From pathogen to a commensal: modification of the Microbacterium nematophilum-C. elegans interaction during chronic infection by the absence of host insulin signalling.

The nematode worm Caenorhabditis elegans depends on microbes in decaying vegetation as its food source. To survive in an environment rich in opportunistic pathogens, C. elegans has evolved an epithelial defence system where surface-exposed tissues such as epidermis, pharynx, intestine, vulva and hindgut have the capacity of eliciting appropriate immune defences to acute gut infection. However, it is unclear how the worm responds to chronic intestinal infections. To this end, we have surveyed C. elegans mutants that are involved in inflammation, immunity and longevity to find their phenotypes during chronic infection. Worms that grew in a monoculture of the natural pathogen Microbacterium nematophilum (CBX102 strain) had a reduced lifespan and vigour. This was independent of intestinal colonisation as both CBX102 and the derived avirulent strain UV336 were early persistent colonisers. In contrast, the long-lived daf-2 mutant was resistant to chronic infection, showing reduced colonisation and higher vigour . In fact, UV336 interaction with daf-2, resulted in a host lifespan extension beyond OP50, the E. coli strain used for laboratory C. elegans culture. Longevity and vigour of daf-2 mutants growing on CBX102 was dependent on the FOXO orthologue DAF-16. Our results indicate that the interaction between host genotype and strain-specific bacteria determines longevity and health for C. elegans .


INTRODUCTION
Bacteria associated with the animal gut are important for gastrointestinal function (Fischbach, 2018). Intestinal bacteria contribute to metabolic activities and are involved in the absorption of nutrients, protection of mucosal surfaces and the regulation of the immune function of the gut (Fischbach, 2018). Quantitative and/or qualitative alteration of the intestinal microbiota underline many inflammatory diseases as well as chronic gastrointestinal infections (CGIs), the latter being amongst the most common chronic diseases worldwide (Drossman et al, 2016). In the short term, CGIs can lead to altered mucosal and immune function (Drossman et al, 2016). In the longer term, CGIs cause impaired epithelial barrier function (a major factor of reduced health span in old age) and changes in intestinal microbiota (dysbiosis) that can "drive" constitutive inflammation in conditions like intestinal bowel disease and enterocolitis (Caravan et al, 2014). Moreover, sustained inflammation can lead to intestinal cancer or may accelerate age-dependent neurodegeneration (Caravan et al, 2014). In this context, understanding how host genetics interacts with the intestinal microbiota in health and disease is an important aspect in managing long-term health span. However, it is also a complex problem with many biological parameters.
Accumulating evidence indicates that the health-disease balance in CGIs is determined by the interaction of four components (Stecher et al, 2015). These are 1) the infectious agent inducing the disease, 2) host genetics that will influence mucosal barrier function and pro or anti-inflammatory responses, 3) the intestinal microbiota that can drive the disease when its composition change and 4) diet, which influences all other components. Negative interaction of these factors can abolish normal intestinal barrier function leading to constant mucosal inflammation and reduced health and life expectancy (Finch, 2010). In contrast, non-inflammatory management can lead to extension of lifespan and healthspan (Hooper and Gordon 2001). It is evident that interactions of the above 4 components generate a complex set of conditions, which makes it hard to untangle the layers of chronic disease and arrive at causality. In the simplified system that the nematode worm, Caenorhabditis elegans (C. elegans), is used in research laboratories around the world, the animal develops, feeds and ages in a bacterial monoculture. This means that food=microbiota=pathogen (or commensal) depending on the choice of bacterium.
This condition ensures the ability to modify host genetics in vivo by keeping all other parameters important for CGIs in control. When the pathogen changes, so will the function of diet and microbiota, and thus the system enables in principle to find the host genes that interact with a specific bacterial strain. In the wild, C. elegans is a bacterial feeder spending much of its life in decomposing vegetable matter and depends on microbes for food (Frezal and Felix 2015). These microbes are ground by the pharynx before they subsequently enter the gut. To survive in an environment rich in potentially damaging microorganisms, C. elegans has evolved an epithelial defence system coupled with the ability to discriminate between pathogenic vs. edible bacteria (reviewed in Kim and Ewbank, 2018).
Important antimicrobial molecules participating in these defences include a group of proteins called invertebrate lysozymes (ILYS) and in particular ILYS-3, which is expressed in both the pharynx and the intestine (O'Rourke et al, 2006).
ILYS-3 (invertebrate-specific but related to human epithelial antimicrobial peptides) contributes to the digestion of the large amount of peptidoglycan fragments generated by the worm's bacterial diet (either pathogenic or non-pathogenic) Biology Open • Accepted manuscript The isolation of natural bacterial pathogens of C. elegans has permitted a glimpse of the defence mechanisms employed by the worm as well as the hostpathogen interactions triggering such mechanisms (see Hodgkin et al, 2000;Nicholas and Hodgkin 2004;Hodgkin et al 2013). One such pathogen is Microbacterium nematophilum (Hodgkin et al, 2000). This Gram-positive bacterium adheres to the rectal and anal cuticle (Hodgkin et al, 2000) and induces inflammation, anal-region infection and tail swelling (Hodgkin et al, 2000;Parsons and Cipollo, 2014). Despite the fact that the most obvious response to infection is rectal colonization and the induction of inflammation in the rectal tissues, this bacterium also establishes itself in the gut of the worm. In fact, host lethality caused by M. nematophulum is due to gut infection rather than rectal inflammation (Parsons and Cipollo 2014). This makes it a good system to investigate effects that occur in the digestive tract associated with long-term gut colonization. In particular, to identify how long-term survival and health of the organism are influenced in the face of chronic intestinal infection.
To explore this question, we tested C. elegans mutants induced by chemical mutagenesis or targeted deletion in signalling pathways known to be involved in immunity to M. nematophilum infection and/or C. elegans longevity. These mutant worms were grown using solely the M. nematophilum strain CBX102 (where CBX102 is the sole source of food=microbiota=pathogen). Using CBX102, we were able to separate estimated host survival probabilities into four categories in relation to ilys-3 and wild type (N2) worms. We identified daf-2 as long-lived in conditions of chronic infection. Bacterial colonisation of CBX102 in N2 worms was increased compared to the laboratory E. coli strain OP50. However, colonisation in N2 per se was not the reason for pathogenesis as the non-virulent M. nematophilum strain UV336 did not curtail lifespan despite being able to colonise at the same levels as CBX102.
Nevertheless, daf-2 worms were healthier and had reduced colonisation compared to normal worms. daf-2 health and longevity on CBX102 involved the canonical insulin signalling pathway and were thus dependent on the FOXO orthologue daf-16, like many other daf-2-mediated effects. Finally, the non-pathogenic UV336 was able to support an extended lifespan for daf-2 beyond that observed when using OP50.
These results indicate the complex and strain-specific interactions between intestinal bacteria and host genetics.

Chronic Gastrointestinal Infection (CGI) curtails lifespan, reduces
health and accelerates ageing in N2 worms. In our experimental CGI set-up, C.  1A) and health, measured by vigour of movement in liquid assays (Fig. 1B).
The avirulent M. nematophilum UV336 strain (derived from CBX102 by UV mutagenesis, Akimkina et al, 2006), had the same level of bacterial colonisation as CBX102 ( Fig. 1C) but in contrast to the latter, presented no negative impact on median lifespan (Fig. 1A) or health span (Fig. 1B) both of which were largely comparable to OP50. In this context, two strains of the same species behaved one as a pathogen (CBX102) and one as a commensal (UV336). Moreover, CBX102 accelerated mitochondrial fragmentation (Fig S1), a sign of age-dependent stress in worms (Han et al, 2017).

CGI defines four lifespan groups of C. elegans mutants
To find worms that could outlive N2 under CGI while retaining their health, we tested C. elegans mutants induced by chemical mutagenesis, in signalling pathways known to be involved in immunity to infection and/or longevity. Our tests pertained to studying intestinal colonization, lifespan and health/vigour. All strains were cultured from eggs in pure CBX102 and tested for bacterial colonization.

daf-2 mutant is long-lived and healthier than N2 under CGI.
From the mutants tested, only one mutant, in the insulin receptor, daf-2 was found to be living longer than N2 under CGI (Fig. 2D). This confirmed and extended observations for daf-2 longevity in OP50 (Kenyon et al, 1993) as well as acute infections by S. aureus, P. aeruginosa or E. faecalis (Garsin et al, 2003) and Salmonella typhimurium (Portal-Celhay et al, 2012). Bacterial colonisation of daf-2 was reduced compared to N2 (Fig. S2). It was also reduced compared to other normally long-lived mutants such as age-1 (Fig S2). The latter is long-lived on OP50 (Friedman and Johnson, 1988) but had lifespan indistinguishable to N2 on CBX102.
Despite the adverse effects of CBX102 on N2 lifespan (when compared to OP50), N2 median lifespan on UV336 vs. OP50 was statistically indistinguishable (Fig. 3A). The survival pattern of daf-2 mutants on CBX102 was statistically comparable to that of daf-2 on E. coli OP50 (Fig. 3B). Compared to N2 on CBX102, daf-2 worms were still longer-lived (compare Fig. 3A and 3B). Notably, daf-2 lifespan was extended on UV336 compared to daf-2 on CBX102 even beyond the TD50 and maximum lifespan limits defined by OP50 (Fig 3B). This boosting effect on lifespan by UV336 over and above OP50 was not observed in N2 (Fig. 3A). This result showed that the genotype of the host can modify the effect of a bacterial strain and this interaction determines lifespan. Conversely, any effect of a bacterial species is strain specific.

Daf-16 is required for the longevity and health of daf-2 mutants under CGI
Lifespan extension through the DAF-2 insulin-signalling pathway in C. elegans occurs by de-repression of the fork-head transcription factor DAF-16, which is normally under negative regulation by DAF-2. Strong loss-of-function alleles of daf-16 such as mgDf47 and mu86 suppressed the long-lived phenotype of daf-2 under CGI with CBX102 making the double daf-16; daf-2 statistically indistinguishable from N2 (Fig 4). Moreover, loss of DAF-16 suppressed the vigorous thrashing ability of daf-2 making again the double daf-16; daf-2 statistically indistinguishable in its vigour compared to N2 ( Fig S4). As expected from the above, daf-16 on its own, exhibited a comparable degree of survival to CGI as N2 worms.
Therefore, in C. elegans, the DAF-2/DAF-16 axis is important for maintaining longevity and health under CGI by a natural pathogen.

DISCUSSION
We wanted to develop a simple model to test host longevity and health under CGI. C. elegans is such a model since microbiota=pathogen=food as the worm is a bacterial feeder and its laboratory culture is typically a bacterial mono-association.
Our work shows where longevity and immunity converge under CGI. Our data indicate that the insulin signalling pathway modulates intestinal colonisation to affect long-term host survival. How long the host will live however, is also dependent on the strain-specific pathogenicity of the bacteria on which C. elegans is feeding. The natural pathogen M. nematophilum strain CBX102 curtailed lifespan and health of N2 wild type worms but strain UV336 was statistically indistinguishable from E. coli OP50, the "normal" lab food. Inactivation of the insulin receptor in daf-2, made worms live longer and be healthier and physiologically Biology Open • Accepted manuscript younger on CBX102. This correlated with reduced colonisation (Fig S3; Fig S5). In addition, UV336 extended daf-2 lifespan even beyond what has been seen with E.
coli OP50, acting as a lifespan-extending bacterium when interacting with this host genetic background. More work is needed to identify the genetic differences between the two M. nematophilum strains and how lack of insulin host signalling modifies these bacterial strains and their properties.
The insulin pathway-mediated modification of a pathogen to a commensal (CBX102) or to a lifespan-extending bacterium (UV336) may have parallels in other model organisms. Recent evidence in mice has shown that inducing insulin resistance through dietary iron drove conversion of a pathogen to a commensal.
Specifically, insulin resistance converted the enteric pathogen Citrobacter to a commensal (Sanchez et al, 2018). There, reduced intestinal glucose absorbance was crucial for Citrobacter to be a commensal (Sanchez et al, 2018). More work is needed to determine if systemic glucose levels and/or intestinal glucose absorption play a role also in C. elegans and how this relates to the worm insulin pathway.
Reduced glucose levels increase lifespan in worms (Watts and Ristow, 2017).
Reducing glycolysis has been shown to induce mitochondrial OXPHOS to generate a lifespan-extending reactive oxygen species (ROS) signal (Schulz et al, 2007) while increased levels have the opposite effect (Schulz et al, 2007;Zarse et al, 2012). Limitations in this comparison include the fact that our one bacterium microbiota is different than the complex one in mice. Moreover, the insulin pathway in worms and mammals may have differences in biochemical terms (reviewed in Watts and Ristow, 2017).

Biology Open • Accepted manuscript
Taken together, our results and recent data from mice (Sanchez et al, 2018) show that the consequences a bacterium will cause to a host exist as a continuum.
Thus, host genetics is important to determine where a bacterium may lie in this continuum. Our data show that interaction between the worm and its bacterial food will be shaped by both host genes as well as the bacterium at the strain level. In our system, the most prominent host proponent shaping this interaction is the insulin-FOXO-dependent signalling pathway. In this context, C. elegans is an excellent model to design genetic screens and identify worm mutants that suppress the UV336-dependent extension of the daf-2 longevity phenotype.

ACKNOWLEDGMENTS This work was funded by the EP Abraham Cephalosporin
Trust grant no. CF 319 (to PL). of Two-Tukey's multiple comparisons one-way ANOVA tests, 99% CI. All panels: ***p<0.0001, NS=non-significant and n=25 animals/treatments/group. Results are Table S1. Statistics for Lifespan and Health span assays and mutants tested. For group categories see Fig. 2. WT is wild type (strain N2). Measurements: A* and B* relative to ilys-3 and C** and D** relative to the reference strain (N2). C. elegans mutants without a numerical value in the health span column were not moving at all and therefore we were unable to film their vigour. All lifespan experiments above were done in parallel.