Cryptosporidiosis modulates gut microbiome metabolism and the immune response in an infected host

Abstract


Background
Enteric protozoal infections are a major human health concern globally, causing malnutrition through the loss of appetite, decreased nutrient absorption, and increased catabolism of nutrient reserves due to in ammation and diarrhoea. Cryptosporidiosis is a globally endemic infection with about 4.7 million annual reported cases [1] with high impact on children aged 4 years or below and immunocompromised individuals [2]. The infection is caused by Cryptosporidium parvum (henceforth indicated as Cryptosporidium), a major waterborne enteric pathogen [1]. Cryptosporidium is a highly specialised obligate apicomplexan parasite that is transmitted via the faecal-oral route from sources such as drinking water, or recreational waters contaminated with sewage and/or animal faeces [3]. Due to its ability to infect humans and other mammals, it is considered a ubiquitous zoonotic parasite [2,3].
Cryptosporidium infections are limited to the epithelial lining of the gastrointestinal tract, causing minimal invasion and penetration through mucosal layers, and are known to be autophagic [4]. Due to its highly specialised lifecycle, Cryptosporidium lacks numerous metabolic systems and must interact with its host to compensate for these de ciencies [5][6][7]. This signi cant intertwined relationship between a host and Cryptosporidium is, therefore, highly complex but is only partially understood. It has been shown that cryptosporidiosis causes long-term pan-body effects such as weight loss, abdominal, eye, and joint pain and, very likely, irritable bowel syndrome (IBS) [8]. However, the molecular mechanisms and host responses that result in these broad disease effects are not well understood.
In this context, a multi-omics approach will provide broader systems biology information relating to cryptosporidiosis. Omics platforms such as genomics, proteomics, and metabolomics, alone or in combination, have provided new insights that have been valuable in preventative health [9,10], toxicology, and medicine [11,12]. The high sensitivities and speci cities of multi-omics platforms provide excellent discrimination between samples and treatment types and have been applied to study environmental, clinical, and natural medicine systems [13][14][15]. Metabolomics and genomics have been independently applied to understand the Cryptosporidium life cycle in aquatic systems [16,17], and invitro studies [4,6,7]. Metabolomics or gut microbial community genomics studies have been performed separately on the infected host [18,19]. Furthermore, a multi-omics approach will provide broader systems biology information relating to cryptosporidiosis effects and host reponses.
In this study, we investigated interactions between C. parvum and a murine host during cryptosporidiosis using multi-omic platforms (metabolomics and proteomics). To de ne the host-speci c interactions of C. parvum, infection with the bacterium uropathogenic Escherichia coli (UPEC) [20], and the eukaryotic pathogen Giardia lamblia was used for comparative purposes. Gut infection with UPEC does not appear to cause disease symptoms in humans or mice [21], while Giardiasis results in similar symptoms to those seen in cryptosporidiosis [2]. Untargeted metabolomics, proteomics, and microbiome 16S rRNA gene sequencing were applied to numerous body tissues and gut washes after infection with C. parvum or abovementioned infectious agents. We examined how cryptosporidiosis modulates the enteric microbial community pro le and, affects host protein levels and metabolic processes throughout the mouse intestinal tract and in extra-intestinal tissues. This work also provided new insights into previously unreported dynamics of microbiome, parasite and host relationships in different regions of the gastrointestinal tract during cryptosporidiosis.

Animal Ethics and husbandry
All experiments were approved by the Monash University Animal Ethics Committee (Monash University AEC no. MARP/2018/055) following the guidelines of Victorian State Government and The National Health and Medical Research Council, Australian Government. Mice were housed in Optimice cages containing sterile sawdust at 18-24ºC, 40-70% humidity, and 12:12 hour light/dark cycle. Mice were provided with sterile water and feed (Ridely AgriProducts Pty. Ltd., Melbourne, VIC, Australia) ad libitum.

Mouse infection model
For the study, groups of ve, three-week-old C57BL/6J female mice were acclimatised for one week before infection with either 1×10 5 C. parvum or G. lamblia oocysts (C. parvum, Cat. Number: C10E7; G. lamblia, cat. Number: G10E6; BTF Pty Ltd., North Ryde, NSW, Australia) or 1×10 8 CFU of UPEC (ST131 lineage strain EC958) via oral gavage. An additional group of uninfected mice (n = 5) was also included for comparison against the infected groups. Cryptosporidium and Giardia infection was monitored for 10 days via daily faecal collection and detection of oocysts by uorescent microscopy using the EasyStain kit TM (Biopoint Pty Ltd., Sydney, Australia). UPEC infection was quanti ed using a previously described method [20]. Mice were euthanised at 10 days post-infection by CO 2 exposure and liver tissue, serum, and faeces collected. The luminal contents of the duodenum, jejunum, ileum, caecum, and colon were sampled by ushing 1.0 mL of sterile phosphate buffer saline through each section of the gut and collecting the contents ( Figure 1). Serum (collected by cardiac bleed) and liver tissue samples were used as representatives for indirect and cross-organ effects of C. parvum infection. The luminal contents represented the direct effects of infection. All samples were immediately stored on dry ice and then at -80°C until further analysis.
The protein concentration of the cleaned extract was determined by Bradford assay and a volume equivalent to 5 µg of protein was taken. The volume was adjusted to 10 µL by adding urea (8 M). Dithiothreitol (1µL, 15% v/v) and acrylamide (1 µL, 40% v/v) were added, each followed by a 30-min incubation (25°C). After trypsin (0.1 µg in 20 mM ammonium bicarbonate, 47 µL) was added, the mixture was incubated for 3 h at 37°C. Formic acid (1µL, 10% v/v) was added to terminate trypsin activity.
Genomic extraction, analysis, and processing Mouse faeces and luminal contents (n = 5 each) were homogenised and DNA was extracted using the manufacturer's instructions (ZymoBiomics DNA miniprep kit, Zymo Research Corp., Irvin, CA, USA), followed by DNA ampli cation, sequencing, and analysis (Supplementary Materials). Sequence analysis was performed using QIIME 2 (Release no. 2019.7) pipeline [25], as previously described [15]. Multivariate statistics using METAGENassist analysis [26] were performed to investigate the metabolic nature of the microbial community detected in each sample group.

Multi-omics integration and statistical analysis
The metabolomics and proteomics data were adjusted for batch-effect, log transformed and multivariate data analysis conducted with the software SIMCA (version 16, Sartorius Stedim Biotech, Umeå, Sweden) and MetaboAnalyst 4.0 [27]. The cut-off level for signi cant metabolites was a signal-to-noise (S/N) ratio of 10, while for proteins, it was a relative abundance of 1 × 10 5 . For statistical analysis of both metabolome and proteome, a fold change of ≤ 0.5 (downregulation) or ≥ 2.0 (upregulation), and a Benjamini-Hochberg adjusted p-value of ≤ 0.05. Metabolic and proteomic outputs were integrated using the 'Joint-pathway analysis tool' of Metaboanalyst 4.0 and Paintomics 3 [28].
The metabolic pathway networks obtained after statistical analyses were manually curated in Omix visualization software (Version 1.9.34; Omix Visualisation GmbH and Co. KG, Lennestadt, Germany).

Microbiome distribution and, protein and metabolic expression in the gut
Interaction dynamics in the infected host's gastrointestinal tract during cryptosporidiosis were assessed by (i) the response of the host's system and gut microbiome, and (ii) the effects on non-gut organs, as detailed below.
The genomic analysis of luminal contents and faeces performed via rarefaction analysis and Good's coverage index (Table S1) (Figures S3B and S3C). Searches against 43 microbial UniProt databases showed that the number of expressed microbiome proteins increased from the duodenum (30 proteins) through to colon (815 proteins) and faeces (956) ( Figure S3A). The metabolome output showed the presence of 162 metabolites across all the analysed samples ( Figure S4, Table S2Table S3).

Gut microbiome response during cryptosporidiosis
The 16S rRNA gene analysis by Greengenes database assigned OTUs to 71 bacterial genera. Of these, 22 genera were represented in all luminal contents and faeces and were indicated as the core microbial community (Figure 2A). Whilst Faecalibaculum, Barnesiella, and Lactobacillus were abundant in the small intestine, the Ruminococcaceae population increased in the caecum and colon ( Figure 2B). During cryptosporidiosis, in the small intestine, bene cial bacteria such as Faecalibaculum and Lachnospiraceae showed considerable depletion, while Lactobacillus, Lachnospiraceae, Desulfovibrio, and Coriobacteria populations increased ( Figure 2B).
The role of the microbiome in the production of short-chain fatty acid (SCFAs) in the gut [29] is known, especially during induced gut disorder stress [15]. We examined whether cryptosporidiosis-induced changes in the microbiota composition affected SCFA production in the gut of infected mice. We observed that among SCFAs, formate had a higher metabolic expression in the small intestine (duodenum > jejunum > ileum), while acetate, propanoate, and butanoate accumulation increased in the caecum and colon during cryptosporidiosis ( Figure 3A). In addition to SCFAs, the accumulation of Damino acids, such as D-alanine and D-proline, in the small intestine ( Figure 3B) re ected an increased abundance of Lactobacillus (Figures 2B), supporting the observations of Sasabe et al [30].
An increase in protective/stress response (microbiome) proteins with the increase in Lactobacillus (or similar bacterial) population was also indicated by the increased levels of proteins responsible for glycolysis and fatty acid metabolism ( Figure 4A The caecum and colon showed the expression of different prokaryotic proteins. The expressed proteins mainly related to the glycolysis pathway, leading to fatty acid synthesis and oxidative stress protection proteins such as rubrerythrins ( Figure 4B).

Host-response in the gut during cryptosporidiosis
The activation of host defence systems showed a considerable increase in the jejunum where numerous immune-type proteins were expressed during infection. Host response proteins associated with protective and in ammatory responses (actins, myosins, keratins, heat shock proteins, apoptosis-associated proteins), oxidative stress (glutathione S-transferase, GTPases, selenium binding proteins), and glycolysis/gluconeogenesis-associated enzymes were observed (Table S9, Supporting dataset 2). Immunity-related heat shock proteins, namely actins and tubulins, were expressed along with glycolysisrelated enzymes such as glyceraldehyde-3-P dehydrogenase and dihydrolipoyl dehydrogenase. Cryptosporidium proteins such as actin (FC = 3065.1), tubulin (FC = 1040.6), and heat shock proteins (FC of HSP90 = 2483; HSP 70 = 197.2) showed considerable upregulation in the ileum, indicative of increased expression during infection. Additionally, 3064 host response proteins were expressed in the ileum, the highest among all the intestinal regions ( Figure S3).
On the contrary, the number of elevated metabolites increased in the caecum and colon during infection. In the caecum, amino acids such as glycine, methionine, creatinine, tyrosine, alanine, lysine, and cysteine were increased ( Figure 5A, Table S6). Other increased metabolites in the colon included fatty and organic acids, such as malate, 3-aminoisobutyrate, fumarate, 3,4-dihydroxymandelate, and citrate ( Figure 5A, Table S7). The metabolic composition of the faeces was similar to that of the colon, with addition of increased abundances of organic acids and non-digestible sugars, such as cellobiose ( Figure 5A, Table  S8). Overall, metabolic activity, combined with protein expression, indicated that Cryptosporidium metabolic activity was most prominent in the small intestine, followed by a decline in the caecum and a spike in the colon and faeces.
The integrated joint-pathway analysis of metabolic-proteomic datasets showed 69 key metabolic pathways being expressed, of which 10 were statistically signi cant with respect to uninfected mice (Holm adjusted p-value ≤ 0.05) ( Table 1).  Figure 5A, 6, and S7). Proteins related to the citrate cycle and oxidative phosphorylation were expressed across the mouse intestine during cryptosporidiosis ( Figure 5B). The regression analysis indicated increased catabolism of orthophosphate throughout the intestinal tract ( Figure 5A), except in faeces. Among proteins, the highest expressions (cFC > 2) were related to oxidative phosphorylation and glycolysis (Table S9). Other energy generation pathways such as glutamate metabolism possibly assisted Cryptosporidium to create a proxy-citrate cycle. These involved host mitochondrial NADH dehydrogenases [ubiquinone] (Uniprot IDs: D3YUK4, Q99LY9, Q9Z1P6, and Q9D6J6; cFC = 1.35). These results indicate that considerable oxidative phosphorylation is necessary to maintain highly upregulated citrate cycle activities ( Figure 6) during cryptosporidiosis [5].
Changes in fatty acid metabolism, especially medium-chain and long-chain fatty acids (MCFAs and LCFAs), were observed in the intestine upon Cryptosporidium infection. Although fatty acid oxidation and glycerolipid metabolism was observed in the duodenum, the latter was more prominent throughout the small intestine, as indicated by a signi cant decrease of glycerol in the jejunum (FC = 0.07) and the ileum (FC = 0.04), and palmitate (FC = 0.26) and palmitoleate (FC = 0.22) in the ileum.
Proxy-citrate cycle protein expression during cryptosporidiosis compared to G. lamblia and UPEC infection To ascertain if the protein pro le observed in the gut was speci c to cryptosporidiosis, we compared the proteomic output during cryptosporidiosis to that obtained from a UPEC gut infection or Giardia infection. Immune precursor proteins such as actins showed similar expression across the intestine (cFC 1 -6.2) across all three infections. However, proteins related to oxidative phosphorylation and glycolysis had higher expression during cryptosporidiosis when compared to UPEC infection and Giardiasis (Table S9). Additionally, some proteins with high expression during cryptosporidiosis such as ADP/ATP translocase (cFC = 3.59), electron transfer avoproteins (cFC = 2.5 -2.63) and acyl CoA binding proteins (cFC = 2.08) were either non-signi cantly (p-value ≥ 0.05) different or were downregulated (p-value ≤ 0.05) during Giardia or UPEC infection. Also, related proteins such as glutathione peroxidase (cFC = 2.38) and phosphoglycerate kinase (cFC = 2.07) had signi cantly higher expression during cryptosporidiosis (Table  S9). The analysis indicated that the proxy-citrate cycle was speci cally upregulated during cryptosporidiosis concerning other gut infections.

Extra-intestinal effects of cryptosporidiosis
Few studies have focused on the effects of enteric infection on non-gut organs and, to our knowledge, no studies have addressed this for cryptosporidiosis. For this study, serum and liver were used as representative samples for measuring extra-intestinal effects, such as nutrient absorption, detoxi cation, and immune response.
During cryptosporidiosis, we observed downregulation of fatty acid metabolism in the serum; the major fatty acids affected were palmitoleate (FC = 0.07), oleate (0.05), and myristate (0.02) when compared to the uninfected mice (Table S10). In the liver, we observed a similar decrease, speci cally of 6-hydroxy caproic acid and succinic acid (Table S11).
During cryptosporidiosis, of 1320 and 3016 expressed proteins in serum and the liver, respectively, 327 were signi cantly upregulated across both (Supporting Data 2). These included immune response proteins, such as myosins and selenium binding proteins. Complement factors H (cFC = 3.97) and B (cFC  (Table S12, Supporting dataset 3).

Discussion
Cryptosporidiosis dynamics in the gut The microbiome generally forms the rst line of defence upon the onset of parasite infection. Numerous mechanisms play a role in the Cryptosporidium-microbiome relationship during cryptosporidiosis that can be used to further elaborate on the dynamics of this zoonotic infection in humans [31]. Considered among the core microbiome population, Faecalibaculum promotes higher SCFA metabolism, maintaining ionic balance and controlling virulence factors secreted by bacterial pathogens [32]. Synthesis and metabolism of SCFAs by the gut microbiome modulate in ammatory cytokine activity [33], especially by increased butanoate and propionate production in the caecum and colon [29]. Among the gut microbial community, Faecalibaculum and members of the Erysipelotrichaceae are known to produce high levels of lactate and formate [34,35]. In the current study, formate showed considerable contribution towards both pyruvate and glycolysis metabolism (Figures 2A and S5, and S7). A depletion of Faecalibaculum ( Figure  2) due to Cryptosporidium infection in this study would, therefore, explain the microbiome dysbiosis and decreased pyruvate and glycolysis metabolism in the small intestine.
An increased Desulfovibrio population, on the other hand, has been shown to perturb microbiota during protozoal gut infections. Desulfovibrio is known to cause enteritis through mucosal or epithelial damage by releasing toxic compounds such as hydrogen sulphide in the gut environment [36]. We observed that bacteria such as Coriobacteriaceae and Lactobacillus, increased in abundance during Cryptosporidium infection, especially in the small intestine. Coriobacteriaceae have been demonstrated to modulate glucose metabolism [37]. Similarly, the increased level of Lactobacillus may represent a microbial response for countering the mucosal/epithelial damage caused by Cryptosporidium infection. Reportedly, lactate metabolising bacteria are highly active in the production of D-amino acids [38] and, provide an elevated microbial response to balance the mucosal/epithelial damage caused by Cryptosporidium infection [36].
Microbial carboxylase transporters are part of tripartite ATP-independent periplasmic (TRAP) and Tripartite Tricarboxylate Transporters (TTT) protein families. These proteins induce pathogenicity and colonisation of bacteria such as Haemophilusin uenzae and Salmonella Typhimurium. This activity uses energy sources, such as glutamate, and causes increased levels of dicarboxylic acids, for instance, acetate or hexanoate ( Figure 3A) [39]. Cryptosporidium excystation in the duodenum has been documented [3] and may be responsible for the observed upregulation of proteins assoaciated with glycolysis, glutaminolysis, and citrate cycle in the small intestine. We found that the citrate cycle was more active across the intestine during cryptosporidiosis ( Figure 5A, Figure 3A, 4A, 6, and S7). C. parvum reportedly lacks the machinery for the citrate cycle pathway and salvages it from the host [40]. However, the current study shows, for the rst time, a greater role of yeasts in driving this proxy-citrate pathway.
Additionally, the pathway may have also contributed to fatty acid metabolism. In particular, malonate catabolism (FC = 0.35) and fatty acid synthase activity (A0A0U1RNJ1, P19096; cFC = 1.69) indicate the presence of glycolysis and proxy-citrate cycle derived fatty acid synthase I (FAS I) system in the jejunum and the ileum, as reported previously [41].
Compared to the small intestine, glutamine/glutamate metabolism was upregulated in the infected caecum. Glucose depletion in the caecum and the colon is known to trigger glutamate utilisation as the primary carbon source, by both Cryptosporidium and host defence cells [6,42]. In the context of cryptosporidiosis, we observed glutamate utilisation typical of parasitic activity for generating αketoglutarate, catalysed by glutamine synthetase, glutamate kinase, and glutamate-5-semialdehyde dehydrogenase, as previously documented [6].
We observed upregulated host and yeast transketolases, followed by yeast polyubiquitin proteins, indicative of these proteins/enzymes catalysing ubiquinone biosynthesis in the jejunum-ileum tract. The preliminary step of the ubiquinone biosynthesis pathway begins with E4P metabolism and is catalysed by glucose-6-phosphate dehydrogenase (G6PDH) in trypanosomatid [43] and Plasmodium [44] metabolism, and during C. parvum infection in rats [45]. The utilisation of enzyme systems from the host, yeasts, and the parasite indicates a host-parasite-microbiome association in the small intestine. This association may have compensated the de cient Cryptosporidium metabolic machinery for synthesising ubiquinone (coenzyme Q), which is a critical element of the electron transport chain. Such associations and their bene ts to Cryptosporidium multiplication have been reported in aquatic systems [17] and neonatal mice gut dysbiosis [18].
High yeast ubiquitin-related activity was observed across the small intestine, especially in the ileum. While 11 mouse ubiquitin transfer or conjugating enzymes were observed to be upregulated throughout the host response (FC = 1.1 -10.4), yeast ubiquitin/polyubiquitin proteins were also upregulated (FC = has been reported that Cryptosporidium salvages the host ubiquinone system [6], our study indicated signi cant salvaging of this system from the yeast population of the microbiome. To our knowledge, the current study is the rst to suggest that enhanced Cryptosporidium colonisation may depend more on the parasite-microbiome relationship than the host-parasite relationship.

Extra-intestinal effects of cryptosporidiosis
In non-gut organs such as the liver, oxalic acid upregulation is indicative of likely hyperoxaluria or a hyperoxaluria-like condition. In this condition, glyoxylate metabolism is negatively affected due to the de ciency of hepatic alanine glyoxylate aminotransferase (AGT) and cytosolic glyoxylate reductase (GR) [46,47]. However, the relationship to hyperoxaluria as an indirect effect of Cryptosporidium infection in the gut remains to be determined.
Mitochondrial pyruvate carboxylase was possibly one of the most interesting of the expressed proteins in the liver. This zinc-containing protein, in the presence of allosteric activators such as acetyl-CoA, catalyses the pyruvate " oxaloacetate reaction towards both Krebs cycle replenishment and gluconeogenesis [48]. However, an excessive accumulation of oxalate (caused by oxaloacetate accumulation) in the liver of Cryptosporidium-infected mice may be attributed to the high expression of Llactate dehydrogenase (LDH). The role of hepatic LDH in converting glyoxylate to oxalate has recently been reported for primary hyperoxaluria mouse models [49] and blood-based protozoal infections such as that with Plasmodium [50]. However, its indirect hepatic activity, especially as a follow-up pyruvate carboxylase activity, due to gut infection has not been reported and its dynamics require further study.

Conclusions
We utilised a mouse model to study the direct (gut) and indirect (serum and liver) effects of cryptosporidiosis using a multi-omics approach. Energy pathways such as glycolysis and glutaminolysis were signi cantly impacted in the jejunum and ileum during cryptosporidiosis. The proteomic and metabolic outputs indicated an underdeveloped proxy-citrate cycle in Cryptosporidium, partially salvaged from the host, with additional input of yeast citrate cycle enzymes. Instead of the commonly reported G6PDH-catalysed route, the ubiquinone (CoQ) biosynthesis system in the ileum began with host transketolase activity, followed by the salvation of the yeast ubiquinone biosynthetic system. The gut microbiome response to cryptosporidiosis was detected via increased metabolism of D-amino acids and SCFAs. Similarly, high oxalate accumulation in the liver indicated enteric hyperoxaluria as a likely indirect effect of cryptosporidiosis. Our study shows the ability of multi-omics to contribute a robust understanding of gut infections and demonstrate the previously unreported microbiome interaction dynamics during cryptosporidiosis. The study highlights the comprehensive dynamics of gut infection and may serve as a model that can be used for the in-depth study of other enteric infections. These results provide a platform from which new avenues of precision medicine and improved treatment methods for cryptosporidiosis may be devised.

Competing interests/Con ict of interest statement
The authors declare no con ict of interest.