Abstract
Metabolism and gut microbiota are essential for newborn health,
influencing immune function, energy balance, and growth. Breast milk
provides IgA, crucial for shaping gut microbiota in infants. In
non-breastfed newborns, we observe the presence of IgM antibodies that
can recognize various bacteria, influence bacterial clustering, and
alter bacterial metabolism, such as carbon source utilization in vitro
within small bacterial communities. Based on these findings, we
developed a monoclonal IgM, M291, derived from a newborn-B cell, which
mimics naturally occurring antibodies and could serve as a surrogate
tool to modulate intestinal bacterial functions and metabolism. Oral
administration of M291 alters the metabolome of germ-free mice
colonized with a defined bacterial consortium or an infant gut
microbiota, by modulating the bacterial transcriptome, while
maintaining microbial abundance and diversity. This study establishes
proof of concept for the design and application of newborn-derived
antibodies to modulate microbial and host metabolism, including lipid
metabolism and bile acid secretion, without significantly altering
microbiota composition.
Subject terms: Mucosal immunology, Microbiome, Metagenomics, Antibodies
__________________________________________________________________
Here, the authors characterize a newborn-derived monoclonal IgM,
showing it influences bacterial clustering, gene expression, and
metabolic activity, thereby modulating gut bacterial functions and host
metabolism without altering microbiota composition.
Introduction
The intestinal metabolome, consisting of various low-molecular-weight
metabolites including sugars, organic acids, and amino acids, is shaped
by the combined effects of host metabolism and gut microbiota
activities^[38]1. This metabolomic profile enhances microbiota’s
functionality and represents the intricate interactions between host
physiology and microbial metabolism. Consequently, the intestinal
metabolome acts as a comprehensive biomarker, mirroring the gut
ecosystem’s dynamic state^[39]2,[40]3.
The initial microbial colonization in the gut is crucial for the
development and maturation of the intestinal mucosa and the immune
system in neonates^[41]4. Gut colonization starts at birth, with
aerobic and facultative anaerobic bacteria such as Enterobacteria and
Streptococci playing a pivotal role during this initial phase^[42]5.
The consumption of oxygen by these pioneer microbes creates a
microenvironment conducive to the successive colonization of obligate
anaerobes, including Bifidobacterium, Bacteroides, and
Clostridium^[43]4,[44]6. The establishment of the gut microbiota and
metabolome is influenced by multiple factors, including mode of
delivery and feeding practices^[45]7–[46]9. Microbiota utilize
nutrients to generate energy and a myriad of metabolites that are
essential for physiological processes. Early seeders are essential for
the maintenance of health, as dysbiosis is implicated in a spectrum of
diseases such as obesity, inflammatory bowel disease, and asthma across
both infant and adult populations^[47]10–[48]12. Recent studies have
documented that gut bacteria and metabolites play pivotal roles in the
maintenance of homeostasis by regulating the gut-systemic metabolic
interplay, which may influence immune and metabolic development in
early life^[49]10,[50]13. Indeed, an alteration can impact cognitive
development^[51]14, asthma^[52]15, or allergies^[53]16. Thus, a
balanced gut microbiota is essential for metabolic functions, nutrient
absorption, and immune responses, highlighting its significance in
early-life development and long-term health.
Immunoglobulins (Igs) have been found to coat a significant portion of
the gut microbiota^[54]17,[55]18. These Ig secreted in the intestinal
lumen play multifaceted roles, including elimination of microorganisms
and facilitation of communication between the host and microbiota,
thereby promoting homeostasis. Thus, Ig functions can be categorized
into elimination, neutralization, colonization, and sculpting of the
microbiota^[56]19–[57]22. The interaction of Igs with the microbiota is
capable of modulating bacterial gene expression. For example, a
monoclonal antibody (mAb) that targets a capsular polysaccharide
antigen has been demonstrated to downregulate the expression of
specific epitopes, consequently impairing bacterial fitness^[58]23.
Furthermore, IgA antibodies have been shown to influence bacterial
fitness through multiple mechanisms, such as inhibiting bacterial
motility or impeding the uptake of essential sugars, with the specific
effects being dependent on the surface components recognized by the
respective IgA. This indicates that the immune system possesses the
capacity to concurrently impact the commensal gut bacteria through
diverse mechanisms^[59]24.
IgA is the predominant isotype present at mucosal surfaces in mammals,
including the small intestine, where it is actively
secreted^[60]25,[61]26. In addition to IgA, IgM, albeit to a lesser
extent, is also detected in the gut lumen^[62]18. Both are transported
into the gut lumen via the polymeric Ig receptor, with IgM forming
pentamers and IgA forming dimers, to coat microbiota^[63]27. At birth,
IgA is undetectable in the meconium and takes several weeks to be
synthesized and appear in stools^[64]28,[65]29. Studies show that serum
IgM levels in newborns are much higher than IgA^[66]30. In breastfed
infants, intestinal IgM-positive plasma cells are present at
birth^[67]31. IgA detection in stool initiates between 2 to 4 weeks of
age in full-term infants, mediated through both T cell-dependent and T
cell-independent pathways^[68]32. However, this detection depends on
mucosal tissue^[69]33. This developmental trajectory reflects the
gradual maturation of mucosal immunity after birth. We recently
identified a perinatal-specific population of B cells that produce
antibodies capable of recognizing a broad spectrum of microbiota
species^[70]34. In vitro, these antibodies modulated carbon source
utilization and influenced the dynamics of bacterial
communities^[71]34.
In the present study, we characterized intestinal antibodies during the
first month of life by analyzing their microbial recognition and their
impact on bacterial behavior both individually and within communities.
Our findings show that neonatal IgM antibodies can modulate bacterial
clusters and influence microbial abundance. We developed a newborn
B-cell-derived monoclonal IgM pentameric antibody and investigated its
interactions with microbial consortia in vitro as well as in vivo.
Notably, we show that a single monoclonal IgM can alter the in vivo
metabolome of germ-free mice colonized with neonatal microbiota—not by
changing the overall microbial community structure, but by modulating
bacterial functions. This study opens the scope for using mAbs as a
novel therapeutic tool for shaping the intestinal milieu at the
metabolic level.
Results
Characterization of newborn intestinal Ig during the first month of life
In the intestinal lumen of formula-fed neonates during the first month
of life, IgM was found to be the predominant Ig, whereas IgA levels
increased over time (Fig. [72]1A, Supplementary Fig. [73]1A). There was
a notable shift in the ratio of intestinal Ig isotypes, from IgM
predominance to a more balanced IgA presence. Specifically, IgM
accounted for about 80% of total intestinal Igs in the first week,
dropping to 50% by the fourth week (Fig. [74]1B). As gut colonization
begins shortly after birth, we examined the binding affinity of
intestinal Igs to gut bacteria throughout the first month. In the first
two weeks, a significantly higher proportion of bacteria from neonatal
fecal samples were coated with IgM than IgA (Fig. [75]1C, Supplementary
Fig. [76]1B), aligning with the early IgM prevalence. By month’s end,
IgA-coated bacteria became more prevalent, mirroring increased IgA
secretion (Fig. [77]1B, [78]1C). Analyzing Ig binding to peptidoglycan
(PGN) of Gram-positive bacteria and lipopolysaccharide (LPS) of
Gram-negative bacteria, IgM showed stronger PGN recognition, while IgA
recognized both PGN and LPS equally (Fig. [79]1D).
Fig. 1. Characterization of Intestinal Secreted Neonatal Immunoglobulins
During the First Month of Life.
[80]Fig. 1
[81]Open in a new tab
A Intestinal secreted IgA (blue) and IgM (red) levels were quantified
over the first month of life using ELISA. B Weekly proportions of
secreted IgA and IgM in total immunoglobulins within fecal samples were
assessed. w1 (Days 1 to 7, n = 10), w2 (Days 8 to 14, n = 9), w3 (Days
15 to 21, n = 7), and w4 (Days 22 to 28, n = 7). C The frequency of
bacteria-bound intestinal IgA and IgM was measured weekly by FACS. D
Total secreted intestinal immunoglobulins (normalized to 9 µg/ml) were
tested by ELISA for reactivity against PGN and LPS. Recognition of six
primary bacterial colonizers by intestinal secreted immunoglobulins: E.
coli (E) E. faecalis (F) S. epidermidis (G) S. salivarius (H) B.
bifidum (I) and B. thetaiotaomicron (J). K Cumulative bacterial
recognition summary of E–J. L Average recognition of Gram-positive
versus Gram-negative bacteria is represented as a heat map. All data
are presented as cumulative area under the curve (CAUC). Each point
represents an individual biological sample (n = 4–10). Box-and-whisker
plots: center line = median; box = 25th–75th percentiles; whiskers =
5th–95th percentiles (C–L). Statistical analyses were conducted
according to the respective panels: p values were calculated with the
two-sided Mann-Whitney U test for panels B–J. Significance levels are
indicated as follows: *p < 0.05; **p < 0.01; ***p < 0.001;
****p < 0.0001; n.s, not significant.
Intestinal Igs are known to coat bacteria^[82]18. We subsequently
evaluated the binding specificity of neonatal IgA and IgM to
early-colonizing bacterial species, including Gram-positive
(Enterococcus faecalis, Staphylococcus epidermidis, Streptococcus
salivarius, Bifidobacteria bifidum) and Gram-negative bacteria
(Escherichia coli, Bacteroidetes thetaiotaomicron) (Fig. [83]1E–J).
Despite the observed differences in PGN and LPS binding between IgM and
IgA (Fig. [84]1D), both Igs demonstrated comparable binding to these
bacteria. The recognition of all bacterial species tested increases
significantly after the second week (Fig. [85]1E–J), as demonstrated by
the cumulative area under the curve (CAUC) (Fig. [86]1K). Furthermore,
an analysis of IgM binding across 20 bacterial species reveals a
progressive increase in bacterial recognition from week 1 to week 4
(Fig. [87]S1). Notably, strong recognition is observed early for C.
perfringens and E. cloacae, followed by S. aureus and S.
epidermidis—three of the four Gram-positive bacteria tested. By week 3,
additional species, such as L. crispatus and E. faecium, are
recognized. Interestingly, S. salivarius, B. thetaiotaomicron, and B.
dorei consistently show low recognition. By week 4, only B.
thetaiotaomicron remains at low recognition levels (Supplementary
Fig. [88]1C). The average CAUC of bacterial recognition indicated that
Gram-positive bacteria are generally more recognized than Gram-negative
bacteria (Fig. [89]1L), aligning with the higher binding to PGN
compared to LPS (Fig. [90]1D). During the first month of life, we
concurrently analyzed the gut microbiota using 16S rRNA sequencing
(Supplementary Fig. [91]1D–F). Our findings revealed a notable increase
in the abundance of Firmicutes relative to Actinobacteria, while the
proportion of Proteobacteria remained relatively stable at
approximately 20% throughout this period (Supplementary Fig. [92]1D).
From the second week, which coincided with a decline in Bifidobacteria
(Supplementary Fig. [93]1E). Notably, we observed a higher prevalence
of Gram-positive bacteria compared to Gram-negative bacteria
(Supplementary Fig. [94]1F), which aligns with the increased capacity
of intestinal immunoglobulins to recognize PGN over LPS, thereby
enhancing the recognition of Gram-positive bacteria (Fig. [95]1D–L).
These results demonstrate that IgM are the primary antibodies in the
neonatal gut shortly after birth, with a preferential PGN recognition
over LPS. Both intestinal IgA and IgM broadly recognized bacterial
species throughout the first month of life.
Newborn intestinal Igs affect bacterial community metabolism
We subsequently evaluated the impact of early intestinal antibody
isotypes on the neonatal bacterial community. We isolated secreted IgA
(Nint_pA) and IgM (Nint_pM) from samples collected at week 3, and the
binding capacity to PGN and LPS was assessed (Supplementary
Fig. [96]2A, B). Notably, Nint_pM exhibited stronger binding to PGN,
while Nint_pA showed an increased recognition of both bacterial
antigens relative to Nint_pM (Supplementary Fig. [97]2A, B), indicating
similar reactivity patterns between bulk and purified Igs. Further
analysis of bacterial recognition across six selected species—E. coli,
E. faecalis, S. epidermidis, S. salivarius, B. bifidum and B.
thetaiotaomicron—reveals binding by both Nint_pA and Nint_pM, with
Nint_pA showing a higher binding capacity (Supplementary Fig. [98]2C).
To ascertain the influence of neonatal intestinal Igs on bacterial
growth, we cultured individual bacterial strains in the presence of
either Nint_pM (Fig. [99]2A) or Nint_pA (Fig. [100]2B). In the presence
of either Igs, E. coli, E. faecalis, S. epidermidis, S. salivarius, and
B. thetaiotaomicron display significantly enhanced growth compared to
controls (Fig. [101]2A, B).
Fig. 2. Impact of newborn intestinal immunoglobulins on bacterial clustering
and microbial consortia.
[102]Fig. 2
[103]Open in a new tab
Indicated bacterial strains were cultured individually in AF medium in
the presence of neonatal intestinal IgM (Nint_pM) (A) or IgA (Nint_pA)
(B) at a concentration of 5 mg/ml. Growth was assessed by OD600nm and
expressed as a percentage relative to the average of control cultures
without immunoglobulins (PBS). C The MIC6 microbial community was
incubated with Nint_pA or Nint_pM, and immunoglobulin-bound bacterial
clustering was analyzed by FACS. D Supernatants from MIC6 cultured in
the presence of Nint_pA or Nint_pM were used to culture individual
bacteria. Growth was assessed based on OD600nm and expressed as a
percentage of the MIC6 supernatant without immunoglobulins, serving as
a reference control. E Utilization of indicated carbon sources by the
MIC6 consortium for growth was analyzed using Eco-plates in the absence
(gray) and presence of 5 mg/ml Nint_pM (red) or Nint_pA (blue).
Box-and-whisker plots: center line = median; box = 25th–75th
percentiles; whiskers = 5th–95th percentiles (A, B, D. Box plots
represent median and maximum and minimum of variation (n = 3) E.
Statistical analyses were conducted according to the respective panels:
p values were calculated with the two-sided Mann-Whitney U test for A,
B, and D, or the two-sided unpaired Student’s t-test for E.
Significance levels are indicated by ****, ***, **, and * for p
values < 0.0001, <0.001, <0.01, and <0.05, respectively; n.s = not
significant.
We investigated the capacity of dimeric IgA and pentameric IgM in the
gut to form diverse bacterial clusters through cross-species binding.
Bacteria were labeled with fluorescent amino acids, and clustering was
analyzed by FACS in the presence or absence of Igs (Supplementary
Fig. [104]2D–F). Clustering occurred without Igs, but pentameric
Nint_pM enhanced multi-species clustering more than dimeric Nint_pA,
which mainly formed smaller clusters (Fig. [105]2C). Nint_pM
facilitated the formation of 3- to 4-member clusters involving E. coli,
E. faecalis, S. epidermidis, and S. salivarius, while Nint_pA
predominantly bound single species, especially E. faecalis, S.
salivarius, and B. bifidum. Nint_pA induced two-species clusters with
E. faecalis associating with S. epidermidis or S. salivarius. Notably,
both Nint_pA and Nint_pM reduced clustering with B. bifidum and B.
thetaiotaomicron (Fig. [106]2C).
Bacterial clustering can influence microbial interactions, affecting
growth, gene expression, and metabolic functions^[107]35,[108]36. To
further investigate the effects of Nint_pA and Nint_pM, we cultured a
microbial consortium of 6 bacterial species (MIC6) overnight with each
isotypic antibody preparation. Neither IgA nor IgM impacted bacterial
composition of MIC6 culture (Supplementary Fig. [109]2G). We then used
supernatants from MIC6 cultures treated with IgA or IgM preparations to
individually culture the 6 bacterial strains. Comparisons with
untreated supernatant (sMIC_PBS) revealed that while B.
thetaiotaomicron and S. salivarius growth remained stable, S.
epidermidis growth decreased with sMIC_pA, and B. bifidum growth
increased with sMIC_pM (Fig. [110]2D). Additionally, sMIC_pM
significantly enhanced the growth of E.coli, E.faecalis, S.epidermidis,
and B.bifidum compared to sMIC_pA. Subsequently, we examined if
metabolism of MIC6 modulated by IgA or IgM could influence bacterial
functioning through differential nutrient utilization. MIC6 carbon
source usage was analyzed with or without Nint_pA or Nint_pM
(Fig. [111]2E). Nint_pM increased glycogen (GLG) usage and decreased
glucose-3-phosphate (G3P), Polysorbate 80 (T80), and γ-hydroxybutyrate
acid (GHB) uptake compared to controls. Nint_pA only decreased G3P
uptake (Fig. [112]2E). In summary, neonatal intestinal IgM,
particularly the pentameric form, preferentially binds to peptidoglycan
(PGN) and induces bacterial aggregation. Although purified IgM isotypes
affected growth of individual bacterial cultures, this effect is not
seen when bacteria are grown as part of a complex microbial community.
However, community metabolism was influenced, impacting nutrient usage
and secretion, thereby affecting bacterial growth. These results
indicate a subtle, indirect influence of secretory intestinal
antibodies on microbial community dynamics.
In summary, we found that IgA and IgM are able to modulate individual
bacterial cultures and modulate differentially bacterial metabolism.
However, IgM have a stronger impact on bacterial aggregation and
metabolism compared to IgA.
Designing a newborn mAb to modulate bacterial functions
We explored the impact of neonatal B cell-derived microbiota-reactive
monoclonal mAbs, specifically BRG291-derived monomeric IgG
(G291)^[113]34 and pentameric IgM (M291) (Supplementary Fig. [114]3A),
on bacterial communities, focusing on their binding to bacterial
surface antigens PGN (Fig. [115]3A) and LPS (Fig. [116]3B). M291
demonstrated robust PGN binding, exceeding that of the polyreactive
ED38 reference mAb (Fig. [117]3A). Neither M291 nor G291 bound to LPS
(Fig. [118]3B). A subsequent analysis of 20 bacterial species, selected
for their representation of gut microbiota diversity, revealed that
M291 recognized bacteria more broadly and more strongly than G291
(Fig. [119]3C–E, Supplementary Fig. [120]3B). These findings suggest
that the pentameric M291 has enhanced bacterial recognition, especially
for PGN, akin to neonatal intestinal secreted IgM.
Fig. 3. Microbial reactivity of neonatal G291 and M291 mAbs.
[121]Fig. 3
[122]Open in a new tab
Monoclonal antibodies derived from BRG291, produced as monomeric IgG
[G291] and pentameric IgM [M291], were evaluated by ELISA for their
reactivity toward PGN (A) and LPS B. Included as controls were the
mGO53 mAb (negative control) and the polyreactive ED38 mAb (positive
control). C, D The same antibodies, G291 and M291, were assessed for
their reactivity with six primary bacterial colonizers in the MIC6
consortium: E. coli, E. faecalis, S. epidermidis, S. salivarius, B.
bifidum, and B. thetaiotaomicron. Results are displayed as
concentration-dependent binding curves (A, B) or cumulative area under
the curve (CAUC) values (C, D). E Comparative recognition of 17
bacterial strains by G291 and M291 assessed by ELISA. Results are shown
as AUC calculated on concentration-dependent binding curves.
We assessed effects of G291 and M291 on individual bacterial cultures
and the MIC6 consortium. No significant growth differences were
observed with G291 on single bacterial cultures (Fig. [123]4A), whereas
M291 significantly enhanced the growth of E. faecalis, S. salivarius,
B. bifidum, and B. thetaiotaomicron (Fig. [124]4B). Before analyzing
the impact on the MIC6 consortium, we first examined bacterial cluster
formation within the MIC6 community using FACS (Supplementary
Fig. [125]4A and [126]4B). M291 or G291 did not significantly affect
the overall bacterial clustering, with clusters of four bacteria
observed under both conditions. However, a greater diversity among
clusters containing at least two bacteria was obtained in the presence
of M291. Specifically, among the 12 clusters composed of two different
bacteria, 8 were enhanced with M291 treatment, compared to only 5 with
G291 (Fig. [127]4C). Overnight culture of MIC6 with M291 showed a
reduction in the number of S. epidermidis and B. thetaiotaomicron
compared to the control and G291 (Supplementary Fig. [128]4C). When
supernatants from the mAb treated MIC6 cultures were applied to single
bacterial cultures, M291 supernatant inhibited the growth of E. coli,
E. faecalis, S. epidermidis, S. salivarius and B. thetaiotaomicron,
whereas the G291 supernatant had no effect compared to the control
(Fig. [129]4D). Further evaluation revealed that G291 and M291 impacted
carbon source utilization differently. Both antibodies significantly
affect the usage of β-Methyl-D-Glucoside (βMDG), T80, and lactose
(Lac), with M291 enhancing βMDG and T80 utilization over G291, though
lactose was unaffected. M291, but not G291, significantly impacted
L-Arginine (L-Arg) and GHB utilization, while G291 uniquely enhanced
N-acetylglucosamine (GlcNac) usage (Fig. [130]4E).
Fig. 4. Comparative effect of the pentameric and monomeric forms of a
neonatal monoclonal antibody on bacterial clustering and carbon source
utilization in vitro.
[131]Fig. 4
[132]Open in a new tab
Indicated bacterial strains were cultured individually in AF medium
with 5 mg/ml of G291 (A) or M291 (B). Growth was measured via OD600 nm
and expressed as a percentage relative to controls without
immunoglobulins. C MIC6 was incubated with G291 or M291 (5 mg/ml), and
bacterial complexes were analyzed by FACS. Changes relative to MIC6
without antibodies are displayed in a heat map. D MIC6 culture
supernatants from incubations with G291 or M291 were used to grow
individual bacteria. Growth was calculated as a percentage of the
control (supernatant without immunoglobulin; gray) based on OD600 nm. E
Carbon source utilization by MIC6 was assessed with Eco-plates, in the
absence (gray) and presence of G291 (green) and M291 (orange), done in
triplicates. A, B and D Each point represents an individual technical
replicates of one of 3 to 4 biological samples (n = 6-12).
Box-and-whisker plots: center line = median; box = 25th–75th
percentiles; whiskers = 5th–95th percentiles. E Box plots represent the
median, as well as the minimum and maximum values (n = 3). Statistical
analyses were conducted according to the respective panels: p values
were calculated with the two-sided Mann-Whitney U test for A, B, and D,
or the two-sided unpaired Student’s t-test for E. Significance levels
are indicated by ****, ***, **, and * for p values < 0.0001, <0.001,
<0.01, and <0.05, respectively; n.s = not significant.
Given that complement components in the intestinal lumen can be
mobilized with or without antibody action to target
bacteria^[133]37,[134]38, we evaluated the combined effect of
complement with M291 on the MIC6 bacterial community (Supplementary
Fig. [135]4D and [136]4E). While bacterial CFU remained unchanged
(Supplementary Fig. [137]4D), carbon source utilization was altered,
with enhanced usage of α-cyclodextrin (α-CD), βMDG, L-Arg, GHB, and
decreased GLG usage (Supplementary Fig. [138]4E).
These findings indicate that a pentameric mAb, like M291, had a greater
influence over bacterial community behavior, affecting growth, carbon
source utilization, and metabolite production, compared to its
monomeric counterpart.
Influence of M291 treatment on MIC6 colonization and metabolome
To investigate the in vivo effects of M291 on gut colonization, we
colonized germ-free (GF) mice with the MIC6 bacterial consortium. Mice
were orally inoculated with MIC6, with a booster dose given on day 3.
Colonization was confirmed on days 7 and 10 using quantitative PCR
(qPCR), and the mice were euthanized 11 days after the initial gavage
(Fig. [139]5A). Experimental design included, one control group of GF
mice without MIC6 (GF_UC), one group receiving the MIC6 consortium
without treatment (GF_Co_UN), and one group treated with M291 via
drinking water, both before and after MIC6 administration (GF_Co_M291)
(Fig. [140]5A). By day 7 post-colonization, mice were predominantly
colonized by B. bifidum and B. thetaiotaomicron, followed by S.
epidermidis, E. faecalis, and S. salivarius, with E. coli remaining
less prevalent (Fig. [141]5B). This bacterial hierarchy was maintained
at day 10. Following M291 treatment, the overall hierarchical
distribution among species was preserved at both day 7 and day 10
(Fig. [142]5B), with a stable relative abundance across bacterial
communities at day 7 (Supplementary Fig. [143]5A). However, M291
treatment significantly reduced E.coli and B. thetaiotaomicron at day
7, and B. bifidum at day 10 (Fig. [144]5B). Altogether, these data
suggest that the MIC6 colonization can be slightly modulated by M291
treatment.
Fig. 5. Effect of M291 treatment on the metabolome of MIC6 colonized GF mice.
[145]Fig. 5
[146]Open in a new tab
A Schematic representation of the experimental design. GF mice were
either left uncolonized (GF_UC) or colonized with the MIC6 bacterial
consortium, comprising E. coli, E. faecalis, S. epidermidis, S.
salivarius, B. bifidum, and B. thetaiotaomicron. Colonized mice were
divided into two groups: untreated (GF_Co_UN) or treated with M291
administered via drinking water (GF_Co_M291). B Fecal samples were
collected, and bacterial growth was assessed by qPCR at day 7 (left
panel) and day 10 (right panel). Each point represents an individual
technical replicates of one of 4 biological replicates (n = 12).
Box-and-whisker plots: center line = median; box = 25th–75th
percentiles; whiskers = 5th–95th percentiles. C Differential metabolite
profiles among GF_UC (green), GF_Co_UN (blue), and GF_Co_M291 (red)
were visualized using a circular heat map (left panel) and PCA (right
panel). D–H Metabolite profile analysis of GF_Co_UN and GF_Co_M291
groups. D Volcano plot comparing metabolites highlighting upregulated
metabolites in red and downregulated metabolites in green. E Heat map
of the 50 most significantly altered metabolites. F Relative
distribution of metabolite superclasses. G Ontology of significantly
upregulated metabolites within the lipids and lipid-like molecules
superclass in GF_Co_M291. H Dynamic Autocorrelation (DA) score of
pathways related to secondary metabolites. Statistical analyses were
calculated with the two-sided Mann-Whitney U test. Significance levels
are indicated by **, and * for p values < 0.01, and <0.05,
respectively.
We next analyzed the metabolites induced by MIC6 colonization. We
identified nearly 1, 000 metabolites, revealing distinct metabolite
profiles that enabled clear differentiation among the three groups
(Fig. [147]5C). Following bacterial colonization, 609 metabolites
showed significant differences compared to GF mice (Supplementary
Fig. [148]5B), with 308 upregulated and distributed across various
classes, in particular lipids and organic acids (Supplementary
Fig. [149]5C). MIC6 bacterial colonization primarily triggered
metabolic processes at the cellular level and within the digestive
system, indicating an overall boost in intestinal activity
(Supplementary Fig. [150]5D). M291 treatment induced a significant
response, resulting in 490 metabolites, with 307 exhibiting significant
differences (Fig. [151]5D). Metabolite variety is further illustrated
among the top 50 metabolites, showing greater ontological diversity for
M291 treatment group compared to the untreated group (Fig. [152]5E).
Metabolites are distributed across various classes, with two major
super classes, organic acids (n = 148) and lipid-related molecules
(n = 141) supporting cellular growth, energy production, tissue
remodeling, and immune system maturation, making them vital for healthy
growth (Fig. [153]5F).
Among lipids, fatty acyls were heavily promoted as they are key in
energy storage or in hormone synthesis. Notably, steroids and their
derivatives were found to be predominantly induced upon M291 treatment,
which are essential for healthy development and functioning
(Fig. [154]5G). A detailed analysis of the secondary metabolic pathways
revealed a general upregulation in response to M291 treatment, except
pyruvate metabolism pathway (Fig. [155]5H). Interestingly, bile acids
(BAs) secretion was induced by M291 treatment (Fig. [156]5H and
Supplementary Fig. [157]5E). Collectively, these results demonstrate
the distinct impact of M291 on the interplay between host and bacterial
metabolisms within a well-defined bacterial consortium. Importantly,
M291 treatment influenced the metabolome without disrupting the
microbial community structure.
Metabolites produced in M291-treated MIC6 colonized mice can affect bacterial
functions
To evaluate the overall impact of metabolomic changes on bacterial
growth, we cultured individual bacterial strains with cecal content
from GF mice. Growth rates were calculated relative to the GF_UC
baseline. Application of the GF_Co_UN metabolome significantly enhanced
the growth of B. bifidum, S. epidermidis, E. faecalis, and E. coli
compared to GF_Co_M291 (Fig. [158]6A). These variations may reflect
differences in metabolic pathways, where 27 pathways were activated
with M291 treatment (Fig. [159]5H). This overall enrichment of pathways
may indicate an elevated metabolic activity that limits nutrient or
secondary metabolite availability for bacterial growth. The metabolic
effects of cecal content were dependent on bacterial concentration and
the quantity of cecal content. Increasing the quantity of cecal content
used to culture bacteria significantly influenced the growth of B.
bifidum, S. epidermidis, E. faecalis, E. coli, and B. thetaiotaomicron
(Supplementary Fig. [160]6A), suggesting that metabolites at lower
concentrations can impact B. thetaiotaomicron growth. Additionally,
when reducing initial bacterial load, growth of E. coli and S.
salivarius decreased significantly (Supplementary Fig. [161]6B).
Interestingly, lower metabolite concentrations had no impact on E. coli
growth but did reduce the growth potential of S. salivarius
(Supplementary Fig. [162]6C). When testing metabolite effects in our
MIC6 setup, we did not observe any significant changes in growth
dynamics induced by the GF_Co_UN metabolic content compared to the one
of GF_UC group. However, GF_Co_M291 significantly enhanced the growth
of E. coli, E. faecalis, and B. thetaiotaomicron (Fig. [163]6B). Cecal
metabolomic changes, particularly those induced by M291 treatment,
significantly influenced the growth of specific gut bacterial species,
highlighting the complex interplay between host metabolism and gut
microbiota. These findings suggest that targeted metabolic
interventions could be used to modulate gut microbial community
functions, potentially offering benefits for various gastrointestinal
conditions.
Fig. 6. Impact of metabolome from mAb treated colonized GF mice on in vitro
bacterial culture.
[164]Fig. 6
[165]Open in a new tab
Indicated bacterial strains were cultured individually (A) or as the
MIC6 consortium (B) in 1 mL of cecal metabolite content derived from
GF_Co_M291, GF_Co_UN, or GF_UC. Growth was assessed by measuring the OD
at 600 nm and expressed as a percentage relative to the control
(GF_UC), represented by the dashed line. A, B Each point represents an
individual technical replicates of one of 4 biological replicates
(n = 12). Box-and-whisker plots: center line = median; box = 25th–75th
percentiles; whiskers = 5th–95th percentiles. Statistical analyses were
conducted using the two-sided Mann-Whitney U test. Significance levels
are indicated as follows: **** for p values < 0.0001, *** for p
values < 0.001, ** for p values < 0.01, and * for p values < 0.05.
Influence of M291 treatment on newborn FMT and associated metabolome
Using microbiota from non-breastfed newborns, we performed a fecal
microbiota transplant (FMT) in GF mice, which were then treated with
either M291 (GF_FMT_M291) or Nint_pM (GF_FMT_NintpM) as previously
described, and compared them to untreated control (GF_FMT_UN) and
uncolonized (GF_UC) groups (Fig. [166]7A). We conducted an analysis of
FMT sequencing and the gut microbiota composition of colonized GF mice
(GF_FMT). Bacterial composition was severely modified at the phylum,
family and genera levels upon colonization (Fig. [167]7B–D,
Supplementary [168]7A, B) reflecting a heavy loss in diversity. Among
all GF_FMT samples, seven species accounted for up to 99.9 % of the
total detected bacteria, with E. faecalis, K. aerogenes, and C.
butyricum emerging as the predominant species while C. paraputrificum,
B. breve, E. fergusonii and K. pneumoniae were present at lower
abundance (Fig. [169]7D, Supplementary Fig. [170]7A). Evenness was
higher in GF_FMT_NintpM compared to GF_FMT_UN or GF_FMT_M291 at day 3
suggesting uniformity in species abundance, that no single species
overwhelmingly dominates the community. This significant difference
diminishes by day 7 and is no longer observed by day 10 (Supplementary
Fig. [171]7B). The global analysis revealed a distinct separation among
the three groups on day 3, which gradually converged as the
colonization process progressed (Supplementary Fig. [172]7C).
Fig. 7. Bacterial abundance and gene expression in GF mice following FMT
colonization.
[173]Fig. 7
[174]Open in a new tab
A Experimental design procedure for GF mice, left uncolonized (GF_UC),
or colonized through Fecal Matter Transfer (FMT) under three
conditions: without treatment (GF_FMT_UN), under M291 treatment
(GF_FMT_M291), and under Nint_pM treatment (GF_FMT_NintpM). B Relative
abundance of phyla, C and genera of bacteria identified in the FMT
input and in GF mice after receiving FMT samples. D Relative abundance
of main bacterial species identified across all conditions
post-colonization. Metatranscriptomic analysis performed at Day 10.
Heat maps of taxonomic analysis (E) and differentially expressing genes
(DEG) identified (F) of input (green), GF_FMT_UN (blue) and GF_FMT_M291
(red). G Pathways identified from differentially expressing bacterial
genes in untreated vs. M291 treated mice represented as donuts
corresponding to major KEGG pathway process (upper panel) and gene
classes (lower panel) involved in Metabolism pathways.
Monitoring of microbial colonization over the experimental period
showed a clear dominance of C. butyricum, K. aerogenes, and E.
faecalis, which together constituted 98–99% of the population by day 7.
Meanwhile, lower-abundance species, including C. paraputrificum, B.
breve, and E. fergusonii displayed increasing colonization over time.
By day 10, E. fergusonii and K. pneumoniae showed a tendency to decline
(Supplementary Fig. [175]7D). Following antibody treatment,
colonization of C. butyricum, K. aerogenes, and E. faecalis remained
unaffected. However, B. breve and K. pneumoniae exhibited trends
opposite to those observed in GF_FMT_UN-colonized mice, with B. breve
decreasing by day 10, while K. pneumoniae increased (Supplementary
Fig. [176]7D). Nevertheless, no significant changes in species relative
abundance were observed.
As shown by bacterial gene sequencing analysis, the bacterial
composition at the genus level in FMT samples differs from that in
GF_FMT_UN. However, no significant differences were observed between
GF_FMT_UN and GF_FMT_M291, as evidenced by the lack of clustering
between these groups (Fig. [177]7E). Metatranscriptomic analysis
revealed differential expression of 348 bacterial genes (Fig. [178]7F,
Supplementary Fig. [179]7E). FMT samples were clearly distinct from
GF_FMT_UN, indicating active bacterial metabolism. In contrast,
GF_FMT_M291 samples exhibited a distinct pattern of gene regulation
upon M291 treatment (Fig. [180]7F). Collectively, these data
demonstrate that M291 treatment does not alter the overall bacterial
species composition at the community level but significantly affects
gene expression. Notably, of the 348 differentially expressed genes,
264 are involved in metabolic pathways, with 93 genes implicated in
energy or carbohydrate metabolism (Fig. [181]7G). These findings
highlight the modulation by M291 of bacterial functions, particularly
those related to metabolic activity and carbohydrate utilization.
We therefore performed metabolite analysis of cecal content, revealing
distinct clustering patterns between the four different groups
(Fig. [182]8A). We analyzed the effects of the metabolome from GF
colonized mice on bacterial growth in vitro. MIC6 bacteria were
cultured individually or as a consortium (Supplementary
Fig. [183]8A–C). Cecal content from uncolonized GF mice served as a
control for bacterial growth. Notably, E. faecalis growth was enhanced
only in the GF_FMT_UN group, while treatments with GF_FMT_M291 and
GF_FMT_NintpM did not significantly affect its growth. M291 treatment
promoted the growth of B. thetaiotaomicron while inhibiting E. coli
growth, whereas Nint_pM treatment specifically enhanced E. coli growth
(Supplementary Fig. [184]8A). The impact of cecal content on bacterial
growth was concentration-dependent. (Supplementary Fig. [185]8B).
Furthermore, the effects of the metabolome were tested using the entire
MIC6 community. Interestingly, S. epidermidis growth remained
unaffected, whereas E. coli, E. faecalis, S. salivarius, B. bifidum,
and B. thetaiotaomicron were all significantly impacted by GF_FMT_M291
and GF_FMT_NintpM treatments, which notably decreased their growth
(Supplementary Fig. [186]8C). Additionally, M291 treatment associated
metabolome altered the carbohydrate usage profile within the bacterial
consortium, with GF_FMT_M291 exhibiting a more diversified carbohydrate
utilization compared to GF_FMT_UN (Supplementary Fig. [187]8D). Taken
together, our findings suggest that microbiota-targeting antibody
treatments have a modest influence on bacterial composition in vivo.
However, the associated evenness is linked to more substantial changes
in the metabolome, as evidenced by its significant impact on the growth
of bacterial species.
Fig. 8. Differential impact of FMT and M291 on the metabolome of colonized
germ-free mice.
[188]Fig. 8
[189]Open in a new tab
A Metabolomic analysis across GF_UC (light blue), GF_FMT_UN (green),
GF_FMT_M291 (dark blue), and GF_FMT_NintpM (red) groups as defined in
Fig. [190]7A, displayed as a circular HMP and a PCA plot. B Volcano
plot comparing metabolites between GF_FMT_UN and GF_FMT_M291, with
upregulated and downregulated metabolites. C Heat map highlighting the
50 most significant metabolites in the comparison of GF_FMT_UN and
GF_FMT_M291. D Relative distribution of metabolite superclasses in
GF_FMT_M291 and GF_FMT_UN. E Ontology of significantly upregulated
metabolites within the lipids and lipid-like molecules superclass in
GF_FMT_M291. F Pathway impact and statistical significance of the
metabolic pathways identified through pathway enrichment analysis of
GF_FMT_M291 and GF_FMT_U. Note that the y-axis shows the negative
logarithm of p-values;,therefore, pathways with higher statistical
significance are plotted higher in the graph. G Metabolites of linoleic
acid metabolism pathway analysis LA (Linoleic Acid); γ-LA(gamma
Linoleic Acid); AA (Arachidonic Acid). Each point represents a
biological replicates (n = 4). Box-and-whisker plots: center line =
median; box = 25th–75th percentiles; whiskers = 5th–95th percentiles.
Statistical analyses were conducted using the two-sided Mann-Whitney U
test. Significance levels are indicated as follows: ** and * for p
values < 0.01, and <0.05, respectively; n.s.= not significant. ** for p
values = 0.0079, and * for p values = 0.0159.
Following FMT colonization, 1360 metabolites were identified in GF
mice, of which 510 were differentially expressed, with 148 metabolites
being upregulated as a result of the FMT colonization (Supplementary
Fig. [191]8E). We analyzed the metabolome of the FMT colonized mice.
Among these 510 metabolites induced in GF_FMT_UN, 182 were classified
as lipids and lipid-like molecules and 83 organic acids and derivatives
(Supplementary Fig. [192]8F). These metabolites primarily influence
intestinal and cellular functions, with notable activation in the
biosynthesis of secondary metabolites and amino acid metabolism,
signaling an active intestinal microbiome (Supplementary Fig. [193]8G).
Interestingly, GF mice colonized with distinct pioneer microbiota
shared 349 metabolites, with lipids and lipid-like molecules
constituting approximately one-third of the common metabolites
(Supplementary Fig. [194]8H). Among this superclass, half were fatty
acyls, and a quarter were steroids and derivatives (Supplementary
Fig. [195]8H). Treatment with M291 led to the differential expression
of 613 metabolites between the GF_FMT_UN and GF_FMT_M291 groups, with
202 metabolites upregulated in the presence of M291 (Fig. [196]8B).
Among the top 50 differentially expressed compounds, metabolite
diversity was notably greater in the absence of M291 (Fig. [197]8C).
Analysis of these 613 metabolites indicated that lipids and lipid-like
molecules, along with organoheterocyclic compounds, remained the
dominant superclasses, with lignans, neolignans, and related compounds
emerging as the third largest superclass (Fig. [198]8D). Ontology
analysis within these superclasses highlighted a significant
representation of prenol lipids and steroids among lipid and lipid-like
molecules (Fig. [199]8E).
Upon analyzing the secondary class pathways following M291 treatment,
we observed the up-regulation of 17 pathways, with 2 remaining
unchanged and 11 being down-regulated. Among the down-regulated
pathways, we focused on the linoleic acid (LA) metabolism
(Fig. [200]8F), which is significantly influenced by various bacterial
genera within the gut, such as Bifidobacteria and
Roseburia^[201]39,[202]40. LA metabolites significantly influence gut
health, immunity, and host physiology. LA is a precursor to arachidonic
acid (AA), which undergoes metabolism via three primary enzymatic
routes: cyclooxygenase, lipoxygenase, and cytochrome P450 pathways.
These pathways generate bioactive lipids—prostaglandins, leukotrienes,
and epoxyeicosatrienoic acids—that are essential for cell
differentiation, tissue development, organ function, and disease
pathogenesis^[203]41. Upon FMT colonization, we observed a decrease in
LA levels. This reduction in LA directly impacts the entire metabolic
pathway, extending from LA to AA. Notably, the downstream metabolites
derived from AA metabolism exhibit differential alterations.
Specifically, while prostaglandins and leukotrienes remain unchanged,
the levels of 14(15)-epoxyeicosatrienoic acid (14(15)EpETE) are
decreased (Fig. [204]8G). Upon treatment with M291, there is a
significant enhancement in the downregulation of the LA metabolic
pathway, extending from LA to AA metabolites (Fig. [205]8G).
Concurrently, while 14(15)EpETE levels remain unchanged, there is an
upregulation of leukotriene production and a downregulation of
prostaglandin synthesis (Fig. [206]8G). Collectively, the observation
underscores the intricate relationship between LA metabolism and the
modulation of gut health and immune responses following M291 treatment
and FMT colonization.
Metabolomic analysis indicates that M291 treatment significantly alters
the metabolic profile, of the gut particularly enhancing lipid
metabolism. The treatment also differentially impacts the growth of
specific bacterial species within the MIC6 community, suggesting a
complex interplay between bacterial growth and metabolic activity.
These findings underscore the potential of targeted metabolic
interventions in modulating gut microbiota functions, with implications
for therapeutic strategies in gut-related disorders.
Discussion
This study provides novel insights into the role of neonatal IgM
antibodies with broad microbial reactivities in modulating the
interplay between gut microbiota and host metabolism. We developed the
pentameric M291 mAb that exerts distinct effects on bacterial growth
and functions, metabolic pathways, without significantly disrupting the
microbial community structure. This suggests its therapeutic potential
as a modulator of gut metabolism, with implications for early-life
development and long-term health outcomes.
Upon birth, neonates are exposed to microorganisms that initiate the
colonization. The balance between pathogen defense and minimal
inflammation is critical for establishing symbiotic relationships with
commensal bacteria. We characterized gut-secreted antibodies, their
binding capacity to various microbial species and their potential
influence on gut colonization^[207]42. Secretory Igs production
correlates with the proportion of coated bacteria in the neonatal gut
and their ability to recognize bacterial antigens and species
throughout the first month of life, with a clear preference for
Gram-positive bacteria. Such coated bacteria form clusters that may
prevent direct bacterial interaction with epithelial cells, thereby
protecting the host; modulating bacterial gene expression in bacteria
such as B. thetaiotaomicron; and potentially promoting microbial
community homeostasis through effects on motility, bacterial fitness,
or bacteriophage susceptibility^[208]43. IgM predominates in the gut
immediately after birth, in agreement a late IgA
production^[209]28,[210]29. Our previous work demonstrated that pools
of IgG mAbs derived from newborn B cells can modulate the metabolic
activities of synthetic microbial communities in in vitro bacterial
cultures^[211]34. The binding properties of M291 to PGN and its
enhanced bacterial recognition compared to its monomeric counterpart
(G291) underscore the importance of polymeric antibody structures in
mediating host-microbiota interactions. M291 promoted bacterial cluster
formation and impacted bacterial growth within the MIC6 community,
indicating that multivalency of IgM antibodies is well-suited to
modulate complex microbial consortia. Nevertheless, the lack of a
disease model in this study restricts our ability to extrapolate these
findings to pathological contexts where such interventions might be
most needed.
Recently, a mouse monoclonal IgA (W27) has been demonstrated to
modulate gut microbiota by selectively inhibiting the growth of certain
bacteria, such as E.coli, without affecting beneficial species like L.
casei. This indicates that IgA has a regulatory role in gut microbial
balance and health^[212]44. Our study shows that M291 treatment
influenced bacterial growth and metabolism. Cecal content from
M291-treated mice, when added to cultures, especially affected growth
of B. bifidum and B. thetaiotaomicron in relation with carbon source
availability. For example, growth of B. thetaiotaomicron is inhibited
by B. breve in the presence of glucose but not affected by B.
longum^[213]45. B. fragilis supports B. longum survival and changes
organic acid production in co-culture. Bifidobacterium can utilize
complex carbohydrates like human milk oligosaccharides and
plant-derived glycans through transport systems, degrading them into
simpler sugars for fermentation. This metabolic strategy supports
cross-feeding interactions among Bifidobacterium strains and other gut
bacterial spp., processes modulated by host dietary inputs and critical
for maintaining gut homeostasis^[214]46. In addition, Bifidobacteria
can compete for nutrients, which can help reduce their bacterial growth
and colonization^[215]47. This competition can also lead to changes in
the composition of the gut microbiome, favoring the growth of
beneficial bacteria like Bacteroidetes^[216]47.
Our findings suggest that the metabolites produced in response to M291
treatment serve as modulatory factors, shaping bacterial physiology and
interspecies interactions. Utilizing a GF mouse model, we demonstrated
the impact of mAb treatment on the metabolome while bacterial
composition remained similar between groups. With two distinct
consortia, we observed a differential impact of mAb treatment on the
metabolome, adjusted according to the bacterial species used as
colonizers of GF mice.
Using a well-defined community composed of E. coli, E. faecalis, S.
salivarius, S. epidermidis, B. bifidum, and B. thetaiotaomicron, we
identified a significant number of metabolites and pathways that are
modulated. Lipid and protein metabolism were upregulated in the
presence of mAb, indicating a stimulation of the intestinal tract
involving both host and microbiota. We observed a substantial
upregulation of the BAs secretion pathway, crucial for digestion and
absorption of dietary fats and fat-soluble vitamins, as well as
regulation of cholesterol and glucose metabolism, contributing to
overall nutritional health^[217]48. BAs are crucial for digesting fats,
absorbing vitamins, and shaping the gut microbiota, impacting long-term
host health by regulating bacterial growth and gut
health^[218]49,[219]50.
Effect of M291 on lipid-related molecules like prenol lipids and
steroids shows it can adjust the interaction between the host and
microbes’ metabolisms. M291 also affected LA metabolism, which is
important for gut health. It lowered LA levels and changed the AA
pathway. AA is a key fatty acid that helps with gut health by producing
substances like prostaglandins and leukotrienes, which control
inflammation and immune reactions^[220]51. Gut bacteria can change AA
into substances that affect gut health^[221]52. The modulation of lipid
mediators, particularly by M291 treatment, could have a complex impact
on gut homeostasis and immune responses. This suggests that LA levels
can influence the stress response and metabolic activity of certain gut
bacteria. A reduction in LA levels, has been shown to induce metabolic
stress in intestinal bacteria like B. breve, affecting their metabolic
capabilities and growth rate^[222]39. LA metabolites, found in
fermentation pellets, indicate their incorporation into bacterial
cells, implying that LA levels could alter the composition and
functionality of the gut microbiota^[223]39. Decreased prostaglandin A1
in gut metabolism can significantly affect gastrointestinal protection,
inflammation, and immune responses, while the upregulation of
leukotriene D4 is associated with exacerbated effects. These findings
indicate that fecal FMT and M291 treatment profoundly influence gut
homeostasis and immune responses, highlighting the intricate interplay
between dietary lipid metabolism and gut microbiota in modulating host
health and disease progression. M291 treatment significantly alters the
metabolic profile of the gut, enhancing lipid metabolism and
differentially impacting the growth of specific bacterial species
within the MIC6 community, underscoring the potential of targeted
metabolic interventions in modulating gut microbiota functions with
implications for therapeutic strategies in gut-related disorders.
These metabolites are important for immune control and keeping tissues
healthy, so effect of M291 on LA metabolism might be important for gut
health and overall immunity^[224]53.
The relevance of our findings is underscored by the analysis of
metabolites generated from consortia comprising bacteria not typically
associated with FMT. This is illustrated by the effect of metabolites
from M291-treated GF mice on the utilization of carbon sources by the
MIC6 community. This shows that M291 treatment can influence the
metabolism of bacteria, directly or indirectly, through metabolite
interactions. Metatranscriptomic analysis demonstrates a change in
bacterial metabolism, due to a differential bacterial gene expression
upon treatment. A capsular polysaccharide Ab has been shown to change
the gene expression of S. pneumoniae during colonization, affecting
bacterial fitness^[225]54. These adaptations could affect the microbial
ecosystem’s dynamics, showing the complex relationship between
metabolites, bacterial physiology, and community function.
Our results show that antibodies can change how the gut microbiota
works, as seen by increased activity in metabolic pathways, including
those for lipids and proteins, when treated with mAb. The activation of
BAs secretion pathways also shows the complex interaction between the
host’s factors and microbial metabolism, which is important for
nutrient absorption and health. Mechanisms involve microbiota-dependent
changes in bile acid metabolism, affecting signaling through bile acid
receptors such as FXR and TGR5^[226]55. These changes influence lipid
metabolism by altering bile acid recycling and enterohepatic signaling
pathways. Adaptive responses to dietary lipids are regulated by
epithelial processes involved in both digestion and
absorption^[227]56,[228]57. The mechanisms underlying these
interactions appear to involve bacteria-derived small molecules that
upregulate lipid absorption genes in small intestinal epithelial cells.
L-lactate from L. paracasei promotes lipid storage by inhibiting
beta-oxidation, while acetate from E. coli enhances lipid oxidation,
highlighting their roles as key regulators of enterocyte lipid
metabolism^[229]58. Duodenal bacterial isolates decrease lipid
absorption in both cultured enterocytes and mice, providing a mechanism
by which they may exacerbate environmental enteric dysfunction and
stunting in children^[230]59. Together, these insights highlight the
central role of regulating lipid metabolism and its broader
implications for addressing overnutrition, undernutrition, and related
disorders.
Monoclonals have greatly developed highlighting their therapeutic
applications across various diseases, including cancer and infectious
diseases^[231]60,[232]61 and here we present a new path, a new field of
use that influences metabolism.
We demonstrate the potential of neonatal-derived antibodies as tools
for therapeutic intervention in the gut. By modulating specific
bacterial behaviors and metabolic pathways, mAb like M291 offer a
targeted approach to restoring or enhancing gut homeostasis. This
approach is particularly relevant in the context of microbiota gut
colonization, where interventions to prevent dysbiosis could have
lifelong health benefits. The ability of M291 to selectively influence
metabolomic profiles without disrupting microbial diversity further
underscores its potential for safe and effective modulation of the gut
environment. However, because only a single monoclonal antibody (M291)
was evaluated, we cannot yet generalize these observations to other
antibody clones or isotypes, and systematic screens will be required to
define structure–function relationships across the broader neonatal
antibody repertoire. Therefore, antibodies targeting microbiota act as
modulators of microbial metabolism, affecting bacterial dynamics and
ecosystem balance. Future studies should explore the applicability of
such interventions in models of neonates, dysbiosis, disease, as well
as the potential synergistic effects of combining mAb treatments with
dietary or probiotic strategies.
Methods
Inclusion and ethics statements
This research follows the guidelines of inclusion and ethics by Nature
Communications. All authors have fulfilled the authorship criteria.
Preparation of newborn fecal samples
Fecal samples for intestinal IgM/IgA analysis were obtained from
formula-fed healthy newborns during the first month of life with the
written consent of the parents under the approval No. 2022-IPID-01.
Fecal samples were resuspended at 100 mg/ml in cold sterile PBS and
homogenized by pipetting and vortexing. Fibers were spun down at 500 ×
g for 5 min. Supernatants were collected, filtered through 100 μm, and
spun down at 10,000 × g or 10 min. Supernantants were collected,
filtered at 0.22 μm, and directly frozen at −80 °C (fecal water). The
bacterial pellets were washed twice with cold PBS and resuspended in
30% sterile glycerol at a concentration of 1 g/mL prior to storage at
−80 °C and use for FMT germ-free mouse experiments.
Newborn intestinal Ig purification
Newborn fecal samples with an IgA:IgM ratio of 1:2 from three samples
of Week 2 and Week 3 (days 12, 15 and 19) were pooled, and IgA were
purified by affinity chromatography using Peptide M agarose (gel-pdm-5)
stack into Poly-Prep® Chromatography Columns (7311550; Bio-Rad). After
IgA purification, the flow-through from the Peptide M agarose column
was used for IgM precipitation. An equal volume of saturated ammonium
sulfate solution was added to the flow-through to achieve a final
concentration of 50% ammonium sulfate. The mixture was incubated
overnight at 4 °C with continuous stirring. The precipitate was
collected by centrifugation at 10,000 × g for 30 min, and the resulting
pellet was resuspended in PBS. The resuspended pellet was dialyzed
against PBS using a dialysis bag with a molecular weight cut-off (MWCO)
of 100 kDa overnight at 4 °C. The purity of IgA and IgM was assessed by
cross-ELISA (IgM detection on purified IgA fraction; IgA detection on
purified IgM fraction) reaching 99% and 93%, respectively.
ELISA
Bacterial recognition was measured by ELISA as previously
described^[233]62. Briefly, high-binding 96-well ELISA plates were
coated overnight with 250 ng/well of purified LPS (L2637-10MG;
Sigma-Aldrich), PGN (69554-10MG; Sigma-Aldrich), or with PFA 0.2% fixed
bacteria on precoated Poly-L-lysine (3ug/ml) plates. After blocking and
washing steps with 0.001% Tween 20-PBS, mAbs or purified intestinal
antibodies were adjusted and tested at 4 μg/ml and 3 consecutive 1:3
dilutions in PBS. ELISA plates were developed using goat HRP-conjugated
anti-human IgG, IgM, or IgA antibodies. Results are given as calculated
AUC from titration curves in the graphs. ED38 and mGO53 mAb were used
as polyreactive and non-polyreactive controls, respectively. Results
were calculated as the area under the curve (AUC) from titration
curves. First, the curve was created by plotting the ERL data over
time. Then, AUCs were divided into multiple small trapezoid areas,
which are measured individually, using the trapezoid rule [area = ½
(base a + base b) x height], and added up to get the total AUC.
Production of G291 and M291 monoclonal antibodies
BRG291 sequences were selected from our previous study^[234]34 to
produce monomeric IgG1 and pentameric IgM forms. Recombinant antibodies
were produced on selected sequences cloned into human Igγ1-
Igλ-expressing vectors as previously. Recombinant G291 (IgG1) was
produced by transient co-transfection of adherent HEK 293 T using PEI-
precipitation method as previously described^[235]34, and purified by
affinity chromatography using Protein G Sepharose® 4 Fast Flow
(Cytiva). Purified antibodies were dialyzed against PBS^[236]34. M291
was produced by co-transfection with J chain-containing plasmid of
Freestyle™ 293-F suspension cells. M291 was purified using
size-exclusion chromatography with multi-angle light scattering
(SEC-MALS). The molecular weight was confirmed by sodium dodecyl
sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and
high-performance liquid chromatography (HPLC), achieving an 85.3%
purity of the pentameric form of M291.
Bacterial strains
All bacterial strains were procured from the Chinese Human Gut
Microbial Biobank (hGMB). E. coli, E. faecalis, S. epidermidis, S.
salivarius, B. bifidum, and B. thetaiotaomicron were used^[237]34.
Bacterial medium and growth conditions
Bacterial strains were obtained from different sources and cultured in
AF media adapted from^[238]63, including brain heart infusion 18 g/L,
trypticase Soy Broth 15 g/L, yeast extract 5 g/L, KH2PO4 2.5 g/L,
heamin 0.5 mg/L D-glucose 0.5 g/L) for synthetic consortia design
culture in aerobic or anaerobic conditions. Anaerobic bacterial strains
and synthetic consortia were culture in anaerobic conditions (7% H2,
10% CO2, 83% N2). Each bacterial strain individual culture condition
was assessed to accurately define the starting CFU number and the
growth culture duration required to analyze growth at exponential
phase. For individual culture experiments, the effect of
immunoglobulins was calculated as follows:
[MATH: Growth(%
)=bacteria growth(OD)*100averag
eofbacterialgrowthinPBS(<
mrow>OD) :MATH]
Synthetic MIC6 bacterial consortia cultures
Synthetic bacterial community was cultured as previously
described^[239]63. MIC6 is composed of E. coli, E. faecalis, S.
epidermidis, S. salivarius, Bifidobacterium bifidum, and B.
thetaiotaomicron. Briefly, bacterial strain monocultures were prepared
from an overnight culture and subculture and were diluted to OD[600]
0.1 in AF medium for overnight culture. Monoculture were were mixed
after reaching the exponential phased in the following proportions: E.
coli (OD 0.01), E. faecalis (OD 0.01), S. epidermidis (OD 0.1), S.
salivarius (OD 0.01), B. bifidum (OD 0.01), and B. thetaiotaomicron (OD
0.1) before co-culturing in AF medium. After overnight culture the
co-culture was diluted 10 times in AF medium for 2 consecutive days and
100 times for 2 consecutive days incubated at 37 °C without shaking
under anaerobic conditions. Synthetic consortia species stability was
assessed by qPCR and stored in glycerol at −80 °C for further
experiments. In indicated experiments, bacterial consortia were
incubated with 50 µl of mAbs (60 µg/ml) or with purified intestinal Ig
(60 µg/ml) 30 min at 4 °C. After wash, bacteria were incubated either
with 50 µL of guinea pig complement (Cedarlane) or anti-IgG1
crosslinking Abs (clone HP6069), then O.N. cultured. at 37 °C in 1 mL
AF media.
Microbial ecological cultures
Community–level physiological profiling has been performed to monitor
changes in MIC culture performed in an anaerobic chamber using the
Biolog EcoPlate system, which contains 31 independent carbon sources
that are useful for their metabolism. The metabolic fingerprint
reaction patterns is monitored by following the respiration that
rapidly reduces the tetrazolium dye included with the carbon source and
was measured by OD at 590 nm.
Fluorescent bacterial labeling
Bacterial strains were grown individually until reaching the
exponential phase and diluted to OD[600] 0.2 and stained with
fluorescent D-amino Acids (FDAAs)^[240]64. Briefly, bacteria were
cultured in AF media in the presence of 0.5 mM of either RADA (# 6649
TOCRIS), sBADA (# 7860 TOCRIS), YADA (# 6650 TOCRIS) or Rf470DL (# 7406
TOCRIS) with a final concentration of DMSO 1%. Bacteria were cultured
for a few hours at 37 °C before washing in cold sterile PBS and fixed
30mn PFA 4%. Bacterial fluorescent labeling was assessed by FACS. FAADs
bacteria were stored −80 °C in 30% glycerol and adjusted to OD1 for
further analyzes.
Analysis of bacterial coated antibodies
Fluorescent bacteria were pooled at 2 µL for each strain. To measure
IgM- and IgA-coated intestinal bacteria, 5 µL of microbial pellets were
resuspended in PBS. The bacterial suspensions were passed through a
40 µm cell strainer to eliminate aggregates and ensure uniformity.
Fluorescent bacterial mixes (E. coli + E. faecalis + S.
epidermidis + S. salivarius and B. bifidum + B. thetaiotaomicron) were
then treated with 1 µg of G291, M291, Int_pM, Int_pA, or PBS as a
control at 4 °C for 30 min. Following treatment, bacteria were washed
twice with sterile, cold PBS before proceeding to FACS cluster
analysis. Staining included IgM PerCP-Cy5.5 (clone: MHM-88, 1:100),
anti-human IgA Viogreen (clone: IS11-8E10, 1:100), or anti-human IgG
BV605 (clone: G18-145, 1:100) for 30 min at 4 °C. The bacteria were
washed again and directly analyzed using the Spectral Cytek Aurora. For
the analysis of IgM- and IgA-coated intestinal bacteria, 5 µL of
microbial pellets were resuspended in PBS, filtered through a 40 µm
strainer, and incubated at 4 °C for 30 min with anti-human IgM
PerCP-Cy5.5 (clone: MHM-88, 1:100) and anti-human IgA Viogreen (clone:
IS11-8E10, 1:100) before FACS analysis. To investigate fluorescent
bacterial clustering, we initially analyzed bacterial clusters in the
absence of antibodies to establish a baseline of natural bacterial
interactions. Subsequently, we applied the same gating strategy to
bacterial mixtures in the presence of antibodies. By comparing the
proportions of each cluster, we observed variations between the PBS
control and antibody-treated samples. These differences reflect the
potential of the antibodies to modulate bacterial interactions, either
by enhancing or reducing them.
Extraction of DNA from bacteria
DNA from 1 ml of a single bacterial culture or from synthetic consortia
was extracted using the E.Z.N.A bacterial kit (Omega, D3350) following
the manufacturer’s instructions. Briefly, bacteria pellets were
incubated with 100 µl of TE buffer and 10 µl of lysozyme for 10 min at
37 °C. For Gram-positive bacteria and synthetic consortia, this step
was directly followed by a glass bead lysis 5 min at a maximum of
speed. Supernatants were incubated with 100 µl of BTL and 20 µl
protease K at 55 °C for 1 h to disrupt the cell wall, and 5 µl of RNase
A was added, mix gently and incubate RT for 5 min. After spinning,
supernatants were incubated 10 min at 60 °C with 220 µl of BDL buffer
and transfer into the provided column t after adding 220 µl of ethanol
75%. After spinning, flow through were discard and 500 µl of HBC was
added to the column before 2 steps of wash. DNA were then eluted using
elution buffer pre-warmed at 65 °C and the DNA amount was measured at
260 and 280 nm.
Real-time quantitative PCR
Real-time qPCR was performed using an ABI Quantstudio Sequence
Detection System apparatus. The qPCR reaction mixture (20 µl) was
composed of 10 µM of each primer, 2x Power SYBR Green PCR Master Mix
(AidLab, PC3302), and 2 µl of extracted DNA. The amplification program
consisted of one cycle of 95 °C for 10 min, followed by 40 cycles of
95 °C for 1mn, 55 °C for 50 s, and 72 °C for 1 min followed by eluting
curve program consisting on 50 °C for 15 s and 95 °C for 15 s. Primers
used for MIC6 qPCR were as follows: E. coli-F
(5’-CATTGACGTTACCCGCAGAAGAAGC-3’);
E.coli-R(5’-CTCTACGAGACTCAAGCTTGC-3’)^[241]65; E. faecalis-F
(5’-ATCAAGTACAG TTAGTCT-3’); E. faecalis-R
(5’-ACGATTCAAAGCTAACTG-3’)^[242]66; S. epidermidis -F
(5’-CATTGGATTACCTCTTTGTTCAGC-3’); S. epidermidis-R (5’-CAAGCGAAATCTGTT
GGGG-3’)^[243]67; S. salivarius-F (5’-GTTCCAGCAGCTAAAGAGGAAG-3’); S.
salivarius-R (5’-CCGGTGCTACTTTAGCTACTGG-3’)^[244]68; B.
thetaiotaomicron-F(5’-GGARCATGT GGTTTAATTCGATGAT-3’); B.
thetaiotaomicron -R (5’-AGCTGACGACAACCATGC AG-3’)^[245]69; B.
bifidum-F(5’-AGGGTTCGATTCTGCTCAG B. bifidum -R(5’-CATCCG
GCATTACCACCC-3’)^[246]70.
Determination of copy number
Bacterial suspensions were plated on AF agar plates ON 37 °C with
serial dilutions from OD[600] 10^–3 to 10^–8 to count CFU/ml, and DNA
from the same dilutions was extracted to do qPCR. Linear curves from OD
and CT values were used to calculate the copy numbers for each
strain/growth condition.
Germ-free mouse colonization and treatment
Animal experiments were approved by the ethical committee (Ethical
Approval No. A2023025). 6 weeks old Germ free male C57Bl/6 mice were
kept in cage setups with autoclaved food and water to minimize unwanted
cross-contaminations in 12 light/12 dark cycles with 18-23 °C with
40-60% humidity. Mice and the environment were tested to confirm
germ-free conditions. Mice were orally administered with 150 µl of MIC6
consortium (10^9 CFU of each bacterial strain composing MIC6 (6 × 10^9
CFU / mouse)) or FMT (equivalent to 30 µg fecal matter). M291 or
Nint_pM was provided in the drinking water at 20 µg/ml 2 days before
and all along the experiment^[247]71. Mice were euthanized 11 days
after the first gavage.
Preparation of cecal content for metabolome analysis and bacterial culture
Caecum from mice were retrieved and weighted. Cecal content was
homogenized in cold sterile PBS at 1 g/ml. Homogenate was spun down at
10,000 × g for 10 min at 4degres. Supernatant were collected and
analyzed for metabolome content. Pellets and 0.22 µm filtered
supernatants were aliquoted individually and immediately frozen at
−80 °C for bacterial growth experiments.
Metabolomics
Supernatants were prepared from the caecum as described above. The
sample (100 µL) was thoroughly mixed with 400 µL of cold methanol
acetonitrile (v/v, 1:1) via vortexing. Then, the mixture was processed
with sonication for 1 h in an ice bath. The mixture was then incubated
at −20 °C for 1 h and centrifuged at 4 °C for 20 min at a speed of
14,000 g. The supernatants were then harvested and dried under vacuum
for LC-MS analysis. UHPLC-MS/MS analysis. Metabolomics profiling was
analyzed using a UPLC-ESI-Q-Orbitrap-MS system (UHPLC, Shimadzu Nexera
X2 LC-30AD, Shimadzu, Japan) coupled with Q-Exactive Plus (Thermo
Scientific, San Jose, USA). For liquid chromatography (LC) separation,
samples were analyzed using the ACQUITY UPLC® HSS T3 column
(2.1×100 mm, 1.8 μm) (Waters, Milford, MA, USA). The flow rate was
0.3 mL/min, and the mobile phase contained A: 0.1% FA in water and B:
100% acetonitrile (ACN). The gradient was 0% buffer B for 2 min and
linearly increased to 48% in 4 min, then up to 100% in 4 min and
maintained for 2 min, then decreased to 0% buffer B in 0.1 min, with a
3 min re-equilibration period employed. The electrospray ionization
(ESI) with positive and negative modes was applied for MS data
acquisition separately. The HESI source conditions were set as follows:
Spray Voltage: 3.8 kv (positive) and 3.2 kv (negative); Capillary
Temperature: 320 °C; Sheath Gas (nitrogen) Flow: 30 arb (arbitrary
units); Aux Gas Flow: 5 arb; Probe Heater Temp: 350 °C; S-Lens RF
Level: 50. The instrument was set to acquire over the m/z range
70-1050 Da for full MS. The full MS scans were acquired at a resolution
of 70,000 at m/z 200, and 17,500 at m/z 200 for the MS/MS scan. The
maximum injection time was set to 100 ms for MS and 50 ms for MS/MS.
The isolation window for MS2 was set to 2 m/z, and the normalized
collision energy (stepped) was set as 20, 30, and 40 for fragmentation.
Data preprocessing and filtering. The raw MS data were processed using
MS-DIAL for peak alignment, retention time correction, and peak area
extraction. The metabolites were identified by accurate mass (mass
tolerance <10 ppm) and MS/MS data (mass tolerance <0.02 Da), which were
matched with HMDB, MassBank, and other public databases and our
self-built metabolite standard library. In the extracted-ion features,
only the variables having more than 50% of the nonzero measurement
values in at least one group were kept. KEGG enrichment analysis. To
identify the perturbed biological pathways, the differential metabolite
data were subjected to KEGG pathway analysis using the KEGG database
([248]http://www.kegg.jp). KEGG enrichment analyses were conducted with
Fisher’s exact test, and FDR correction for multiple tests was
performed. Enriched KEGG pathways were nominally statistically
significant at the p < 0.05 level.
Sequencing of fecal samples
Total microbial genomic DNA was extracted from mouse fecal samples
using the FastPure Stool DNA Isolation Kit (MJYH, Shanghai, China)
according to manufacturer’s instructions. The quality and concentration
of DNA were determined by 1.0% agarose gel electrophoresis and a
NanoDrop® ND-2000 spectrophotometer (Thermo Scientific Inc., USA) and
kept at −80 °C prior to further use. The hypervariable region V3-V4 of
the bacterial 16S rRNA gene were amplified with primer pairs 338 F
(5’-ACTCCTACGGGAGGCAGCAG-3’) and 806 R (5’-GGACTACHVGGGTWTCTAA T-3’) by
a T100 Thermal Cycler (BIO-RAD, USA). The PCR reaction mixture
including 4 μL 5 × Fast Pfu buffer, 2 μL 2.5 mM dNTPs, 0.8 μL each
primer (5 μM), 0.4 μL Fast Pfu polymerase, 10 ng of template DNA, and
ddH2O to a final volume of 20 µL. PCR amplification cycling conditions
were as follows: initial denaturation at 95 °C for 3 min, followed by
27 cycles of denaturing at 95 °C for 30 s, annealing at 55 °C for 30 s,
and extension at 72 °C for 45 s, and a single extension at 72 °C for
10 min, and end at 4 °C. All samples were amplified in triplicate. The
PCR product was extracted from 2% agarose gel and purified. Then
quantified using Synergy HTX (Biotek, USA). Purified amplicons were
pooled in equimolar amounts and paired-end sequenced on an Illumina
NextSeq 2000 PE300 platform (Illumina, San Diego, USA) according to the
standard protocols by Majorbio Bio-Pharm Technology Co. Ltd. (Shanghai,
China).
Sequencing for metatranscriptomics
Total RNA was extracted from the tissue using TRIzol® Reagent according
to the manufacturer’s instructions. Then RNA quality was determined by
the 5300 Bioanalyser (Agilent) and quantified using the ND-2000
(NanoDrop Technologies). Only high-quality RNA sample
(OD260/280 = 1.8 ~ 2.2, OD260/230 ≥ 2.0, RQN ≥ 6.5, 28S:18S ≥ 1.0,
>1 μg) was used to construct sequencing library. RNA purification,
reverse transcription, library construction, and sequencing were
performed at Shanghai Majorbio Bio-pharm Biotechnology Co., Ltd.
(Shanghai, China) according to the manufacturer’s instructions. The
RNA-seq transcriptome library was prepared following Illumina® Stranded
mRNA Prep, Ligation (San Diego, CA) using 1 μg of total RNA. Shortly,
messenger RNA was isolated according to the polyA selection method by
oligo (dT) beads, and then fragmented by the fragmentation buffer
first. Secondly, double-stranded cDNA was synthesized using a
SuperScript double-stranded cDNA synthesis kit (Invitrogen, CA) with
random hexamer primers. Then the synthesized cDNA was subjected to
end-repair, phosphorylation, and adapter addition according to library
construction protocol. Libraries were size-selected for cDNA target
fragments of 300 bp on 2% Low Range Ultra Agarose, followed by PCR
amplification using Phusion DNA polymerase (NEB) for 15 PCR cycles.
After quantification by Qubit 4.0, the sequencing library was performed
on the NovaSeq X Plus platform (PE150) using the NovaSeq Reagent Kit.
The raw paired-end reads were trimmed and quality controlled by fastp
with default parameters. Then clean reads were separately aligned to
the reference genome with orientation mode using HISAT2 software. The
mapped reads of each sample were assembled by StringTie in a
reference-based approach. To identify DEGs (differential expression
genes) between two different samples, the expression level of each
transcript was calculated according to the transcripts per million
reads (TPM) method. RSEM was used to quantify gene abundances.
Essentially, differential expression analysis was performed using the
DESeq2 or DEGseq. DEGs with |log2FC | ≧1 and FDR <0.05(DESeq2) or FDR <
0.001(DEGseq) were considered to be significantly different expressed
genes. In addition, functional-enrichment analysis, including GO and
KEGG were performed to identify which DEGs were significantly enriched
in GO terms and metabolic pathways at a Bonferroni-corrected P-value <
0.05 compared with the whole-transcriptome background. GO functional
enrichment and KEGG pathway analysis were carried out by Goatools and
Python scipy software, respectively. All the alternative splice events
that occurred in our sample were identified by using the recently
releases program rMATS. Only the isoforms that were similar to the
reference or comprised novel splice junctions were considered, and the
splicing differences were detected as exon inclusion, exclusion,
alternative 5′, 3′, and intron retention events.
Data and statistical analysis
Analyses were performed using GraphPad Prism (v 8.3.0). AUC of
bacterial recognition of mAbs was calculated using R function
smplot2::sm_auc_all (v 0.1.0)82 8 9, and the background AUC, which was
calculated as the mean of the three lowest values in each assay, was
subtracted. R function sva::ComBat (v 3.48.0)83 was applied to remove
the batch effect of experiments14. Heatmaps were plotted using R
package 12 pheatmap (v 1.0.12) or using Prism GraphPad v9. The number
of donors and experiments and methods of tests were indicated in each
Fig. legend. Statistical significance of difference between groups were
tested using Student’s t test or Mann-Whitney test according to test
requirements. Error bars represent SD and P values < 0.05, **P < 0.01).
Statistics and reproducibility
Sample sizes were determined by the availability of specimens. All
statistical analyses were conducted using GraphPad Prism 9.5.0. Sample
sizes (n) are specified in each figure legend and represent biological
replicates unless otherwise indicated. We applied a two-sided
Mann-Whitney U test. Statistical significance was defined as follows:
P < 0.05 (*), P < 0.01 (**), P < 0.001 (***), P < 0.0001 (****); ns
denotes not significant. For data visualization, box-and-whisker plots
display the median as a center line, with the box spanning the 25th to
75th percentiles and whiskers extending to the 5th and 95th
percentiles. Heatmaps present z-scored values unless otherwise stated,
and volcano plots illustrate log₂ fold-change against −log₁₀ P-value.
Biological and technical replicates are explicitly distinguished in the
figure legends and throughout the manuscript to ensure clarity and
reproducibility. No statistical method was used to predetermine sample
size. No data were excluded from the analyses. The experiments were not
randomized. The Investigators were not blinded to allocation during
experiments and outcome assessment.
Reporting summary
Further information on research design is available in the [249]Nature
Portfolio Reporting Summary linked to this article.
Supplementary information
[250]Supplementary Information^ (1.7MB, pdf)
[251]Reporting Summary^ (188.6KB, pdf)
[252]Transparent Peer Review file^ (1.7MB, pdf)
Acknowledgements