomocysteinemia; HMLF, high
methionine low folate; LEfSe, linear discriminant
analysis effect size; NAFLD, non-alcoholic fatty liver disease;
NMDS, non-metric multi-dimensional scaling; OTU, operational
taxonomic unit; PCA, principal
component analysis; SCFA, short-chain fatty acids; TC, total
cholesterol; TG, triglyceride.
* Corresponding author. 223A, Lo Kwee-Seong Integrated
Biomedical Sciences Building, Area 39, The Chinese University
of Hong Kong, China.
E-mail address: [email protected] (Y. Huang).
Contents lists available at ScienceDirect

Biochemistry and Biophysics Reports
journal homepage: www.elsevier.com/locate/bbrep
https://doi.org/10.1016/j.bbrep.2021.100921
Received 10 November 2020; Received in revised form 10
January 2021; Accepted 11 January 2021
mailto:[email protected]
www.sciencedirect.com/science/journal/24055808
https://www.elsevier.com/locate/bbrep
http://crossmark.crossref.org/dialog/?doi=10.1016/j.bbrep.2021.
100921&domain=pdf
http://creativecommons.org/licenses/by-nc-nd/4.0/
2
randomization study suggested no such causal relationship
among levels
of Hcy, fasting glucose and fasting insulin in European
individuals [14].
In addition, the effect of HMLF diet and HMLF diet-induced
HHcy on
glucose homeostasis are scarcely investigated. Before gut
absorption and
hepatic metabolism of Hcy, the dietary HMLF content might
have
exerted an effect on gut microbiome, which emerges as a new

organ
system in the host [15]. Whether HMLF content affects gut
microbiome,
in terms of relative abundance and biodiversity of gut bacteria,
remains
to be determined. Therefore, in the present study, we aim to
provide
novel insights into the predisposing effects of HMLF diet on
glucose
homeostasis and gut microbiome in HHcy-related
complications.
2. Materials and methods
2.1. Animal protocol
The current study was conducted under the approval of the
Animal
Experimentation Ethics Committee, the Chinese University of
Hong
Kong (CUHK). All animal experiments complied with the
ARRIVE
guidelines [16], and the guide for the care and use of laboratory
animals
established by the National Institutes of Health. Male C57BL/6
mice (8
weeks old, 22–25 g) were supplied by the Laboratory Animal
Services
Center, CUHK. Randomization was undertaken to lower the risk
of bias
in this study. All C57BL/6 mice (n = 10) were randomized
before
treatment. Before changing the diet, the male C57BL/6 mice
were orally
administrated 200 μL antibiotic cocktail (1 g L-1 ampicillin, 1 g
L-1

metronidazole, 0.5 g L-1 neomycin and 0.5 g L-1 vancomycin)
for 3
consecutive days [17]. The mice were subsequently fed on
either normal
chow (n = 5) or HMLF diet (n = 5) (Table 1; Envigo Teklad,
Madison,
WI, USA) for 8 weeks. Thereafter, fecal samples were collected
for
subsequent bacterial 16S rRNA sequencing (Fig. 1A). The mice
were
kept in individual and well-ventilated caging systems under a 12
h light:
12 h dark cycle. The mice were guaranteed free access to
laboratory
diets and water, at a constant temperature (22 ± 1 ◦C) and
humidity
(55% ± 5%). Body weights of mice were monitored weekly.
2.2. Plasma Hcy measurement by ELISA
In the presence of heparin, mouse blood was collected by
cardiac
puncture. The blood was centrifuged for 10 min at 1000×g at 4
◦C for
plasma collection. The plasma Hcy level was evaluated by a
commer-
cially available ELISA kit (Cell Biolabs, San Diego, CA, USA)
following
the manufacturer’s protocol [18].
2.3. Lipid profile determination
The levels of total cholesterol (TC), triglyceride (TG) and high-
density lipoprotein (HDL) cholesterol in mouse plasma were
measured
by commercial assay kits (Stanbio, Boerne, TX), according to

the man-
ufacturer’s instructions. In addition, the level of non-HDL
cholesterol
was determined by using the following formula: non-HDL
cholesterol =
TC - (TG/5) – HDL [19].
2.4. Blood glucose determination
After 4 and 8 weeks of HMLF diet, glucose tolerance test (GTT)
and
insulin tolerance test (ITT) were performed in C57/BL6 mice 6
h post-
fasting. By intraperitoneal injection of glucose (1 g kg− 1) or
insulin (1
unit kg− 1), the glucose levels were evaluated in venous blood
of mouse
tail at different time points (0, 15, 30, 60, 90 and 120 min).
2.5. DNA extraction from fecal samples
After 8 weeks, fecal pellets were collected from mice fed on
normal
chow and HMLF diet and stored at − 80 ◦C for later processing.
Bacterial
DNA was isolated from the stored fecal samples by using the
Nucleo-
Spin® DNA Stool kit (Macherey-Nagel, Düren, Germany). In
brief,
180–220 mg of fecal samples were homogenized in bead beating
tubes
by vortex at a maximal speed for 10 min. Total genomic DNA
was
subsequently captured on the silica membrane and NucleoSpin®
In-
hibitor Removal Column was applied to remove impurities that

interfere
with downstream procedures. At last, purified DNA was eluted
by 80 μL
Elution Buffer SE [20]. DNA concentrations were determined
by
maximum absorbance at 260 nm.
2.6. 16S rRNA sequencing: library preparation and sequencing
High-throughput sequencing of bacterial 16S rRNA gene was
con-
ducted by Novogene Technology Co. (Beijing, China). Briefly,
after DNA
quality was determined by agarose gel electrophoresis, the
verified DNA
was amplified by the 515F forward
(GTGCCAGCMGCCGCGGTAA) and
806R reverse (GGACTACVSGGGTATCTAAT) primers, which
aim at the
V4 hypervariable region of bacterial 16S rRNA gene. The PCR
products
were purified by gel extraction kit (Qiagen, Germany).
Sequencing li-
braries were constructed by NEBNext® Ultra™ DNA Library
Pre Kit for
Illumina (New England Biolabs), according to manufacturer’s
in-
structions. In the final step, the library was sequenced via an
Illumina
sequencing platform to generate 250 bp paired-end reads [21].
2.7. Data analysis for bacterial 16S rRNA sequencing
Bacterial 16S rRNA sequencing analysis was performed by
using
QIIME (v1.7.0; http://qiime.org/index.html) [22]. Quality

filtering on
successfully merged reads were conducted based on the default
settings
of QIIME. Sequences were assigned to operational taxonomic
units
(OTUs) according to the similarity to annotated bacterial
sequences by
using Uparse (v7.0.1001; http://drive5.com/uparse/; 97%
sequence
similarity cut-off) [23]. The most abundant sequence was
selected to
represent each OTU and classified to a taxonomic group by RDP
clas-
sifier (v2.2; http://sourceforge.net/projects/rdp-classifier/) [24].
Chimeric OTUs were discarded based on the analysis against
the
sequence in SILVA database
(http://www.mothur.org/wiki/Silva-refere
nce-files) [25]. OTUs were intended for beta diversity analysis
where
unweighted Unifrac distances were computed by QIIME.
Consequently,
these distances were displayed by both principal component
analysis
(PCA) and non-metric multi-dimensional scaling (NMDS). Both
PCA and
NMDS are intended to obtain principal coordinates for
subsequent
visualization from the complex and multidimensional data [26].
In
addition, the difference in microbiota composition among
groups was
analyzed by both MetaStat and linear discriminant analysis
effect size
(LEfSe) [27], in which the results of LEfSe were displayed in a
cladogram.

2.8. Statistical analysis
All data were expressed as mean ± SD of 5 animals. Differences
be-
tween the 2 experimental groups (normal chow and HMLF diet
groups)
were assessed by GraphPad Prism 6 software by two-tailed non-
para-
metric Mann-Whitney test. A P-value ≤ 0.05 indicates statistical
Table 1
The compositions of normal chow and HMLF diet.
Formula (g kg− 1) Normal chow HMLF diet
Casein 200.0 200.0
L-Methionine 3.0 20.0
Sucrose 524.747 506.7475
Maltodextrin 100.0 100.0
Soybean oil 30.0 30.0
Anhydrous milkfat 25.0 25.0
Cellulose 80.0 80.0
Mineral Mix 35.0 35.0
Succinylsulfathiazole 0 1.0
C.K. Cheng et al.
http://qiime.org/index.html
http://drive5.com/uparse/
http://sourceforge.net/projects/rdp-classifier/
http://www.mothur.org/wiki/Silva-reference-files
http://www.mothur.org/wiki/Silva-reference-files

3
significance. Statistical difference in bacterial family between
groups
based on 16S rRNA sequencing data was determined by T-test,
using P-
value ≤ 0.05 as a cut-off.
Fig. 1. HMLF diet induced HHcy in C57BL/6 mice.
(A) Schematic representation of the present study. (B)
Plasma Hcy concentration of C57BL/6 mice fed with
either normal chow or HMLF diet for 8 weeks. (C)
Changes in body weights of mice during 8 weeks of
feeding. (D) Lipid profile of two groups of mice (i.e.
normal chow and HMLF diet groups). HDL-C: high-
density lipoprotein cholesterol; HMLF: high methio-
nine low folate; non-HDL-C: non-HDL cholesterol; TC:
total cholesterol; TG: triglyceride. *P < 0.05 vs
normal chow group. n = 5 per group. All data are
expressed as mean ± SD.
Fig. 2. HMLF diet impaired glucose homeostasis in C57BL/6
mice. (A) OTT of C57BL/6 mice at (A) week 4 and (B) week 8
of HMLF diet feeding. (C) ITT of C57BL/6
mice at (C) week 4 and (D) week 8 of HMLF diet feeding.
AUCs of (E) OTT and (F) ITT at weeks 4 and 8 of HMLF diet
feeding. HMLF: high methionine low folate. *P
< 0.05 vs normal chow group. n = 5 per group. All data are
expressed as mean ± SD.
C.K. Cheng et al.

4
3. Results
3.1. A HMLF diet induced HHcy in C57BL/6 mice
To evaluate the effect of HMLF diet on glucose homeostasis and
gut
microbiome, C57BL/6 mice were fed either normal chow or
HMLF diet
for 8 weeks (Fig. 1A). The 8-week feeding of HMLF diet
elevated the
plasma Hcy concentration in C57BL/6 mice to a level higher
than 200
μmol L− 1, approximately 3.5 folds higher than that in normal
chow-fed
mice (Fig. 1B). Moreover, the HMLF diet slightly lowered the
body
weight gains of C57BL/6 mice when compared to normal chow
(Fig. 1C). However, the HMLF diet did not alter the lipid
profiles (i.e. TC,
TG, HDL-C and non-HDL-C) in C57BL/6 mice (Fig. 1D).
3.2. HMLF diet impaired glucose homeostasis in diet-induced
HHcy mice
We performed GTT and ITT to determine glucose absorption by
metabolic organs and insulin sensitivity, respectively, at weeks
4 and 8
of diet feeding. At week 4, no significant glucose intolerance
(Fig. 2A
and E) and insulin resistance (Fig. 2C and F) were observed in
diet-
induced HHcy mice when compared to normal chow-fed mice.
How-

ever, diet-induced HHcy mice were slightly but significantly
glucose
intolerant (Fig. 2B and E) and insulin resistant (Fig. 2D and F)
when
comparted to normal chow-fed mice at week 8.
3.3. HMLF diet altered the gut microbiome profile in diet-
induced HHcy
mice
To compare the gut microbiota compositions between normal
chow
and HMLF diet groups, 16S rRNA sequencing of V4
hypervariable region
was performed. Beta diversity analysis was displayed on PCA
and NMDS
plots, which were based on the calculations of Bray-Curtis
distance. Both
PCA and NMDS plots revealed that the community
compositions of in-
testinal microbiota in HMLF diet group was distinct from that
of normal
chow group (Fig. 3A and B). Furthermore, the relative
abundance of
predominant taxa in the two groups were compared. At the
phylum
level, both Bacteroidetes and Firmicutes were dominant phyla,
contrib-
uting to about 70% of relative abundance (Fig. 3C). Of note, the
relative
abundance of Bacteroidetes was higher (P = 0.0340), whereas
that of
Proteobacteria was lower in HMLF diet group (Fig. 3C, P =
0.0204). The
species abundance heatmap was constructed to compare the
predomi-

nant genera of two groups. Notably, the HMLF diet-fed mice
were
associated with slight enrichment in Tyzzerella (P = 0.0647),
Odoribacter
(P = 0.0512) and Allobaculum (P = 0.0684) (Fig. 3D). Lower
abundance
of Roseburia (P = 0.0361) and Romboutsia (P = 0.0491) were
observed in
HMLF diet-fed mice (Fig. 3D).
3.4. HMLF diet increased the abundance of
porphyromonadaceae family
of bacteria
T-test analysis revealed significant variation between two
groups,
where the HMLF diet group was characterized by a higher
abundance of
Porphyromonadaceae family (P < 0.05) (Fig. 4A). MetaStat
method was
used to analyze the microbial species abundance data for fecal
samples
of normal chow and HMLF diet groups. Based on the q value at
family
level, a significant variation among the species (Q < 0.05) w as
observed
in the Porphyromonadaceae family between two groups (Fig.
4B). LEfSe
analysis was performed to study the enrichment of bacterial taxa
upon
different dietary contents. Cladogram generated from LEfSe
analysis
demonstrated that both Porphyromonadaceae and
Erysipelotrichaceae
families of bacteria were enriched in the HMLF diet group (Fig.
4C).

4. Discussion
A HMLF diet causes the accumulation of non-proteinogenic
amino
acid Hcy in the circulation [28]. In the present investigation, we
demonstrated that a dietary pattern of HMLF content brought
about
multiple effects in vivo. In addition to the induction of HHcy,
which is
considered as an independent risk factor for several
cardiovascular
diseases [29], we provided novel evidence that the HMLF diet
altered
the glucose homeostasis and gut microbiome profile in C57BL/6
mice.
An 8-week feeding of HMLF diet successfully induced HHcy in
C57BL/6 mice by increasing the concentration of plasma Hcy to
approximately 200 μmol L− 1, a level comparable to human
patients with
severe HHcy (>100 μmol L− 1) [30]. Body weight gains of
HMLF diet-fed
mice were observed to be lower than that of normal chow -fed
mice. This
is consistent with a previous clinical finding that an elevated
Hcy level is
associated with body weight loss in humans [31]. A lower body
weight
gain can be partly attributed to appetite loss caused by folate
deficiency
in diet [32]. In addition, HMLF diet did not elevate levels of
plasma
cholesterol and triglycerides in C57BL/6 mice, implying that
HHcy

represents a risk factor different from hypercholesterolemia and
obesity.
A gradual impairment of glucose homeostasis was observed, as
revealed by GTT and ITT at weeks 4 and 8 of HMLF diet
feeding. A
previous study suggested that high insulin level raises Hcy level
in rats
[33]. Human body generally produces more insulin, i.e. hyper-
insulinemia, in response to insulin resistance [9]. Under
diabetic con-
ditions, elevated Hcy level is often observed [34]. The present
study
provides novel findings that HMLF diet can gradually induce
insulin
resistance and hyperglycemia in non-diabetic C57BL/6 mice,
implying
that HMLF diet could alter glucose homeostasis.
Both PCA and NMDS plots indicate the compositional
difference in
gut microbiome between the HMLF diet and normal chow
groups.
Bacteroidetes represent the bacterial phylum responsible for
protein
synthesis and polysaccharide absorption. In brief, Bacteroidetes
decom-
pose complex polysaccharides into short-chain fatty acids
(SCFAs),
which mediate intestinal permeability and inflammation [35].
Higher
Bacteroidetes abundance in HMLF diet group implies a higher
risk for
various diseases, such as inflammatory bowel disease, alcoholic
liver
disease and steatohepatitis, due to leaky gut and low-grade

inflamma-
tion [36]. Meanwhile, the phylum Proteobacteria consists of
bacteria
contributing to carbohydrate and protein metabolism, and
oxygen ho-
meostasis inside the intestinal tract [37]. Often overrepresented
in obese
individuals, the Proteobacteria phylum is a marker of dysbiosis
and hence
dysbiosis-related diseases such as colitis and metabolic
disorders [38].
The sequencing results indicates a lower risk of Proteobacteria -
related
dysbiosis in the HMLF diet group. Weight loss is often
associated with
reduced Proteobacteria abundance [38]. Notably, the lower
abundance
of Proteobacteria correlates with the lower body weight gain in
HMLF
diet group. HMLF diet induced gut microbiome alterations
distinct from
obesity.
Correlated to higher risk of cardiovascular disease [39], the
spore-forming Tyzzerella and non-spore-forming Allobaculum
bacteria
are slightly enriched in the HMLF diet group. Notably, the
butyrate-producing Roseburia bacteria was shown in lower
frequency in
the HMLF diet group, consistent with previous case-control
studies that
lower Roseburia abundance correlates with glucose intolerance
and
higher risk of type 2 diabetes [40]. Our results therefore imply
that the
HMLF dietary pattern may pose a higher risk of cardiometabolic

diseases
by altering the gut microbiome profile.
More importantly, an expansion in the family
Porphyromonadaceae
was observed in the fecal samples of HMLF diet group.
Porphyr-
omonadaceae represents the bacterial family under
Bacteroidetes phylum,
potentially exerting harmful effects on the host [41].
Enrichment of
Porphyromonadaceae family is associated with non-alcoholic
fatty liver
disease (NAFLD), characterized by hepatic inflammation and
risk of
nonalcoholic steatohepatitis [42]. Previously, another study
correlated
the higher frequency of Porphyromonadaceae with the
development of
GLP-1 resistance, resulting in the impairment on glucose-
induced insulin
secretion [43]. Our results provide novel hint that the HHcy-
inducing
HMLF diet might impair glucose homeostasis, as revealed by
slight
glucose intolerance and insulin resistance, by increasing the
C.K. Cheng et al.
5
Fig. 3. HMLF diet caused compositional alterations in gut

microbiota of C57BL/6 mice. Beta diversity of gut microbiome,
as represented by (A) PCA and (B) NMDS
plots. In these two plots, samples of two groups are denoted by
data points of different colors and shapes. The further the
distance between points, the more distinct in
species composition. (C) Microbiota composition of fecal
samples at phylum level. Distinct species are denoted by
different colors. Proportion of each species is
represented by the length of columns. (D) Species abundance
heatmap of the fecal samples of the two groups at genus level.
HMLF: high methionine low folate;
NMDS: non-metric multi-dimensional scaling; PCA: principal
component analysis. (For interpretation of the references to
color in this figure legend, the reader is
referred to the Web version of this article.)
C.K. Cheng et al.
6
Porphyromonadaceae abundance, possibly through the induction
of
GLP-1 resistance. In short, a HMLF diet might have posed pro-
diabetic
threats by firstly altering gut microbiome before subsequent
absorp-
tion and metabolism to induce HHcy. However, detailed
mechanism
in-between the alterations in gut microbiome and metabolic
parameters
after a HMLF diet remains elusive. Further studies are needed
to confirm

which gut-derived metabolites contribute to the altered
metabolic pa-
rameters after a HMLF diet.
Data availability statement
Data are available from the authors upon reasonable request.
Author Statement
Cheng Chak Kwong: Conceptualization, Methodology,
Investigation,
Writing – original draft, Visualization, Wang Chenguang:
Investigation,
Resources, Shang Wenbin: Investigation, Resources, Lau Chi
Wai:
Investigation, Resources, Luo Jiang-Yun: Writing – review &
editing,
Wang Li: Writing – review & editing, Huang Yu: Writing –
review &
editing, Supervision, Project administration, Funding
acquisition.
Declaration of competing interest
The authors declare that they have no known competing
financial
interests or personal relationships that could have appeared to
influence
the work reported in this paper.
Acknowledgements
This study was financially supported by the Vice-Chancellor’s
One-
off Discretionary Fund [Grant number 4930709] and Hong Kong

RGC
Senior Research Fellow Scheme 2020/21 [Grant number
SRFS2021-
4S04].
References
[1] A.M. Troen, E. Lutgens, D.E. Smith, I.H. Rosenberg, J.
Selhub, The atherogenic
effect of excess methionine intake, Proc. Natl. Acad. Sci. U. S.
A 100 (2003)
15089–15094, https://doi.org/10.1073/pnas.2436385100.
[2] A. Kumar, H.A. Palfrey, R. Pathak, P.J. Kadowitz, T.W.
Gettys, S.N. Murthy, The
metabolism and significance of homocysteine in nutrition and
health, Nutr. Metab.
14 (2017) 78, https://doi.org/10.1186/s12986-017-0233-z.
[3] V. Fratoni, M.L. Brandi, B vitamins, homocysteine and bone
health, Nutrients 7
(2015) 2176–2192, https://doi.org/10.3390/nu7042176.
[4] M. Shinohara, C. Ji, N. Kaplowitz, Differences in betaine-
homocysteine
methyltransferase expression, endoplasmic reticulum stress
response, and liver
injury between alcohol-fed mice and rats, Hepatology 51 (2010)
796–805, https://
doi.org/10.1002/hep.23391.
[5] L. Brattström, D.E. Wilcken, Homocysteine and
cardiovascular disease: cause or
effect? Am. J. Clin. Nutr. 72 (2000) 315–323,
https://doi.org/10.1093/ajcn/
72.2.315.

[6] A. Rozycka, P.P. Jagodzinski, W. Kozubski, M. Lianeri, J.
Dorszewska,
Homocysteine level and mechanisms of injury in Parkinson’s
disease as related to
MTHFR, MTR, and MTHFD1 genes polymorphisms and L-dopa
treatment, Curr.
Genom. 14 (2013) 534–542, https://doi.org/10.2174/
1389202914666131210210559.
[7] G. Seghieri, M.C. Breschi, R. Anichini, A. DeBellis, L.
Alviggi, I. Maida, F. Franconi,
Serum homocysteine levels are increased in women with
gestational diabetes
mellitus, Metabolism 52 (2003) 720–723,
https://doi.org/10.1016/s0026-0495
(03)00032-5.
Fig. 4. HMLF diet increased the abundance of
porphyromonadaceae family of bacteria. (A) Between-group T-
test analysis and (B) MetaStat of fecal samples of
normal chow and HMLF diet groups. *P < 0.05 or *Q < 0.05 vs
normal chow group. n = 5 per group. (C) Cladogram diagram
showing the microbial species with
significant differences. Circles radiating from inside to outside
represent the taxonomic rank: phylum, class, order, famil y, and
genus. Red nodes in the phylogenetic
tree denote microbial species that play crucial role in the HMLF
diet group. Yellow nodes stand for those microbial species with
no significant difference. HMLF: high
methionine low folate. (For interpretation of the references to
color in this figure legend, the reader is referred to the Web
version of this article.)
C.K. Cheng et al.

https://doi.org/10.1073/pnas.2436385100
https://doi.org/10.1186/s12986-017-0233-z
https://doi.org/10.3390/nu7042176
https://doi.org/10.1002/hep.23391
https://doi.org/10.1002/hep.23391
https://doi.org/10.1093/ajcn/72.2.315
https://doi.org/10.1093/ajcn/72.2.315
https://doi.org/10.2174/1389202914666131210210559
https://doi.org/10.2174/1389202914666131210210559
https://doi.org/10.1016/s0026-0495(03)00032-5
https://doi.org/10.1016/s0026-0495(03)00032-5
7
[8] T. Huang, J. Ren, J. Huang, D. Li, Association of
homocysteine with type 2
diabetes: a meta-analysis implementing Mendelian
randomization approach, BMC
Genom. 14 (2013) 867, https://doi.org/10.1186/1471-2164-14-
867.
[9] M.F. McCarty, Insulin secretion as a potential determinant
of homocysteine levels,
Med. Hypotheses 55 (2000) 454–455,
https://doi.org/10.1054/mehy.1999.1008.
[10] S.H. Kim, G.M. Reaven, Insulin resistance and
hyperinsulinemia: you can’t have
one without the other, Diabetes Care 31 (2008) 1433–1438,
https://doi.org/
10.2337/dc08-0045.
[11] J.B. Meigs, P.F. Jacques, J. Selhub, D.E. Singer, D.M.

Nathan, N. Rifai, R.
B. D’Agostino, P.W. Wilson, Framingham Offspring Study,
Fasting plasma
homocysteine levels in the insulin resistance syndrome: the
Framingham offspring
study, Diabetes Care 24 (2001) 1403–1410,
https://doi.org/10.2337/
diacare.24.8.1403.
[12] E.P.I. Chiang, Y.C. Wang, W.W. Chen, F.Y. Tang, Effects
of insulin and glucose on
cellular metabolic fluxes in homocysteine transsulfuration,
remethylation, S
-adenosylmethionine synthesis, and global deoxyribonucleic
acid methylation,
J. Clin. Endocrinol. Metab. 94 (2009) 1017–1025,
jc.2008-2038.
[13] N. Yang, Z. Yao, L. Miao, J. Liu, X. Gao, H. Fan, Y. Hu,
H. Zhang, Y. Xu, A. Qu,
G. Wang, Novel clinical evidence of an association between
homocysteine and
insulin resistance in patients with hypothyroidism or subclinical
hypothyroidism,
PloS One 10 (2015), e0125922,
https://doi.org/10.1371/journal.pone.0125922.
[14] J. Kumar, E. Ingelsson, L. Lind, T. Fall, No evidence of a
causal relationship
between plasma homocysteine and type 2 diabetes: a mendelian
randomization
study, Front. Cardiovasc. Med. 2 (2015) 11,
fcvm.2015.00011.

[15] M. Rastelli, P.D. Cani, C. Knauf, The gut microbiome
influences host endocrine
functions, Endocr. Rev. 40 (2019) 1271–1284,
https://doi.org/10.1210/er.2018-
00280.
[16] C. Kilkenny, W. Browne, I.C. Cuthill, M. Emerson, D.G.
Altman, NC3Rs Reporting
Guidelines Working Group, Animal research: reporting in vivo
experiments: the
ARRIVE guidelines, Br. J. Pharmacol. 160 (2010) 1577–1579,
https://doi.org/
10.1111/j.1476-5381.2010.00872.x.
[17] C. Bárcena, R. Valdés-Mas, P. Mayoral, C. Garabaya, S.
Durand, F. Rodríguez, M.
T. Fernández-García, N. Salazar, A.M. Nogacka, N. Garatachea,
N. Bossut,
F. Aprahamian, A. Lucia, G. Kroemer, J.M.P. Freije, P.M.
Quirós, C. López-Otín,
Healthspan and lifespan extension by fecal microbiota
transplantation into
progeroid mice, Nat. Med. 25 (2019) 1234–1242,
s41591-019-0504-5.
[18] M. Siervo, O. Shannon, N. Kandhari, M. Prabhakar, W.
Fostier, C. Köchl, J. Rogathi,
G. Temu, B.C.M. Stephan, W.K. Gray, I. Haule, S.M. Paddick,
B.T. Mmbaga,
R. Walker, Nitrate-rich beetroot juice reduces blood pressure in
Tanzanian adults
with elevated blood pressure: a double-blind randomized
controlled feasibility
trial, J. Nutr. 150 (2020) 2460–2468,
https://doi.org/10.1093/jn/nxaa170.

[19] M. Justice, A. Ferrugia, J. Beidler, J.C. Penprase, P.
Cintora, D. Erwin, O. Medrano,
S.M. Brasser, M.Y. Hong, Effects of moderate ethanol
consumption on lipid
metabolism and inflammation through regulation of gene
expression in rats,
Alcohol Alcohol 54 (2019) 5–12,
https://doi.org/10.1093/alcalc/agy079.
[20] K. Fiedorová, M. Radvanský, E. Němcová, H.
Grombǐríková, J. Bosák,
M. Černochová, M. Lexa, D. Šmajs, T. Freiberger, The impact
of DNA extraction
methods on Stool bacterial and fungal microbiota community
recovery, Front.
Microbiol. 10 (2019) 821,
https://doi.org/10.3389/fmicb.2019.00821.
[21] L. Wang, L. Jin, B. Xue, Z. Wang, Q. Peng, Characterizing
the bacterial community
across the gastrointestinal tract of goats: composition and
potential function,
Microbiologyopen 8 (2019), e00820,
https://doi.org/10.1002/mbo3.820.
[22] J.G. Caporaso, J. Kuczynski, J. Stombaugh, K. Bittinger,
F.D. Bushman, E.
K. Costello, N. Fierer, A.G. Peña, J.K. Goodrich, J.I. Gordon,
G.A. Huttley, S.
T. Kelley, D. Knights, J.E. Koenig, R.E. Ley, C.A. Lozupone,
D. McDonald, B.
D. Muegge, M. Pirrung, J. Reeder, J.R. Sevinsky, P.J.
Turnbaugh, W.A. Walters,
J. Widmann, T. Yatsunenko, J. Zaneveld, R. Knight, QIIME
allows analysis of high-

throughput community sequencing data, Nat. Methods 7 (2010)
335–336, https://
doi.org/10.1038/nmeth.f.303.
[23] R.C. Edgar, UPARSE: highly accurate OTU sequences
from microbial amplicon
reads, Nat. Methods 10 (2013) 996–998,
https://doi.org/10.1038/nmeth.2604.
[24] Q. Wang, G.M. Garrity, J.M. Tiedje, J.R. Cole, Naive
Bayesian classifier for rapid
assignment of rRNA sequences into the new bacterial taxonomy,
Appl. Environ.
Microbiol. 73 (2007) 5261–5267,
https://doi.org/10.1128/AEM.00062-07.
[25] C. Quast, E. Pruesse, P. Yilmaz, J. Gerken, T. Schweer, P.
Yarza, J. Peplies, F.
O. Glöckner, The SILVA ribosomal RNA gene database project:
improved data
processing and web-based tools, Nucleic Acids Res. 41 (2013)
D590–D596, https://
doi.org/10.1093/nar/gks1219.
[26] P. Wang, J. Gao, W. Ke, J. Wang, D. Li, R. Liu, Y. Jia, X.
Wang, X. Chen, F. Chen,
X. Hu, Resveratrol reduces obesity in high-fat diet-fed mice via
modulating the
composition and metabolic function of the gut microbiota, Free
Radic. Biol. Med.
156 (2020) 83–98,
https://doi.org/10.1016/J.FREERADBIOMED.2020.04.013.
[27] L. yanPei, Y. shiKe, H. huZhao, L. Wang, C. Jia, W.
zhiLiu, Q. huiFu, M. niShi,
J. Cui, S. chunLi, Role of colonic microbiota in the

pathogenesis of ulcerative
colitis, BMC Gastroenterol. 19 (2019) 1–11,
https://doi.org/10.1186/s12876-019-
0930-3.
[28] H. Škovierová, E. Vidomanová, S. Mahmood, J. Sopková,
A. Drgová, T. Červeňová,
E. Halašová, J. Lehotský, The molecular and cellular effect of
homocysteine
metabolism imbalance on human health, Int. J. Mol. Sci. 17
(2016), https://doi.
org/10.3390/ijms17101733.
[29] C. Tinelli, A.D. Pino, E. Ficulle, et al.,
Hyperhomocysteinemia as a Risk Factor and
Potential Nutraceutical Target for Certain Pathologies, Front
Nutr. 6 (2019), 49,
https://doi.org/10.3389/fnut.2019.00049.
[30] B.R. Price, D.M. Wilcock, E.M. Weekman,
Hyperhomocysteinemia as a risk factor
for vascular contributions to cognitive impairment and
dementia, Front. Aging
Neurosci. 10 (2018), https://doi.org/10.3389/fnagi.2018.00350.
[31] J.B. Dixon, M.E. Dixon, P.E. O’Brien, Elevated
homocysteine levels with weight
loss after Lap-Band® surgery: higher folate and vitamin B12
levels required to
maintain homocysteine level, Int. J. Obes. 25 (2001) 219–227,
https://doi.org/
10.1038/sj.ijo.0801474.
[32] M. Namdari, A. Abadi, S.M. Taheri, M. Rezaei, N.
Kalantari, N. Omidvar, Effect of
folic acid on appetite in children: ordinal logistic and fuzzy

logistic regressions,
Nutrition 30 (2014) 274–278,
https://doi.org/10.1016/j.nut.2013.08.008.
[33] M.F. Gursu, G. Baydas, G. Cikim, H. Canatan, Insulin
increases homocysteine levels
in a dose-dependent manner in diabetic rats, Arch. Med. Res. 33
(2002) 305–307,
https://doi.org/10.1016/S0188-4409(01)00379-4.
[34] D.E. Platt, E. Hariri, P. Salameh, M. Merhi, N. Sabbah, M.
Helou, F. Mouzaya,
R. Nemer, Y. Al-Sarraj, H. El-Shanti, A.B. Abchee, P.A.
Zalloua, Type II diabetes
mellitus and hyperhomocysteinemia: a complex interaction,
Diabetol. Metab.
Syndrome 9 (2017) 1–7, https://doi.org/10.1186/s13098-017-
0218-0.
[35] C. Yu, Z. Su, Y. Li, Y. Li, K. Liu, F. Chu, T. Liu, R. Chen,
X. Ding, Dysbiosis of gut
microbiota is associated with gastric carcinogenesis in rats,
Biomed. Pharmacother.
126 (2020) 110036,
https://doi.org/10.1016/j.biopha.2020.110036.
[36] H. Fukui, Increased intestinal permeability and decreased
barrier function: does it
really influence the risk of inflammation? Inflamm. Intest. Dis.
1 (2016) 135–145,
https://doi.org/10.1159/000447252.
[37] C.D. Moon, W. Young, P.H. Maclean, A.L. Cookson, E.N.
Bermingham,
Metagenomic insights into the roles of Proteobacteria in the
gastrointestinal

microbiomes of healthy dogs and cats, Microbiologyopen 7
(2018), https://doi.
org/10.1002/mbo3.677.
[38] G. Rizzatti, L.R. Lopetuso, G. Gibiino, C. Binda, A.
Gasbarrini, Proteobacteria: a
common factor in human diseases, BioMed Res. Int. 2017
(2017), https://doi.org/
10.1155/2017/9351507.
[39] S. Ascher, C. Reinhardt, The gut microbiota: an emerging
risk factor for
cardiovascular and cerebrovascular disease, Eur. J. Immunol. 48
(2018) 564–575,
https://doi.org/10.1002/eji.201646879.
[40] M. Gurung, Z. Li, H. You, R. Rodrigues, D.B. Jump, A.
Morgun, N. Shulzhenko, Role
of gut microbiota in type 2 diabetes pathophysiology,
EBioMedicine 51 (2020)
102590, https://doi.org/10.1016/j.ebiom.2019.11.051.
[41] L. Yu, N. Qiao, T. Li, R. Yu, Q. Zhai, F. Tian, J. Zhao, H.
Zhang, W. Chen, Dietary
supplementation with probiotics regulates gut microbiota
structure and function in
Nile tilapia exposed to aluminum, PeerJ 2019 (2019),
peerj.6963.
[42] E. Blacher, M. Levy, E. Tatirovsky, E. Elinav,
Microbiome-modulated metabolites at
the interface of host immunity, J. Immunol. 198 (2017) 572–
580, https://doi.org/
10.4049/jimmunol.1601247.

[43] E. Grasset, A. Puel, J. Charpentier, X. Collet, J.E.
Christensen, F. Tercé, R. Burcelin,
A specific gut microbiota dysbiosis of type 2 diabetic mice
induces GLP-1 resistance
through an enteric NO-dependent and gut-brain Axis
mechanism, Cell Metabol. 25
(2017) 1075–1090, https://doi.org/10.1016/j.cmet.2017.04.013,
e5.
C.K. Cheng et al.
https://doi.org/10.1186/1471-2164-14-867
https://doi.org/10.1054/mehy.1999.1008
https://doi.org/10.2337/dc08-0045
https://doi.org/10.2337/dc08-0045
https://doi.org/10.2337/diacare.24.8.1403
https://doi.org/10.2337/diacare.24.8.1403
https://doi.org/10.1210/jc.2008-2038
https://doi.org/10.1210/jc.2008-2038
https://doi.org/10.1371/journal.pone.0125922
https://doi.org/10.3389/fcvm.2015.00011
https://doi.org/10.3389/fcvm.2015.00011
https://doi.org/10.1210/er.2018-00280
https://doi.org/10.1210/er.2018-00280
https://doi.org/10.1111/j.1476-5381.2010.00872.x
https://doi.org/10.1111/j.1476-5381.2010.00872.x
https://doi.org/10.1038/s41591-019-0504-5
https://doi.org/10.1038/s41591-019-0504-5
https://doi.org/10.1093/jn/nxaa170
https://doi.org/10.1093/alcalc/agy079
https://doi.org/10.3389/fmicb.2019.00821
https://doi.org/10.1002/mbo3.820
https://doi.org/10.1038/nmeth.f.303
https://doi.org/10.1038/nmeth.f.303
https://doi.org/10.1038/nmeth.2604
https://doi.org/10.1128/AEM.00062-07

https://doi.org/10.1093/nar/gks1219
https://doi.org/10.1093/nar/gks1219
https://doi.org/10.1016/J.FREERADBIOMED.2020.04.013
https://doi.org/10.1186/s12876-019-0930-3
https://doi.org/10.1186/s12876-019-0930-3
https://doi.org/10.3390/ijms17101733
https://doi.org/10.3390/ijms17101733
https://doi.org/10.3389/fnut.2019.00049
https://doi.org/10.3389/fnagi.2018.00350
https://doi.org/10.1038/sj.ijo.0801474
https://doi.org/10.1038/sj.ijo.0801474
https://doi.org/10.1016/j.nut.2013.08.008
https://doi.org/10.1016/S0188-4409(01)00379-4
https://doi.org/10.1186/s13098-017-0218-0
https://doi.org/10.1016/j.biopha.2020.110036
https://doi.org/10.1159/000447252
https://doi.org/10.1155/2017/9351507
https://doi.org/10.1155/2017/9351507
https://doi.org/10.1002/eji.201646879
https://doi.org/10.1016/j.ebiom.2019.11.051
https://doi.org/10.7717/peerj.6963
https://doi.org/10.7717/peerj.6963
https://doi.org/10.4049/jimmunol.1601247
https://doi.org/10.4049/jimmunol.1601247
https://doi.org/10.1016/j.cmet.2017.04.013A high methionine
and low folate diet alters glucose homeostasis and gut
microbiome1 Introduction2 Materials and methods2.1 Animal
protocol2.2 Plasma Hcy measurement by ELISA2.3 Lipid
profile determination2.4 Blood glucose determination2.5 DNA
extraction from fecal samples2.6 16S rRNA sequencing: library
preparation and sequencing2.7 Data analysis for bacterial 16S
rRNA sequencing2.8 Statistical analysis3 Results3.1 A HMLF
diet induced HHcy in C57BL/6 mice3.2 HMLF diet impaired
glucose homeostasis in diet-induced HHcy mice3.3 HMLF diet

altered the gut microbiome profile in diet-induced HHcy
mice3.4 HMLF diet increased the abundance of
porphyromonadaceae family of bacteria4 DiscussionData
availability statementAuthor StatementDeclaration of competing
interestAcknowledgementsReferences
ARTICLE
The lytic polysaccharide monooxygenase CbpD
promotes Pseudomonas aeruginosa virulence in
systemic infection
Fatemeh Askarian 1✉, Satoshi Uchiyama2,14, Helen
Masson3,14, Henrik Vinther Sørensen 4, Ole Golten1,
Anne Cathrine Bunæs1, Sophanit Mekasha1, Åsmund Kjendseth
Røhr1, Eirik Kommedal 1,
Judith Anita Ludviksen5, Magnus Ø. Arntzen 1, Benjamin
Schmidt2, Raymond H. Zurich2,
Nina M. van Sorge 6,7,8, Vincent G. H. Eijsink1, Ute Krengel
4, Tom Eirik Mollnes5,9,10,11,
Nathan E. Lewis 2,3,12, Victor Nizet 2,13✉ & Gustav Vaaje-
Kolstad 1✉
The recently discovered lytic polysaccharide monooxygenases
(LPMOs), which cleave
polysaccharides by oxidation, have been associated with
bacterial virulence, but supporting
functional data is scarce. Here we show that CbpD, the LPMO
of Pseudomonas aeruginosa, is a

chitin-oxidizing virulence factor that promotes survival of the
bacterium in human blood. The
catalytic activity of CbpD was promoted by azurin and
pyocyanin, two redox-active virulence
factors also secreted by P. aeruginosa. Homology modeling,
molecular dynamics simulations,
and small angle X-ray scattering indicated that CbpD is a
monomeric tri-modular enzyme
with flexible linkers. Deletion of cbpD rendered P. aeruginosa
unable to establish a lethal
systemic infection, associated with enhanced bacterial clearance
in vivo. CbpD-dependent
survival of the wild-type bacterium was not attributable to
dampening of pro-inflammatory
responses by CbpD ex vivo or in vivo. Rather, we found that
CbpD attenuates the terminal
complement cascade in human serum. Studies with an active site
mutant of CbpD indicated
that catalytic activity is crucial for virulence function. Finally,
profiling of the bacterial and
splenic proteomes showed that the lack of this single enzyme
resulted in substantial re-
organization of the bacterial and host proteomes. LPMOs
similar to CbpD occur in other

pathogens and may have similar immune evasive functions.
https://doi.org/10.1038/s41467-021-21473-0 OPEN
1 Faculty of Chemistry, Biotechnology and Food Science,
Norwegian University of Life Sciences (NMBU), Ås, Norway. 2
Division of Host-Microbe Systems &
Therapeutics, Department of Pediatrics, UC San Diego, La
Jolla, CA, USA. 3 Department of Pediatrics, University of
California, San Diego, School of
Medicine, La Jolla, CA, USA. 4 Department of Chemistry,
University of Oslo, Oslo, Norway. 5 Research Laboratory,
Nordland Hospital, Bodø, Norway.
6 Department of Medical Microbiology, University Medical
Center Utrecht, Utrecht University, Utrecht, The Netherlands. 7
Department of Medical
Microbiology and Infection Prevention, Amsterdam University
Medical Center, University of Amsterdam, Amsterdam, The
Netherlands. 8 Netherlands
Reference Laboratory for Bacterial Meningitis, Amsterdam
University Medical Center, Amsterdam, The Netherlands. 9
K.G. Jebsen TREC, Faculty of Health
Sciences, UiT- The Arctic University of Norway, Tromsø,
Norway. 10 Department of Immunology, Oslo University
Hospital, and K.G. Jebsen IRC, University
of Oslo, Oslo, Norway. 11 Center of Molecular Inflammation
Research, Norwegian University of Science and Technology,
Trondheim, Norway. 12 Novo
Nordisk Foundation Center for Biosustainability at UC San
Diego, University of California, San Diego, School of Medicine,
La Jolla, CA, USA. 13 Skaggs School
of Pharmacy and Pharmaceutical Sciences, UC San Diego, La
Jolla, CA, USA. 14These authors contributed equally: Satoshi
Uchiyama, Helen Masson.
✉email: [email protected]; [email protected]; [email protected]

NATURE COMMUNICATIONS | (2021) 12:1230 |
https://doi.org/10.1038/s41467-021-21473-0 |
www.nature.com/naturecommunications 1
12
3
4
5
6
7
8
9
0
()
:,;
http://crossmark.crossref.org/dialog/?doi=10.1038/s41467-021-
21473-0&domain=pdf
21473-0&domain=pdf
21473-0&domain=pdf
21473-0&domain=pdf
http://orcid.org/0000-0002-7438-4671
http://orcid.org/0000-0002-7438-4671
http://orcid.org/0000-0002-7438-4671
http://orcid.org/0000-0002-7438-4671
http://orcid.org/0000-0002-7438-4671
http://orcid.org/0000-0003-3057-5668
http://orcid.org/0000-0003-3057-5668
http://orcid.org/0000-0003-3057-5668
http://orcid.org/0000-0003-3057-5668
http://orcid.org/0000-0003-3057-5668
http://orcid.org/0000-0002-7502-4184
http://orcid.org/0000-0002-7502-4184

http://orcid.org/0000-0002-7502-4184
http://orcid.org/0000-0002-7502-4184
http://orcid.org/0000-0002-7502-4184
http://orcid.org/0000-0002-1803-727X
http://orcid.org/0000-0002-1803-727X
http://orcid.org/0000-0002-1803-727X
http://orcid.org/0000-0002-1803-727X
http://orcid.org/0000-0002-1803-727X
http://orcid.org/0000-0002-2695-5863
http://orcid.org/0000-0002-2695-5863
http://orcid.org/0000-0002-2695-5863
http://orcid.org/0000-0002-2695-5863
http://orcid.org/0000-0002-2695-5863
http://orcid.org/0000-0001-6688-8151
http://orcid.org/0000-0001-6688-8151
http://orcid.org/0000-0001-6688-8151
http://orcid.org/0000-0001-6688-8151
http://orcid.org/0000-0001-6688-8151
http://orcid.org/0000-0001-7700-3654
http://orcid.org/0000-0001-7700-3654
http://orcid.org/0000-0001-7700-3654
http://orcid.org/0000-0001-7700-3654
http://orcid.org/0000-0001-7700-3654
http://orcid.org/0000-0003-3847-0422
http://orcid.org/0000-0003-3847-0422
http://orcid.org/0000-0003-3847-0422
http://orcid.org/0000-0003-3847-0422
http://orcid.org/0000-0003-3847-0422
http://orcid.org/0000-0002-3077-8003
http://orcid.org/0000-0002-3077-8003
http://orcid.org/0000-0002-3077-8003
http://orcid.org/0000-0002-3077-8003
http://orcid.org/0000-0002-3077-8003

www.nature.com/naturecommunications
L
ytic polysaccharide monooxygenases (LPMOs) are copper-
dependent enzymes that cleave glycosidic linkages by oxi -
dation1–4. Most characterized LPMOs act on recalcitrant
polysaccharides such as cellulose (β-1,4(Glc)n) and chitin (β-
1,4
(GlcNAc)n)5. Hence, current LPMO research focuses mainly on
biomass degradation from fundamental and applied
perspectives.
All LPMOs possess a conserved active site with two histidine
residues coordinating a copper ion in a so-called “histidine
brace”3. Catalysis entails the reduction of the copper followed
by
activation of an oxygen-containing co-substrate to oxidize the
C1
or C4 carbon of the polysaccharide substrate (reviewed in ref.
5).
The co-substrate may be either dioxygen or hydrogen peroxide,
the latter yielding catalytic rates several orders of magni tude
higher than the former6–8. LPMOs target multiple soluble and
insoluble substrates, demonstrating their catalytic versatility
(reviewed in ref. 9). Despite being primarily associated with
biomass degradation machineries9,10, genes encoding LPMOs
are
found in many pathogenic organisms and have been speculated
to
contribute to bacterial physiological processes or pathogenicity
phenotypes11–19. However, scant data exist to substantiate
LPMO
function during bacterial infection.

One important pathogen that expresses a putative LPMO is
Pseudomonas aeruginosa (PA), a frequently multidrug-resistant
microorganism associated with nosocomial infections of
surgical
sites, urinary tract, lung, and blood20. PA accounts for ~20% of
hospital-associated Gram-negative bacteremia cases, with high
mortality (>30%), particularly in patients receiving
inappropriate
initial antimicrobial treatment21,22. PA is known to explicitly
disarm several aspects of the innate host defense and erodes
efficacy of first-line antibiotics through an array of immune
evasion factors and antibiotic-resistant determinants (reviewed
in
refs. 23,24).
The putative LPMO expressed by PA is called “chitin-binding
protein D” (CbpD). Many publications refer to CbpD as a viru-
lence factor (e.g., ref. 25 and references within), yet no direct
functional evidence exists. Prevalent across PA clinical isolates
and cystic fibrosis-associated clones26,27, CbpD is expressed
under quorum-sensing control28–32, secreted through the Type
II
secretion machinery33,34, and modified by several post-
translational mechanisms25,35,36. The induction of cbpD tran-
scription by human respiratory mucus in mucoidal PA37 and its
high abundance in the secretome of an acute transmissible
cystic
fibrosis-associated strain38, are two observations that strongly
suggest an important function of this LPMO in disease
pathogenesis.
In this work, we combine studies of CbpD structure and
enzymatic activity with analysis of its immune evasion
properties
ex vivo and in vivo to unravel the function of this protein in PA
bloodstream infection. We show that this highly conserved and

prevalent tri-modular LPMO is a chitin-oxidizing virulence
factor
that enhances resistance of the bacterium to whole blood-
mediated killing and virulence during ex vivo and in vivo infec-
tion by attenuation of the terminal complement cascade.
Results
CbpD has a monomeric, tri-modular structure. Analysis of the
CbpD amino acid sequence (Uniprot ID Q9I589) revealed a
multidomain architecture comprising an N-terminal signal
peptide
(residues 1–25) followed by an auxiliary activity family 10
LPMO
(residues 26–207, AA10), a module of unknown function
(residues
217–315, annotated as “GbpA2” domain in the Pfam
classification,
here referred to as module X or MX) and finally a C-terminal
family 73 carbohydrate-binding module (CBM73; residues
330–389) (Fig. 1a). The protein is highly conserved and
prevalent
among PA clinical and non-clinical isolates (Supplementary Fig.
1
and Supplementary Results). Phylogenetic analysis of pre-
dominantly pathogenic Gram-negative and Gram-positive
bacteria
placed the CbpD AA10 sequence in a cluster containing
homologs
from Legionella spp. (Supplementary Fig. 2). Most of the
closely
related LPMOs possess a modular architecture similar to CbpD.
To gain further insight into the CbpD structure, homology mod-
eling, molecular dynamics simulations, and small-angle X-ray
scattering (SAXS) analysis were performed. Combined, these
approaches revealed that CbpD is a monomeric, elongated

protein,
whose conformation may be affected by post-translational mod-
ifications (Fig. 1a–c and Supplementary Figs. 3−5,
Supplementary
Tables 1–3, and Supplementary Results).
CbpD enzymatic performance is influenced by other redox-
active virulence factors. The CbpD active site resembles those
found in other LPMO10s, containing a copper ion coordinated
by
two histidine residues (H26 and H129) and a semi-conserved
glutamic acid residue (E199) (Fig. 1b). The recombinant,
copper-
saturated full-length enzyme produced in Escherichia coli
(rCbpDEC) was active toward β-chitin, resulting in the release
of
chitooligosaccharide aldonic acids ranging from 2 to 8 in the
degree of polymerization (Fig. 1d). The CbpD catalytic module
(AA10) alone was also active toward β-chitin, whereas no
breakdown products were generated by the MX or CBM73
modules (Fig. 1d). CbpD chitin oxidation activity required an
intact active site, as the mutation of one of the active site histi -
dines (H129: rCbpDEC-H129A) to alanine (Ala) abolished
CbpD
activity (Supplementary Fig. 6a, b). When secreted by PA,
CbpD
is post-translationally modified (refs. 25,35,36 and
Supplementary
Table 1), which may influence enzymatic performance. Full -
length CbpD variants produced and purified from heterologous
or homologous expression in E. coli (rCbpDEC) or wild-type
(WT) PA14 (P. aeruginosa UCBPP-PA14) (rCbpDPA), respec-
tively, were equally active on the β-chitin model substrate (Sup-
plementary Fig. 6c).
Experiments using full-length and truncated forms of CbpD

demonstrated that all variants bound chitin, albeit with different
binding kinetics (Supplementary Fig. 7a). To explore other
potential CbpD substrates, the binding properties of rCbpDEC
and rCbpDPA were screened using a mammalian glycan
array that contains 585 glycan structures present in mammals,
including a variety of structures found in mucins such as core 1,
type 1, blood group, and Lewis type glycans39. Binding
was not detected to any glycan for either CbpD variant at 5
and 50 µg ml−1 (Supplementary Data 1), nor was binding to
GlcNAc6 (water-soluble chitooligosaccharide) observed
(Supple-
mentary Fig. 7b and Supplementary Data 1).
In addition to CbpD, PA secretes the redox-active virulence
factors azurin (AZU), a copper protein that contributes to
electron transfer reactions40, and pyocyanin (PCN), a
zwitterionic
secondary metabolite capable of oxidizing and reducing other
molecules, including reduction of O2 to H2O241,42. We
investi-
gated the influence of azurin and pyocyanin on CbpD enzymatic
performance. Pyocyanin boosted CbpD activity towards β-chitin
using ascorbate as an external electron donor (Fig. 1e), perhaps
due to pyocyanin catalyzing the formation of H2O2, which is an
LPMO co-substrate. Excessive pyocyanin concentrations (1
mM)
led to rapid inactivation of CbpD, similar to what is commonly
observed for LPMOs exposed to high H2O2 concentrations
(Fig. 1e). Azurin, on the other hand, has been proposed to
protect
PA from H2O243, but the addition of this copper protein to a
reaction containing pyocyanin, CbpD, and ascorbate, did not
abolish CbpD activity (Fig. 1f). Rather, 1 μM azurin activated
1 μM copper-free CbpD, which otherwise was inactive (Fig. 1f),
suggesting that CbpD may be able to obtain copper from azurin.

ARTICLE NATURE COMMUNICATIONS |
https://doi.org/10.1038/s41467-021-21473-0
2 NATURE COMMUNICATIONS | (2021) 12:1230 |
https://doi.org/10.1038/s41467-021-21473-0 |
Transcription of cbpD is subject to environme ntal variations.
Absolute quantification of cbpD transcription in various condi -
tions revealed several noticeable features. First, cbpD is tran-
scribed during growth in Luria–Bertani medium (LB). When
PA14 (WT) was grown in M9 minimal medium supplemented
with glucose and casamino acids (M9PA; a growth medium
mimicking nutritional deprivation), cbpD abundance was ~1500
copies/100 ng RNA in mid-exponential phase (OD600 = 0.6),
approximately twice the levels seen in LB medium (Fig. 2a).
The
addition of normal human serum (NHS; 10% v/v, 30 min) to
M9PA (M9PA/NHS) at OD600 nm = 0.6 did not change cbpD
transcript abundance (Fig. 2a). Thus, under the conditions tested
here, cbpD is constitutively transcribed, with expression levels
being subject to environmental variations. Transcription of
LPMO-encoding genes in two other pathogenic bacteria was
also
influenced by the growth condition (Supplementary Fig. 8a, b
and
Supplementary Results).
Loss of cbpD distinctly influences the PA proteome under host-
mimicking conditions. Following verification of cbpD tran-
scription, the functional impact of CbpD on PA pathophysiology
was scrutinized. This was achieved by comparing the proteomic
profiles of the WT parent strain PA and its isogenic mutant

PA14ΔcbpD (hereinafter referred to as “ΔCbpD”) grown to mid-
logarithmic phase (OD600 = 0.6) in LB medium or in a tissue
culture medium (RPMI supplemented with 10% LB) that better
mimics in vivo conditions, with or without supplemental normal
human serum (NHS). Secretion of CbpD in LB and RPMI was
verified by quantitative proteomic analysis of the secretome
(Supplementary Fig. 8c).
Our analysis identified 2128 proteins, of which 1202 were
shared for both strains under all three conditions (Fig. 2b and
Supplementary Data 2). Principal component analysis (PCA)
(Fig. 2c) and hierarchical clustering (Fig. 2e) showed strong
coherence between biological replicates and a different protein
AA10 Module X CBM73
1
26
SP
207
217
315
330
Y37
K48
Y65
K97

S104
S108
T109
Y143
S152
S161
S164
K185
S195
S197
S208
S210 S251
T253
K262
Y308
Y287
S276
T277
a
T183
H26 H129 E199
AA10

Module X
H129
E199
Active site
b
CBM73
Standard
rCbpDEC (FL)
No protein
c
Time (min)
4 6 8 1210
0
25
50
75
100
125
150
DP2ox DP3ox
DP4ox

DP5ox
DP6ox
m
A
U
/Azu
Time (min)
4 6 8 1210
0
40
80
120
160
200
DP2ox DP3ox
DP4ox
DP5ox
DP6ox
d
rCbpDPA (control)

Azu (control)
rCbpDPA
e
Standard
f
S28 S49
K53 K124 K147 K166
R173 K182
Y335
K354
K359
K368
rCbpDEC (AA10)
rCbpDEC (MX+CBM73)
Time (min)
8 9 11107
DP6ox
DP5ox
DP4ox
0
100

200
300
400
500
m
A
U
100 µM PCN
1000 µM PCN
10 µM PCN
0 µM PCN
m
A
U
Fig. 1 Multidomain structure of CbpD and evaluation of its
activity on a model substrate. a Schematic representation of the
domain architecture of
CbpD. The full-length protein contains a signal peptide (SP), an
N-terminal AA10-type LPMO domain (AA10) followed by a
module with unknown function
(module X or MX) and a C-terminal CBM domain (CBM73).
The residue boundaries for each domain, as well as potential
post-translational modification
(PTM) sites based on refs. 25,35,36 and Supplementary Table 1,
are labeled; phosphorylation sites are indicated above and
lysine/arginine modifications

below the domain illustration. PTMs identified in this study
(Supplementary Table 1) are colored red. b Homology model of
CbpD generated with Raptor-
X92 showing flexible linkers in an extended conformation. The
active site is indicated by a yellow partially transparent square.
c SAXS model of monomeric
CbpD (χ2 = 2.35; produced with Pepsi-SAXS114; SASBDB ID:
SASDK42), superimposed onto the ab initio SAXS model
“envelope” (produced with
DAMMIF111,112 based on an average of 20 calculated models).
Panels b and c were generated using PyMol. d Product
formation by 1 µM of rCbpDEC variants
(FL: full-length; AA10: AA10 module only; MX + CBM73: the
MX and CBM73 domains only) after a 2 h reaction at 37 °C
with 10 mg ml−1 β-chitin in
20 mM Tris-HCL pH 7.0, with 1 mM ascorbate as reducing
agent, analyzed by HILIC. The degrees of polymerization (DP)
of oxidized chitooligosaccharide
aldonic acids in a standard sample are indicated. Control
reactions without ascorbate did not show product formation. e
HILIC analysis of reaction products
emerging from a reaction of 1 µM rCbpDEC with 10 mg ml−1
β-chitin, 250 µM ascorbate in 20 mM Tris-HCl pH 7.0, and
serial dilutions of pyocyanin (PCN)
for 2 h at 37 °C. The chromatograms have been offset on the y
axis to enable visual interpretation of their quantitative
magnitude. Chromatograms for
reactions containing serial dilutions of PCN and with or without
various reductants that were sampled at different time points are
shown in Supplementary
Fig. 7c. f HILIC analysis of reaction products emerging from a
reaction of 1 μM of copper-free full-length rCbpDPA with 10
mg ml−1 β-chitin in 20 mM Tris-
HCL pH 7.0, 100 µM pyocyanin (PCN), 250 µM ascorbate, in
the presence or absence of 1 µM azurin (Azu) after incubation
for 2 h at 37 °C. Control

reactions without added ascorbate did not show product
formation (Supplementary Fig. 7d).
NATURE COMMUNICATIONS |
https://doi.org/10.1038/s41467-021-21473-0 ARTICLE
https://doi.org/10.1038/s41467-021-21473-0 |
expression response of ΔCbpD compared to WT, particularly in
the presence of human serum (RPMI/NHS). CbpD is part of
cluster V (Fig. 2e, Supplementary Fig. 8e, and Supplementary
Data 5), which is enriched in proteins related to transporter
activity (Supplementary Table 4). Volcano plot analyses (Fig.
2d
and Supplementary Data 3) showed that exposure to NHS
resulted in significant differential regulation of 589 proteins in
ΔCbpD vs. WT (Fig. 2d, right panel), compared to a more
modest
number of differentially expressed proteins in LB (n = 50)
(Fig. 2d, left panel) or RPMI (n = 136) (Fig. 2d, middle panel).
KEGG Pathway enrichment analysis of the differentially
regulated
proteins (ΔCbpD vs. WT) revealed that loss of CbpD alters
https://doi.org/10.1038/s41467-021-21473-0
https://doi.org/10.1038/s41467-021-21473-0 |

metabolism and protein synthesis (RPMI/NHS) and impairs the
protein secretion apparatus (RPMI) of the ΔCbpD strain under
host-mimicking conditions, hinting at the potential importance
of
the LPMO during infection (Fig. 2f). The most profound
difference between WT and ΔCbpD in presence of NHS were
reflected in clusters II and IV (Fig. 2e, Supplementary Fig. 8e,
and
Supplementary Data 5), which were also associated with
decreased metabolic processes and increased translational
activity, respectively (Fig. 2f and Supplementary Table 4). No
significant pathway enrichments were observed when comparing
the two strains grown in LB medium.
Several proteins associated with outer membrane homeostasis
in PA44–48 were significantly downregulated (Fig. 2g and
Supplementary Data 3, 4, and 6) upon loss of CbpD or only
detected in WT parent strain under all examined conditions.
Channel proteins (OprM, OprD, OprQ, OprE, OprB, and OprF),
LPS-modifying/assembly proteins (PagL, LptE, LptD), outer
membrane protein assembly Bam complex (BamB, BamE,
BamD)
and VacJ (Fig. 2g and Supplementary Data 6) were among those
proteins. In addition, several proteins associated with the type
II
secretion system (SecD, SecY, and XcpQ)49 were commonly
downregulated or not detected upon loss of CpbD (Supplemen-
tary Data 6). Despite the roles of these proteins in structural
stability of the PA outer membrane, both strains displayed
comparable detergent resistance upon growth in LB (Supple-
mentary Fig. 9a). Given the function of LptE, LptD and PagL in

lipopolysaccharide (LPS) biogenesis or cell surface localization,
LPS was extracted from mid-log phase WT and the ΔCbpD
mutant. Compositional and structural analysis revealed neither
difference between the lipid-A structures (Supplementary Fig.
9d)
nor in the O-antigen and outer core monosaccharide
composition
(Supplementary Fig. 9c). However, the inner core monosacchar -
ide composition had a ~50% higher amount of 3-deoxy-D-
manno-oct-2-ulosonic acid (Kdo) in the extracted LPS from
ΔCbpD compared to WT (Supplementary Fig. 9b), indicating
that CbpD has a direct or indirect influence on LPS
composition.
When grown in the tissue culture medium RPMI, several
virulence-related proteins (Supplementary Data 3; marked in
bold), including proteases (e.g., LasB), pyoverdine-related bio-
synthesis (e.g., FpvA, PvdL, PvdH, PvdA), quorum-sensing-
related regulators (e.g., PqsH), virulence-associated transcrip-
tional factors (e.g., FleQ), and pilus synthesis (e.g., ChpC,
PilA),
were significantly repressed in the ΔCbpD strain compared to
WT (Fig. 2h). Finally, upon exposure to NHS, a wider set of
functionally interconnected regulators (e.g., Vfr, AlgU, MvaT,
AlgR, AlgQ, Fur, FliA) and proteins associated with virulence
or
environmental versatility of PA (Supplementary Data 3; marked
in bold), were differentially up- or downregulated in the mutant
lacking cbpD (Fig. 2i, j). Several super-regulators of quorum-
sensing (QS) in PA (AlgR, DksA, MvaT, and Vfr)50, or QS by-
products such as phenazine synthesis (PhzM, PhzB1, and PhnB),
T3SS regulator (Vfr), or virulence-associated transcriptional
factors (AlgU, AlgR)51 were upregulated (Fig. 2j, I) in ΔCbpD
compared to WT in the presence of NHS. In addition, several
proteins involved in the biosynthesis of antibiotics (BkdB,

AcsA,
Zwf, PurH, TpiA, GcvH, GlyA) or flagellar-associated proteins
(FliD, FliC, FliA) were among the differentially (up/down)
regulated proteins in ΔCbpD vs. WT (Fig. 2i, j). Proteome
modulation could be a protective strategy to aid the pathogen
survival within a hostile host environment. The different
modulation strategies employed by WT and ΔCbpD PA strain
suggest that CbpD action directly or indirectly affects several
cellular processes associated with metabolism and pathogenicity
upon exposure to host environmental conditions.
CbpD promotes PA resistance to human whole blood killing
ex vivo. PA is an important cause of nosocomial bacteremia
with
high associated mortality52, and as NHS resulted in distinct
proteome modulation in ΔCbpD compared to WT, we next
explored if the deletion of CbpD affected PA survival in freshly
collected human blood. While the deletion of cbpD did not alter
PA growth in standard media (Supplementary Fig. 10), it sig-
nificantly reduced PA survival in human blood compared to the
parent strain (Fig. 3a, left panel). Complementation of the
ΔCbpD mutant by plasmid-mediated expression of cbpD
(ΔCbpD:CbpD) restored CbpD expression (Supplementary
Fig. 11a) and significantly enhanced human blood survival
compared to the ΔCbpD mutant empty vector control (ΔCbpD:
Mock) (Fig. 3a, right panel). Control experiments showed that
the
contribution of CbpD to PA survival in human blood was not
attributable to differences in bacterial resistance to neutrophil
phagocytosis, H2O2-mediated killing, the cytotoxicity of the
recombinant full-length protein (rCbpDPA) against immune
cells
(Supplementary Fig. 11b–d and Supplementary Results). CbpD
expression did neither change in the presence of increasing
concentrations of succinate (Supplementary Fig. 8f), an
important

LPS-controlled signaling metabolite produced by phagocytes
(reviewed in ref. 53).
While CbpD did not interact with 40 receptors associated
with innate immune cell function (Supplementary Fig. 11e),
Fig. 2 CbpD expression and proteomics analysis. a Expression
of cbpD in wild-type (WT) PA14 was evaluated for the mid-
exponential growth phase
(OD600 = 0.6) in LB medium, and M9PA or M9PA
supplemented with NHS (10%, 30 min incubation). The data are
plotted as the mean ± SEM, representing
three experiments performed in duplicate and analyzed by two-
way ANOVA (Tukey’s multiple comparisons test; M9PA vs LB
P = 0.0279, M9PA vs NHS
P = 0.0410). b Histogram showing the total number of the
quantified proteins per condition in WT and its isogenic mutant,
ΔCbpD. c Principal component
analysis (PCA) performed on the entire proteome. For each
individual replicate, the quantified proteins were plotted in two-
dimensional principal
component space by PC1 = 43.8092% and PC2 = 31.5492% and
clustered according to growth condition and strain. d Volcano
plots demonstrating
differentially abundant proteins and q values of significance
identified by comparing the ΔCbpD vs. WT proteomes. The red
dotted line(s) crossing the
y axis and x axis indicate significance cutoff at q = 0.05 (log10
= 1.3) and (+/−) 1.5-fold change (log2 = 0.58) in protein
abundance. Cutoff values for
significance were set to fold change ≥1.5 and q ≤ 0.05 in a two-
tailed paired t test. e Hierarchical clustering of proteins
expressed by WT and ΔCbpD in LB,
RPMI, and RPMI-NHS using agglomerative hierarchical
clustering analysis. For visualization, rows (genes) have been
standardized, so that the mean is 0

and the standard deviation is 1. f Heatmap showing KEGG
pathways that were significantly enriched in ΔCbpD vs. WT
upon growth in RPMI or RPMI/NHS.
g Heatmaps showing the average fold change values (log2) for
significantly regulated proteins belong to shared proteome in
the ΔCbpD compared to WT
strain in all growth conditions. h, i Heatmaps showing the
average fold change values (log2) for selected regulators and
virulence factors in the unique
proteome that are significantly regulated in ΔCbpD relative to
WT during growth in RPMI (h) or RPMI-NHS (i). The selected
proteins are marked in bold in
Supplementary Data 3. j Functional analysis of protein-protein
interaction networks revealed some highly interconnected
proteins listed in (i). The heatmap
(i) was mapped to the nodes using a blue–white–red gradient
that indicates fold change log2 LFQ (ΔCbpD/WT). Proteins
without any interaction partners
(singletons) or chains with no interaction with the main network
have been omitted from the visualization. Source data are
provided as a Source Data file
(a) or Supplementary Data.
https://doi.org/10.1038/s41467-021-21473-0 |
www.nature.com/naturecommuni cations
exogenous administration of rCbpDPA restored survival of the
ΔCbpD mutant in human blood (Fig. 3b), suggesting that the

secreted form of the protein may protect PA from immune
clearance. To pursue this point further, we hypothesized that
CbpD might also promote blood survival of other bacterial
strains that lack LPMOs in their genome. According to the
CAZy
database54, neither Gram-negative E. coli nor Gram-positive
Staphylococcus aureus harbor LPMO-encoding genes in their
genomes. We found that exogenous rCbpDPA promoted the
survival of E. coli temporarily (1 h post infection) but not S.
aureus in human blood (Fig. 3c, d), suggesting that the
protective
function of CpbD is neither restricted to the producing organism
nor universal.
Lastly, to investigate whether the catalytic activity of CbpD is
required for its role in promoting blood survival, we developed
a
construct encoding a catalytically inactive CbpD variant for in
trans complementation of ΔCbpD (ΔCbpD:CbpDH129A). The
active site mutant ΔCbpD:CbpDH129A complemented strain
had
reduced whole blood survival compared to the WT protein
ΔCbpD:CbpD complemented strain, indicating that CbpD
catalytic activity is important for its protective role in blood
survival (Fig. 3e).
CbpD contributes to PA resistance to lysis by the terminal
complement pathway. The complement system plays a crucial
role in killing Gram-negative bacteria through the assembly and
subsequent insertion of the terminal membrane attack complex
(MAC; C5b-9) in the bacterial cell envelope55. The importance
of
the complement system and the roles of C5 and MAC in defense
against PA are well documented56,57. Complement activation is
commonly measured by quantification of the activation products
C3bc (proximal complement pathway), C5a, and C5b-9

(terminal
complement pathway). Thus, the effect of CbpD on PA-induced
complement activation was investigated by quantification of
these
products 30 min after the addition of WT PA (PA14) or ΔCbpD
to human blood. Both strains significantly induced complement
activation in whole blood as measured by high levels of C5a and
fluid-phase C3bc compared to the buffer control (Fig. 3f). The
addition of rCbpDPA to whole blood showed that the pseudo-
monal LPMO did not by itself trigger complement activation
(Fig. 3f). CbpD did not inhibit the WT/ΔCbpD-induced
proximal
complement pathway activation in whole blood since C3bc
concentrations were similar for both PA variants (Fig. 3f, left
panel). The addition of WT PA to human blood resulted in
lower
levels of C5a compared to ΔCbpD and supplementation of
https://doi.org/10.1038/s41467-021-21473-0
https://doi.org/10.1038/s41467-021-21473-0 |
rCbpDPA to ΔCbpD restored the C5a level to that observed for
WT PA (Fig. 3f, middle panel). CbpD did not inhibit the for -
mation of the WT/ΔCbpD-induced fluid-phase C5b-9, the
terminal pathway end product (Fig. 3f, right panel). Of note,
rCbpDPA slightly reduced the classical and alternative comple-
ment pathway activities when screened by measurement of the
surface-bound C5b-9 complex (Supplementary Fig. 11f and
Supplementary Results). Since C5a is a potent inflammatory

mediator, we also quantified 27 key cytokines and chemokines
in
blood with PA added. Indeed, PA (WT/ΔCbpD) addition
strongly induced several inflammatory cytokines (e.g. TNF, IL-
1β,
RANTES, PDGF-BB, IL-8) compared to control. Some of these
cytokines were less prominent in blood with ΔCbpD added
compared to WT PA, such as TNF and RANTES (1 and 3 h) and
PDGF-BB (1 h only) (Fig. 3g and Supplementary Fig. 11j, k).
Cleavage of the C5 molecule to the C5a anaphylatoxin and the
C5b MAC subunit is mediated by the proteolytic C5 convertase
complex, which assembles on the bacterial surface as a result of
complement activation58. We asked if the increase in C5a levels
associated with the loss of CbpD may be linked to C5
convertase
assembly. Indeed, incubation of WT and ΔCbpD PA with a
dilution series of C5‐ depleted serum (ΔC5), showed higher
labeling of ΔCbpD cells with alternative complement pathway
C5
convertases components, C3b and Bb (Fig. 3h). This difference
was not observed when the bacteria were pre-incubated with
heat-inactivated ΔC5 serum or buffer (RPMI/HSA), which lacks
the potential to deposit C3b and to convert the C5 molecule
(Fig. 3h).
To further scrutinize the impact of CbpD on C5 convertase
assembly, we hypothesized that lower convertase activity could
lead to less deposition of MAC (C5b-9) on the bacterial surface.
This can be quantified by using antibodies recognizing the C9
neoantigen of the C5b-9 complex59. Indeed, experiments with
NHS (10 and 30%) showed higher C9 deposition on the ΔCbpD
surface compared to WT PA (Fig. 3i) and on the complemented
mutant (ΔCbpD:CbpD) compared to mock (Supplementary
Fig. 11g). Next, we evaluated MAC deposition on the PA
surface

after incubation of bacterial cells with 10% NHS using
fluorescence microscopy. C5b-9/MAC deposition was detected
on the surface of the majority of ΔCbpD cells, but only to a
limited
extent in WT cells (Fig. 3j). Furthermore, a membrane integrity
assay was used to assess outer membrane damage caused by
MAC
in WT PA, ΔCbpD mutant, and in trans-complemented
constructs. No significant difference was observed between WT
and ΔCbpD in the absence of serum, but exposure to 10% NHS
resulted in a significant increase in permeability of the ΔCbpD
mutant compared to WT PA or the ΔCbpD:CbpD complemented
strain compared to ΔCbpD:Mock (Supplementary Fig. 11h).
Transmission electron microscopy of serum-exposed bacteria
showed predominantly intact cells with well-preserved cell
envelopes for the WT PA strain (Supplementary Fig. 11i),
whereas
images of serum-exposed ΔCbpD showed cellular debris (amor-
phous materials) and marked morphological changes (e.g., loss
of
shape) indicative of cell death (Supplementary Fig. 11i). Taken
together, these results show the importance of CbpD in
protecting
PA against bacterial lysis by the terminal complement pathway.
Fig. 3 Ex vivo and in vitro analysis of the effect of CbpD
deletion on PA virulence. a Survival of PA WT, ΔCbpD, and
ΔCbpD trans-complemented with a
plasmid expressing CbpD or empty plasmid upon incubation in
human blood (hirudin used as anti-coagulant). The data are
plotted as the mean ± SEM,
representing six (left panel)/three (right panel) experiments
performed in triplicate. Data are analyzed by two-way ANOVA
(Sidak’s multiple
comparisons). Left panel: 1 h P = 2.1E-14, 3 h P = 5E-15. Right
panel: 1 h P = 2.185E-12, 3 h P = 1.292E-4. b–d Survival of PA

WT (b), E. coli (ESBL) (c), and
S. aureus (MRSA USA 300) (d) in human whole blood in the
absence or presence of added rCbpDPA (20 µg ml−1). Survival
was calculated relative to
inoculum. Untreated PA WT- (b), ESBL- (c), and MRSA- (d)
infected blood was arbitrarily set equal to 100%, and bacterial
survival in blood treated with
rCbpDPA is represented as survival in percentage (%). The data
are plotted as the mean ± SEM, representing five (panel b)/three
(panels c and d)
experiments performed in triplicate. Data are analyzed by two-
way ANOVA (Tukey’s multiple comparisons). b 1 h: WT vs
ΔCbpD P = 0.0030, ΔCbpD:
rCbpDPA vs ΔCbpD P = 0.0022; 3 h: WT vs ΔCbpD P =
2.882E-7, ΔCbpD:rCbpDPA vs ΔCbpD P = 3.201E-5, c 1 h: P =
0.0020; 3 h: P = 0.898. e Survival of
ΔCbpD complemented with catalytically inactive CbpD
(ΔCbpD:CbpDH129A) or active CbpD in blood. The data are
plotted as the mean ± SEM,
representing n = 7 (1 h)/n = 8 (3 h) samples that were examined
over three experiments. Data were analyzed by a two-tailed t
test (1 h: P = 0.0001, 2 h:
P = 0.0062). f Quantification of soluble complement factors
C3bc, C5a, and C5b-9/MAC in hirudin-treated human blood 30
min post infection with WT or
ΔCbpD. When indicated, rCbpDPA was added together with the
bacteria to a final concentration of 20 µg ml−1. The results are
given in complement
arbitrary units (CAU) per ml. The data are plotted as the mean ±
SEM, representing three experiments performed in triplicate.
Data were analyzed by two-
way ANOVA (Tukey’s multiple comparisons). Left panel
(C3bc): WT vs buffer P = 0.0008, WT vs rCbpDPA P = 0.0195,
ΔCbpD vs buffer P = 0.0001,
ΔCbpD vs rCbpDPA P = 0.0039; Middle panel (C5a): WT vs
ΔCbpD P = 2.84096083E-7, WT vs buffer 9.6E-14, WT vs

rCbpDPA 9.6E-14, ΔCbpD vs
ΔCbpD+rCbpDPA P = 7.5720483E-8, ΔCbpD vs buffer P =
9.6E-14, ΔCbpD vs rCbpDPA P = 9.6E-14; right panel (C5b-9):
WT vs ΔCbpD P = 0.0583, WT
vs buffer P = 9.6E-14, WT vs rCbpDPA P = 9.6E-14, ΔCbpD vs
ΔCbpD+rCbpDPA P = 0.0204, ΔCbpD vs buffer P = 9.6E-14,
ΔCbpD vs rCbpDPA P = 9.6E-
14. g Cytokine profiling of whole human blood. Plasma
harvested from blood samples infected with WT or ΔCbpD. The
categorical heatmap shows the
concentration of cytokines, chemokines or growth factors at 1 h
and 3 h post infection. The data are depicted as the geometric
mean of the cytokine values
in each cell, representing three experiments performed in
triplicate. Individual values are plotted in Supplementary Fig.
11j and 11k, in which the mean ± SEM
is depicted. Data were analyzed by two-way ANOVA (Tukey’s
multiple comparisons) and the significant difference between
WT and ΔCbpD is indicated
by asterisks. Left panel (1 h): TNF P = 5.282E-8, RANTES P <
1E-15, PDGF-BB P = 0.0007; Left panel (3 h): TNF P < 1E-15,
RANTES P = 0.0029. h, i Relative
quantification of complement factors C3b and Bb (h) and C9 (i)
bound to WT or ΔCbpD by flow cytometry. Bacteria were
incubated with buffer or
different concentrations of NHSΔC5, NHS or heat-inactivated
serum (HI; 10 %) 45 min to 1 h prior to analysis. C3b, C5b-9,
and Bb were detected using
antibodies against C5b-9/C3b (both Alexa‐ fluor 488 labeled) or
Bb following subsequent incubation with a secondary Alexa
Fluor 488-conjugated
antibody. The anti-C5b-9 binds to the C9 neoantigen of the
C5b-9/MAC complex. The data are plotted as the geometric
mean of fluorescence intensity
(GMFI) ± SEM, representing n = 5 samples that were examined
over two experiments (h) or three biological replicates (i). The

gating strategy is provided
in Supplementary Fig. 16. Data were analyzed by two-way
ANOVA (Sidak’s multiple comparisons) and the significant
difference between WT and ΔCbpD
is indicated by asterisks. h Bb: NHSΔC5 (3%) P = 1.484E-9,
NHSΔC5 (10%) P < 1E-15; C3b: NHSΔC5 (10%) P < 1E-15; i
NHS (10%) P = 0.0021, NHS
(10%) P = 0.0021, NHS (30%) P = 0.0029. j Representative
fluorescence microscopy images of C5b-9 complex deposition
on PA WT or ΔCbpD upon
incubation with 10% NHS or HI-NHS. The complex was
detected using anti-C5b-9 followed by incubation with a
secondary Alexa Fluor 488-conjugated
antibody. Bacterial membranes were stained red using FM5‐ 95.
Two separate channels were merged into a single image by Fiji.
The scale bar represents
10 µm. The imaging was performed twice. The significance is
indicated by asterisks (*): *P ≤ 0.05; **P ≤ 0.01; ***P ≤ 0.001;
****P ≤ 0.0001. Source data are
provided as a Source Data file (a–h).
https://doi.org/10.1038/s41467-021-21473-0 |
CbpD contributes to PA virulence in murine systemic infection
in vivo. Given the contribution of CbpD in resistance of PA to
whole blood-mediated killing, we evaluated the impact of CbpD
on PA systemic virulence in a murine intravenous (IV)

challenge
model. CD1 mice (8-week old) infected with 2 × 107
CFU/mouse
WT PA experienced 100% mortality within 48 h post infection
(Supplementary Fig. 12b and Fig. 4a, left panel). In contrast to
the
universal mortality of mice challenged with the WT strain, 80%
of
isogenic mutant ΔCbpD-infected mice survived during the 48 h
observation period (Fig. 4a, left panel). To investigate the effect
of
CbpD on bacterial clearance, mice were euthanized at 4 h post
infection and the CFU in blood, liver, kidney, and spleen were
https://doi.org/10.1038/s41467-021-21473-0
https://doi.org/10.1038/s41467-021-21473-0 |
enumerated. Strikingly, the mice infected with ΔCbpD mutant
had significantly reduced bacterial loads in blood and tissues
compared to the mice infected with WT strain (Fig. 4b). Con-
sidering the significance of PA in respiratory infections, the
effect
of CbpD virulence was also studied in a murine intratracheal
challenge model. WT PA infection of CD1 mice with 5 × 106
CFU/mouse experienced ~60% mortality within 4 days post
infection while the ΔCbpD-infected mice had 10% mortality at
the same time point (Fig. 4a, right panel). These data reveal an

explicit requirement of CbpD for full PA virulence in these two
in vivo models.
Serum cytokine/chemokine responses were profiled 4 h post
bacterial systemic challenge. Significantly higher levels of IL-6,
IL-
12 (p40), G-CSF, keratinocyte chemoattractant (KC; a
functional
homolog of human IL-8/CXCL1), and MCP-1 (CCL2) were
detected in mice challenged with WT PA compared to mice
infected with the ΔCbpD mutant (P < 0.05) (Fig. 4c and
supplementary Fig. 12a). This result suggests a heightened
inflammatory response that tracks the increased bacterial burden
seen in the fully virulent WT strain. Correspondingly, histo-
pathological evaluation of splenic tissue samples revealed a
slightly higher infiltration of neutrophils in WT-infected mice
(Fig. 4d, left panel) compared to mock-infected (control) and
ΔCbpD-infected mice (4 out of 8) (Fig. 4d, right panel).
Proteomics was used to gauge the effect of CbpD deletion on
host-specific responses to PA systemic infection. The response
of
the spleens from CD1 mice infected with WT PA, ΔCbpD, or
PBS
(mock-infected) were analyzed. In total 2496, 2428, and 2271
proteins were quantified in WT-, ΔCbpD-, and mock-infected
mice, respectively, and 2187 of these proteins were present
commonly across all treatments (Supplementary Data 7). Hier-
archical clustering and principal component analysis revealed a
distinct infectious agent-specific segregation of the splenic
proteomes into WT, ΔCbpD, and mock-infected (control)
clusters (Fig. 4e, f). The volcano plots of splenic proteomes
showed 42 significantly regulated proteins in WT-infected vs.
mock-infected mice (28 up- and 14 downregulated), whereas
160

differentially regulated proteins were detected in ΔCbpD-
infected
vs. mock-infected mice (77 up- and 83 downregulated; FC ≥ 1.5
and q ≤ 0.05; Fig. 4g and Supplementary Data 8 and 10).
Infected
spleens shared 200 regulated proteins that had comparable
relative expression irrespective of the infecting strain (Supple -
mentary Data 8–11). The shared proteome consists of 32 sig-
nificantly regulated proteins (FC ≥ 1.5, q ≤ 0.05) (Fig. 4h and
Supplementary Data 8 and 10) and 168 proteins that were only
detected in the infected compared to control mice (no FC value,
Supplementary Data 9 and 11). Functional analysis of this
shared
proteome (Supplementary Data 10 and 11) revealed high
enrichment of several immune responses, including cellular
response to IFNβ, RIG-I-like receptor signaling pathway, and
Toll-like receptor signaling as a general response to PA
infection
(Supplementary Fig. 13a). Evaluation of the unique proteome
signature associated with the deletion of CbpD revealed that 10
and 128 proteins were regulated (Supplementary Data 8 and 10)
in response to WT PA or ΔCbpD infection, respectively (FC ≥
1.5
and q ≤ 0.05). In addition, 44 and 65 proteins were only
detected
in the WT and ΔCbpD-infected compared to control mice,
respectively (Supplementary Data 9 and 11). The ΔCbpD
associated unique proteome had a scattered expression with
relative changes from +3.2 (Serpina3g) to −9.7 (Amy2)
Fig. 4 The effect of CbpD on PA pathogenesis in vivo. a CD1
mice were inoculated intravenously (IV) with 2 × 107 CFU (left
panel, 10 mice/group) and
intratracheally (IT) with 5 × 106 CFU (right panel, 10
mice/group) PA WT or ΔCbpD per mouse. Survival is
represented by Kaplan–Meier survival curves and

analyzed by log-rank (Mantel–Cox) test (left panel: P = 1.118E-
5, right panel: P = 0.0250). b Bacterial loads in the kidney,
liver, spleen and blood (CFU/g for
organs and CFU/ml for blood) of 8-week-old CD1 mice were
enumerated 4 h post systemic infection (IV) with PA WT or
ΔCbpD. The data are plotted as
the mean ± SEM, representing one experiment performed with
eight mice/group and were analyzed by two-tailed t test (kidney
P = 0.0129, blood P =
0.0002, liver P = 5.827E-5, spleen: P = 1.910E-5. Mock-
infected mice (n = 4) were included as a control. c The
categorical heatmap shows the concentration
of cytokines, chemokines or growth factors in the serum of CD1
mice 4 h post infection (as described in b). The data are
depicted as the geometric mean of
the cytokine values, representing one experiment performed
with eight mice/group. Mock-infected mice (n = 3) were
included as a control. Individual values
are plotted in Supplementary Fig. 12a, in which the mean ±
SEM of each cytokine is depicted. The data were analyzed by
two-way ANOVA (Dunnett’s
multiple comparisons test) and the significant difference
between WT and ΔCbpD is indicated by asterisks (*). IL-6: P =
0.0497, IL-12 (p40): P = 0.0018, G-
CSF P < 0.0001, KC: P = 0.0063, MCP-1: P < 0.0001. d
Representative hematoxylin and eosin-stained sections of spleen
tissues from infected (WT or
ΔCbpD, n = 8 mice/group) and uninfected mice (control, n = 4
mice) collected 4 h post infection (as described in b), analyzed
by light microscopy. White
pulp (WP) and red pulp (RP) regions are marked.
Polymorphonuclear (PMN) leukocytes are indicated with
arrows. The scale bar represents 50 and 20 µm in
the upper and lower panels, respectively. The histopathologic
scoring analysis of PMN infiltration of the spleen tissue is
shown as a histogram. e Hierarchical

clustering of spleen proteome based on Spearman rank
correlation coefficients. The infection status (as described in b)
is indicated by color and R stands for
replicate, (indicating number of mice/group). f Principal
component analysis (PCA) of the identified proteins, showing
segregation of the spleen proteomes
into two different infected (WT or ΔCbpD) and one uninfected
group. The quantified proteins are plotted in two-dimensional
principal component space by
PC1 = 46.9724% and PC2 = 36.0129%. g Volcano plot showing
differentially abundant proteins in the spleens of WT- or
ΔCbpD-infected mice relative to
mock-infected mice. The red dotted line(s) through the y axis
and x axis, indicate significance cutoff values at q = 0.05 (log10
= 1.3) and (+/−) 1.5-fold
change (log2 = 0.58) in protein abundance, respectively.
Significance was determined using the two-tailed paired t test. h
Heatmap of log2 fold change
values of the 32 significantly regulated proteins in infected (WT
or ΔCbpD) vs. non-infected spleen. The proteins belong to the
shared category. i Heatmap
of enrichment score (−log10 P value, cutoff = 0.01) showing
pathways/cellular processes that were enriched in the spleen
proteome (Supplementary
Data 10 and 11, unique category) of ΔCbpD-infected mice
(down- or upregulated; arrows pointing down/up). Enrichment
analysis was performed by
Metascape and the P value was calculated based on the
cumulative hypergeometric distribution. j Depiction of the
Ingenuity Pathway Analysis (IPA) of the
unique proteome showing canonical pathway webs that are
significantly altered in ΔCbpD-infected mice. The P value was
calculated by right-tailed fisher’s
exact test and is shown in the figure. Pathways without
interaction with the web (singletons) are omitted from the
visualization. The average fold change

value of up- or downregulated proteins associated with some of
the top-scored pathways are presented as heatmap and is
visualized using a blue–white–red
gradient. k Heatmap of enrichment score (−log P value, cutoff =
0.01) showing pathways/cellular processes that were enriched in
the unique splenic
proteomes (Supplementary Data 10 and 11) of WT-infected vs.
uninfected mice Enrichment analysis was performed by
Metascape and the P value was
calculated based on the cumulative hypergeometric distribution.
l Depiction of the IPA of the unique proteome showing
canonical pathway webs that are
significantly altered in WT-infected mice. The P value was
calculated by right-tailed fisher’s exact test and is shown in the
figure. Pathways without
interaction with the web (singletons) are omitted from the
visualization. When applicable, the significance is indicated by
asterisks (*): *P ≤ 0.05; **P ≤ 0.01;
***P ≤ 0.001; ****P ≤ 0.0001. Source data are provided as a
Source Data file (a–c).
https://doi.org/10.1038/s41467-021-21473-0 |
compared to the control mice (Supplementary Data 8 and 10).
String analysis revealed these proteins were highly
interconnected
(Supplementary Fig. 13b).

Functional exploration of the distinct markers (193 proteins,
Supplementary Data 10 and 11) unique to the ΔCbpD-infected
spleens showed enrichment of several cellular responses,
includ-
ing apoptosis-induced DNA fragmentation and autoimmune
thyroid disease by the upregulated proteins, and synthesis of
leukotrienes (LT) and eoxins (EX) and engulfment of apoptotic
cells by the downregulated proteins (Fig. 4i). Inflammatory
responses are well known to alter metabolism in host cells,
which
was corroborated by the observed enrichment of pathways
associated with several terms coupled to metabolic processes
such as “Glycine, serine and threonine metabolism”, “Branched-
chain amino acid catabolism”, and “alpha-amino acid biosyn-
thetic processes” in the downregulated proteins (Fig. 4i). Some
of
the proteins associated with the enriched cellular processes
(e.g.,
isoforms of 14–3–3 proteins: Ywhae, Ywhag, Ywhaz) were
localized in the central part of the STRING network analysis or
clustered together (e.g. Hist1h1a, Hist1h1e, Hmgb2, Hmgb2)
(Supplementary Fig. 13b). To further explore the interactions
among regulated proteins (128) unique to the ΔCbpD-infected
mice (Supplementary Data 10), an ingenuity pathway analysis
(IPA)-based protein network was performed, and nine different
disease-based and molecular networks were algorithmically
generated (Supplementary Figs. 13c and 14). The “Inflammatory
response, lipid metabolism, small-molecule biochemistry” was
ranked (score 42) as the top enriched function- and disease-
based
protein network (Supplementary Fig. 13c). Several of the
proteins
associated with this network showed high connectivity to the
extracellular signal-regulated protein kinases 1 and 2 (ERK 1/2)
hub (Supplementary Fig. 14, network 1), which was also among

the enriched pathways associated with upregulated proteins in
ΔCbpD-infected mice (Supplementary Fig. 4i). On the single
protein level, some of the upregulated proteins of this network
are
involved in immune responses e.g., C4b (representative of
complement factor C4), Serpina3g, and H2-Ab1 (H-2 class II
histocompatibility antigen). The other enriched networks with a
score over 20 were “Cancer, Cell-To-Cell Signaling and Interac-
tion, Cellular Movement”, “Cancer, Cellular Assembly and
Organization, Neurological Disease”, “Cancer, Endocrine
System
Disorders, Protein Trafficking” and “Cancer, Molecular Trans -
port, Organismal Functions” (Supplementary Figs. 13c and 14).
Canonical pathway analysis of the unique proteome of the
ΔCbpD-infected mice revealed significant enrichment of 15
different canonical pathways (Supplementary Fig. 13d), in
which
cell cycle: G2/M DNA damage checkpoint regulation (e.g.,
Top2a,
Ywhae, Ywhag, Ywhaz), valine degradation I (e.g., Aldh1l1,
Bcat2,
Dbt), and oxidative phosphorylation (e.g., Atp5e, Cox72a,
Ndufa11, Ndufa9, Ndufb10) were among the top three (Supple-
mentary Fig. 13d and Fig. 4i). Several of these pathways were
highly interconnected and formed two main canonical pathway
webs that were associated with either metabolic processes or
host
immune responses (Fig. 4i).
Functional analysis of the regulated or detected markers (in
total 54 proteins) (Supplementary Data 10 and 11) unique to the
WT-infected spleens showed that these proteins were associated
with negative regulation of immune responses (e.g., Arg1, Arg2,
Ppp3cb, Clec2d, Cfh), negative regulation of cell proliferation
(e.g., Cnn1, Hdac2, Pdcd4, Ddah1, Cers2), and several catabolic
processes (Fig. 4i). IPA analysis of the regulated proteins

unique
to the WT-infected mice (Supplementary Data 10) revealed that
90% of the proteins were categorized under a network
associated
with “energy production, lipid metabolism, small-molecule
biochemistry” (score 27) (Supplementary Fig. 13e). Several of
these proteins (Cstb, Slc25a3, Clec2d, Pdcd4, Ppp3cb) were
directly or indirectly associated with the Myc hub, a master
transcription factor of multiple proliferative genes (Supplemen-
tary Fig. 13e), which aligned with our enrichment analysis
(Fig. 4i). Canonical pathway analysis of the unique proteome
(Supplementary Data 10) showed necroptosis (e.g., Ppp3cb,
SLC25A3) and myo-inositol biosynthesis (e.g., Impa1) as the
profoundly enriched pathways in WT-infected mice (Fig. 4l).
Ultimately, analysis of the proteins involved in the complement
cascade in the splenic proteome revealed the identification of
several proteins associated with this system (e.g., Cr2, Cfh, C4b
(Slp), Cfb, C3, Vtn, and C1qbp) (Supplementary Data 7).
However, only the C4b molecule (representative of complement
factor C4) (Supplementary Data 10) and complement factor H
(Cfh) (Supplementary Data 11) were identified among the
upregulated proteins unique to the ΔCbpD and WT-infected
mice, respectively. Vitronectin (Vtn), which is involved i n the
regulation of the terminal complement cascade, was upregulated
(Supplementary Data 11) in the shared proteome of infected vs.
mock-infected mice.
Collectively, these data suggest that CbpD contributes to PA
virulence in vivo and its loss leads to changes in the host
proteome during systemic infection in vivo.
Discussion
A predominant focus on the function and applications of
LPMOs

in biomass conversion has overshadowed the putative
importance
of these enzymes in bacterial virulence. The high prevalence of
lpmO genes in pathogenic bacteria and their increased
expression
in multiple omics-type driven analyses in the context of host-
pathogen interactions (reviewed in refs. 16,17) suggest a
potential
biological function during infection. Our finding of the decline
in
bacterial survival in vivo in a murine model of PA systemic
infection (Fig. 4), and in whole human blood ex vivo (Fig. 3)
confirms that this LPMO can promote bacterial resistance to
immune clearance, suggesting its function as a virulence factor.
Aligned with our results, deletion of lpmO (lmo2467) in
Listeria
monocytogenes attenuated bacterial loads in the spleen and liver
of mice15.
The major impact of CbpD was reflected in our proteome
studies that showed significant effects caused by cbpD deletion
on
both the bacterial and the host proteome (Figs. 2 and 4). The
specific changes in the ΔCbpD proteome in response to a low
concentration of NHS (which contains complement compo-
nents), included downregulation of several metabolic processes
and differential regulation of virulence factors, including an
upregulation in the expression of several key regulators or tran-
scriptional factors (e.g., AlgR, DksA, MvaT, AlgU, and Vfr).
These changes reflect the struggle of the isogenic mutant to
achieve physiological adaptation to the host environment and to
overcome the complement-mediated stress (Fig. 2). Alteration
in
carbon metabolism and virulence has been suggested as an
adaptive mechanism for PA to promote viability in human
blood60. Despite the presence of strong crosstalk among

virulence-associated regulators in PA51, the proteome
modulation
strategy employed by the mutant was not sufficient to produce a
functional impact in ΔCbpD defense during sepsis. This under -
scores the importance of CbpD in PA pathogenesis and adapt-
ability during infection ranging from acute bacteremia to
respiratory tract infection (Figs. 3 and 4).
To cope with sepsis, the systemic activation of the innate
immune system triggers a variety of interconnected signaling
pathways for a successful host response. Although WT- and
ΔCbpD-infected mice showed similar trends in the enrichment
of
several pro-inflammatory pathways in their shared splenic pro-
teomes (e.g., Toll-like, Rig-1-like, IFNβ), comparison of the
https://doi.org/10.1038/s41467-021-21473-0
https://doi.org/10.1038/s41467-021-21473-0 |
unique proteomes revealed that ΔCbpD and WT elicited
different
host immune responses in the infected mice. The differences in
the splenic unique proteomes were reflected by the ΔCbpD-
infected mice showing modulation of a larger number of
proteins
and enrichment of distinct cellular processes or pathways such
as
apoptosis, cell cycle, and ERK 1/2 pathways (Fig. 4). Indeed,
the

role of apoptosis, one of the top enriched pathways, as a means
to
aid the host during infection is well established (reviewed in
ref. 61) and is suggested to operate by removing the
intracellular
niches for selected pathogens61. Moreover, the cellular debris
resulting from apoptosis can trigger activation of innate immune
responses (reviewed in ref. 61) such as the complement system,
which subsequently enhances clearance of the infection
(reviewed
in ref. 62). The different host immune response elicited by the
ΔCbpD strain compared to the WT may reflect the hypovirulent
nature of the ΔCbpD strain and underpins the contribution of
CbpD significantly to the extent and consequence of PA-
associated infection. From the perspective of understanding PA
systemic infections in general, one important host marker
unique
to the WT-infected mice was the complement-negative regulator
factor H (Supplementary Data 11). This observation indicates
that the enrichment of cellular processes such as “negative reg-
ulation of immune responses” in the WT-infected mice may be
partly responsible for the development of intense bacteremia
observed for this strain. Importantly, increased expression of
factor H during bacterial infection is associated with prolonged
hospitalization of the patients63.
Although PA strains show different levels of resistance to
complement-mediated lysis64, the importance of the
complement
system, particularly the membrane attack complex (MAC; C5b-
C9), in the eradication of PA is well documented65,66. In
addition,
complement-deficient mice show an exacerbated inflammatory
response56 and high susceptibility to PA infection56,57. The
role
of MAC in killing/lysis of Gram-negative bacteria has been

Biochemistry and Biophysics Reports 25 (2021) 100921Availa

Biochemistry and Biophysics Reports 25 (2021) 100921Availa

Recommended

Recommended

More Related Content

Similar to Biochemistry and Biophysics Reports 25 (2021) 100921Availa

Similar to Biochemistry and Biophysics Reports 25 (2021) 100921Availa (20)

More from ChantellPantoja184

More from ChantellPantoja184 (20)

Recently uploaded

Recently uploaded (20)

Biochemistry and Biophysics Reports 25 (2021) 100921Availa