High dose saccharin supplementation does not induce gut microbiota dysbiosis or glucose intolerance in healthy humans and mice

43 Background: Non-caloric artificial sweeteners (NCAS) are widely used as a substitute for dietary 44 sugars to control body weight or glycemia. Paradoxically, saccharin and other NCAS have been 45 reported to induce glucose intolerance in mice fed a high-fat diet and in a subset of humans by 46 directly inducing unfavorable changes in gut microbiota. These findings have raised concerns 47 about NCAS and called into question their broad use. Whether these results can be generalized 48 to healthy populations consuming conventional diets is unknown. It is also unclear how different 49 NCAS, that do not share a common chemical structure, can produce identical direct effects on gut microbiota. A common feature of all NCAS is their strong affinity for sweet taste receptors 51 (STRs) which are expressed in the intestine. However, their role in mediating NCAS-induced 52 effects has not been addressed. 53 Results: We conducted a double-blind, placebo-controlled, parallel arm study exploring the effects 54 of saccharin on gut microbiota and glucose tolerance in healthy men and women. Participants 55 were randomized to placebo, saccharin, lactisole (STR inhibitor), or saccharin with lactisole 56 administered in capsules twice daily to achieve the maximum acceptable daily intake for two 57 weeks. In parallel, we performed a ten-week study administering high-dose saccharin in the 58 drinking water of chow-fed mice with genetic ablation of STRs (T1R2-KO) and wild-type (WT) 59 littermate controls. In humans and mice alike, none of the interventions affected glucose or 60 hormonal responses to a glucose tolerance test, nor ex vivo glucose absorption in mice. Similarly, 61 saccharin supplementation did not alter microbial diversity or abundance at any taxonomic level 62 in humans or mice. No treatment effects were also noted in readouts of microbial activity such as 63 fecal metabolites or short chain fatty acids (SCFA). However, compared to WT, T1R2-KO mice 64 were protected from age-dependent increases in fecal SCFA and the development of glucose 65 intolerance. 66 Conclusions: In the absence of other permissive conditions, short-term saccharin consumption at 67 the maximum recommended levels does not alter gut microbiota or induce glucose intolerance 68 and, thus, it may be safely included in the diet of healthy individuals who wish to substitute sugars 69 for weight management or caloric control. caloric control. Our findings also do not contradict previous

6 are summarized in Supp. Table.1. At baseline, no differences in basic anthropometric and 128 metabolic parameters were noted between treatment groups (Table.1). The remaining 129 participants of all groups met the expected dose requirement for the treatment period (Supp. 130 Table.2). No adverse effects of the treatments were reported. 131

Glucose tolerance and ex vivo intestinal function 132
Two weeks of continuous saccharin supplementation at a dose equal to ADI [11] did not alter 133 glucose responses to a 75g oral glucose tolerance test (OGTT) among participants ( Figure.1A). 134 To test for possible delayed effects of the treatment, we assessed glucose tolerance after a two-135 week recovery period during which all groups received placebo. No differences in glucose 136 excursions were observed between the post-treatment and recovery (washout) periods (ANCOVA 137 repeated measures p=0.99; Supp. Figure.2). Similar to glucose responses, plasma excursions 138 of insulin, C-peptide, glucagon or glucagon-like peptide 1 (GLP-1) were not different between 139 groups with treatment or after the wash out period ( Figure.1B-E and Table.2) (Supp. Figure.2). 140 Next, we addressed the long-term effects of high-dose saccharin supplementation on glucose 141 tolerance in mice and specifically explored the role of NCAS sensing by intestinal STRs. Ad libitum 142 chow-fed WT and T1R2-KO mice were supplemented with saccharin in the drinking water for 10 143 weeks to achieve daily consumption equal to 4 times the human ADI adjusted for mouse body 144 surface area [12]. The actual saccharin consumption slightly exceeded the target consumption 145 for both genotypes (Supp. Figure.3A), but without affecting food intake (Supp. Figure.3B). 146 Saccharin consumption did not cause differences in body weight gain compared to water alone 147 in either genotype (Supp. Figure.3C). As we observed in humans, saccharin treatment had no 148 effect on glucose tolerance in WT or T1R2 mice assessed after two or ten weeks of treatment 149 ( Figure.2A-B). However, we did observe age-dependent increases in intra-gastric GTT (IGGTT) 150 responses in WT mice. Notably, these effects were absent in T1R2-KO mice, which also had 151 reduced IGGTT responses compared to WT littermates [13] ( Figure.2A).

7
Although saccharin treatment was unsuccessful in modifying IGGTT responses, it may have 153 induced localized intestinal changes that may contribute to long-term metabolic susceptibility. To 154 address this possibility we assessed post-treatment ex vivo glucose transport using intact 155 intestines (Ussing chamber) and found no effect of saccharin supplementation in the transport of 156 the non-metabolizable glucose analog 3-O-methyl-glucose (3-OMG) ( Figure.2C), but we 157 observed decreased glucose transport in T1R2-KO intestines, consistent with the IGGTT data 158 and our previous studies [13]. In addition, saccharin supplementation did not change the 159 expression of glucose transporters or of STRs (Supp. treatments when we performed pre-post analysis of variance to account for between subjects 175 differences in microbial communities within a treatment group (Two-way ANOVA; p>0.05 for each 176 treatment). In mice, we did not observe a genotype or gender effect on gut microbiota composition 177 or taxonomic diversity at baseline (Supplemental Figure. Similar to humans, no clustering effect was observed for post-treatment groups by multivariate 181 analysis ( Figure.3F). Also, no changes were noted in within-subject microbial abundances, as 182 assessed by pre-post analysis of variance (Two-way ANOVA; p>0.05 for each treatment at family 183 level). 184

Fecal metabolomics 185
Although the interventions did not induce substantial shifts in the gut microbial communities in 186 either humans or mice, we tested whether saccharin might have instead altered the intestine's 187 metabolic profile by performing untargeted metabolomics of fecal samples. and identification (S-plot analysis) revealed that the presence of saccharin itself in the feces was 197 the only metabolite responsible for the clustering effect ( Figure 4D-E). Hence, removal of 198 saccharin from the model abolished the clustering effects, eliminating any independent treatment 199 effects on the fecal metabolome ( Figure 4F). In addition, we specifically assessed fecal glucose 200 content in all human and mouse samples, but found no treatment or genotype differences 201 excluding major defects in glucose absorption. (Supplemental Figure.4I-J). Finally, we 202 independently measured short-chain fatty acids (SCFA) in feces and found no treatment effect in 203 human participants ( Figure.4G). However, we noticed an age-dependent increase in SCFA in 204 WT mice, but these effects were absent in T1R2-KO mice ( Figure.4H). 205

Discussion 206
Concerns and confusion about the general safety of NCAS can be attributed, in part, to the 207 amount and quality of the available evidence. A critical knowledge gap has been the lack of 208 interventional studies designed to rigorously investigate whether consumption of NCAS per se is 209 sufficient to cause deterioration of glucose homeostasis in healthy individuals. Using a 210 randomized, placebo-controlled design, we clearly show that daily consumption of saccharin at 211 maximum ADI for 2 weeks is inadequate to alter fecal microbiota composition and metabolites or 212 affect glucose tolerance in healthy participants. Notably, identical results were recapitulated in 213 chow fed mice that consumed saccharin equal to 4-times the human ADI for 10 weeks. 214 Over the past 30 years, a number of cross-sectional and observational studies have reported 215 positive correlations between NCAS consumption and outcomes such as metabolic syndrome 216 and weight gain (reviewed in [6,15]). These findings have alarmed both consumers and health 217 care professionals, despite the fact that health and other lifestyle-related characteristics of the 218 populations might have influenced these outcomes through reverse causality or residual 219 confounding. For instance, positive associations between NCAS consumption (estimated from 220 soda consumption) and metabolic syndrome were noted in a recent cross-sectional study [16], 221 but after careful adjustment for age, dietary quality and physical activity these associations 222 disappeared. A paucity of well-controlled interventional studies has also contributed to confusion 223 in the field. 224 In this regard, an elegant report by Suez et al (2014) [8] appeared to establish a causative 225 relationship between the consumption of NCAS (i.e. saccharin) and the development of glucose 226 intolerance through direct modification of gut microbiota composition. This report, mainly 227 conducted in mice, revived concerns about the use of NCAS and long-term health implications. 228 However, in this study only 3 out of the 7 human participants developed glucose intolerance in 229 response to 3-7 days of NCAS use. In contrast, we exposed 23 healthy lean participants in 2 230 separate cohorts (Saccharin, or Saccharin plus Lactisole groups) to 15 days of daily saccharin 231 consumption at the maximum ADI levels. None of the treated subjects, who were also not regular 232 NCAS users, developed glucose intolerance or showed altered endocrine responses during an 233 OGTT, but it is reasonable to speculate that the treatment effects of NCAS supplementation may 234 be delayed. However, OGTT responses remained unaltered after 2 additional weeks of placebo 235 treatment following the main intervention. In agreement with our findings in healthy lean 236 participants, 12 weeks of NCAS supplementation using sweetened beverages did not change 237 glucose tolerance in healthy overweight or obese individuals [17]. This suggests that the 238 development of glucose intolerance in response to NCAS use is independent of obesity status 239 per se and may instead require the presence of other, yet unknown, underlying risk factors. For 240 instance, the saccharin responders in Suez et al (2004) [8] had different baseline microbiome 241 compared to non-responders; a factor shown to confound outcomes of dietary interventions [18]. 242 We circumvented these issues since all participants contained similar basal gut microbiota 243 composition. This similarity is partially due to the enforcement of comprehensive inclusion and 244 exclusion criteria including dietary habits that were consistent with the typical macronutrient intake 245 of healthy US adults. Thus, saccharin treatment did not alter gut microbiota composition 246 compared to other interventions, but also did not induce any relative changes in treated 247 participants (i.e. within-subject pre-post analyses). Although gut microbiota abundances were 248 mainly unaltered by the treatments, marginal shifts in some species or changes in microbial 249 metabolism [19] might predispose the host to dysbiosis [20]. This effect is unlikely, as neither 250 saccharin nor any other treatment significantly altered fecal metabolite profiles or induced any 251 relative changes in treated participants. The microbiota-induced pathophysiology is often linked 252 to SCFA changes in microbial production and availability [21], but saccharin did not alter fecal 253 SCFA in humans and mice alike, mirroring the null effect observed in untargeted metabolite 254 profiles. However, the age-dependent increase in SCFA in the WT mice is consistent with the 255 age-dependent development of glucose intolerance in the same mice and it is in agreement with 256 findings showing that increased fecal SCFA correlate with age, obesity and metabolic 257 dysregulation [22]. Notably, in T1R2 mice the absence of SCFA increases with aging correlates 258 with the absence of glucose intolerance. These associations require further investigation since suggesting that saccharin bioavailability was not a limiting factor for gut microbes in our 268 population. Nevertheless, even in high saccharin consumers (>90 th percentile) the average intake 269 is only about 2mg/kg/d, a minor fraction of the ADI (5mg/kg/d) [26]. Taken together with our 270 findings, it is reasonable to suggest that typical saccharin use is unlikely to induce adverse 271 alterations in the gut microbiota of the general healthy consumer. 272 On the other hand, the absence of effects following short-term NCAS supplementation in our 273 study cannot exclude the possibility that the deleterious consequences of NCAS consumption 274 might require higher doses and/or longer durations. Because of safety limitations regarding the 275 dose and duration of treatment involving human participants, we supplemented C57Bl\6J mice 276 with saccharin for 10 weeks using a target dose that exceeded the human ADI by 4 times adjusted 277 for mouse body surface area to discern possible mechanistic effects that might have not been 278 apparent in the human study. Surprisingly, but in agreement with the human findings, glucose 279 tolerance, gut microbiota composition and fecal metabolite profiles were unaffected by the higher 280 saccharin dose and extended treatment in chow fed mice. As in humans, saccharin appeared in 281 the feces of almost all treated mice, confirming saccharin's bioavailability for microbial 282 metabolism. In contrast to our findings, mice fed chow diet and supplemented with 10% solution 283 of commercial saccharin, which contained 95% glucose by mass, or mice fed high-fat and 284 supplemented with pure saccharin, developed glucose intolerance mediated by unfavorable 285 changes in gut microbiota [8]. Similarly, 12 weeks of saccharin supplementation in chow-fed 286 ICR/HaJ mice caused marginal glucose intolerance, but responsive mice also showed increased 287 food intake and weight gain [27]. Our saccharin-fed mice consumed similar amount of chow and 288 experienced the same age-related increases in body weight compared to water control 289 littermates. Taken together, these findings suggest that high saccharin consumption may exert In the presence of other permissive dietary factors, saccharin may be able to cause glucose 299 intolerance by directly altering gut microbiota. However, it is still perplexing how other NCAS, such 300 as aspartame or sucralose, can demonstrate identical effects [8] considering that they share no 301 structural similarities to suggest their intersection of common pathways of microbial metabolism. 302 NCAS are bona fide ligands for STRs, so it is reasonable to speculate that if consumption of all 303 NCAS leads to specific metabolic effects, such as glucose intolerance, a common underlying 304 mechanism should exist. Thus, a secondary aim of our studies was to test whether STR partially 305 mediate the effects of NCAS feeding. Participants that consumed lactisole, a human specific 306 inhibitor of STRs, or mice with genetic ablation of STRs had no differences in glucose tolerance 307 or gut microbiota in response to saccharin feeding, which suggests that in the absence of a 308 primary effect of NCAS consumption the role of STR signaling is not apparent. Nevertheless, we 309 observed a genotype effect in mice independent of treatment. T1R2-KO mice had reduced IGGTT 310 responses and ex vivo glucose transport compared to WT littermates, confirming our previous 311 findings [13]. Interestingly, although WT mice developed mild age-related glucose intolerance, 312 T1R2-KO mice were resistant to these effects. We previously showed that T1R2-KO mice were 313 also protected against metabolic derangements induced by high-fat diet [29], suggesting that STR 314 signaling may be involved in age-and diet-dependent deterioration of glucose homeostasis. 315 Although we report no adverse effects of short-term NCAS consumption on the glycemic 316 responses in healthy lean participants and mice, our study has some notable limitations. First, we 317 tested saccharin as a representative NCAS but it is unknown whether our results can be 318 extrapolated to all NCAS. Since the six FDA-approved NCAS have different metabolic fates and 319 bioavailability [30], potential health implications relevant to their consumption must be addressed 320 separately. Second, the duration of treatment in humans was limited to two weeks, which may 321 have been inadequate to induce physiological effects in a healthy young population. This does 322 not preclude the possibility that years of chronic high use of saccharin or of other NCAS may 323 eventually lead to slow maladaptive responses or predispose consumers to the development of 324 disease. Third, we focused on a number of outcomes based on previous reports and specific 325 objectives. Thus, saccharin might have altered other physiological parameters that, if measured, 326 may have helped identify other adverse health conditions linked to NCAS consumption. 327

Conclusions 328
We clearly show that short-term saccharin supplementation per se is insufficient to alter gut 329 microbiota or induce glucose intolerance in apparently healthy humans and mice on conventional 330 diets. The clinical significance of our findings should not be underestimated since it emphasizes 331 that the recommended saccharin use is safe for healthy consumers that wish to substitute dietary 332 sugars for weight management or caloric control. Our findings also do not contradict previous 333 reports showing harmful effects of saccharin. On the contrary, together they highlight that the 334 potential harmful effects of chronic NCAS use are likely contingent upon permissive physiological 335 or lifestyle features in vulnerable populations. Therefore, for individuals who lack these 336 characteristics -such as those studied here -consumption of NCAS is likely innocuous, but for 337 susceptible populations NCAS use may be contraindicated. Consequently, it is imperative that 338 future studies concentrate in isolating and identifying the critical underlying pathophysiology or 339 conditions that may render specific NCAS as harmful. 340

Human Studies 343
We conducted a randomized, placebo-controlled, double-blind, interventional study 344 Healthy men and women 18-45 years of age were recruited from volunteer lists and by social 351 media to participate in the study. Only subjects who consumed less than a can of diet beverage 352 or a spoonful of NCASs weekly (or the equivalent from foods) during the past month, whose body 353 mass index (BMI) ≤ 25.0 kg/m 2 , and who were weight stable (± 3 kg) during the 3 months prior to 354 enrollment were included. Subjects with acute or chronic medical conditions that would 355 contraindicate participation in the research testing or that were taking medications that could 356 potentially affect metabolic function were excluded. Specifically, individuals with diabetes, 357 bariatric surgery, inflammatory bowel disease or a history of malabsorption and pregnant or 358 nursing women were excluded. A complete list of inclusion and exclusion criteria are available 359 (Supp. methods). 360 Participants were randomized into four treatment groups and were instructed to consume 361 capsules containing: 1) Pulp filler/placebo (1000mg/day 1) Sodium saccharin (400mg/day), 3) 362 Lactisole (670mg/day) or 4) Sodium saccharin (400mg/day) + lactisole (670mg/day) twice daily 363 for two weeks. A sealed envelope with the randomization allocation sequence (SAS procedure 364 PROC PLAN) was given to the pharmacist who prepared and provided the appropriate treatment. 365 The pharmacist was the only un-blinded member of the study. Diet-related instructions were 366 provided to avoid additional consumption of NCASs for the duration of the study. Participants 367 were asked to give blood samples and stool samples during their visits. The investigation agents, 368 saccharin and lactisole, were formulated in capsules for oral delivery (Compounding Pharmacy, 369 Advent-Health) at the maximum acceptable daily intake (ADI) [11]. (300mL) to assess glucose tolerance (i.e. OGTT); 8) OGTT blood sampling (t = 10, 20, 30, 45, 376 60, 90, 120, 180 min); 9) Participants were provided with 2-week supply of treatment capsules 377 and were instructed to consume 2 capsules a day (morning and evening) with water until the night 378 before their next visit. At visit 2 (post-treatment), the same procedures as listed above were 379 repeated. All groups were subjected to additional 2 weeks of pulp filler/placebo capsule treatment 380 (blinded for participants) and at visit 3 (recovery) the same procedures were performed.  Biosystems). Each sample was diluted to 10 nM, and equal volume from each sample was pooled. 420 The quality of the library was checked by Bio-Rad Experion bioanalyzer (Bio-Rad). Illumina MiSeq 421 platform was used for sequencing (Novogene Bioinformatics Technology Co., Ltd). 422 Raw FASTQ sequences were quality checked with FastQC v0.11.5. Raw sequences were 423 trimmed with 'cutadapt' v2.6 to remove low quality bases and adaptor sequences. The trimmed 424 FASTQ files were converted into a Qiime2 v2019.1 file format PairedEndFastqManifestPhredd33. 425 The imported forward and reverse reads were merged using 'vsearch' with a minimum sequence 426 length of 200 base pairs. Joined pairs were quality trimmed using Qiime2 'quality filter' with an 427 average quality score of 20 (Q20) over a 3 base pair sliding window and removing trimmed reads 428 having less than 75% of their original length. 'Deblur 16S rRNA positive filter' was used as a final 429 quality control step by dereplicating and removing chimera sequences from each sample; reads 430 were trimmed to a final length of 195 base pairs. Taxonomic analysis and Operational Taxonomic 431 Unit (OTU) tables were created with Qiime2 and converted using biom format is Qiime1. All 432 statistics were ran in Graphpad Prism v8 unless specified otherwise. Alpha and beta diversity 433 measurements were calculated using Microbiomeanalyst.ca with no filtering. Alpha diversity 434 calculations were based on Shannon diversity index with Mann-Whitney test and figures were 435 plotted in Graphpad Prism. All boxplot data were evaluated with median and minimum/maximum 436 values. Statistical analysis of the multiple group comparisons was performed using one-way 437 analysis of variance (ANOVA) followed by Tukey post-hoc test; when two groups were compared, 438 the nonparametric t-test was performed. For mouse genotypes at 0 weeks, a one-way ANOVA 439 with Tukey post-hoc test was performed to determine initial genotypic effects on microbiome. 440 Results were considered significant with P-value < 0.05. Beta diversity was calculated on 16S 441 rRNA OTU data using Bray-Curtis dissimilarity and NMDS figure created using R package 442 'vegan'. Permutational multivariate ANOVA based on NMDS ordination distances was used to 443 calculate community composition. Based on OTU data produced by Qiime2, a relative abundance 444 bar chart was created using Microbiomeanalyst.ca. For abundances statistical analysis, each 445 individual in human and mouse population was tested with a t-test and two-way ANOVA for each 446 family level classification for pre and post treatment. 447

Fecal Metabolomics 448
The nuclear magnetic resonance (NMR) spectra of aqueous fecal extracts were acquired at 298K 449 on a Bruker Avance III 800 MHz spectrometer equipped with a TCI probe (Bruker Biospin, 450 Germany). The 1 D 1 H NMR experiments were conducted using the first increment of the nuclear 451 Overhauser enhancement spectroscopy (NOESY) pulse sequence with presaturation for water 452 suppression (Relaxation delay-90-t1-90-mixing time-90-Free induction decay). The acquisition 453 parameters were as follows: 64 scans and 4 dummy scans, 64K data points, 90° pulse angle 454 (11.3 us), relaxation delay of 3 s and a spectral width of 14 ppm. The spectra were acquired 455 without spinning the NMR tube in order to avoid spinning side bands artifacts. The free induction 456 decays were multiplied by a decaying exponential function with a 1 Hz line broadening factor prior 457 to Fourier transformation. The 1 H NMR spectra were corrected for phase and a polynomial fourth-458 order function was applied for base-line correction. Chemical shifts are reported in ppm as 459 referenced to Trimethylsilylpropanoic acid (δ = 0). NMR signal were assigned using a range of 460 2D NMR spectra, namely 1 H− 1 H correlation spectroscopy, 1 H− 1 H total correlation spectroscopy , 461 1 H− 13 C edited heteronuclear single quantum correlation, and 1 H− 13 C heteronuclear multiple bond 462 correlation spectra. 1D and 2D NMR spectra were processed using TopSpin 3.2 (Bruker Biospin, 463 Germany). 464 The spectral region δ 0.50−10.0 was integrated into regions with equal width of 0.005 ppm using 465 the AMIX software package (V3.8, Bruker-Biospin). The region δ 4.70−4.90 was discarded due 466 to imperfect water saturation. Prior to statistical data analysis, each bucketed region was 467 normalized to the total sum of the spectral intensities to compensate for the overall concentration 468

differences. 469
Multivariate statistical analysis was carried out with SIMCA-P+ software (version 14.1, Umetrics, 470 Sweden). Data were mean-centered and scaled using the Pareto method, while log-471 transformation was applied to achieve an improved normal distribution of the data. Principal 472 component analysis (PCA) and orthogonal projection to latent structures with discriminant 473 analysis (OPLS-DA) were conducted on the scaled data. The OPLS-DA model's confidence level 474 for membership probability was set to 95% and was validated using a 7-fold cross validation 475 method. The quality of the model was assessed by the values of R 2 Y and Q 2 . The R 2 Y metric 476 describes the percentage of variation explained by the model; Q 2 shows the predictive ability of 477 the model. The difference between these metrics describes the model's fitness. 478

Gene expression 494
Gene expression of scraped mucosa from mouse intestines was performed as described [13]  The clinical study was performed in accordance with the requirements of Good Clinical Practice 537 and the Revised Declaration of Helsinki. All participants provided written informed consent to 538 participate after receiving verbal and written information about the study. The protocol was 539 approved by the Institutional Review Board of Advent-Health and registered at IRBNet (#982524). 540 The study was registered on ClinicalTrials.gov on the 26 th of January of 2017 (NCT03032640). 541 All the studies in mice were performed in accordance to NIH and institutional guidelines of the 542 Ohio State University Institutional Animal Care and Use Committee. 543

Consent for publication 544
Not applicable 545

Figure 3
Taxonomic abundances and diversity of gut microbiota in response to treatments in humans and mice (A and D) Bar chart summary showing relative abundance at the family level post-treatment inhuman participants or in WT and T1R2 mice. Each bar represents abundances of one subject. (B and E) Alpha diversity box plot (Shannon diversity metric) showing community richness between groups posttreatment in human participants (Mann-Whitney U Test; p=0.156, U= 5.22) or in WT and T1R2 mice (Mann-Whitney U Test; p=0.987, U=152). (C and F) Nonmetric multidimensional scaling (NMDS) plot showing community similarities between groups post-treatment in human participants (p< 0.999, NMDS stress = 0.2274) or WT and T1R2 mice (p< 0.111, NMDS stress = 0.209). NMDS ordination was derived from pairwise Bray-Curtis distances and statistical inferences made using PERMANOVA. N=11-13 for human studies, n=8-11 for mouse studies.

Figure 4
Fecal metabolomics in response to treatments in humans and mice (A) Differences in human fecal metabolites between treatment groups using orthogonal partial least squares discriminant analyses (OPLS-DA). (B) Post-treatment saccharin presence in human fecal samples. Dashed lines represent average noise ± SD. (C) Differences in WT andT1R2 fecal metabolites following saccharin treatment using OPLS-DA. (D) Post-treatment saccharin presence in mouse fecal samples. Dashed lines represent average noise ± SD. (E) Metabolite distribution (S-plot) in fecal mouse samples. Metabolites attributed to saccharin shown in red. (F) Differences in WT and T1R2 fecal metabolites following saccharin treatment using OPLS-DA after removal of saccharin signals. (G) Assessment of short chain fatty acids (SCFA) following treatments in human samples. One-way ANCOVA baseline as covariate. (H) SCFA in mouse feces before (pre) and after (post) treatment. Two-way ANOVA repeated measures with post-hoc. N=11-13 for human studies, n=8 for mouse studies.