With Probiotics, Resistance Is Not Always Futile

Abstract

Probiotics and other bacteriotherapies are actively being explored and applied as symptom- and disease-modifying agents. In a recent issue of Cell, two papers contribute to our understanding of how live bacterial therapies variably affect individuals and the short- and longer-term impact of these therapies on colonization and host response.

The idea of using fecal bacterial mixtures and probiotics to treat gastrointestinal (GI) diseases has existed for centuries, although not until 1958 was the first fecal microbe transplantation (FMT) for pseudomembranous colitis formally reported by Eiseman et al. (for review, see Baktash et al., 2018). These remedies are based on the central premise that restoring the gut microbiome to its original state after a perturbation (called gut dysbiosis) will relieve symptoms and potentially be curative. Specific methods of treatment include probiotics (adding living, non-pathogenic bacteria thought to be generally healthful), prebiotics (dietary supplements that promote growth of beneficial bacteria), and FMT (as capsules, through endoscopy, or via an enema). Moving clinical microbiome research from a correlative to mechanistic science is an ongoing challenge. Work by Suez et al. (2018) and Zmora et al. (2018) in the September 6^th issue of Cell takes us one step closer to understanding how the gut microbiota and the host respond to probiotics alone or after dysbiosis from antibiotics with the addition of probiotics or auto-FMT (fecal transplant from the same individual at a later time).

The probiotics industry is a rapidly growing market that is projected to be over US $65.9 billion by 2024 (Zion Market Research, 2018). Most probiotics are available over the counter, and live microbes, such as Lactobacillus spp. and Bifidobacterium, are frequently found in common food products such as yogurt and miso. Thus, a large proportion of the world is exposed to live bacterio-therapies. Despite this broad use, little is known about the probiotic persistence, patterns of colonization, and biological host responses. Having undertaken an impressive and quite comprehensive set of interventional and descriptive experiments in mouse models and humans, Suez et al. and Zmora et al. inform our understanding of interindividual variations in responses to probiotics, the impact of prior antibiotic use on these responses, and the eventual host transcriptomic response to these therapies.

Most microbiome studies focus on the fecal microbiota as a representation of the entire gut microbiota. Using stool samples, washes of the intestinal lumen during endoscopy, and tissue biopsy of the gut, Zmora et al. (2018) show differences in bacterial composition along the GI tract. While mucosal biopsy and lumen washes were similar in bacterial composition at the same site, subtle differences in bacterial quantity are identified. Further-more, they demonstrate that the mucosal microbiome varies from the stool both quantitatively and compositionally, as determined by multidimensional analytical methods. Despite this finding, a general concordance is observed between the mucosal microbiome and stool, with bacterial species seen in the stool as representative, in aggregate, of those seen at various sites along the GI tract. Collectively, this suggests that stool is a reasonable approximation of the relative abundance of bacteria averaged over the entire GI tract, but similar to previous studies (Eckburg et al., 2005), the stool does not precisely represent a specific local environment along the GI tract where microbes are likely to interface with the immune system. Whether this has an impact on how the immune system interacts with the microbiota remains an unanswered question.

In healthy individuals, antibiotic exposure can induce changes in the stool microbiome that may last for several months (Dethlefsen and Relman, 2011), and even subtle perturbations may alter the microbiota. For example, the colonic environment and mucus are disrupted during even mild osmotic diarrhea (Tropini et al., 2018). The field is striving to under-stand the complex community within the gut lumen, including biofilm layers and specific secreted substrates that are then used by another member of the community, as well as how the host immune system interfaces with the microbiota. With probiotics, whether there is long-term replacement of indigenous microbes by other species or inherent advantages given to the existing microbiota to repopulate is unclear.

The therapeutic intent of probiotics, prebiotics, and FMT is to help restore the gut to its original pre-dysbiosis state or to a new balance whereby an underlying disease is eliminated or lessened. Primary research studies and meta-analyses have suggested that probiotics decrease the duration of antibiotic-associated diarrhea (Hickson et al., 2007; Goldenberg et al., 2015; Bernaola Aponte et al., 2013), yet evidence from these studies is of variable quality, and the mechanism underlying the clinical benefits remains unknown. A major challenge to the field is understanding if bacteriotherapies such as probiotics and FMT modulate the existing microbiome. Furthermore, it is critical that we discern the underlying mechanisms of bacterio-therapy and the personalization that may be required to determine which populations would benefit most from this therapy. Suez et al. (2018) show that there is a delay in returning the microbiome to its pre-antibiotic state after probiotics when compared to watchful waiting and that an auto-FMT induces the most rapid and complete recovery in previously healthy subjects (this is also recapitulated in a mouse model). Several key questions arise from these data, including whether there is an inherent resistance to repopulation by the probiotic species (and if so, what causes this resistance), or if there are specific patient populations that are more likely to benefit from bacteriotherapy.

The companion paper (investigating empiric probiotics without antibiotics) by Zmora et al. (2018) demonstrates that the baseline microbiota composition in the stool quickly returns to near-normal after cessation of probiotics for most patients, with a sub-population of individuals being persistently colonized with bacteria found in the probiotics. Furthermore, host steady-state mRNA levels from gut biopsies suggest there is a different signature in individuals who are persistently colonized by probiotic strains compared to those who return more quickly to their baseline microbiome. The authors note that the steady-state mRNA signature on a complex set of tissues correlates with being permissive or resistant to colonization, suggesting that host cells determine whether or not colonization occurs. While an intriguing concept, data presented thus far do not allow us to answer whether there is an inherent signal from the microbes or from cells of the host immune system that induces a transcriptional response in the cells of the gut. Supporting the idea that microbes may play a prominent role in signaling, recent studies have shown that feces from young animals transplanted into older animals can induce reversal of some aging phenotypes (Smith et al., 2017). Establishing the directionality of the relationship between microbes and the host will likely require careful modeling experiments and may also be informed by a detailed mechanistic dissection of the host-microbe interface.

Understanding the mechanisms that underlie an individual’s propensity for colonization versus resistance will be valuable in developing and potentially ‘‘individualizing’’ future bacteriotherapies. At present, the scientific basis for developing a ‘‘personalized’’ probiotic does not yet exist; however, auto-FMT, which Suez et al. (2018) demonstrate can return a microbiota to a baseline state after perturbation, is also a compelling potential therapeutic modality. Of course, a major limitation in bringing auto-FMT into broader clinical application is that an appropriate stool sample must be collected in advance; perhaps unsurprisingly, the concept of ‘‘self-banking’’ of stool is starting to gain favor. Of course, many open questions exist, such as the optimal timing and storage for a baseline stool collection. Finally, despite the generally held notion that auto-FMT is ‘‘safe’’ because one is introducing indigenous bacteria back to the host, alterations in microbial composition can occur ex vivo (including horizontal gene transfer and contamination) and can thus compromise the ‘‘safety’’ of this product. Thus, concerns such as emergence of pathogenic strains must be considered, especially for the broader application of such an approach in developing children or immunocompromised patients.

Fundamentally, Suez et al. (2018) and Zmora et al. (2018) demonstrate a convincing, statistically significant difference in how bacteriotherapy may restore the microbiome after dysbiosis in a healthy human population, but whether these data have broader health implications in individuals who are unwell remains unknown. Populations that may be most likely to benefit from bacterio-therapy are those who are unwell: for example, those with prolonged anti-biotic-associated diarrhea or individuals who are immunocompromised and thus at risk to acquire opportunistic infections. To evaluate the impact of bacteriotherapies on these individuals, prospective interventional studies will be particularly informative. Notably, such efforts are underway and will likely be reported in the near future (such as ClinicalTrials.gov identifier NCT02269150: Auto-FMT for Prophylaxis of Clostridium difficile Infection in Recipients of Allogeneic Hematopoietic Stem Cell Transplantation).

The work by Suez et al. (2018) and Zmora et al. (2018) helps to unravel the mechanisms of colonization and persistence of and host responses to probiotics and FMT, strengthening the possibility of future ‘‘personalized microbiome’’ bacteriotherapies. Looking forward, research focused on microbial interactions—both within the microbiota and between the microbiota and host—should include careful analysis of the relevant biogeography of microbial colonization as well as host responses. Furthermore, future research will also benefit from considering factors such as the ability of microbes to form bio-films, ecological exclusion from specific locations (be it microbe induced, host-induced, or a combination), signaling molecules (including whether direct contact between microbes and the immune system is required, or secreted compounds locally or more distally), metabolomics, and host-barrier functions from the immune system. Taken together, these and many more measurable features will help us better understand this complex system.