Clinical translation of microbe‐based therapies: Current clinical landscape and preclinical outlook

Abstract Next generation microbe‐based therapeutics, inspired by the success of fecal microbiota transplants, are being actively investigated in clinical trials to displace or eliminate pathogenic microbes to treat various diseases in the gastrointestinal tract, skin, and vagina. Genetically engineered microbes are also being investigated in the clinic as drug producing factories for biologic delivery, which can provide a constant local source of drugs. In either case, microbe‐therapeutics have the opportunity to address unmet clinical needs and open new areas of research by reducing clinical side effects associated with current treatment modalities or by facilitating the delivery of biologics. This review will discuss examples of past and current clinical trials that are investigating microbe‐therapeutics, both microbiome‐modulating and drug‐producing, for the treatment of a range of diseases. We then offer a perspective on how preclinical approaches, both those focused on developing advanced delivery systems and those that use in vitro microbiome model systems to inform formulation design, will lead to the realization of next‐generation microbe‐therapeutics.


| I N TR ODU C TI ON
The human body coexists with microbiota, or communities of microbes, within the gastrointestinal (GI) tract, mouth, skin, vagina, and other tissues. 1 Each distinct microbiome, which encompasses the microbiota and their genetic material, balances key microbial populations in these tissues to regulate both health and disease. 2 An imbalance in these populations, or dysbiosis, in the GI tract may contribute to or result from cancer, obesity, diabetes, Clostridium difficile infection, or depression, among others. [2][3][4] Vaginal dysbiosis can lead to recurrent infections, increased risk of HIV transmission, preterm birth, or pelvic inflammatory disease. 5 Skin diseases such as dermatitis, and oral diseases such as caries are also significantly impacted by the microbiota. 6,7 Efforts to identify and describe the key role specific microbes have in these conditions are at the forefront of biological and medical research. 4 This knowledge will be essential to translate mechanistic understandings of the impact of commensal microbes on human health to the effective implementation of microbes as therapeutics.
Two main therapeutic uses of microbes are being investigated in the clinic. The first involves displacing pathogenic microbes and restoring symbiosis in patients via the delivery of living therapeutic bacteria.
The second involves genetically programming microbes to secrete therapeutics, either locally at sites of disease or through biological barriers for systemic absorption. In either case, the delivery of these microbes must occur appropriately to provide a therapeutic benefit. Therefore, their design must account for delivery challenges of live microbe therapeutics, which include: (a) environmental factors (e.g., acid, enzymes, UV-light) that can impair microbe viability, deactivate the secreted biologic, or induce damage that limits their efficacy, (b) biological barriers (e.g., mucus, existing microbiota, lumen contents) that physically prevent interactions (e.g., engraftment, drug diffusion), and (c) achieving a suitable residence time at the site of action (e.g., duodenum for drug absorption 8 ). Unfortunately, the interactions between the commensal microbiota, the delivered microbe-therapeutic, and the host environment remain opaque and stand as a bottleneck to the rational design of delivery approaches for microbe-based therapeutics. Future research in microbe-therapeutics will require a focus on elucidating these mechanisms of action in order to rationally design delivery approaches.
In this review, we will give an overview of the current approaches to therapeutic microbiome modulation and the advantages that microbe-based therapeutics may have over current treatment options.
The current clinical landscape of microbe-therapeutics will be highlighted by reviewing clinical trials that utilize bacteria as therapeutics, which includes examples of bacteria both as tools to modulate the microbiome and as drug-producing factories. Next, we will focus on recent examples of formulation approaches that have improved microbe delivery. Finally, we will end with a perspective on how microbiome model systems can be used to inform the rational design of next-generation microbe-based therapeutics.

| M I CR OBE -BA SED TH E RA P EU T I CS F OR M I CR OBI OM E M ODU L A TI ON
Here, we will highlight current clinical studies where bacteria are used to modulate the GI, skin, and vaginal microbiomes. It is worth noting that oral probiotics regulated as dietary supplements, rather than as therapeutics, do not require extensive clinical data to support functional claims. 9 While investigational clinical trials aimed at understanding the action of these dietary supplements and probiotics are underway, they will not be discussed here as they have been reviewed elsewhere. 10

| Current approaches to microbiome modulation
The most effective and established method for altering microbiota compositions are antibiotics, which are often a first-line treatment for bacterial infections. 11 Antibiotics have prevented countless deaths and are mainstays in clinical care. However, instances of antibiotic use have recently been linked to negative clinical outcomes. For example, the use of broad-spectrum antibiotics can lead to dysbiosis by disrupting the commensal microbiota 12 and their overuse has contributed to the rise of antibiotic-resistant pathogens. 13,14 By creating a commensalfree environment containing antibiotic-resistant pathogens, antibiotics often promote more-severe, recurring infections 15,16 as is the case for recurrent C. diff infections (RCDI). 17,18 These risks, particularly with RCDI, have generated significant interest in developing alternative therapies that mitigate the killing of commensal bacteria and the evolution of antibiotic-resistant pathogens. 15,19 One potential alternative are bacterial viruses (phage), which infect bacteria, propagate in their bacterial-hosts, lyse the bacteria, and are then released into the local environment (e.g., intestinal lumen) to continue this cycle. 20 Phages are highly specific to bacterial strains and can be used to exclusively eliminate enteric pathogens, while sparing commensal bacteria; this has motivated research into their use for the treatment of antibioticresistant pathogens. [21][22][23] However, the clinical translation of phagebased therapies has been minimal due to challenges related to their purification, characterization, and regulation. [24][25][26][27] Furthermore, due to the complex evolutionary dynamics between phage and bacteria, pathogens may become resistant to phage infection and lysis, which limits their long-term and repeated use. 28 Other alternative approaches, such as inorganic metals, antimicrobial peptides, and gene editing enzymes (e.g., CRISPR-Cas9) are also being developed, 29 but will not be reviewed here as they are not yet widely used in the clinic.

| Fecal transplant-based approaches for gut microbiome modulation
Microbe-therapeutics for microbiome modulation aim to displace colonized pathogens through competitive metabolic interactions, niche exclusion, or initiation of host immune responses. 30 In doing so, microbe-therapeutics have the potential to address the challenges facing antibiotics as outlined above. 31 The best example of these therapies are fecal microbiota transplants (FMTs), which take fecal bacteria from a healthy donor and transplant it into the GI tract of a dysbiotic or diseased individual, typically through colonoscopy or nasal tube infusion. 32,33 FMTs are one of the only clinical methods for treating RCDI, 34 which occurs in 15-30% of patients after taking the standard regimen of antibiotics, 19 and have been up to 90% effective in multiple clinical studies. 27,[35][36][37] FMTs are currently being investigated in the clinic to treat RCDI, Crohn's disease, and colitis. It should be noted that antibiotics are almost, if not always, administered prior to FMTs. For the purpose of this review, traditional FMTs will not be discussed in detail, as they have been reviewed in depth previously. 19,38,39 Here, we will highlight microbe-based clinical trials that have been inspired by the success of FMTs (Table 1). These efforts are focused on formulating FMTs as an oral pill, which is a promising administration route that improves compliance, acceptance, and accessibility of FMTs by shifting away from rectal administration.
Rebiotix's RBX7455, a lyophilized oral formulation for microbiota restoration isolated from fecal donor samples, recently began a Phase 1 proof of concept trial for treating RCDI. 40,41 Distinct from other oral FMTs, RBX7455 is based off of Rebiotix's established enema formulation, RBX2660. 42 RBX2660 has been shown to significantly reduce patient incidences of C. diff associated diarrhea, 43

vancomycin resistant
Enterococcus infection, 42 and C. diff recurrence 44,45 in previous trials and is being clinically investigated for other indications (Table 1). Since Rebiotix will have clinical data from both standard enema and oral formulations, direct comparison of these studies may provide insight into the importance of the administration route for microbe-therapeutics.
Furthermore, since RBX7455 is lyophilized and thus processed for storage, these comparisons will have additional implications in the processing, handling, and formulation of FMT-based oral therapeutics. Similar to RBX7455, Finch Therapeutics' CP101 is a lyophilized oral formulation consisting of fecal donor-derived microbiota. An initial clinical trial described the development of a lyophilization protocol that enabled reproducible encapsulation in terms of donor bacteria stability, viability, and physicochemical properties. When tested in humans for the treatment of RCDI, 88% of patients achieved clinical success (no CDI recurrence after 2 months). Furthermore, it was shown that a small dose of 2-4 capsules was as effective as a high dose of 24-27 capsules in terms of clinical efficacy. This clearly shows that a high pill burden for oral FMTs is not necessary to achieve clinical success or microbiome VARGASON AND ANSELMO | 125 modulation. Additionally, the authors conducted phylum-level classification of microbiota engraftment to confirm that the patient's microbiota compositions following treatment shifted towards the donor's composition ( Figure 1a). 46 Microbe engraftment was determined on multiple days in the first month and was monitored for up to a year after the study. By using multiple comparative points within the study, this is a stronger assessment of engraftment and cannot be attributed to the formulation residence time in the GI tract. This data, and other data not highlighted here, 47,48 were used to validate a predictive model of FMT microbe engraftment, which included factors such as the composition of donor samples, the elapsed time since the transplant, type and duration of antibiotics, and route of administration. 49 The model concluded that antibiotic type and use did not significantly affect microbe engraftment, despite conflicting clinical evidence. 50 This discrepancy may indicate that engraftment does not always predict efficacy. In general, the model was in agreement with clinical trial outcomes and thus it can be useful in identifying the bacterial strains responsible for therapeutic efficacy.
In other clinical work, Seres Therapeutics is evaluating bacterial spores for the treatment of RCDI, primary C. diff infection, and colitis (Table 1). Their most advanced therapeutic, SER-109, is an oral capsule of 50 species of bacteria spores, differentiating it from  Currently SER-262 is in a Phase 1b clinical trial for RCDI (Table 1). 55  inhibition of clinical recurrence. 66 Overall, these studies and clinical

| BA CTE R IA A S D R U G P R ODU C IN G AN D DE L IV E RI N G V E HI C LE S
The use of bacteria to produce drugs has been a longstanding, essential cornerstone of the pharmaceutical industry 70,71 and has been investigated in clinical trials for in vivo therapeutic production and delivery (Table 3). Since the genetic engineering of bacteria for therapeutic applications has been reviewed elsewhere, 72,73 we will focus on clinical examples and discuss opportunities for a formulationbased approach to improve delivery by considering microenvironment interactions.
In 2006, to the best of our knowledge, the first clinical trial utilizing genetically engineered bacteria to deliver drugs in humans described an engineered Lactococcus lactis (L. lactis) strain that secreted IL-10 for the treatment of Crohn's disease. Results from the trial showed that the oral capsule-delivered therapy was well tolerated and that multiple patients showed complete remission of Crohn's disease (Figure 2a). 74 An important consideration in this study was to ensure biological containment to avoid the potential health-risks that could occur if this strain were to stably colonize the patient, be excreted, and subsequently enter the environment. As such, the strain was engineered to require a thymine-rich environment for survival, thus it would pose little risk if the bacteria were to escape the human host. In a follow up Phase 2 clinical study, this strain did not show a statistically significant benefit compared to a placebo. 75 The low efficacy in the follow up study may be attributed in part to DNA degradation during GI transit, observed during the Phase 1 trial, 74 or the inability for IL-10 to penetrate intestinal mucosal barriers. The prior concern may be mitigated with a more advanced delivery strategy, such as an enteric capsule, Oragenics is developing a genetically engineered L. lactis strain designed to secrete trefoil factor (AG013) that is being investigated in a Phase 2 clinical trial as a mouth rinse formulation for the treatment of oral mucositis. A Phase 1b trial with this product showed a 35% reduction of ulcerative mucositis following mouth rinse administration up to six times daily (Figure 2b). 76 Importantly, extensive preclinical data demonstrated that both the L. lactis and secreted trefoil factor were limited to the site of administration, and were undetectable systemically, indicating a low risk of systemic exposure and toxicity. 77 79 In other clinical studies, genetically engineered strains for cancer treatment or prevention are also being investigated ( Table 3).
The delivery requirements are much clearer for drug-producing bacteria therapies, as compared to their microbiome-modulating counterparts, since the site of action and properties of the delivered drug are well known. As such, formulation-based approaches that can increase resistance to environmental challenges (e.g., an enteric capsule), residence time (e.g., mucoadhesive formulations), and localization to either the diseased tissue or the site of absorption will improve delivery. Since the majority of these genetically engineered strains secrete biologics that have been notoriously difficult to stabilize and deliver in vivo, 80 formulation approaches can also be used to protect both the bacteria and biologic drug. Furthermore, if biologics are to be absorbed systemically, approaches to increase residence time at the relevant absorption site (e.g., duodenum) will also improve biologic delivery.

| P RE CL I NI CA L A P P ROA CH ES TO I M P ROV E M IC ROB E-D EL I VE RY
While delivery approaches for microbes are still in their infancy, methods that improve survival, control transit and residence time, and target specific sites can ensure that microbes arrive at the right place, at the right time, and in the right concentration. In the case of drug-secreting bacteria, these functions will enable better drug transport either to the local pathology or across biological barriers for systemic absorption.
Similarly, for bacteria that modulate the microbiome, advanced formulations can offer improved delivery to the target site; however, whether these advantages lead to enhanced efficacy remains an open question as these formulations have not been explored rigorously and not enough is known about the microbe's mechanism of action. Here, we will highlight preclinical studies that have demonstrated how formulation approaches can improve the delivery of microbes. We will then offer a perspective on how preclinical in vitro models can aid in informing formulation design, especially for microbiome modulation applications.

| Formulation for improved delivery
There  tissues will need to consider residence time, a critical parameter that will dictate therapeutic efficacy and is mediated by environmental conditions such as self-cleaning in the vagina, 82 enzymatic degradation and saliva production in the oral cavity, 83  whether improved resistance to acid and bile salts or the enhanced binding to, and growth on, mucus was predominantly responsible for improved delivery. 84 In any case, improved delivery was achieved using a formulation approach that modified the surface of the microbetherapeutic. It is reasonable to assume that these microbe modifications can be combined with the standard formulation, an oral capsule.
This work clearly highlights the potential for using pharmaceutical formulation approaches to better control interactions with both the chemical and physical environments to improve live-microbe delivery.
In a separate work, E. coli Nissile 1917 (Eda) was genetically engineered to treat colorectal cancer (CRC) locally in the GI tract. 85 The authors considered the CRC microenvironment, such as surface receptors on cancer cells, and the GI tract environment, such as ingested food, to optimize their formulation. The final formulation (Figure 4a),   addition to the static features listed above. These models are essential to determine the therapeutic efficacy of a certain formulation or combination treatment. For example, a microfluidic gut-on-a-chip ( Figure   5c) was used to investigate how antibiotics and therapeutic microbes can treat intestinal inflammation from enteroinvasive E. coli (Figure 5d).
The chip mimicked the key features of the GI tract such as the intestinal barrier properties, intestinal morphology (Figure 5e), anaerobic conditions, shear stress, and peristaltic forces. 92,93 Concomitant administration of the therapeutic microbes and antibiotics protected against lesion formation caused by pathogenic E. coli (Figure 5f). 94 Furthermore, the system was used to show how colonization of specific therapeutic microbes under physiological conditions prevents the inflammation caused by exposure to pathogenic bacteria. Individual aspects of this physiological model could be turned on or off ( Figure   5f), which allows for isolation of the key contributing factors; in this case, the distinct beneficial contributions of therapeutic microbes and antibiotics could be tested independently.

| Preclinical outlook
Few formulation-based approaches have been tested for microbe-

| CON CL U S I ONS
Clinical trials have proven the potential for bacteria to offer alternative clinical treatment for a variety of diseases, through the secretion and delivery of challenging therapeutics, as well as the modulation of the microbiota composition toward symbiosis. As the development of livemicrobe therapeutics progresses, it will become necessary to consider the interactions these therapies have with the host microenvironment.
Since the importance of having control over where, when, and how a drug interacts with the diseased site has been shown to be a defining success criteria for all other forms of drugs, it should be a primary consideration for microbe-therapeutics as well. Preclinical work has already proven that protecting the microbes from environmental challenges, directing their action toward mucosal surfaces, and targeting them to addressing these open questions is also unclear. Therefore, this knowledge gap must be addressed, potentially through static and dynamic in vitro models, before rational formulation design can be used to increase therapeutic microbe efficacy. As understandings of relevant microenvironment interactions and challenges increase, opportunities to translate this knowledge to delivery platforms that can increase microbe viability, residence time, stability, and efficacy will become clearer. We envision that current research will enable (a) the determination of which strains are responsible for displacing specific pathogens, (b) the use of in vitro model systems to study phenomena that can inform therapy design, and (c) the development of a toolkit to functionalize, engineer, and package bacteria such that they interact in specific ways with the local microenvironment. This new area will require a fundamental understanding of how these therapies treat disease and a simultaneous effort to improve delivery.