Experimental infection with the hookworm, Necator americanus , promotes gut microbial diversity in human volunteers with relapsing multiple sclerosis

Background Helminth-associated changes in gut microbiota composition have been hypothesised to contribute to the immune-suppressive properties of parasitic worms. Multiple sclerosis is an immune-mediated autoimmune disease of the central nervous system whose pathophysiology has been recently linked to alterations of gut microbial communities. Results In the present study we investigated, for the first time, qualitative and quantitative changes in gut microbial composition of human volunteers with remitting multiple sclerosis (RMS) prior to and following experimental infection with the human hookworm, Necator americanus ( N+ ), and following anthelmintic treatment, and compared the findings with data obtained from a cohort of RMS patients subjected to placebo treatment ( PBO ). Bacterial 16S rRNA high-throughput sequencing data revealed significantly decreased microbial alpha diversity in the gut microbiota of PBO compared to N+ subjects over the course of the trial; additionally, we observed significant differences in the abundances of several bacterial taxa with putative immune-modulatory functions between study cohorts. Parabacteroides were significantly expanded in the gut microbiota of N + individuals for which no relapses were recorded at the end of the trial. Conclusions Overall, these data lend support to the hypothesis of a contributory role of parasite-associated alterations in gut microbial composition to the immunomodulatory properties of hookworm parasites.

quantitative changes in gut microbial composition of human volunteers with remitting multiple sclerosis (RMS) prior to and following experimental infection with the human hookworm, Necator americanus ( N+ ), and following anthelmintic treatment, and compared the findings with data obtained from a cohort of RMS patients subjected to placebo treatment ( PBO ). Bacterial 16S rRNA high-throughput sequencing data revealed significantly decreased microbial alpha diversity in the gut microbiota of PBO compared to N+ subjects over the course of the trial; additionally, we observed significant differences in the abundances of several bacterial taxa with putative immune-modulatory functions between study cohorts. Parabacteroides were significantly expanded in the gut microbiota of N + individuals for which no relapses were recorded at the end of the trial.
Conclusions Overall, these data lend support to the hypothesis of a contributory role of parasite-associated alterations in gut microbial composition to the immunomodulatory properties of hookworm parasites. Background A growing body of evidence supports a key role of infections by gastrointestinal (GI) helminth parasites in shaping the composition and function of the human gut microbiota, with significant implications for local and systemic host immunity, and metabolic potential 3 [reviewed by 1]. Notably, helminth-driven quantitative and qualitative modifications in the overall make-up of gut microbial populations have been proposed to contribute to the immune-suppressive properties of parasites [2][3][4][5][6][7]. For instance, in a milestone study conducted by Broadhurst et al. [8], experimental infections of a primate model of chronic idiopathic diarrhoea (CID) with the human large intestinal whipworm, Trichuris trichiura, were followed by a significant improvement of clinical signs and weight gain; these were accompanied by a notable increase in microbial alpha diversity in the colonic mucosal microbiota of worm-infected macaques. This observation led the authors to speculate that the onset of Th2-mediated host immune responses against the parasites might have resulted in significant changes of the mucosal environment, such as a reduced bacterial attachment to the intestinal mucosa post-worm colonisation and significant contractions of populations of potentially pathogenic bacteria to levels comparable to those of healthy controls [8]. Another study conducted in a cohort of human volunteers with Coeliac Disease (CeD) experimentally infected with the hookworm of the small intestine, Necator americanus, reported increases in gut microbial richness (observed in both faecal samples and biopsy tissues) that followed parasite colonisation and subsequent challenge with increasing doses of gluten [4,5,9]; whilst the detected differences did not reach statistical significance (likely due to sample size limitations), the increased gluten tolerance observed in infected CeD volunteers was postulated to contribute to the antiinflammatory properties of N. americanus via the restoration of microbial and immune homeostasis [5,9]. These data point towards a possible role of helminth-associated changes in gut microbial community composition and function in parasite-mediated suppression of chronic inflammation; nonetheless, thus far, studies of the role(s) that the gut microbiota plays in the therapeutic properties of helminth parasites have been carried out in human models of gut diseases [5,9]. Consequently, the dysbiotic state of the gut 4 microbiota of these individuals at baseline makes the determination of the mechanisms of microbiota-driven helminth immune-suppression challenging.  [12]. Notably, the clinical outcome of the WIRMS study provided further support to the promise of helminth-based therapy for treatment of RMS; indeed, at the end of the study, 51% of RMS patients experimentally infected with N. americanus showed no detectable new CNS lesions, as assessed by magnetic resonance imaging (MRI) scans, vs. 28% of placebo-treated volunteers [10]. Underpinning the immune-modulatory properties of N. americanus infections, the percentages of eosinophils and of CD4+CD25 high CD127 neg T cells in peripheral blood of worm-colonised individuals was significantly increased 9 months post-infection compared to placebo-treated subjects [10].
Given the existence of robust communications between the gut and the CNS by means of immunological, neural and endocrine mechanisms (i.e. gut-systemic-CNS axis; Rhee et al., 2009), as well as recent evidence demonstrating causal relationship between gut microbiome phenotype and the onset of MS [13], it is conceivable that the beneficial properties of N. americanus in RMS might be associated, at least in part, to the direct or 5 indirect effects that the parasites exert on the composition of the gut microbiota and relative abundances of individual bacterial species. Thus, building on the availability of unique biological specimens (i.e. faecal samples) collected over the course of the WIRMS trial, we explore, for the first time, the longitudinal changes in gut microbial profiles of human volunteers with RMS, prior to and following experimental infection with N. americanus, and subsequent administration of anthelmintic treatment, and compare the findings with data obtained from a cohort of uninfected, placebo-treated RMS patients. In particular, we show that, unlike the gut microbiota of placebo-treated RMS patients, that of N. americanus-infected volunteers was characterised by unaltered microbial diversity throughout the course of the trial, and by significant expansions in populations of bacteria with known immune-modulatory properties (e.g. Tenericutes/Mollicutes) with potential roles in parasite-mediated suppression of autoimmunity.

Ethics statement
This phase 2, single centre, randomised, double-blinded, placebo-controlled clinical trial (WIRMS; Clinicaltrials.gov identifier NCT00630383) aimed to assess the therapeutic efficacy of live hookworm (N. americanus) infective larvae in patients with RMS [10]. The trial was conducted at the Queen's Medical Centre, University of Nottingham, UK. The study was approved and carried out in strict accordance and compliance with the National Research Ethics Service Committee East Midlands (reference 11/EM/0140). Written informed consent was obtained from all subjects enrolled in the study.

Trial design
For details of patient recruitment, inclusion and exclusion criteria and trial design we refer to the original publication by Tanasescu et al. [10]. Briefly, a total of 73 clinically stable 6 RMS patients aged 18-64 (51 females and 22 males), who suffered at least 1 relapse over the prior 12 months or two over the prior 24 months and who were not subjected to immunomodulatory treatment were randomised and assigned to the two treatment groups, i.e. percutaneous infection with 25 N. americanus infective third-stage larvae (N+; n = 36), or placebo treatment with pharmacopoeial grade water (PBO; n = 37). Stool samples were collected one week prior to infection/placebo-treatment (= T pre ), as well as 1 (T1), 5 (T5), and 9 (T9) months post-infection/placebo-treatment (together referred to as T treatm ent ), and two months post-anthelminthic treatment (= T post ) and stored at -20°C until nucleic acids extraction (Fig. 1). Only study subjects who provided samples for all of these time points were included in this study (n = 50). Infections were confirmed for each N + patient via PCR and qPCR-guided N. americanus DNA detection performed using the latest available faecal sample prior to anthelmintic treatment, and following previously established protocols [10,14,15]. Two volunteers did not complete the WIRMS trial and thus were excluded from this study.

Bioinformatics and statistical analyses
Raw paired-end Illumina reads were trimmed for 16S rRNA gene primer sequences using Cutadapt (https://cutadapt.readthedocs.org/en/stable/) and sequence data were processed using the Quantitative Insights Into Microbial Ecology 2 (QIIME2-2019.1; https://qiime2.org) software suite [17]. Successfully joined sequences were quality filtered, dereplicated, chimeras identified, and paired-end reads merged in QIIME2 using DADA2 [18]. Sequences were clustered into Operational Taxonomic Units (OTUs) on the basis of similarity to known bacterial sequences available in the SILVA reference database (https://www.arb-silva.de/download/archive/qiime; Silva_132) sequences that could not be matched to references in the SILVA database were clustered de novo based on pair-wise sequence identity (99% sequence similarity cut-off). The first selected cluster seed was considered as the representative sequence of each OTU. The OTU table with the assigned taxonomy was exported from QIIME2 alongside a weighted UniFrac distance matrix.
Singleton OTUs were removed prior to downstream analyses. Cumulative-sum scaling (CSS) was applied, followed by log2 transformation to account for the non-normal 8 distribution of taxonomic counts data. Statistical analyses were executed using the Calypso software [19] (cgenome.net/calypso/); samples were ordinated in explanatory matrices using supervised Canonical Correspondence Analysis (CCA) including 'infection status' as explanatory variables. Differences in bacterial alpha diversity (Shannon index) between study groups (N + and PBO) were evaluated based on rarefied data (read depth of 8 712) and using analysis of variance (ANOVA); F-Tests were used to statistically assess the equality of assessed means (i.e. effect size). To take into account the paired nature of samples from N + and PBO across timepoints, differences between these sets were assessed using linear mixed effect regression. Differences in beta diversity (weighted UniFrac distances) were identified using Analysis of Similarity (ANOSIM) and effect size indicated by an R-value (between − 1 and + l, with a value of 0 representing the null hypothesis [20]). Differences in the abundance of individual microbial taxa between groups were assessed using the Linear Discriminant Analysis Effect Size (LEfSe) workflow (21). Bacterial taxa predictive of the study cohort either responding to hookworm treatment (N + responders ) or having MS disease activity (N + non− responders ) were identified via Wilcoxon rank test (19). The discriminatory power of microbial community profiles to distinguish between two biological conditions is characterized by a Support Vector Machine evaluated by leave-one-out cross-validation [19].

Results
The faecal microbiota composition of parasite-infected RMS patients 9 differs from that of placebo-treated RMS volunteers 68 180 ± 70 000) paired-end reads; of these, 9 100 255 high-quality sequences (per sample mean 38 397 ± 31 519) were retained following quality control. Rarefaction curves generated following in silico subtraction of low-quality sequences indicated that the majority of faecal bacterial communities were represented in the remaining sequence data, thus allowing us to undertake further analyses (Fig. S1).
Tenericutes/Mollicutes are expanded in the faecal microbiota of parasiteinfected RMS volunteers LEfSe analysis revealed significant differences in the relative abundances of individual microbial taxa (phylum to species level) between N + and PBO patients at T pre , T treatm ent , and T post (Fig. 4), as well as between the two groups at each timepoint within T treatm ent (Fig. S7A). Of these bacterial taxa, five genera (Roseburia, Dorea, Tyzzerella, Fusicatenibacter, and Agathobacter) belonging to the family Lachnospiraceae, Peptostreptococcaceae, Carnobacteriaceae, and Coriobacteriaceae were significantly more abundant in PBO subjects compared to N + subjects over the course of helminth infection (T treatm ent ). Conversely, three orders (RF39, Izimaplasmatales, and Anaeroplasmatales) belonging to Tenericutes/Molicutes, the families VadinBB60, Clostridiaceae (genus Hungatella), Ruminococcaceae (genera UCG005 and UCG010), Flavobacteriaceae, and the genera Sutterella, Barnsiella, and Coprobacter were significantly more abundant in the faecal microbiota of N + subjects than in that of PBO (Fig. 4). Notably, differences in Mollicutes abundance resulted from expanded populations of these bacteria in the microbiota from N + and a simultaneous contraction of the same taxa in faecal samples from PBO (Fig. S7B). Finally, Coriobacteriaceae remained more abundant in the microbiota of PBO subjects at T post , whilst Tenericutes/Mollicutes and VadinBB60, remained more abundant in N + subjects at the same time point (Fig. 4).
A summary of the findings from this study, including populations of gut bacteria expanded or reduced upon experimental infection with N. americanus, as well as fluctuations in gut microbial alpha-and beta-diversity over the course of the WIRMS trial, is available from the MICrobiome HELminth INteraction database (MICHELINdb) at www.michelindb.com [22].
Relapses are associated with differences in the abundances of selected bacterial populations in parasite-infected RMS volunteers Differences in gut microbiota composition between N + patients who presented significant MRI activity throughout the study (N + non− responders ; n = 10) vs. N + who did not show disease activity (N + responders ; n = 14) [cf. 10] were also investigated. Whilst no substantial differences in overall microbiota composition were detected between faecal samples from these sub-cohorts at T pre and T post (Figs. S8 and 5A), CCA analysis at T treatm ent detected significant differences between the microbial profiles of N + non − responders and N + responders ( P = 0.001; Fig. 5B). No significant differences in Shannon and beta diversity were detected between these groups at any timepoint (Figs. S9 and S10); nevertheless, gut microbial beta diversity was significantly elevated in the N + responders cohort at T treatm ent , albeit with a small effect size (R = 0.08, P = 0.001; Fig. 6).
This difference resulted from a significant increase in beta diversity in the faecal microbiota of N + responders , as well as a decrease in that of N + non− responders (Fig. S11).
Parabacteroides also remained the best predictor at T treatm ent , with the abundances of Roseburia NK4A136, Ruminococcus, Oscillibacter Eubacterium ( coprostanoligenes) also proving to be associated to treatment outcome in RMS patients (Fig. 8).

Discussion
In the present MHRA approved study we investigated, for the first time, the quantitative and qualitative changes in gut microbial profiles of human volunteers with RMS prior to and following experimental infection with N. americanus, and following administration of anthelmintic treatment, and compared the findings with data obtained from an age-and gender-matched cohort of RMS patients subjected to placebo treatment.

Changes in gut microbial diversity
No substantial differences in gut microbiota composition were detected between N+ and PBO patients prior to experimental infection based on CCA analysis, while significant alterations in faecal microbial profiles were detected in the infected cohort compared with uninfected subjects post-worm colonisation. Of note, these differences were no longer significant post-anthelminthic treatment, thus suggesting a direct modulating effect of live hookworm infection on the microbial communities inhabiting the host gut. In particular, microbial alpha diversity was significantly higher in the gut of N+ subjects compared to PBO over the course of helminth treatment. It must however be pointed out that this observation was predominantly linked to a marked decrease of gut microbial alpha diversity, and specifically of bacterial evenness, in PBO over time. Parasite removal via anthelmintic treatment did not result in microbial alpha diversity alterations in both N+ and PBO. Notably, compared with values recorded prior to hookworm infection, this parameter was significantly increased in N+ subjects at end the of the trial, whilst the opposite trend was observed in the gut microbiota of PBO. This finding is of particular interest, since elevated levels of microbial alpha diversity are typically associated with a 'healthier' gut microbiome and overall host health [reviewed by 23]. Similarly, increases in gut microbial alpha diversity were reported in CeD subjects experimentally infected with N. americanus, which led the authors to speculate that this mechanism might be at least partially responsible for the therapeutic effect of deliberate helminth infections in individuals affected by selected allergic and autoimmune disorders [5,9]. On the other hand, a decrease in gut microbial alpha diversity has previously been reported during MS relapses [24]. This matches our observations of a significantly lower microbial alpha diversity being detected at T treatm ent in the faecal microbiota of PBO volunteers who suffered relapses over the course of the trial, as well as decreased alpha diversity following specific relapse events in both N+ relapse and PBO relapse patients.

Gut microbial taxa expanded in the PBO cohort
The relative abundances of several gut bacterial taxa were also significantly altered in N+ subjects compared with the PBO cohort, both prior to N. americanus experimental infections and, more markedly, post-helminth colonisation. Indeed, despite subject randomisation prior to the beginning of the study, we detected minor differences in gut microbial composition between study cohorts at T pre . No prior characterisation of faecal microbial communities was conducted prior to group assignments; in addition, due to the intrinsic heterogeneity of gut microbial communities across any given population [25,26], differences in the abundances of gut microbial taxa are frequently detected between groups of individuals enrolled in randomized, double-blinded, placebo-controlled trials [27,28], with varying functional significance [28]. Nevertheless, in our study, most differences between N+ and PBO gut microbial profiles were detected post-infection, peaking at 9 months post-helminth colonisation. Notably, in PBO patients, bacterial taxa that have been previously associated with the gut microbiota of relapsing MS patients [24] were significantly expanded; in particular, a family of anaerobic bacteria, the Lachnospiraceae, including the genera Roseburia, Dorea, and Tyzzerella (amongst others), were significantly increased in the faecal microbiota of PBO compared to N+ subjects postinfection. Lachnospiraceae is a key family of the human gut microbiome that degrades complex polysaccharides to short-chain fatty acids (SCFAs), i.e. acetate, butyrate, and propionate, that are used for energy by the host. Notably, SCFAs are known for their antiinflammatory properties [29]. This is of note, since Lachnospiraceae have been reported to be substantially expanded in the gut microbiota of individuals affected by pathological conditions, such as inflammatory bowel disease (IBD) [30]. In particular, the abundance of Dorea increases in the gut of irritable bowel syndrome patients [31], as well as during intestinal inflammation [32]. Whilst Dorea are known to produce the SCFA butyrate, this genus of bacteria also metabolises sialic acids, which are commonly found at the terminal ends of mucins; release of these acids is implicated in mucin degradation, with consequences ranging from increased gut permeability to compromised gut homeostasis; this cascade of events has been suggested to result in proinflammatory responses that promote chronic inflammation in MS [33,34]. The expansion of Lachnospiraceae and its respective genera in the PBO cohort over the course of the trial is likely associated to MS disease progression and an immune shift towards a pro-inflammatory phenotype [34]. Of note, no significant expansion of Lachnospiraceae could be detected in the gut microbiota of N+ (as a whole) over time, with the exception of N+ non-responders (see below).
Gut microbial taxa expanded in the N+ cohort Bacteria belonging to the phylum Tenericutes were substantially and consistently increased in the gut microbiota of N+ patients post-infection. This phylum consists of the sole class Mollicutes, Gram-negative, small and wall-less bacteria that fulfil a diverse array of roles within the mammalian microbiome [reviewed by 35,36]. The relative abundance of these bacteria has been reported to differ between the normal gut microbiota and that found in a wide range of autoimmune conditions, such as IBD [37,38], Type 1 Diabetes [39][40][41], MS [42], and experimental autoimmune encephalomyelitis (EAE; a murine model of MS) [43]. Albeit inconsistently [40,41], Tenericutes/Mollicutes are often reduced in the gut microbiota of the diseased cohort, when compared to healthy controls [37-39, 42, 43]. Notably, whilst one of these studies reported higher Tenericutes abundance in the gut microbiota of healthy controls than in paediatric MS patients, exposure to immunomodulatory drugs reversed this trend [42]. Indeed, these bacteria have been suggested to proliferate in Th2-dominant environments [8,44].
Tenericutes/Mollicutes were also expended in the gut microbiota of vertebrates infected by Th2-inducing helminth parasites, human cohorts naturally infected with roundworms (i.e. Trichuris and/or Ascaris and/or hookworm) [45], rats infected with the tapeworm Hymenolepis diminuta [44], and primates with CID experimentally infected with T. trichiura [8]. This is in accordance with data from the WIRMS trial, that reported a markedly increased eosinophilia in N+ compared to PBO subjects [10]. Nevertheless, the functional consequences that expanded populations of Tenericutes/Mollicutes associated to infection by parasitic helminths play in the pathophysiology of the abovementioned chronic inflammatory and autoimmune disorders, as well as in RMS, remains to be determined. Alongside the evidence from previous studies, our data lends further support to the need for mechanistic investigations of the interactions between helminth and/or their products and this bacterial taxon in vitro and/or in vivo, to assist untangling its potential role in disease progression and prevention. Gut microbial differences between N+ r e s p o n d e r s and N+ n o n -r e s p o n d e r s In addition, we investigated the differences in gut microbial composition between N+ responders and N+ non-responders , with the aim to identify potential bacterial candidates with roles in MS disease activity. Whilst the overall gut microbiota of these sub-cohorts did not differ substantially prior to helminth infection and or post-anthelmintic treatment, differences were recorded in the gut microbiota of these sub-groups post-helminth colonisation. In particular, while gut microbial alpha diversity remained similar between N+ responders and N+ non-responders, beta diversity was significantly higher in the former (albeit with a small effect size), indicating a 'diversified' gut microbiota in N+ responders, either directly associated to N. americanus infection and/or, indirectly, to the immune responses mounted against the worms [1]. Meanwhile, the gut microbial beta diversity of N+ non-responders was reduced over the course of the trial, possibly resulting from downstream effects of disease progression on the host gut microbiota [10,[46][47][48].
We detected significantly increased populations of Tenericutes/Mollicutes in the gut microbiota of N+ responders , and expanded populations of Lachnospiraceae in N+ nonresponders , further supporting the hypothesis of a functional role of these taxa in the immune-modulatory properties of N. americanus. Furthermore, Flavobacteriaceae, the largest family in the phylum Bacteroidetes [49], was consistently increased in the gut microbiota of N+ responders . This taxon has been repeatedly reported to be depleted in people suffering from autoimmune conditions, such as rheumatoid arthritis [50] and myasthenia gravis [51], although the functional importance of this taxon in the pathophysiology of these conditions remains elusive. Finally, in this study, we asked the question of whether some of the minor differences in gut microbial profiles observed between N+ responders and N+ non-responders prior to hookworm experimental infection might be associated with positive or negative clinical outcomes. Amongst others, the genus Parabacteroides was significantly more abundant in the gut microbiota of N+ responders compared to N+ non-responders . Notably, a recent study conducted in murine models of RMS and chronic-progressive MS identified Parabacteroides as more abundant in the gut microbiota of control mice [52]. Additionally, a previous study detected a significant reduction of Parabacteroides populations in the faecal microbiota of 71 MS patients not undergoing immunesuppressive treatment compared to that of 71 healthy control subjects [53]. Subsequent monocolonisation of antibiotic-treated mice with Parabacteroides distasonis led to significant increases in the CD4+IL-10+ T lymphocyte population in mesenteric lymph nodes and spleens [53]. Furthermore, stimulation of peripheral blood mononuclear cells (PBMCs) from MS patients or healthy controls with total bacterial extracts isolated from the stool samples of the same subjects resulted in the inability of PBMCs from MS patients to differentiate or expand CD25+FoxP3+ Treg populations [53].
This observation led the authors to hypothesise that prior exposure to P. distasonis or other "beneficial" bacteria may have contributed to the expanding regulatory T lymphocyte precursor populations in mice, hence promoting anti-inflammatory responses upon subsequent exposure to the same bacteria [53]. Furthermore, it is known that Parabacteroides metabolise phytoestrogens [54][55][56]. Estrogens have been shown to possess disease-suppressive properties in MS as evidenced in several studies in humans and animal models [57].

Conclusion
In conclusion, we detected significant qualitative and quantitative differences in gut microbial composition between RMS subjects experimentally infected with N. americanus and placebo-treated individuals. A significant decrease in microbial alpha diversity, accompanied by expanded populations of Lachnospiraceae were detected in the faecal microbiota of the PBO cohort, in accordance with published data investigating gut microbial composition during MS relapses. Significantly higher microbial alpha diversity in the gut microbiota of N + patients suggest improved gut microbial and homeostasis. This was further supported by increased relative abundances of Tenericutes. Furthermore, we identified significant differences between the gut microbial profiles of N + responders and N + non− responders , with populations of Parabacteroides being significantly more abundant in the former. Overall, these data lend support to the hypothesis of a contributory role of parasite-associated modulation of host gut microbiota composition to the immunesuppressive properties of hookworms. This work adds valuable knowledge to current understanding of parasite-microbiota associations and will assist future mechanistic studies aimed to unravel the causality of these relationships.

Availability of data and material
The 16S rRNA gene sequence datasets generated and analyzed during the current study are available from Mendeley Data (DOI: 10.17632/pkk4vtc57r.1).

Competing interests
The authors declare that they have no competing interests.

Funding
The clinical trial from which specimens analysed in this study were derived was supported

Supplementary Files
This is a list of supplementary files associated with the primary manuscript. Click to download.