Crohn’s Disease–Like Ileitis and the Inhibitory Effect of Sucralose on Streptococci

To the Editor:

Recently, we assessed the effects of the artificial sweetener Splenda (ingredients: sucralose and maltodextrin) in Crohn’s disease (CD) ileitis¹ after feeding the product to mice prone to CD (SAMP1/YitFc) and (CD-free) control mice (AKR/J). Here, we would like to address criticisms raised by Lynch et al.,² who emphasize that high-quality animal studies are important to patients. Because we agree, it is critical for us to clarify our experimental design to demystify Lynch’s criticisms on the scientific and translational value of our statistically significant findings, which are in synchrony with findings from other studies assessing sucralose in rats,³ a study assessing maltodextrin in cells and mice, which was cited in Rodriguez-Palacios et al.,¹ and an earlier study of Splenda in rats,⁴ which was also criticized by 1 of the authors in Lynch’s letter (V. L. Grotz, who works for the manufacturer of Splenda). Examples of the effect on the gut microbiota and inflammation in other organs also support our findings,^{5, 6} whereas the translational implications of our study to humans are illustrated in recent discussions.^{7, 8}

Overall, Lynch’s letter said that the “conclusions presented are not supported by the collective evidence relevant to the tested product.” However, it is important to stress that (1) the letter presents no original data supporting the criticisms and (2) from a literature search of the first author (PubMed, B. Lynch), it is evident that as for-profit (food industry) consultants,⁹ their (seemingly biased) publications often conclude the same: that the products tested “are safe based on lack of significant effects on body weight, food consumption, hematology, clinical chemistry, organ weights, and histopathological examination,”^10–13 which are nonspecific outcomes of subclinical inflammatory responses in the gut.

Our study design was simple. We offered Splenda in water to mice and compared its effect with that of water alone. The experiment was repeated 3 times, with increasing doses each time. Nothing else was different. We also used an array of highly specific methods, not used by Lynch,^10–13 to determine the subclinical and mechanistic impact of Splenda in gut health (RNA quantification and fluorescence in situ hybridization to identify bacteria and the malX gene in gut tissues, metagenomics, 16S rRNA microbiome, a novel quantitative parallel lanes plating culture method, and biochemical markers of inflammation).

Despite the simplicity and robustness of our study, from Lynch’s letter, at least 2 major aspects called our attention, indicating that the experimental design was not clear to Lynch, raising concerns about the precision of their criticisms. Lynch implied (1) that we ignored the microbiome variability in our mice and (2) that the CCL21 defect (which we discovered and cited²) was an immune deficiency in SAMP that could render our conclusions invalid (we found that Splenda promotes ileum-associated dysbiosis in CD-SAMP mice but not in “healthy” control mice).

On the microbiome: Lynch said that “no data were presented to understand normal variability in the SAMP mouse microbiome with normal dietary fluctuations.” In this context, we cannot see the rationale of the criticism because there is no such as thing as “normal dietary fluctuations” in standard rodent research, nor did fluctuations occur during our study. Our laboratory mice were not fed varying meals. The diet was pelleted, the same, and ad libitum. Readers may notice that we devoted substantial effort to understanding the microbiome’s ecological variability in SAMP at the colony-and-cage level before the Splenda experiments were conducted (refer to Figs. 1 and 2 in our original published study in¹), and we also identified for the first time a novel form of “cyclical bias in microbiome research,”¹⁴ which we controlled for to ensure high research quality. The Methods section¹ also described novel methods we proposed to control for intersubject pre-experimental fecal microbiota heterogeneity.^{15, 16} Lastly, we sampled all animals in the morning, alternating sampling across groups, to control for circadian variation. Contrary to ignoring microbiome variability, as Lynch suggests, others have identified our methods on rodent research as novel (see the Commentary by Taconic Inc.¹⁷).

On CCL21: Lynch said that “SAMP mice also produce little or no CCL21, which is important for normal immune response, and this could predispose the intestinal tissue to bacterial infiltration into the ileal lamina propria.” To date, SAMP is the best and only polygenic inbred mouse model for CD, for which we have already identified numerous polymorphisms.¹⁸ Therefore, we do not understand the rationale for mentioning only 1 of the many biological abnormalities of SAMP mice (CCL21) as a problem in our study, which, as in humans, make them prone to CD. Our surprise with this basic criticism is that regardless of the disease mechanisms that make SAMP mice prone to CD-ileitis, our results arose from comparing 2 identical groups of SAMP mice, 1 of which received Splenda; the other did not. In this context, the significant differences detected can only be certainly attributed to the administration of the product. CCL21 has nothing to do with explaining pronounced differences in the Splenda group; that is, because all SAMP mice had the same genetic/immunological (CCL21) background.

Additional minor points in the letter included histology and that gut microbes do not use sucralose. Lynch criticized that no histological changes were detected in our Splenda mice. Although others have shown that Splenda induces histological alterations,^{3, 19} we were not completely surprised by the lack of significant histological changes in our study because we found other very specific anomalies in the gut wall (quantitative polymerase chain reaction microbial dysbiosis, more invading microbes with increased FISH-malX, more myeloperoxidase ileal reactivity). Further, we recently characterized the variability and limitations of histology in SAMP CD-ileitis.¹⁵ Further, histological changes often cannot be detected in early stages of inflammatory bowel disease in humans.²⁰ In baboons, for instance, changes in the epithelial gut barrier and tight junction protein expression that occur with aging cannot be detected histologically.²¹ This growing body of literature indicates that biochemical markers (and not histology, which is subjectively qualitative and not quantitative) could be sufficient to monitor potential inflammatory response to diets, which may not necessarily cause morphological changes but induce cumulative effects due to persistent inflammation over time.

Lastly, we want to use this opportunity to expand on 1 of the findings not discussed in our study to illustrate that Lynch’s criticisms are outdated by being too narrow in focus (ie, they said that “sucralose is not a substrate for gut microflora”). Lynch implies that ingested sucralose is inert. That is not necessarily true. In our study, we discovered that low-abundance gut commensal streptococci (present in the water-drinking mice) were absent, and replaced by Escherichia coli in the Splenda-drinking mice. Regardless of whether sucralose can be used as a source of carbon/energy for bacterial growth in agar plates, to date, numerous studies have shown that sucralose (a Splenda ingredient) induces clinically relevant and critical physiological changes in bacteria. Sucralose increases antimicrobial resistance and mutation frequency of E. coli to several antimicrobials.²² Mass spectrometry data have long existed to show that C14-labeled sucralose molecules are chemically altered as they pass through the rodent digestive tract,¹⁹ which suggests metabolism and secondary metabolite production. Others have shown that sucralose induces potent inhibitory effects on the bacterial physiology of differentiating filamentous cyanobacteria and polysaccharide sheath induction.²³ Specifically, sucralose inhibits hormogonium differentiation at a concentration approximately 1/10th that of sucrose (table sugar).²³

Others have shown that sucralose prevents the absorption of sucrose (table sugar) in most microorganisms and exerts a potent bacteriostatic effect.²⁴ Mechanistically, it has been determined that sucralose inhibits bacterial invertase and sucrose permease, which are enzymes that appear unable to catalyze the hydrolysis or the effective transmembrane transport of sucralose, which is a complex sugar moiety carrying 3 chloride atoms (Fig. 1A).²⁴ Using another complex sugar analog, fluorescently labeled esculin, others have shown, however, that complex sugars could be incorporated via special intercellular import mechanisms into the cytoplasm of filamentous cyanobacteria without needing hydrolysis, which could also occur in other branching or segmented filamentous bacteria.²⁵ From the widespread use of esculin in the diagnostic “bile-esculin test,” we know that several strains of Enterococcus and Group D Streptococcus have unique means to handle complex sugars to then use glucose-derived moieties as a source of energy, leaving esculetin as a secondary metabolite, which chemically reacts with other unintended chemical elements. Specifically, in the test, esculetin reacts with ferric citrate, which darkens, causing the pigmentation used as a metabolic marker in diagnosis (Fig. 1A).

FIGURE 1.

Molecular conformation of sugar molecules and a microbial modulation model for Splenda. A, Two-dimensional conformation of simple and complex sugar moieties. Source PubChem Open Chemistry Database from the National Center for Biotechnology Information. CID are unique public PubChem Compound Identifiers, accessible at: https://pubchem.ncbi.nlm.nih.gov/compound. Notice that esculin is a complex sugar that yields glucose and esculetin upon microbial hydrolysis. Sucrose (table sugar) can be absorbable by most bacteria. B, Basic microbial modulation model for sucralose/maltodextrin dietary additives. Sucralose (artificial sweetener) is deemed to be nonmetabolizable by microbes, but several effects secondary to exposure in bacteria have resulted in bacteriostatic, sucrose metabolism, and morphological abnormalities in several taxa. Maltodextrin is known to promote the growth of several Proteobacteria (see the text).

Because not all bacteria can use or absorb complex sugars, but some do, the internalization and traffic of complex sugar esculin inside bacterial cells indicate that other complex sugars (eg, sucralose) could be internalized by bacteria or make them susceptible to secondary unintended chemical effects. Although streptococci and enterococci were used for illustration purposes, other esculin utilizers indicate that complex sugars could have an effect in a wide range of novel microbial species not yet realized (eg, lipophilic Corynebacterium kroppenstedtii,²⁶ novel Vibrio hyugaensis,²⁷ enterotoxigenic Bacteroides fragilis associated with ulcerative colitis,²⁸ yeast Saccharomyces spp.,²⁹ and unique Clostridiodes difficile³⁰ strains). Because complex sugar metabolism seems to vary at the strain level, one could mechanistically hypothesize that the effect of certain dietary ingredients could modify the balance of some strains or species in a complex gut community, but not others, in almost undetectable or unpredictable ways if only DNA-based microbiome diagnostic methods were used, and if each ingredient had independent effects on distinct taxa or the host (Fig. 1B). In this context, current 16s rRNA gene approaches provide too little information at the species level and too little power at the strain level to document microbiome variations accurately.

We used culture and single-colony Sanger sequencing to further characterize Splenda-associated changes in gut Streptococcus/E. coli ratios. Supporting the findings in our animal study (Splenda reduced cultivable streptococci), sucralose has been shown to exert a strong bacteriostatic effect on Streptococcus spp. in vivo and in vitro growth assays (including S. sobrinus, S. sanguis, S. salivarius, and Actinomyces viscosus)^{24, 31} and in biofilm formation and lesion depth in dental cariogenic studies using Streptococcus mutans.^32–34 Using diverse environmental bacteria (Stenotrophomonas sp., Rhizobium borbori, Microbacterium sp., Citrobacter murlinlae, and Streptomyces badius), others have shown that sucralose strongly interferes with bacterial growth in a medium containing only glucose in a dose-dependent manner.³¹ This sucralose bacteriostatic effect could have accounted for the microbial switch we noticed by favoring the reduction of gut commensal streptococci and promoting the proliferation of Escherichia coli (Proteobacteria) in the Splenda group either by sucralose or maltodextrin promoting mechanisms (Fig. 1B).

Whether microbes utilize 1 or 2 of the Splenda ingredients in vitro is not medically important as a criticism. What matters to patients is what effect the combined sucralose/maltodextrin ingredients have on the overall physiological balance of the gut microbiome–host responses. Our data conclusively indicate that the sweetener alters the microbiota and suppresses low-abundance cultivable streptococci, supporting the studies listed above. Considering that some streptococci have been associated with IBD (eg, S. mutans, S. sanguinis),^36–39 it is also possible that sucralose could have a modulatory role in Streptococcus-driven IBD, if such microbial-disease connection was valid in humans.

Intrinsically, we agree with Lynch that is important to be academically inquisitive and skeptical, because not all studies are equally stringent. However, it is very important to ensure that well-founded criticisms come from sources free of financial bias and conflict of interests. Further, regulatory agencies and consulting professionals seeking to document the benefit of dietary products, or their negative effects, to obtain marketing approvals should not use only nonspecific outcomes (for instance, “organ weight, body weight, blood cell count”) and claim them to be definitive surrogates of intestinal health, because it has been known since 2000 that sucralose may affect them.³⁵ There is therefore need to use modern tests to infer prehistological inflammation in the intestinal tract, especially important for chronic inflammatory diseases. Of note, only 10% of IBD patients report worsening of symptoms due to sweetened diets. The problem is that no one knows how the next patient would respond to any given diet. The collective of mechanistic studies in rodents indicates that every person indeed responds differently to each diet and that some diets, including Splenda, intrinsically induce well-known basic mechanistic and ecological effects. Our studies provide unique insights to objectively test several elements of the diet directly in humans.

Supported by: Funding originated from National Institutes of Health competitive grants, internal university funds, and the Crohn’s and Colitis Foundation of America. No funding was received from any commercial (food or nonfood) product organization or sugar manufacturer.

Conflicts of interest: No conflict of interests to disclose.

Disclosure: A patent has been filed for a home-based test to assess the impact of diet on digestive health.