An In Vitro Pilot Fermentation Study on the Impact of Chlorella pyrenoidosa on Gut Microbiome Composition and Metabolites in Healthy and Coeliac Subjects

The response of a coeliac and a healthy gut microbiota to the green algae Chlorella pyrenoidosa was evaluated using an in vitro continuous, pH controlled, gut model system, which simulated the human colon. The effect of C. pyrenoidosa on the microbial structure was determined by 16S rRNA gene sequencing and inferred metagenomics, whereas the metabolic activitywas determined by1H-nuclear magnetic resonancespectroscopic analysis. The addition of C. pyrenoidosa significantly increased the abundance of the genera Prevotella, Ruminococcus and Faecalibacterium in the healthy donor, while an increase in Faecalibacterium, Bifidobacterium and Megasphaera and a decrease in Enterobacteriaceae were observed in the coeliac donor. C. pyrenoidosa also altered several microbial pathways including those involved in short-chain fatty acid (SCFA) production. At the metabolic level, a significant increase from baseline was seen in butyrate and propionate (p < 0.0001) in the healthy donor, especially in vessels 2 and 3. While acetate was significantly higher in the healthy donor at baseline in vessel 3 (p < 0.001) compared to the coeliac donor, this was markedly decreased after in vitro fermentation with C. pyrenoidosa. This is the first in vitro fermentation study of C. pyrenoidosa and human gut microbiota, however, further in vivo studies are needed to prove its efficacy.


Introduction
Humans only have the ability to degrade two types of glycosidic linkages found in glycans and therefore cannot degrade most of the complex carbohydrates present in the diet [1]. Collectively known as dietary fibre or microbiota-accessible carbohydrates (MACs), these indigestible plant polysaccharides are degraded and fermented by the gut microbiota as its main energy source [2]. Analyses of microbial genes have shown a variety of carbohydrate-active enzymes (CAZymes) enabling the breakdown of diverse polysaccharide structures in the gut [3,4]. The decline in MAC dietary intake in the urban Western population (lower than the recommended 30 g of fibre/day) has been linked to decreased bacterial diversity, altered immunity and disease progression [5]. MACs also have the ability to modulate the gut microbiota towards a more health-promoting diverse profile. Recent studies have demonstrated the impact of low dietary MACs in mice, and HbA1c, while Clostridium sp. had a negative correlation with FBG, HbA1c and plasma triglycerides [35].
Here, we used an in vitro continuous, pH controlled gut model system that reflected the environmental conditions of the three regions of the human colon (vessel 1, proximal colon; vessel 2, transverse colon; vessel 3, distal colon)to assess whether C. pyrenoidosa has the potential to modulate the coeliac gut microbiome towards a healthier configuration, thereby affecting the production of SCFAs. This study aims therefore to examine the outcome of changes in the microbiota composition (and inferred function) and metabolite profile during an in vitro fermentation with C. pyrenoidosa between coeliac and healthy human donors.

C. pyrenoidosa
C. pyrenoidosa (Sun Chlorella 'A' Powder) individual powder sachets (3 g) pulverized by DYNO-Mill and supplied by Sun Chlorella Corporation (Kyoto, Japan). This remarkable technology liberates the nutrients in Chlorella, whilst breaking up the in-digestible Chlorella cell wall without the need for excessive heat or chemicals. The nutritional composition is reported in Table S1.

Faecal Sample Preparation
Faecal samples were obtained from a 30-year-old female with normal gluten-containing diet (H) and a 30-year-old female with Coeliac disease who had been following a GFD for more than 1 year. Both participants had not taken antibiotics and/or prebiotics or probiotic supplements for at least 3 months before sampling. An aliquot of 20 g of faecal samples from healthy (H) and coeliac (C) volunteer was diluted in 100 mL anaerobic PBS (0.1 mol/L phosphate buffer solution, pH 7.4, w/w) and homogenised (Stomacher 400; Seward, West Sussex, UK) for 2 min at 240 paddle beats per minute.
The University of Roehampton Research Ethics Committee (LSC 18-241) approved this study in accordance with the Declaration of Helsinki. Sample size was in accordance with previous studies [36].

Three-Stage Continuous Culture Gut Model System
A small-scale version of the validated three-stage continuous system was used in this study, consistent with previous studies [37] with vessels (V) representing the proximal (V1, 80 mL, pH = 5.5), transverse (V2, 100 mL, pH = 6.2) and distal colon (V3, 120 mL, pH = 6.8). A schematic chart of the validated three-stage continuous system is reported in Figure S1. Each vessel was magnetically stirred and continually sparged with oxygen free nitrogen gas. Temperature (37 • C) was maintained by a water-cooling system and culture pH was controlled automatically through the addition of 1 N NaOH or HCl [37].
Following inoculation, the colonic model was run as a batch culture for 24 h to stabilize the bacterial populations prior to the initiation of medium flow.
After 24 h, the medium flow was initiated and the system was run for eight full volume turnovers at 15, 16 and 17 days to allow for steady state to be achieved (SS1). The duration of each turnover was 48 h thus this first step was done for 384 h with the flow rate of the system set at 6.25 mL/h. Thereafter, C. pyrenoidosa (Sun Chlorella 'A' Powder) was administered into V1 in a powder form3 g dose at 10.00 am each day until the second steady state (SS2) had been reached at 33, 34, and 35 days. Each steady state was confirmed by stabilization of SCFAs profiles over 3 consecutive days. SS2 was achieved after 384 h; the whole experiment took 792 h. A sample of 4.5 mL was taken through the sample port of each vessel (V1, V2 and V3) subsequent to the first (SS1) and second full turnover (SS2) of medium through the model. Two aliquots of 2.25 mL each were centrifuged at 13,000 rpm, the supernatant was stored at −20 • C for short chain fatty acid profile whereas the pellets were stored at −80 • C for 16S rRNA gene-based next-generation sequencing analysis.
Sample analysis was performed according to the procedure outlined by Grimaldi et al. (2018) [38]. Spectral data was imported into Matlab (Mathworks, version 2018a) using inhouse scripts. Spectral regions containing redundant peaks (H 2 O, trimethylsilylpropionate (internal standard]) were excised from the dataset and the profiles were normalized using a probabilistic quotient approach. Principal components analysis (PCA) was performed on the 1 H-NMR spectral data in Matlab using in-house scripts. Based on the PCA findings, the peak integrals (representing relative abundance) for acetate (δ 1.92, singlet), butyrate (δ 0.90, triplet) and propionate (δ 1.06, triplet) were calculated. Samples were collected in duplicate before intervention for SS1 and SS2 as previously described from each vessel (V1, V2, V3) from both gut models with Coeliac (C) and healthy (H) donors.

Microbial DNA Extraction
Total microbial DNA was extracted from 250 mg of fermentation samples using the QIAamp DNA Stool Mini Kit (QIAGEN, Hilden, Germany) according to the manufacturer's instructions. DNA concentration and quality were evaluated using the NanoDrop ND-1000 spectrophotometer (NanoDrop Technologies, Wilmington, DE, USA).  [39]. In brief, for each sample, the hypervariable V3−V4 regions of the 16S rRNA gene were PCR-amplified using the S-D-Bact-0341-b-S-17/S-D-Bact-0785-a-A-21 primers with Illumina overhang adapter sequences [40]. PCR reactions were performed in a final volume of 25µL, containing 12.5 ng of genomic DNA, 200 nM of each primer, and 2X KAPA HiFi HotStart ReadyMix (Kapa Biosystems, Roche, Basel, Switzerland). The following gradient was used: 3 min at 95 • C for initial denaturation, 25 cycles of denaturation at 95 • C for 30 s, annealing at 55 • C for 30 s and extension at 72 • C for 30 s, followed by a final extension step at 72 • C for 5 min. PCR products were purified by using a magnetic bead-based system (Agencourt AMPure XP; Beckman Coulter, Brea, CA, USA), indexed by limited-cycle PCR using Nextera technology, and further purified as described above. Final libraries were pooled at equimolar concentration (4 nM), denatured with 0.2 N NaOH, and diluted to 6 pM before loading onto the MiSeq flow cell. Sequencing was performed on an Illumina MiSeq platform with the 2 × 250 bp paired-end protocol, according to the manufacturer's instructions (Illumina, San Diego, CA, USA). All sequencing reads were deposited in the National Center for Biotechnology Information Sequence Read Archive (Bioproject: PRJNA685154).

16S rRNA Gene-Based Next-Generation Sequencing and Bioinformatics
Raw reads were processed using a pipeline combining PANDAseq [41] and QIIME 2 [42]. High-quality reads were filtered and clustered into amplicon sequence variants (ASVs) using DADA2 [43]. Taxonomy was assigned using the VSEARCH classifier [44] against Greengenes database as a reference (release May 2013). Alpha diversity was measured using the number of observed ASVs and the Faith's Phylogenetic Diversity (PD whole tree) index. Beta diversity was computed based on weighted and unweighted UniFrac distances and visualized on a Principal Coordinates Analysis (PCoA) plot. Metagenome prediction of Greengenes-picked ASVs was performed with PICRUSt2) [45], using Meta-Cyc [46] as reference for pathway annotation.

Data Analysis
Statistical analyses were performed using GraphPad Prism 8 and RStudio. Matlab was used to plot the PCA of the metabolite dataset.
Data from 1 H-NMR were analysed employing a paired t-test, comparing before (SS1) and after the treatment (SS2) with C. pyrenoidosa for each SCFA (acetate, butyrate, propionate) for both Coeliac and healthy gut models. The statistical significance level was set at p < 0.1 due to the small sample size. As for the microbiota data, statistics were performed using R Studio 1.0.44 on R software version 3.3.2 (https://www.r-project.org (accessed on 13 February 2021)) [47] implemented with the packages stats and vegan (https://cran.r-project.org/web/packages/vegan/vegan.pdf (accessed on 13 February 2021)) [48].
The significance of data separation in the PCoA plot was tested by a permutation test with ANOSIM and pseudo-F ratio statistics using the function adonis in vegan. Bar plots were built using the R packages made4 [47] and vegan. Sample clustering was performed according to the pathway-level abundance profiles, using Kendall's correlation coefficients as metric and Ward-linkage method. Linear discriminant analysis effect size (LEfSe) algorithm with LDA score threshold of 2 (on a log10 scale) was applied to identify the MetaCyc pathways affected by the addition of C. pyrenoidosa [49]. Nonparametric tests (Kruskal-Wallis test or Wilcoxon test, paired or unpaired as needed) were achieved using the stats package. A p ≤ 0.05 was considered statistically significant; a P between 0.05 and 0.1 was considered a tendency.

Impact of C. pyrenoidosa on Faecal-Derived Microbial Communities from Healthy and Coeliac Donors
The faecal-derived microbial communities from a healthy donor (H) and a Coeliac donor (C) were profiled at baseline (steady state 1, SS1) and after the addition of C. pyrenoidosa, upon reaching steady state 2 (SS2). To ensure that effects were due solely to the treatment with C. pyrenoidosa and not to the adaptation of microbes to the in vitro environment, steady-state 1 and steady-state 2 (conditions in terms of microbial community composition and metabolic activity) were needed to be established prior to the actual start of the experiment. The 16S rRNA gene-based next-generation sequencing of all fermentation samples yielded a total of 868,020 high-quality reads, with an average of 36,168 ± 8485 sequences per sample, binned into 2289 amplicon sequence variants (ASVs).
No significant differences were observed in alpha diversity across the entire dataset, regardless of the origin of the faecal sample (C vs. H), simulated intestinal region (vessel 1 vs. 2 vs. 3) and time point (baseline vs. steady state) (p > 0.05, Kruskal−Wallis test) ( Figure 1A). However, a trend towards lower diversity was found after addition of C. pyrenoidosa for both donors. This trend was more marked in vessels 1 and 3 than in vessel 2.
Principal coordinate analysis (PCoA) based on weighted and unweighted UniFrac distance matrices showed separation between the faecal-derived microbial communities from C and H donors, regardless of the time point and the fermentation vessel (p = 0.002 and At baseline, several taxonomic differences were observed between the microbiota of the C versus H donor ( Figure S2). In particular, compared to donor H, the faecal-derived microbial ecosystem of the C donor showed a greater relative abundance of Bacteroidaceae (i.e., Bacteroides), Enterobacteriaceae and Clostridiaceae families (p ≤ 0.2, Wilcoxon test). In contrast, Bifidobacteriaceae (i.e., Bifidobacterium sp.), Alcaligenaceae and Veillonellaceae were more represented in donor H than in C (p ≤ 0.2).
When looking for differences in microbiota composition between baseline and steady state following C. pyrenoidosa supplementation, we observed both common and unique microbial signatures of response (Figures 2 and S3). Among the features shared between C and H donors, it is worth noting that the families Ruminococcaceae and Xanthomonadaceae substantially increased after C. pyrenoidosa addition in all vessels, while a general decrease was found for Bacteroidaceae and Erysipelotrichaceae (p ≤ 0.1, Kruskal−Wallis test) ( Figure S3). On the other hand, some taxa showed distinct trends based on the origin of the faecal sample. Specifically, Prevotellaceae increased only in H samples, while the relative abundance of Enterobacteriaceae decreased in C while it increased in H samples (p ≤ 0.06). Although in the absence of statistical significance, the relative abundance of Clostridiaceae decreased over time, especially in C samples, while that of Alcaligenaceae and Veillonellaceae tended to increase in C samples and decrease in H samples. A trend towards increased proportions in C samples was also observed for Bifidobacteriaceae, especially in vessels 2 and 3 (p = 0.09). Consistent data were obtained at the genus level for Bifidobacterium sp., Bacteroides, Prevotella, the Ruminococcaceae members Faecalibacterium and Oscillospira, and Klebsiella (Enterobacteriaceae family) (p ≤ 0.09) (Figure 2). In addition, significant differences were observed for Lachnospira and Akkermansia, whose proportions decreased overall in all vessels regardless of the faecal sample origin, Ruminococcus sp., whose amounts increased only in H samples, and Megasphaera (Veillonellaceae family), the amounts of which instead increased only in C samples (p ≤ 0.04). To gain insight into the functional variations of the C and H microbiome in the presence of Chlorella pyrenoidosa, we expanded the microbiota data analysis to better elucidate the functional profiles of the microbiota. The respective metagenomes were inferred from the phylogenetic profiles using PICRUSt2, we used the LEfSe algorithm in order to identify the MetaCyc pathways affected by the addition of the C. pyrenoidosa extract. According to this analysis, the faecal-derived microbial communities from H and C donors showed distinct functional profiles that were differentially modulated by C. pyrenoidosa. Samples were then clustered according to the abundance profile of the 238 most abundant identified pathways ( Figure 3A), showing an overall tendency towards segregation by origin (C vs. H). According to a LEfSe analysis ( Figure 3B), C samples showed a distinct functional profile at baseline compared to H samples, characterized by an enrichment in pathways involved in enterobactin biosynthesis, fatty acid beta-oxidation I and the superpathway of glycolysis, pyruvate dehydrogenase, TCA, and glyoxylate bypass. In the contrast, evolutionarily conserved biochemical pathways, such as UMP biosynthesis I, pyrimidine nucleobases salvage I, UTP and CTP de novo biosynthesis were underrepresented in C versus H samples. Among the common effects induced by coincubation with C. pyrenoidosa, we found an increased functional potential involved in guanosine and adenosine nucleotide metabolism (degradation and de novo biosynthesis), together with a reduced contribution of the superpathway of L-tryptophan biosynthesis. In contrast, an increased potential for the biosynthesis of thiazoles I (by facultative anaerobic bacteria), and a greater contribution of the superpathway of thiamine diphosphate biosynthesis II were exclusively observed in C samples. With specific regard to SCFAs production, in C donor, C. pyrenoidosa addition led to the increase of the pathway acetyl-CoA fermentation to butanoate II, and to the depletion of hexitol fermentation to lactate, formate, ethanol and acetate. Collectively, our results confirm the potential of C. pyrenoidosa to modulate the composition and functionality of the human intestinal microbiota with a differential impact closely related to the initial microbial configuration.

Differential Impact of C. pyrenoidosa on SCFAs Production in Healthy and Coeliac Gut Model Samples
Baseline relative abundances of faecal SCFAs in the Coeliac and healthy donors are reported in Figure 4 for all three vessels. In the healthy donor ( Figure 4H), a significant increase in butyrate and propionate relative concentration was seen in vessels 2 and 3 from baseline after the daily administration of C. pyrenoidosa extract (p < 0.001). In the Coeliac donor, a significant decrease in acetate was seen in vessel 3 after the addition of C. pyrenoidosa (p < 0.001), whereas a slight increase in acetate was seen in vessel 1. Furthermore, propionate was significantly increased in vessel 3 (p < 0.001) from SS1 to SS2 and butyrate was increased in vessel 3 between these time-points.

Discussion
The human intestine is more densely populated with microorganisms than any other organ, and is a site where they exert strong influences on human biology. With this in mind, the entire system of the gut microbiota of each individual can be pictured as a 'microbial complex ecosystem' within the intestine, where complicated interaction networks are established, which contributes to diverse mammalian processes including protective functions against pathogens. Here, we are presenting the first exploratory pilot study investigating the effect of C. pyrenoidosa in two in vitro continuous culture gut models, where one was inoculated with a healthy faecal sample and the other with a faecal sample from a Coeliac donor following a gluten-free diet (GFD).
The gut model systems represent an innovative technological platform that permits the modeling of the dynamic nature of the gastrointestinal tract, allowing the adaptation of various parameters, including dilution rate, retention time, pH and temperature, to meet and maintain optimal growth conditions [50][51][52].
This model has the potential to aid in providing an understanding of how the human gut microbiota is affected by C. pyrenoidosa.
The findings of this proof-of-concept study, including recognition of the limitations of the model, the use of a limited number of faecal samples suggest its future development as a tool for investigating the impact of nutrition and dietary supplementation on the gut microbiota.
Models, by definition, are not exact representations of reality. We acknowledge that a model of the human microbiota will never be an exact replica of the real bacterial community within the human large intestine. The model will, however, replicate aspects of the microbial community to allow interrogation and experimental manipulation in a more convenient and less invasive way.
The resemblance to reality is the major feature for a successful simulation and could be based both on the complexity of the mock microbiota or the environmental parameters of the reactors. Although these systems do not incorporate host factors, such as intestinal secretions, immunology, or absorption, it does offer an inexpensive and reliable tool for modeling the microbial ecology and activity of the colon [37,[53][54][55]. The system was run until the microbial communities reached a steady state (this typically took 8 turnovers or 16 days) that was confirmed after obtaining consistent SCFA measurements on three consecutive days [33]. Thereafter, a specific test substrate may be tested until a second steady state (SS2) is reached (again based on stable SCFA measurements), which is then compared to the original steady state or baseline. Therefore, the results of intervention at SS2 can be directly compared to that of SS1; in this way the use of a single system can still provide relevant results [56][57][58]. Moreover, to reduce the potential antagonistic effects from a pool of several donors, a faecal single inoculum from one donor can be justified as the microbiota of one donor might sense and defense the alien microbiota of a different donor.
However, the robustness of each model depends on the certainty that any observed effects on microbial composition and metabolic activity are due solely to the applied experimental treatment and not to gut microbial adaptation to the simulated gut environment. This presumes that a compositional stability of the in vitro gut ecosystem is in place prior to applying any experimental treatment and that this pseudo-steady state can reproducibly generate identical microbial communities which are the basis for comparison with other treatments [59][60][61].
There is an intricate relationship between diet, the gut microbiota and the metabolites it produces. Coeliac disease and the consequent GFD being followed have been shown to increase the risk of metabolic syndrome [62]. C. pyrenoidosa has demonstrated hypoglycaemic properties associated with microbiota changes in mice [27]. However, no studies to date have investigated C. pyrenoidosa and its potential effect on the human gut microbiota. This is the first pilot study to provide potentially new insight into the effects of C. pyrenoidosa, a unicellular green algae being shown much nutritional interest due to the presence of many substances that could be beneficial for human health.
A low relative abundance of Bifidobacterium sp. and higher levels of Proteobacteria, especially Enterobacteriaceae, have been found in subjects with Coeliac disease [63,64] and in healthy adults following a GFD [9]. The aberrant microbial profile observed in Coeliac disease has been associated with alterations in faecal metabolite production with higher concentrations of SCFAs noted in the faeces compared to healthy individuals [10,17]. The results of our study are promising and are suggesting distinct differences in the faecal microbial communities and production of SCFAs between the healthy and Coeliac donor at the baseline. In particular, the Coeliac donor, following a GFD, had significantly higher amounts of acetate, butyrate and propionate, and showed a microbial profile with the changes listed above in addition to increased proportions of Bacteroidaceae and Clostridiaceae, while reduced proportions of Alcaligenaceae and Veillonellaceae.
The addition of C. pyrenoidosa led to several changes in the intestinal microbial ecosystem, some common and others unique to the donor state. It is noteworthy that C. pyrenoidosa supplementation increased the abundance of Ruminococcaceae (especially the butyrate producers Faecalibacterium sp. and Oscillospira) and decreased the proportion of the mucus degrader Akkermansia [65], regardless of the donor and vessel. In addition, C. pyrenoidosa modified the microbial profiles of the Coeliac vessels towards a closer resemblance of the healthy donor, especially for Bifidobacteriaceae, Enterobacteriaceae and Bacteroidaceae. In a previous study, a significant positive correlation between Proteobacteria and the SCFAs, acetate and propionate was observed in children with Coeliac disease [62]. The decrease in Enterobacteriaceae found in the Coeliac gut model after the addition of C. pyrenoidosa, coupled with the significant decrease in acetate (and propionate in the proximal vessel), therefore suggests that Enterobacteriaceae could be responsible for elevated SCFA profiles seen in Coeliac disease. Similarly, the Bacteroidaceae family includes well-known producers of acetate and propionate, and its decrease in Coeliac samples after C. pyrenoidosa supplementation suggests its additional contribution to the altered SCFA profiles. Finally, we cannot exclude that acetate also decreases due to cross-feeding between Bacteroidetes and Firmicutes (especially including Megasphaera, whose increase was unique to Coeliac samples) [66], with increased butyrate production as found in the distal vessel of the Coeliac gut model after C. pyrenoidosa addition. Consistent with this, the inferred metagenomics analysis suggested reduced fermentation to acetate but increased conversion of acetyl-CoA into butyrate, particularly for the Coeliac donor. Other peculiar modifications in the functional profiles of the gut-derived microbial communities included, for both donors, a reduced potential for tryptophan biosynthesis, together with the increased contribution of pathways involved in nucleotides (i.e., guanosine and adenosine) degradation and biosynthesis. Taken together, these data with all the limitations of the study may indicate the propensity of C. pyrenoidosa to influence the fitness of the gut microbiota.
While Coeliac samples showed the greatest change in acetate concentration, significant increases in butyrate and propionate levels were observed for the healthy donor in vessels 2 and 3. The cell wall of C. pyrenoidosa is made up of various sugars, including rhamnose that can be used by Ruminococcus sp. to produce propionate via the propanediol pathway [67]. Interestingly, when rats were given C. pyrenoidosa, an increase in the relative abundance of Ruminococcus sp. was observed. This was negatively associated with fasting blood glucose, suggesting that the hypoglycaemic effect of C. pyrenoidosa could be due to its role in modulating the microbiome. Ruminococcus sp. are mostly found in herbivores and omnivores as they predominantly degrade cellulose [68]. An increase in Ruminococcus sp. was only observed in healthy donor samples after feeding with C. pyrenoidosa. In the healthy donor, substantial changes were also observed in the Bacteroidetes phylum with a decrease in Bacteroides coupled with an increase in Prevotella sp. in all 3 vessels after intervention. This suggests that C. pyrenoidosa can selectively improve the growth of Prevotella over Bacteroides sp. in certain microbiomes. Prevotella sp. produce the SCFA precursor succinate, which can be converted to propionate through the succinate pathway [69].
C. pyrenoidosa contains several different polysaccharides, cellulose, polyphenols and amino acids, and therefore the change seen in bacterial composition and SCFA production could be due to an individual component or a combined synergistic effect [23,66]. As evidenced, the Coeliac microbiota differs in composition and metabolite production from a healthy donor and consequently transforms in different ways in response to Chlorella fermentation. The role and effect of SCFAs on various systems in the body have been gaining recognition. Butyrate acts as a colonocyte energy source as well as improving gut barrier function, reducing inflammation and creating tolerance to commensal bacteria in the gut [69,70]. Increased butyrate production has also been shown to improve β-cell function [19]. The increase in Faecalibacterium sp. and butyrate production observed in both gut models could be beneficial for the host if it can be reproduced in vivo. Faecalibacterium sp. are dominant butyrate producers and can be used as an indicator of a healthy microbiota due to their beneficial effects on gut health, including anti-inflammatory and immunomodulatory properties [71,72], and insulin sensitivity [73]. As people with Coeliac disease usually have a lower abundance of F. prausnitzii, an increase in its proportions could improve gut inflammation and metabolic markers like insulin sensitivity. In all samples, the increase in Faecalibacterium sp. was observed especially in vessels representing the proximal and transverse colon. Butyrate producers are indeed rarely found in the distal part of the colon where the pH increases to 6.5 [14], indicating that fermentable fibre was abundant in this in vitro model or potentially acetate was converted to butyrate by the microbiota [67]. A stable isotope study in healthy individuals determined that 24% of acetate is converted into butyrate by the microbiota and was independent of butyrate-producing capacity [74].

Conclusions
Here, we have reported that the Coeliac-related microbiota differs in composition, predicted function and metabolite production from a healthy donor and consequently responds in different ways to C. pyrenoidosa in in vitro fermentation models. Despite the use of faecal samples from only one volunteer per type, our findings have shown that C. pyrenoidosa may lead to potential improvements in the intestinal microbial ecosystem, some common and others unique to the donor state.
In vitro gut fermentation systems have numerous challenges and limitations, especially the lack of host components. Furthermore, the reproducibility and functional stability of the human gut microbiota in in vitro gut fermentation models is frequently challenged and the subject of much criticism. Nevertheless, these models provide an innovative technological platform on which the greatest advantages are exhibited by the virtually unlimited experimental capacity sincethe experimentation is not restricted by ethical concerns. C. pyrenoidosa may thus manipulate the gut microbial population and alter metabolic activity towards a configuration that might confer health benefits to the host but still further research is required to extend our understanding of human intestinal health and the impact of C. pyrenoidosa.The preliminary data presented in this paper demonstrate all the strengths and limitations of this study and lay the foundations for the design of an in vivo human intervention study that needs to confirm the promising trends herein observed.
Supplementary Materials: Figure S1: A schematic chart of the validated three-stage continuous gut model system, Figure S2: Profiles of faecal-derived microbial communities from a Coeliac (n = 1) and healthy donor (n = 1) in fermentation experiments in the presence of C. pyrenoidosa extract (Sun Chlorella ® 'A' Powder). (A) Family-level relative abundance profiles atbaseline and steady state (i.e., average values at day 8,9,10). (B) Genus-level relativeabundance profiles are shown for each donor (Coeliac, C vs healthy, H), in the followingorder: at the baseline (B) and after C. pyrenoidosa addition(Sun Chlorella ® 'A' Powder)upon reaching steady state (S, day 8, 9 and 10), in vessel (V) 1, 2 and 3 of the gut models. *, unclassified amplicon sequence variants reported at higher taxonomic level, Figure S3: Family-level differences in the faecal-derived microbial communitiesbetween the Coeliac (n = 1) and healthy (n = 1) donor following fermentation with C.pyrenoidosa extract (Sun Chlorella ® 'A' Powder). Boxplots showing the relative abundance distribution of bacterial families differentially represented among the study groups (vessel (V) 1, 2 and 3, inoculated with faeces from Coeliac (C) vs healthy (H) donors, at baseline (B) and steady state (S) after C. pyrenoidosa supplementation (Sun Chlorella ® 'A' Powder) (p ≤ 0.1, Kruskal Wallis test), Table S1: Nutritional composition of C. pyrenoidosa (Sun Chlorella ® 'A' Powder).  Informed Consent Statement: Informed consent was obtained from all subjects involved in the study.

Data Availability Statement:
The datasets sequencing reads were deposited in the National Center for Biotechnology Information Sequence Read Archive (Bioproject PRJNA685154).

Conflicts of Interest:
The authors declare no conflict of interest.
Sample Availability: Samples of the compounds are available from the author on request.