Glycan-based shaping of the microbiota during primate evolution

Genes encoding glycosyltransferases can be under relatively high selection pressure, likely due to the involvement of the glycans synthesized in host-microbe interactions. Here, we used mice as an experimental model system to investigate whether loss of α−1,3-galactosyltransferase gene (GGTA1) function and Galα1-3Galβ1-4GlcNAcβ1-R (αGal) glycan expression affects host-microbiota interactions, as might have occurred during primate evolution. We found that Ggta1 deletion shaped the composition of the gut microbiota. This occurred via an immunoglobulin (Ig)-dependent mechanism, associated with targeting of αGal-expressing bacteria by IgA. Systemic infection with an Ig-shaped microbiota inoculum elicited a less severe form of sepsis compared to infection with non-Ig-shaped microbiota. This suggests that in the absence of host αGal, antibodies can shape the microbiota towards lower pathogenicity. Given the fitness cost imposed by bacterial sepsis, we infer that the observed reduction in microbiota pathogenicity upon Ggta1 deletion in mice may have contributed to increase the frequency of GGTA1 loss-of-function mutations in ancestral primates that gave rise to humans.


Introduction
As originally proposed by J.B.S. Haldane, infectious diseases are a major driving force of natural selection (Haldane, 1949), occasionally precipitating 'catastrophic-selection' events: the replacement of an entire, susceptible, parental population by mutant offspring that are resistant to a given infectious disease (Lewis, 1962). Such an event is proposed to have occurred during primate evolution between 20-30 million-years-ago, possibly due to the selective pressure exerted by an airborne, enveloped virus carrying Gala1-3Galb1-4GlcNAc (aGal)-like glycans (Galili, 2016;Galili, 2019). If proven correct, this would contribute to the evolutionary pressure that led to the selection and fixation of GGTA1 loss-of-function mutations in ancestral primates (Galili et al., 1988).
We recently uncovered a possible fitness advantage associated with loss of GGTA1 function, which acts independently of aGal-specific immunity (Singh et al., 2021). Namely, loss of aGal from immunoglobulin (Ig)G-associated glycan structures increased IgG effector function and resistance to bacterial sepsis in mice (Singh et al., 2021).
Sepsis is a life-threatening organ dysfunction caused by a deregulated response to infection (Singer et al., 2016) that accounts for 20% of global human mortality (Rudd et al., 2020). The pathogenesis of sepsis is modulated by stable host symbiotic associations with microbial communities composed of bacteria, fungi, and viruses, known as the microbiota (Rudd et al., 2020;Vincent et al., 2009). While host-microbiota interactions provide a broad range of fitness advantages to the host (Lane-Petter, 1962;Vonaesch et al., 2018), these also carry fitness costs, for example, when bacterial pathobionts (Chow et al., 2011) translocate across host epithelial barriers to elicit the development of sepsis (Rudd et al., 2020;Vincent et al., 2009). On the basis of this evolutionary trade-off (Stearns and Medzhitov, 2015), it has been argued that the immune system may have emerged, in part, to mitigate the pathogenic effects of the microbiota (Hooper et al., 2012;McFall-Ngai, 2007). Central to this host defence strategy is the transepithelial secretion of copious amounts of IgA natural antibodies (NAb), which target immunogenic bacteria in the microbiota (Macpherson et al., 2000).
IgA recognize a broad but defined subset of immunogenic bacteria in the gut microbiota (Bunker et al., 2017;Bunker et al., 2015;Macpherson et al., 2000), exerting negative or positive selection pressure on these bacteria, shaping the microbiota composition, ecology, and potentially its pathogenicity (Kubinak and Round, 2016). Negative selection can occur, for example, when IgA limits bacterial growth (Moor et al., 2017), while positive selection can occur, for example, when IgA promotes bacterial interactions with the host, favoring bacterial retention, fitness, and colonization (Donaldson et al., 2018;McLoughlin et al., 2016). Moreover, IgA can interfere with cognate interactions between bacteria and tissue resident immune cells at epithelial barriers, regulating microbiota-specific immune responses, including the production of circulating IgM and IgG NAb (Kamada et al., 2015;Zeng et al., 2016).
Here, we provide experimental evidence in mice to suggest that the fixation of GGTA1 loss-offunction mutations during primate evolution exerted a major impact on the composition of their gut microbiota. In support of this notion, mice in which Ggta1 is disrupted (Ggta1 -/-), mimicking human GGTA1 loss-of-function mutations, modulated their gut microbiota composition. This occurs predominantly via an Ig-dependent mechanism, associated with an enhancement of the production of IgA, targeting aGal-expressing bacteria in the gut microbiota. The pathogenicity of the Ig-shaped microbiota is reduced, failing to elicit lethal forms of sepsis upon systemic infection. We propose that GGTA1 loss-of-function mutations conferred a selective benefit during primate evolution, in part, by shaping commensal bacteria in the microbiota to mitigate the pathogenesis of sepsis.

Ggta1 deletion shapes the microbiota composition
We have previously established that Ggta1 -/mice harbor a distinct microbiota composition to that of wild type (Ggta1 +/+ ) mice (Singh et al., 2021). This is illustrated by the relative abundance of specific bacterial taxa, such as an increase in Proteobacteria, Tenericutes, and Verrucomicrobia as well as a reduction in Bacterioidetes and Deferribacteres phyla in Ggta1 -/mice compared to Ggta1 +/+ mice ( Figure 1A, Figure 1-figure supplements 1 and 2; Singh et al., 2021). The relative increase of Proteobacteria, a phylum containing several strains associated with pathogenic behavior, in the gut microbiota of Ggta1 -/mice was not, however, associated with the development of histological lesions in the gastrointestinal tract (Figure 1-figure supplement 3A). Absence of intestinal inflammation was further assessed by quantification of fecal lipocalin-2 (Lcn-2) (Figure 1-figure supplement 3B; Chassaing et al., 2012). There were also no histopathological lesions in the liver, lungs, kidney, and spleen ( Figure 1-figure supplement 3C), suggesting that Ggta1 -/mice maintain symbiotic interactions with these pathobionts, without compromising organismal homeostasis.
To establish whether the differences in the bacterial species present in the gut microbiota of Ggta1 -/vs. Ggta1 +/+ mice are propelled by host genetics, we used an experimental system whereby the microbiota is vertically transmitted over several generations (Ubeda et al., 2012) from Ggta1 +/+ mice to Ggta1 -/and Ggta1 +/+ offspring ( Figure 1B). This approach enables effects exerted by the host genotype on microbiota composition to predominate over those exerted by environmental factors (Gálvez et al., 2017;Vonaesch et al., 2018), diet (Sonnenburg et al., 2016), cohousing or familial transmission (Ubeda et al., 2012), albeit not accounting for putative cage effects or genetic drift (Spor et al., 2011).
Of note, the levels of circulating IgA were reduced in Tcrb -/-Ggta1 -/mice lacking a/b T cells, when compared to Ggta1 -/mice ( Figure 2-figure supplement 1D), suggesting that the production of circulating IgA NAb in Ggta1 -/mice occurs, in part, via a T-cell-dependent mechanism, which is consistent with previous reports in Ggta1 +/+ mice (Bunker et al., 2017;Fagarasan et al., 2010;Macpherson et al., 2000).
We then asked whether Ggta1 -/mice shape their microbiota via a mechanism associated with immune targeting of aGal-expressing bacteria. Consistent with a number of bacteria in the human gut microbiota carrying genes orthologous to the mammalian a1,3-galactosyltransferase Nearly 30% of these aGal + bacteria were immunogenic ( Figure 2D,E), as defined by the detection of surface-bound IgA (Palm et al., 2014), which predominantly targets bacteria in the small intestine (Bunker et al., 2017;Bunker et al., 2015). These IgA + aGal + bacteria accounted for roughly 50% of all the immunogenic (IgA + ) bacteria in the small intestine ( Figure 2D,E). These were also present, although at a lower extent, in the cecum, colon, and feces (Figure 2-figure supplement 1E and F).
Ggta1 -/mice harbored a relatively lower percentage of immunogenic IgA + aGal + bacteria in the small intestine, when compared to Ggta1 +/+ mice ( Figure 2D,E), while the percentage of immunogenic IgA + bacteria was similar in Ggta1 -/vs. Ggta1 +/+ mice ( Figure 2D,E). This is consistent with the idea of a specific mechanism altering the microbiota of Ggta1 -/mice that presumably, at least in part, involves targeting immunogenic aGal + bacteria by IgA. Whether this mechanism involves the recognition of bacterial aGal-like glycans by IgA is not clear.
We then compared the effect of IgA on the levels of systemic IgM and/or IgG NAb directed against antigens expressed by bacteria present in the microbiota (Kamada et al., 2015;Zeng et al., 2016). Induction of dysbiosis, by streptomycin, increased the levels of circulating aGal-specific IgM and IgG in Igha -/-Ggta1 -/vs. Igha +/+ Ggta1 -/mice ( Figure 2F,G). This illustrates again that the IgA response of Ggta1 -/mice is distinct from that of Ggta1 +/+ mice, reducing systemic IgM and IgG responses against antigens expressed by bacteria in the microbiota, as illustrated for aGal-like glycans. Presumably this occurs via a mechanism whereby IgA prevents aGal + bacteria or bacterial products associated to aGal from translocating across epithelial barriers and inducing systemic immune responses against this glycan.
To dissect the relative contribution of the Ggta1 from the Ig genotype in shaping the microbiota, F 2 littermates were interbred to generate F 3 offspring carrying a microbiota targeted by antibodies (Igh-J +/+ Ggta1 +/+ and Igh-J +/+ Ggta1 -/-) or not (Igh-J -/-Ggta1 +/+ and Igh-J -/-Ggta1 -/-) ( Figure 3A). Consistent with our previous observations (Figure 1), there was a marked separation of the microbiota community structure between F 3 Igh-J +/+ Ggta1 -/vs. Igh-J +/+ Ggta1 +/+ mice, as assessed by Principal Coordinate Analyses (PCA) for Weighted and Unweighted Unifrac ( Figure 3B,C). Considering that Weighted Unifrac accounts for the relative abundance of the bacterial taxa while Unweighted Unifrac accounts for only the presence or absence of the taxa in the microbial community (Lozupone and Knight, 2005), they represent differences exerted by high and low abundant bacterial taxa respectively. This suggests that Ggta1 deletion shapes both high and low abundant taxa present in the microbiota when maternal and/or offspring-derived antibodies (i.e. Ig) are present.
LefSe analysis showed that the microbiota from Igh-J +/+ Ggta1 -/mice was enriched in specific bacterial taxa, including Proteobacteria, as compared to the microbiota from Igh-J +/+ Ggta1 +/+ mice (Figure 3-figure supplement 1C). This suggests that Ggta1 deletion favors gut colonization by pathobionts in an Ig-dependent manner.
We then asked whether Ig are functionally involved in shaping the microbiota composition of Ggta1 -/vs. Ggta1 +/+ mice. In the absence of Ig (Igh-J -/-), there were no differences in the microbiota composition of F 3 Igh-J -/-Ggta1 -/vs. Igh-J -/-Ggta1 +/+ mice, as assessed by PCA for Weighted Unifrac ( Figure 3B). This suggests that shaping of highly abundant taxa in the microbiota of Ggta1 -/vs. Ggta1 +/+ mice occurs via an Ig-dependent mechanism. In contrast, the microbiota composition of F 3 Igh-J -/-Ggta1 -/mice remained distinct from that of Igh-J -/-Ggta1 +/+ mice, as assessed by PCA for Unweighted Unifrac ( Figure 3C). This suggests that shaping of low abundance bacterial taxa in the microbiota of Ggta1 -/vs. Ggta1 +/+ mice occurs in an Ig-independent manner.
We then asked whether Ig exert a higher impact on the relative abundance of pathobionts in the gut microbiota of Ggta1 -/vs. Ggta1 +/+ mice. In strong support of this notion, the gut microbiota from Igh-J -/-Ggta1 -/mice, lacking Ig, was enriched with Helicobactereaceae family from Proteobacteria phylum as compared to Igh-J +/+ Ggta1 -/mice expressing Ig ( Figure 3E,F). This is consistent with our previous finding that the gut microbiota of Rag2 -/-Ggta1 -/mice, lacking adaptive immunity,  is highly enriched in Proteobacteria, including Helicobacter (Singh et al., 2021). Of note, this was not observed in Igh-J -/-Ggta1 +/+ vs. Igh-J +/+ Ggta1 +/+ mice (Figure 3-figure supplement 1D). These data suggest that the absence of host aGal favors colonization of the gut microbiota by pathobionts, the expansion of which is restrained by Ig.

Ggta1 deletion reduces microbiota pathogenicity
We then asked whether the Ig-dependent shaping of the microbiota in Ggta1 -/mice affects the pathogenesis of sepsis due to systemic infections emanating from gut microbes. Ggta1 -/mice were protected against systemic infections (i.p.) by a cecal inoculum isolated from Rag2 -/-Ggta1 -/mice, reflecting a microbiota not shaped by Ig (Figure 4-figure supplement 1A and B). This is consistent with our previous finding showing Ggta1 deletion enhances protection against systemic bacterial infections (i.p.) via a mechanism involving IgG NAb (Singh et al., 2021). Moreover, Ggta1 -/mice were also protected against a systemic infection (i.p.) by a cecal inoculum isolated from Ggta1 -/mice, reflecting their own Ig-shaped microbiota (Figure 4-figure supplement 1A and B). This is in keeping with the previously shown enhanced protection of Ggta1 -/mice against cecal ligation and puncture (Singh et al., 2021). Surprisingly Igh-J -/-Ggta1 -/mice lacking B cells (Figure 4-figure supplement 1C), Tcrb -/-Ggta1 -/mice lacking a/b T cells (Figure 4-figure supplement 1D) and Rag2 -/-Ggta1 -/mice lacking B and T cells (Figure 4-figure supplement 1E) remained protected against systemic infections by the cecal inoculum isolated from Ggta1 -/mice. This suggests that the previously described IgG-dependent mechanism that protects Ggta1 -/mice from a systemic infection by a cecal inoculum isolated from Rag2 -/-Ggta1 -/mice (Singh et al., 2021) is distinct from that protecting Ggta1 -/mice from a systemic infection by their own cecal inoculum. We reasoned that this might be explained by a reduction of the overall pathogenicity of the Ig-shaped microbiota of Ggta1 -/mice, compared to the non-Ig-shaped microbiota of Rag2 -/-Ggta1 -/mice. To test this hypothesis, we compared the outcome of Rag2 -/-Ggta1 -/mice upon a systemic infection by Igshaped vs. a non-Ig-shaped microbiota.
We then asked whether the reduction of microbiota pathogenicity imposed by the adaptive immune system of Ggta1 -/mice is also operational in Ggta1 +/+ mice. However, Rag2 -/-Ggta1 +/+ mice succumbed to the same extent to systemic infection (i.p.) with a cecal inoculum isolated from either Rag2 +/+ Ggta1 +/+ vs. Rag2 -/-Ggta1 +/+ mice ( Figure 4D-E), developing similar bacterial loads ( Figure 4F). This suggests that the mechanism via which the adaptive immune system of Ggta1 -/mice shapes and reduces the pathogenicity of its microbiota is not operational in Ggta1 +/+ mice.
Having established that in the absence of adaptive immunity, Ggta1 -/mice are resistant against systemic infection by a low pathogenic Ig-shaped microbiota (Figure 4-figure supplement 1C and E), we asked whether the mechanism of resistance relies on the innate immune system. Depletion of Ly6C + /Ly6G + myeloid cells (i.e. polymorphonuclear cells and inflammatory monocytes) using an anti-GR1 monoclonal Ab (Figure 4-figure supplement 1F; Daley et al., 2008) increased the susceptibility of Ggta1 -/mice to systemic infection by their cecal inoculum ( Figure 4G). This was associated with a 10 2 -10 5 -fold increase in bacterial load, compared to control Ggta1 -/mice ( Figure 4H). Of note, monocyte/macrophage depletion by Clodronate liposomes (Figure 4 (Figure 4-figure supplement 1H). This suggests that polymorphonuclear cells are essential for resistance against systemic infection emanating from the less pathogenic Igshaped microbiota of Ggta1 -/mice.

Discussion
While loss-of-function mutations in genes encoding glycosyltransferases can provide fitness advantages against infection, these can compromise the physiologic functions of the eliminated self-glycan, as illustrated by the occurrence of reproductive senescence upon Ggta1 deletion in mice (Singh et al., 2021). In an evolutionary context, such a trade-off could explain why loss-of-function mutations in these genes are rare, and in some cases unique, to the human lineage. The latter is illustrated by the loss of CMP-N-acetylneuraminic acid hydroxylase (CMAH) function, which eliminated expression of the sialic acid, N-glycolylneuraminic acid (Neu5Gc), in humans (Ghaderi et al., 2011).
Co-evolution of ancestral hominids with commensal bacteria in their microbiota (Dethlefsen et al., 2007;Huttenhower et al., 2012;Moeller et al., 2016), is thought to have provided a series of fitness advantages including, among others, optimization of nutrient intake from diet, regulation of different aspects of organismal metabolism or colonization resistance against pathogenic bacteria (Buffie and Pamer, 2013). Here we propose that the loss of aGal expression, as it occurred during primate evolution, might have exerted a major impact on the nature of these symbiotic associations. In support of this notion, we found that Ggta1 deletion in mice was associated with major changes in the composition of the gut microbiota ( Figure 1). This occurred over several generations under experimental conditions of exposure to environmental-derived pathobionts, and minimum relative impact exerted by other environmental factors on microbiota composition (Ubeda et al., 2012), arguing that the Ggta1 genotype modulates the microbiota composition. On the basis of this observation alone (Figure 1), one cannot exclude the observed divergence in the microbiota bacterial population frequencies harbored by wild type vs. Ggta1-deleted mice (Figure 1) from being a stochastic event. However, the observation that these changes occur via an Ig-dependent mechanism that differs in wild type vs. Ggta1-deleted mice (Figure 3) does suggest that loss of aGal contributes critically to shape the microbiota composition of Ggta1-deleted mice.
While our studies do not provide direct evidence that the loss of aGal played a major role in shaping the microbiota composition of Old-vs. New World primates, this notion is supported, indirectly, by the recent finding that mutations altering the expression of human ABO blood group glycans are associated with shaping of the bacterial composition of the gut microbiota (Rühlemann et al., 2021). In light of the structural similarity between the human B blood group (i. e. Gala1-3(Fuca1-2)Galb1-4GlcNAcb1-R), and aGal (i.e. Gala1-3Galb1-4GlcNAcb1-R) glycans, our findings probably reflect a general mechanism whereby host 'aGal-like' glycans shape the microbiota composition of mice and primates, including that of humans. Whether or not this is the case, considering that microbiota composition is shaped mostly by environmental factors, remains to be formally established.
The mechanism via which Ggta1 deletion shapes the gut microbiota is associated with targeting of aGal-expressing bacteria by IgA (Figure 2). Consistent with this notion, a relatively large proportion of probiotic bacteria as well as bacteria present in the mouse microbiota express aGal-like glycans at the cell surface ( Figure 2). About 30% of these carry IgA at the cell surface in the mouse microbiota and, as such, are considered as immunogenic (Palm et al., 2014). The proportion of IgA + aGal + bacteria was reduced in the microbiota of Ggta1-deleted mice and was presumably eliminated ( Figure 2). This suggests that Ggta1 deletion probably broadens bacterial recognition by IgA to include immunogenic aGal + bacteria, which as a result are probably purged from the microbiota. Whether this occurs via a mechanism involving the recognition of aGal-like glycans, and/or other related epitopes expressed at the surface of these bacteria, by IgA, has not been established. Of note, these are not mutually exclusive possibilities since: (i) IgA can target aGal-like glycans and modulate bacterial pathogenicity (Hamadeh et al., 1995), (ii) IgA are poly-reactive and can target common antigens expressed by bacteria (Bunker et al., 2017;Bunker et al., 2015) and (iii) aGalreactive antibodies present a degree of poly-reactivity against non-aGal related bacterial epitopes (Bernth Jensen et al., 2021). The identity of the aGal + bacteria targeted by IgA remains elusive but likely includes Gram-negative pathobionts from the Enterobacteriaceae family, as demonstrated for Escherichia (E.) coli O86: B7 (Yilmaz et al., 2014), which expresses the aGal-like glycan Gala1-3Gal(Fuca1-2)b1-3GlcNAcb1-4Glc as part of the lipopolysaccharide (LPS) O-antigen (Guo et al., 2005). Of note, this pathobiont can induce a systemic aGal-specific NAb response in humans (Springer and Horton, 1969) as well as in Ggta1-deleted mice, which is protective against infection by pathogens expressing aGal-like glycans (Yilmaz et al., 2014). The finding that several commensal bacteria in the human gut microbiota express aGal-like glycans (Figure 2-figure supplement 2) suggests that other bacteria might contribute to this protective response.
Our findings also suggest that Ggta1 deletion shapes the bacterial community structure among highly abundant bacterial taxa in the microbiota via an Ig-dependent mechanism, and among low abundant bacterial taxa independently of Ig (Figure 3). Presumably, shaping of highly abundant taxa by Ig occurs via a mechanism that involves not only IgA produced by the offspring, but also maternal IgG transferred through the placenta as well as IgM, IgG, and IgA transferred through maternal milk (Koch et al., 2016). These regulate neonatal innate (Gomez de Agüero et al., 2016) and adaptive (Koch et al., 2016) immunity, shaping the offspring microbiota composition (Gensollen et al., 2016;Koch et al., 2016). Whether this occurs via targeting of aGal-like glycans, as discussed above, and/ or via other bacterial antigens expressed by immunogenic bacteria has not been established.
Shaping of lowly abundant bacterial taxa independently of Ig ( Figure 3) suggests that other mechanisms contribute to shaping the microbiota of Ggta1-deleted mice. These probably include the modulation of nutritional or spatial niches due to the loss of aGal from complex glycosylated structures present at epithelial barriers, such as the mucus, as demonstrated for other glycans (Pickard et al., 2014).
The selective pressure exerted by the adaptive immune system of Ggta1 -/mice on the bacterial species present in the microbiota, probably allows for the establishment of a more diverse ecosystem containing pathobionts (Singh et al., 2021), such as Helicobactereaceae (Figure 1-figure supplements 1 and 2). These can elicit the production of antibodies targeting and exerting negative selective pressure on other bacterial symbionts and releasing ecological niches, thus further shaping the microbiota. Expansion of these pathobionts in the microbiota of Ggta1 -/mice is restrained by Ig, which probably explains the lack of associated pathology (Figure 1-figure supplement 3). This suggests that the loss of GGTA1 function in ancestral primates fostered mutualistic interactions with more diverse bacterial ecosystems, incorporating pathobionts such as Helicobacter pylori, associated with fitness costs (Atherton and Blaser, 2009) and gains (Linz et al., 2007).
Our findings suggest that Ig-shaping of the bacterial species present in the gut microbiota of Ggta1 -/mice reduces the microbiota pathogenicity, as illustrated when comparing the lethal outcome of systemic infections using the Ig-shaped vs. non-shaped microbiota ( Figure 4A, Figure 4figure supplement 1A and B). This reduction in pathogenicity means that the effector mechanism underlying resistance against systemic infection by the Ig-shaped microbiota no longer relies on the adaptive immune system (Figure 4-figure supplement 1C and E), but instead on cellular components of the innate immune systems, namely, neutrophils ( Figure 4G,H).
We propose that Ggta1 deletion in mice increases resistance to bacterial sepsis via two distinct antibody-dependent mechanisms. The first involves a relative increase in antibody effector function, when aGal is not present in the biantennary glycan structures of IgG (Singh et al., 2021). The second relies on shaping and reduction of the microbiota pathogenicity by antibodies, most likely from the IgA isotype. To what extent the lack of aGal from the biantennary glycan structures of Ig also contributes to shape and reduce the microbiota pathogenicity, remains to be established. It is possible however, that similar to IgG (Singh et al., 2021), the absence of aGal from the glycan structures of different Ig isotypes, including IgA, modulates their effector function, when targeting immunogenic bacteria in the microbiota.
The notion of host mechanisms shaping the microbiota composition toward a reduction of its pathogenicity is in keeping with host microbial interactions not being hardwired, but instead shifting between symbiotic to pathogenic depending on host and microbe cooperative behaviors (Ayres, 2016;Vonaesch et al., 2018). For example, when restricted to the microbiota, bacterial pathobionts can behave as commensals, posing no pathogenic threat to the host (Vonaesch et al., 2018), while triggering sepsis (Haak and Wiersinga, 2017) upon translocation across epithelial barriers (Caruso et al., 2020). The high fitness cost imposed to modern humans by sepsis (Rudd et al., 2020) suggests that mutations shaping the composition of the microbiota toward a reduced capacity to elicit sepsis should be associated with major fitness advantages. Our findings suggest that loss-of-function mutations in GGTA1 act in such a manner.
When challenged experimentally with a bacterial inoculum or bacterial lipopolysaccharide (LPS), non-human primates appear to be far more susceptible to develop sepsis or septic shock, respectively, as compared to other mammalian lineages (Chen et al., 2019). This may appear at odds with our interpretation that similar to Ggta1 deletion in mice (Figure 4; Singh et al., 2021), the evolutionary loss of GGTA1 function in Old-World primates might have increased resistance to bacterial sepsis (Singh et al., 2021). One likely explanation for this apparent discrepancy may relate to the intrinsic properties of primate immunity, whereby natural selection of traits enhancing immunedriven resistance mechanisms (Olson, 1999;Wang et al., 2006) might be associated with lower capacity to establish disease tolerance to infection (Martins et al., 2019;Medzhitov et al., 2012). This interpretation is consistent with resistance and disease tolerance to infection being negatively correlated (Råberg et al., 2007), such that traits increasing resistance might be associated with a reduction in disease tolerance, as an evolutionary trade-off.
In conclusion, protective immunity against aGal-expressing pathogens was likely a major driving force in the natural selective events that led to the fixation of loss-of-function mutations in the GGTA1 gene of ancestral Old World primates (Galili, 2016;Soares and Yilmaz, 2016). Moreover, in the absence of aGal, the glycan structures associated with the Fc portion of IgG, can increase IgGeffector function and resistance to bacterial infections, irrespectively of aGal-specific immunity (Singh et al., 2021). The net survival advantage against infection provided by these traits came alongside the emergence of reproductive senescence (Singh et al., 2021), a major fitness cost presumably outweighed by endemic exposure to highly virulent pathogens (Singh et al., 2021). Here we provide experimental evidence for yet another fitness advantage against infection, associated with the loss of GGTA1, driven by Ig-shaping and reduction of microbiota pathogenicity. We infer that ancestral Old World primates carrying loss-of-function mutations in GGTA1 probably shaped their microbiota to minimize its pathogenic effect, providing a major fitness advantage against sepsis.

Breeding experiments
Vertical transmission of the microbiota from Ggta1 +/+ mice to Ggta1 -/and Ggta1 +/+ offspring over several generations was achieved, as described (Ubeda et al., 2012;Singh et al., 2021). Briefly, two or more breeding groups were established, consisting of two Ggta1 +/+ females and one Ggta1 -/male per cage. The male was removed after one week and the females were placed in a clean cage until delivery. F 1 Ggta1 +/mice were weaned at 3-4 weeks of age and then co-housed until 8 weeks of age. Two or more F 1 Ggta1 +/groups were established randomly using one male and two females per cage. F 2 pups were weaned at 3-4 weeks of age, genotyped and segregated according to their Ggta1 -/vs. Ggta1 +/+ genotype in separate cages until adulthood. F 3 to F 5 pups were generated in a similar manner. Fecal pellets from two to three cages per genotype were collected (10-12 weeks of age) for microbiota analysis.
The effect of Ggta1 genotype and Ig on microbiota composition derived from Ggta1 +/+ mice was achieved essentially as described (Singh et al., 2021). Briefly, two or more breeding groups were established, consisting of two Igh-J +/+ Ggta1 +/+ females and one Igh-J -/-Ggta1 -/male per cage. The male was removed after 1 week and the females were placed in a clean cage until delivery. F 1 Igh-J -+/-Ggta1 +/pups were kept with mothers until weaning at 3-4 weeks of age and co-housed until 8 weeks of age. Two or more F 1 Igh-J +/-Ggta1 +/breeding groups were established randomly using one male and two females per cage. Littermate F 2 pups were weaned at 3-4 weeks of age, genotyped and segregated according to their Igh-J +/+ Ggta1 -/-, Igh-J -/-Ggta1 -/-, Igh-J +/+ Ggta1 +/+ and Igh-J -/-Ggta1 +/+ genotypes in separate cages until adulthood. F 3 pups were generated in a similar manner. Fecal pellets from two to three cages per genotype were collected (10-12 weeks of age) for microbiota analysis.

Cecal slurry injection
Cecal slurry injection was performed essentially as described (Singh et al., 2021). Microbiota pathogenicity experiments were performed by preparing cecal slurry from Ggta1 -/vs. Rag2 -/-Ggta1 -/mice or from Ggta1 +/+ vs. Rag2 -/-Ggta1 +/+ mice and injecting to recipient mice (i.p. 1 mg/g body weight) in parallel. Mice were monitored every 12 hr for survival for 14 days or euthanized at various time points for analysis of different parameters.

Pathogen load
Quantification of pathogen load in the mice was performed 24 hr after cecal slurry injection, essentially as described (Singh et al., 2021).

Extraction of bacterial DNA from feces
Bacterial DNA was extracted from fecal pellets (QIAamp Fast DNA Stool Mini Kit #50951604) as described (Singh et al., 2021).

Statistical analysis
Statistical tests were performed using GraphPad Prism Software (v.6.0). All statistical details, including statistical tests, exact value of n, what n represents, definition of center, dispersion and precision measures are provided in each figure legend. The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.