**3. Results**

#### *3.1. Dietary Habits and Correlations with the Gut Microbiota in RRMS Patients*

Since diet plays a significant role in gu<sup>t</sup> microbiota composition, we used an FFQ in order to detect differences in dietary habits between RRMS patients and healthy controls. The interviewees reported daily consumption of vegetables (patients (Pt) = 77.8%; controls (Ct) = 61.1%), fruits (Pt = 44.4%; Ct = 27.8%), carbohydrates (Pt = 61.1%; Ct = 61.1%), animalderived proteins (Pt = 50.0%; Ct = 27.8%), saturated/trans fats (Pt = 5.5%; Ct = 16.7%), dairy products (Pt = 55.6%; Ct = 72.2%), and canned products (Pt = 0.0%; Ct = 5.5%). We observed significant differences (*p* < 0.05) among intake of vegetables, fruits, carbohydrates, animalderived proteins, and dairy products when we compared patients and controls. Table 2 summarizes the data obtained from the FFQ, with the frequencies of food consumption per patient and controls and the *p* values.

To find correlations between dietary habits and gu<sup>t</sup> microbiota composition in RRMS patients, we used the consumption frequencies and the reads percentages detected in stool samples from RRMS patients. We detected significant moderate/strong correlation between vegetables consumption by patients and relative abundance of *Roseburia* (*p* = 0.010; *r* = −0.60). We also found negative correlations between animal-derived protein intake and relative abundance of Verrucomicrobiae/Verrucomicrobiales (*p* = 0.041; *r* = −0.50) and *Bacteroides vulgatus* (*p* = 0.014; *r* = −0.58).

#### *3.2. Detection of Intestinal Dysbiosis and Prevalence of Gram-Negative Bacteria in RRMS Patients*

For the purpose to detect intestinal dysbiosis in RRMS patients receiving DMTs, we sequenced the V3/V4 regions from bacterial 16S and determined the alpha and beta diversities by using the annotated operational taxonomic units (OTUs). According to the rarefaction curves, we observed no significant differences (*p* = 0.38) in richness and evenness between samples obtained from RRMS patients and controls (Figure 1A,B). However, when we used the unweighted UniFrac metric with Bonferroni correction, we detected a significant difference (*p* = 0.01) between microbial communities found in RRMS patients and controls (Figure 1D). Figure 1C shows the PcoA plot regarding the weighted UniFrac metric with Bonferroni correction.


**Table 2.** Description of the dietary habits of multiple sclerosis patients and controls.

The consumption of dairy products by patients correlated with the presence of the Bacteroidetes phylum (*p* = 0.015; *r* = −0.58), Bacteroidia/Bacteroidales (*p* = 0.011; *r* = −0.60), Bacteroidaceae/*Bacteroides* (*p* = 0.016; *r* = −0.57), *Bacteroides rodentium* (*p* = 0.044; *r* = −0.49), and *Bacteroides uniformis* (*p* = 0.049; *r* = −0.48). Furthermore, we reported a positive correlation between saturated/trans fat consumption and the abundance of Firmicutes (*p* = 0.044; *r* = 0.49), Clostridia (*p* = 0.039; *r* = 0.50), and Clostridiales (*p* = 0.035; *r* = 0.51).

> To compare the microbiota composition in treated RRMS patients and controls, we sequenced the bacterial 16S in stool samples and analyzed specific bacterial groups by real-time PCR. The prevalent phyla in RRMS patients were Firmicutes (patient reads (Pr) = 43.78%; control reads (Cr) = 50.12%) and Bacteroidetes (Pr = 30.52%; Cr = 14.47%), and the prevalent classes were Clostridia (Pr = 39.29%; Cr = 41.15%) e Bacteroidia (Pr = 25.96%; Cr = 11.99%) (Figure 2A,B). The prevalent orders were Clostridiales (Pr = 35.80%; Cr = 37.16%) and Bacteroidales (Pr = 25.96%; Cr = 11.99%), and the prevalent families were Bacteroidaceae

(Pr = 18.86%; Cr = 9.25%), Ruminococcaceae (Pr = 11.35%; Cr = 16.74%), and Lachnospiraceae (Pr = 10.19%; Cr = 6.24%) (Figure 2C,D). The prevalent genera in RRMS patients were *Bacteroides* (Pr = 18.86%; Cr = 9.25%), *Akkermansia* (Pr = 7.35%; Cr = 6.95%), *Blautia* (Pr = 5.18%; Cr = 2.16%), and *Faecalibacterium* (Pr = 4.31%; Cr = 9.91%). The prevalent species in stool samples from RRMS patients were *Akkermansia muciniphila* (Pr = 7.35%; Cr=7.27%), *Bacteroides vulgatus* (Pr = 4.68%; Cr = 1.07%), *Methanobrevibacter smithii* (Pr = 2.99%; Cr = 10.01%)*, Bacteroides rodentium* (Pr = 1.95%; Cr = 3.43%), *Blautia coccoides* (Pr = 1.33%; Cr = 2.05%), and *Prevotella copri* (Pr = 1.28%; Cr = 1.09%) (Figure 2E,F). Additionally, we found significant differences (*p* < 0.05) in the relative abundances of Bacteroidetes and Actinobacteria phyla, Bacteroidia, Gammaproteobacteria and Actinobacteriia classes, Bacteroidales, Lactobacillales, and Bifidobacteriales orders, Bacteroidaceae, Ruminococcaceae, Flavobacteriaceae, Porphyromonadaceae, and Bifidobacteriaceae families, *Bacteroides*, *Flavobacterium*, *Parabacteroides*, *Streptococcus, Bifidobacterium* genera, *Bacteroides vulgatus* and *Bifidobacterium stercoris* between samples derived from patients and controls (Figure 2). Interestingly, the *Parabacteroides* genus (Pr = 1.31%; Cr = 0%) was detected only in stool samples from RRMS patients, and the *Bifidobacterium* (Pr = 0%; Cr = 4.59%) and *Enterobacter* (Pr = 0%; Cr = 1.12%) genera were found exclusively in stool samples from controls (Figure 2E).

**Figure 1.** Alpha and beta diversity in the gu<sup>t</sup> microbiota of RRMS patients receiving DMTs and that of healthy controls. Rarefaction curves are a representation of species richness for a given number of individual samples: (**A**) Observed and (**B**) Chao 1-estimated OTUs. Principal component analysis (PcoA) is a transformation of weighted or unweighted Unifrac distance, a pair-wise distance between samples based on the calculation of the shared branches of the phylogenetic tree of the representative rRNA genes from OTUs present in at least one sample: (**C**) PcoA plot with weighted and (**D**) unweighted UniFrac metric with Bonferroni's correction.

**Figure 2.** Relative abundances of bacterial taxa in stool samples from RRMS patients and controls. Predominant phyla (**A**), classes (**B**), orders (**C**), families (**D**), genera (**E**), and species (**F**). Bars represent the reads percentages found in metagenomics analyses. \* *p* < 0.05.

Regarding the characterization of the gu<sup>t</sup> microbiota by real-time PCR, we observed similar relative expression units (*p* > 0.05) of *Bacteroides, Lactobacillus, Prevotella,* and *Roseburia* species when we compared patients' and controls' samples (Figure 3). In contrast, we found a significant decrease (*p* = 0.036) in relative expression units of *Bifidobacterium* species detected in stool samples derived from RRMS patients (median = 239.7) compared to controls (median = 7791) (Figure 3B). Moreover, when we classified MS patients based on

different DMTs, there were no significant differences (*p* > 0.05) in relative expression units of *Bacteroides*, *Bifidobacterium*, *Clostridium coccoides*, *Clostridium coccoides-Eubacterium rectale*, *Clostridium leptum*, *Lactobacillus*, *Prevotella*, and *Roseburia* in stool samples from MS patients.

**Figure 3.** Relative abundance of bacterial community in stool samples from RRMS patients and controls. (**A**) *Bacteroides* species, (**B**) *Bifidobacterium* species, (**C**) *Lactobacillus* species, (**D**) *Prevotella* species, and (**E**) *Roseburia* species. Bars represent the median with interquartile range of relative expression units (REU) per 200 mg of stool.

#### *3.3. Detection of Decreased Pro-Inflammatory IL-6 Cytokine in MS Patients*

To determine the serum concentrations of anti- and pro-inflammatory cytokines in RRMS patients, we quantified IL-2, IL-4, IL-6, IL-10, IL-17A, IFN-gamma, and TNF by cytometric bead array. There were no significant differences (*p* < 0.05) in the concentrations of IL-2, IL-4, IL-10, IL-17A, TNF in patients' serum (mean ± standard error IL-2: 0.1867 ± 0.0687 pg/mL; IL-4: 0.3239 ± 0.0743 pg/mL; IL-10: 0.265 ± 0.0429 pg/mL; IL-17A: 2.708 ± 0.8544 pg/mL; TNF: 1.138 ± 0.1372 pg/mL; IFN-gamma: 0.4222 ± 0.1076 pg/mL) when compared with control group (IL-2: 0.4294 ± 0.4051 pg/mL; 233IL-4: 0.2839 ± 0.2244 pg/mL; IL-10: 0.2422 ± 0.18 pg/mL; IL-17A: 4.796 ± 1.43 pg/mL; TNF: 0.7572 ± 0.4383 pg/mL; IFN-gamma: 0.5028 ± 0.158 pg/mL) (Figure 4A–G). IL-6 serum concentrations were decreased (*p* = 0.003) in RRMS patients (0.7261 ± 0.1244 pg/mL) when compared with controls (1.242 ± 0.1601 pg/mL) (Figure 4C). In addition, IL-6 concentrations inversely correlated with Clostridiaceae family members (*p* = 0.001; *r* = −0.70), and TNF levels correlated with Actinobacteria (*p* = 0.025; *r* = 0.48) and *Bacteroides vulgatus* (*p* = 0.001; *r* = −0.70) (Figure 5A–C).

**Figure 4.** Cytokine profile in treated RRMS patients and control subjects. Serum concentrations of (**A**) IL-2, (**B**) IL-4, (**C**) IL-6, (**D**) IL-10, (**E**) IL-17A, (**F**) IFN-gamma, and (**G**) TNF. Statistical analyses were performed by the Mann–Whitney test. Significance was set at *p* < 0.05.

**Figure 5.** Correlations among relative abundances of bacterial taxa and serum concentrations of inflammatory cytokines. (**A**) Negative correlation between relative abundance of Clostridiaceae and IL-6 concentrations in RRMS patients; (**B**) Positive correlation between relative abundance of Actinobacteria and TNF concentrations; (**C**) Positive correlation between *Bacteroides vulgatus* and TNF concentrations. Statistical analyses were performed by Spearman's test. Significance was set at *p* < 0.05.

#### *3.4. Detection of Increased Intestinal Permeability in RRMS Patients*

In order to find whether RRMS patients presented increased intestinal permeability, since alterations in the gu<sup>t</sup> microbiota were detected, we evaluated the serum concentrations of zonulin. Zonulin levels were significantly increased (*p* = 0.017) in MS patients' samples (mean ± standard error: 27.13 ± 2.08 ng/mL) when compared with controls' (mean ± standard error: 19.01 ± 2.98 pg/mL) (Figure 6A). Besides that, zonulin concentrations positively correlated with disease duration (*p* = 0.025; *r* = 0.55; Figure 6B) and with the relative abundance of Bacilli class members (*p* = 0.045; *r* = 0.49; Figure 6C) in MS patients.

**Figure 6.** Zonulin concentrations and correlations with clinical data and gu<sup>t</sup> microbiota. (**A**) Serum zonulin concentrations in RRMS patients and controls (CTRL); (**B**) Positive correlation between zonulin concentrations and disease duration; (**C**) Positive correlation between zonulin concentrations and relative abundance of Bacilli class members.
