Effect of cocoa’s theobromine on intestinal microbiota of rats

1 Cardiovascular Risk and Nutrition Research Group, REGICOR Study Group. Hospital del Mar Research Institute (IMIM), Barcelona, Spain. Spanish Biomedical Research Networking Centre–Physiopathology of Obesity and Nutrition (CIBERobn), Health Institute Carlos III, Madrid, Spain 2 Physiology Section, Department of Biochemistry and Physiology, Faculty of Pharmacy and Food Science, University of Barcelona (UB), Barcelona, Spain


Introduction 19
Although cocoa powder was initially used for medical purposes by Mesoamerican 20 civilizations [1], it is only recently that cocoa has come to be considered a valuable product 21 with healthy properties [2]. Among these beneficial effects, it has been reported that cocoa-22 enriched diets modulate the immune system and the gut microbiota [3]. In particular, a cocoa-23 enriched diet is able to attenuate secretory IgA (S-IgA) in several intestinal compartments [4-24 6] and also the IgA-coated bacteria proportion in the gut [5]. Moreover, a diet containing 10% 25 cocoa for 6 weeks modifies the intestinal microbiota composition in rats by decreasing the 26 proportion of the Bacteroides spp., the Staphylococcus spp., and the Clostridium histolyticum 27 subgroup [5], and thus causing a different short-chain fatty acid (SCFA) production [7]. 28 Similarly, a cocoa diet modulates the intestinal microbiota in orally sensitized rats, as 29 determined by a metagenomics analysis [8]. 30 Cocoa powder contains macronutrients, fiber, minerals, polyphenols (flavonoids, mainly 31 flavanols) and methylxanthines [9]. The most abundant xanthine found in cocoa is 32 theobromine, followed by caffeine. In fact, cocoa is the richest natural source of theobromine 33 [10,11]. While the effects of flavonoids present in cocoa have been thoroughly studied, less 34 attention has been paid to the presence of theobromine in cocoa. Even so, a few studies have 35 related its content to a variety of properties attributed to cocoa powder [10,12]. As 36 theobromine is able to reach the gut [13,14], we hypothesized that this methylxanthine could 37 contribute to the effects of cocoa intake on gut microbiota. Therefore, the purpose of the 38 present work was to establish the role of cocoa theobromine in the composition of gut 39 microbiota and fermentation products after cocoa consumption in rats.

Fecal samples collection and pre-analytical procedures 59
Fresh feces were collected at days 0, 8 and 15 and processed according to the specific 60 variables to be analyzed. Some fresh fecal samples were used to determine fecal pH, using a 61 surface electrode (Crison Instruments, S.A., Barcelona, Spain). The rest of the fecal samples 62 were directly frozen either at -20 ºC until the metagenomics analysis, the bacterial 63 characterization by fluorescence in situ hybridization, and the IgA-coated bacteria 64 quantification, or at -80 ºC until SCFA analysis. For these determinations, fecal homogenates 65 were later obtained following procedures previously described [5]. 66

Quantification of fecal microbiota by fluorescence in situ hybridization (FISH) 68 coupled to flow cytometry (FCM) 69
Quantification of representative groups of gut microbiota was carried out in feces from day 15 70 by FISH coupled to FCM (FISH-FCM), as described by Massot-Cladera et al. [15] were permeabilized with lysozyme (Serva, Heidelberg, Germany) prior to the hybridization 78 process [16]. All samples were hybridized at the specific probe hybridization temperature, as 79 described [15], and kept in the dark at 4 ºC overnight until FCM analysis. 80 To determine the total bacteria number, the samples were mixed with propidium iodide (PI, 81 1 mg/mL; Sigma-Aldrich, Madrid, Spain) prior to FCM analysis [5]. 82 83

Determination of the proportion of bacteria coated with IgA 84
Quantification of IgA-coated bacteria was carried out as previously described [15]. 85 86

Flow cytometry analysis 87
For FISH and IgA-coated bacteria quantification, FCM analysis was performed using a 88 FacsAria SORP sorter (BD, San José, CA, USA) as previously described [5]  Sulfuric acid (2.6 mM) was used as the eluent, and the flow rate was 0.5 mL/min. Peaks were 110 integrated using Agilent ChemStation software (Agilent Technologies, Oxford, UK). 111 Quantification of the samples was obtained through calibration curves of lactic, acetic, 112 propionic, butyric and formic acids (12.5-100 mM). Results were expressed as mM (for total 113 SCFA) and relative increases of the total and individual SCFA with respect to those values 114 found in the RF group. 115

Metagenomics analysis 117
DNA was extracted from two randomly selected samples from each group using a FastDNA® 118 SPIN Kit (MP Biomedicals, Solon, OH, USA) following the manufacturer's protocol. 119 Amplicons of 16S rDNA were purified and diluted in equal concentrations prior to 120 sequencing in Ion Torrent platforms by the Genetic Diagnostic Bioarray facilities (Bioarray, 121 Alicante, Spain), as previously described [8]. Briefly, a massive sequencing using the QIIME 122 software package v1.8.0. and USEARCH v7.0.1090 was carried out and the obtained 123 sequences were assigned into operational taxonomic units (OTUs; sequences that share ≥ 97% 124 similarity) using the UCLUST algorithm and Greengenes reference database (v13_8). Results 125 are expressed as absolute and relative abundance of phyla and number of detected species. 126 The bacterial species found among the experimental conditions, in common or not, were also 127 considered and represented through a Venn diagram. 128

Body weight and food intake 141
Although the initial body weight was similar among the groups, a statistically slower body 142 weight gain was observed during the study for both the CC and TB groups ( Figure 1A). The 143 measurement of the food intake revealed that, even though there was not lower food intake 144 when considering the relative amount per body weight (in all cases it was about 12 g/100g of 145 BW), lower absolute food intake per rat in both CC and TB groups than in RF group was 146 found from the first day of diet ( Figure 1B). 147 148

Gut bacterial populations by FISH-FCM 149
After 15 days of dietary intervention, significant differences in the gut microbiota 150 composition were observed (Figure 2). Concerning total bacteria counts, the CC diet caused 151 the elimination of higher number of bacteria per day than the RF diet. This increase could be 152 associated with the stool amount per day, which was higher in CC rats (3.07 g ± 0.11 g) than 153 that from RF rats (1.78 g ± 0.10 g) (P<0.05). Nevertheless, the total bacteria counts relative to 154 fecal weight from CC fed rats were similar to those in the RF group, whereas the TB group 155 showed lower counts than the other groups (P=0.021 and P=0.055 compared to the RF and 156 CC groups, respectively). 157 Regarding particular bacterial groups, both the CC and TB groups presented lower counts of 158 E. coli than the RF group, with the counts being even lower in the CC group than in the TB 159 one. The TB diet also led to significantly lower counts of Bifidobacterium spp., Streptococcus 160 spp. and Clostridium histolyticum-C. perfringens than the RF group. The decrease in the 161 Clostridium group, together with a reduction in the Bacteroidaceae-Prevotellaceae group, 162 was also significant compared to the CC group. As a result, the Firmicutes counts were lower 163 in feces from the TB group than those from RF rats (P=0.005). Even so, the F/B ratio was not 164 significantly modified in the feces of the studied groups. 165 166

Quantitative metagenomics analysis of gut bacterial populations 167
After the FISH-FCM analysis of microbiota, a metagenomics approach was carried out in 168 representative feces, in order to get an idea about the most modified species. The 169 metagenomics analysis allowed the relative abundance of the OTUs to be obtained ( Figure  170 3). 171 The CC group showed a higher proportion of the Firmicutes and a lower proportion of 172 Bacteroidetes phylum members than the RF group, which was associated with a significantly 173 higher F/B ratio than the RF and TB groups. The TB group displayed no changes in 174 Firmicutes and Bacteroidetes phyla but showed a higher proportion of the Tenericutes 175 phylum than the RF and CC groups. A further analysis also revealed changes in the relative 176 abundance of some species ( Table 2) Table 2). 187

Qualitative metagenomics analysis of gut bacterial populations 189
A total of 71, 80 and 73 different species were detected by metagenomics analysis in feces 190 from the RF, CC and TB groups, respectively (Supplementary Figure 1A). To determine the 191 relation among bacterial species present in each group, a Venn diagram was created 192 (Supplementary Figure 1B). From all the fecal-detected species, 68 were common to all 193 three studied groups. CC intake led to 11 new species; of these, four species were also found 194 in the TB group (species belonging to Bacteroidetes, Firmicutes and Proteobacteria phyla) 195 and seven were exclusively detected in the CC group (including species belonging to the 196 Actinobacteria, Cyanobacteria, Firmicutes and Proteobacteria phyla) ( Table 3). Only 197 "Candidatus Arthromitus" (Firmicutes phylum, Clostridia class) was found exclusively in the 198 TB group. Two species were only detected in the RF group, which belonged to the 199 Paraprevotellacea family (Bacteroidetes phylum) and Coprobacillus genus ( Table 3). In The TB diet led to higher pH values than those found after the RF and CC diets ( Figure 4A). 205 Fecal concentrations of lactic acid were not significantly affected by the experimental diets 206 In vitro, in vivo and clinical studies demonstrate that cocoa is able to modulate the growth of 229 gut microbiota [5,7,15,18]. Previous studies in rats show that the intake of cocoa-enriched 230 diets for at least three weeks modifies the intestinal microbiota pattern [5,7,15]. In the 231 present study, the ingestion of the cocoa diet for two weeks was not able to significantly 232 modify most of the bacterial groups analyzed by FISH-FCM, probably because of the shorter 233 length of this nutritional intervention. However, some changes were observed when 234 theobromine was ingested alone, indicating that theobromine by itself is able to directly or 235 indirectly modify gut microbial populations. The metagenomics analysis, even though it was 236 carried out in a small number of samples, allows to have an idea of particular genera and/or 237 species from gut microbiota modified by CC and TB diets and thus, by using both techniques 238 in a complementary manner, we obtained a wider approach of the gut microbiota changes. 239 According our FISH-FCM results, theobromine seems to exert an inhibitory effect on gut 240 microbiota, mainly on bacteria belonging to the Firmicutes phylum (Clostridium histolyticum-241 C. perfringens group and Streptococcus spp.), Bifidobacterium spp. and E.coli. The effect of 242 TB partially agrees with previously reported effects of a cocoa diet [5,7] on Clostridium spp. 243 and Streptococcus spp. In addition, according to the metagenomics analysis, the decrease in 244 Firmicutes could be associated with the disappearance of Ruminococcus flavefaciens, a 245 cellulolytic bacterium found to be increased by a flavonoid-enriched diet [19,20]. The 246 disappearance of R. flavefaciens after the TB diet, although it contained the same cellulose 247 amount as the RF and CC diets, may reflect a particular effect of theobromine on this species 248 that could be counteracted by the flavonoid content in the cocoa diet. Conversely, 249 theobromine alone seems to be able to increase other bacteria from the same family 250 (Erysipelotrichaceae). This family is decreased by a diet rich in flavonoids [21], which would 251 explain the current changes observed only in the TB group. Furthermore, from the two 252 samples analyzed in the TB group, it can be suggested that theobromine ingested alone 253 induced the presence of "Candidatus Arthomitus", another member of the Firmicutes phylum. 254 This is a segmented filamentous bacterium able to induce adaptive immune responses in the 255 gut [22], and it can adhere to the epithelial cells in the ileum and Peyer's patches, contributing 256 to the prevention of the colonization of the enteropathogenic E.coli O103, Salmonella, and 257 others [23,24]. 258 The cocoa diet, including theobromine, seems to induce the growth of bacteria belonging to 259 Firmicutes, according to the metagenomics analysis. This increase could be associated with a 260 higher relative abundance of one species from the Lachnospiraceae (Butyrivibrio genus) and 261 another from the Ruminococcaceae families, all of them belonging to the Clostridia class. 262 Moreover, the cocoa diet seems to lead to the appearance of new species belonging to the 263 Clostridia class (Dehalobacteriaceae spp., Roseburia faecis and SHA-98 spp), which is in 264 line with the increase of Lachnospiraceae, Clostridiales, and Ruminococcaceae found in pigs 265 fed a grape seed extract [25], and therefore, it could be related to an effect of the cocoa's 266 polyphenol content. 267 In the results of total Bacteroidetes phylum by FISH-FCM and metagenomics analyses, 268 discrepancies were observed, which could be due to the low representation of bacterial 269 members of this phylum in the first analysis and/or the low sample size in the second one. 270 Nevertheless, the metagenomics analysis allowed us to suggest changes inside this phylum. 271 For example, one species from the Paraprevotellaceae family disappeared with both diets, 272 and the cocoa diet decreased in particular the number of species belonging to the 273 Bacteroidales order (Bacteroides sp. and Bacteroides acidifaciens). The B. acidifaciens has 274 been described to be the predominant bacteria responsible for promoting IgA production in 275 the large intestine [26]. This agrees with our current results regarding IgA-coated bacteria and 276 with previous studies showing lower intestinal IgA with a cocoa diet [5,17,27]. On the other 277 hand, CC diet increased the relative abundance of Prevotella sp., which could be due to its 278 polyphenol content since higher numbers in the Prevotella group have been associated with 279 the daily consumption of red wine polyphenols [28]. 280 One important finding of our study is that theobromine (both in the CC and TB groups) 281 lowered the counts of E. coli. This agrees with the reported inhibitory effects of theobromine 282 on Gram-negative bacteria [29], suggesting an inhibitory effect on the growth of potential gut 283 pathogens. This inhibition was enhanced with the CC diet, suggesting the role of polyphenols 284 in this effect [30]. In the same phylum, Ralstonia sp. seems to appear due to the CC and TB 285 diets. Ralstonia sp. was formerly included in the Pseudomonas genera, which includes species 286 able to degrade methylxanthines [31,32]. Therefore, its presence may reflect the adaptation of 287 gut microbiota to diets rich in methylxanthines. 288 The impact of theobromine on gut microbiota was also patent in the Tenericutes phylum, 289 which increased almost fourfold with theobromine ingested alone. This was associated with a 290 higher number of bacteria belonging to the RF39 order (Mollicutes class). A study reported a 291 similar effect with the ingestion of cocoa for 4 weeks [8]. The absence of effects on 292 Tenericutes with the CC diet suggests that other cocoa compounds delayed the theobromine 293 effect on this phylum. 294 With regard to Actinobacteria, a prebiotic effect of cocoa polyphenols in humans [33] and of 295 cocoa fiber in rats [7] by increasing the counts of Bifidobacterium group has been reported. 296 As TB diet, but not CC diet, decreased the proportion of Bifidobacterium spp., it can be 297 suggested that theobromine is counteracting the prebiotic effects of cocoa fiber. However, the 298 metagenomics results suggested no changes in the relative abundance of Actinobacteria 299 species, either in the TB or CC diet, although it seems that CC diet leads the appearance of 300 one species of the Actinomycetales order. In line with these results, blueberries increased the 301 relative abundance of Actinomycetales order in rats, which allows us to suggest the role of 302 polyphenols in such an effect [34]. Finally, the appearance of one species of the Streptophyta 303 order (Cyanobacteria phylum) with the ingestion of cocoa, in agreement with the reported 304 effect of a CC diet for 4 weeks [8], must be related to the cocoa's polyphenol or fiber content. 305 Nevertheless, the role of such bacteria in the intestinal microbiota remains to be elucidated. 306 Overall, this study reveals the impact of theobromine on gut microbiota. The effects were 307 different depending on whether theobromine was ingested alone or when forming part of 308 cocoa, although few common characteristics were found. Some changes observed exclusively 309 in the TB group would have been due to the action of this methylxanthine, which were 310 counteracted by other cocoa compounds, such as fiber and polyphenols. Other changes in the 311 TB group agree with previous results reported with a longer CC diet, suggesting that these 312 other compounds included in the CC diet could delay the TB effect. The modifications 313 exclusively found in the CC group must be related to the cocoa's fiber or polyphenol content. 314 The effect of theobromine on gut microbiota has also been reflected by the changes observed 315 in SCFA in both theobromine-containing diets. The enhanced generation of SCFA was 316 mainly due to the butyric acid. Butyrate is considered the main energy source for colonocytes, 317 and is also important for the regulation of gene expression, the intestinal barrier and the 318 immune system, among others [35,36]. However, whereas butyric acid increased with both 319 diets, the increase in the proportion of acetic acid was only observed after cocoa ingestion. 320 This disagreement could be due to the fermentation of different substrates with both 321 interventions. After cocoa intake, SCFA would come directly from polyphenol and/or fiber 322 fermentation [7], whereas for the TB diet, changes in the generation of SCFA would be 323 indirectly due to the inhibition of some bacterial populations and thus contribute to enhancing 324 the amount of substrate available for other bacteria. The differential patterns in the SCFA 325 generated support the idea that the ingestion of theobromine alone or as part of cocoa has a 326 different impact on gut microbiota. Furthermore, the unexpected higher fecal pH when 327 theobromine was ingested alone deserves further studies focusing on microbial metabolites 328 which could explain the observed fecal pH changes. 329 The current results evidence that theobromine (both in the TB and CC diets) contributes to the 330 lower proportion of bacteria coated with IgA found after the cocoa diet, in line with previous 331 results [5,7,15]. As rats fed the CC diet even showed a lower proportion of IgA-coated 332 bacteria, the combination of cocoa polyphenols with theobromine in the CC diet could have 333 an additive or a synergistic effect on reducing their proportion. On the other hand, the effect 334 of cocoa fiber must be discarded because it was associated with an increase in the percentage 335 of IgA-coated bacteria [7]. 336 Results regarding body weight suggest that theobromine present in cocoa was the main reason 337 for a slower body weight increase produced by the 10% cocoa diet. In fact, there was a lower 338 food intake per animal already in the first day of diet, which could affect the body weight 339 increase and it can also influence gut microbiota. On the other hand, body growth could be 340 affected by TB influence on metabolism. In this sense, it has been demonstrated that caffeine 341 has a stimulatory effect on thermogenesis [37] and has been associated with bone mass loss 342 [38]. 343 In conclusion, here we demonstrate that cocoa theobromine plays a relevant role in some 344 effects related to cocoa intake, such as lower body weight increase and the proportion of IgA-345 coated bacteria. In addition, theobromine modifies gut microbiota, although other cocoa 346 compounds -such as cocoa polyphenols or fiber-also act on the intestinal bacteria, 347 attenuating or enhancing the theobromine effects, that overall leads to the global effect of 348 cocoa on microbiota which differs from that of each particular cocoa component.     The diagram shows the absolute number of detected species that belonged to each of the individual nutritional interventions, the detected species common to each pair of groups and the detected species in common to all the three nutritional interventions (in the center of the representation). RF, reference group; CC, group fed diet containing 10% cocoa; TB, group fed diet containing 0.25 % theobromine.