Microbiome contributions to pain: a review of the preclinical literature

Over the past 2 decades, the microbiome has received increasing attention for the role that it plays in health and disease. Historically, the gut microbiome was of particular interest to pain scientists studying nociplastic visceral pain conditions given the anatomical juxtaposition of these microorganisms and the neuroimmune networks that drive pain in such diseases. More recently, microbiomes both inside and across the surface of the body have been recognized for driving sensory symptoms in a broader set of diseases. Microbiomes have never been a more popular topic in pain research, but to date, there has not been a systematic review of the preclinical microbiome pain literature. In this article, we identified all animal studies in which both the microbiome was manipulated and pain behaviors were measured. Our analysis included 303 unique experiments across 97 articles. Microbiome manipulation methods and behavioral outcomes were recorded for each experiment so that field-wide trends could be quantified and reported. This review specifically details the animal species, injury models, behavior measures, and microbiome manipulations used in preclinical pain research. From this analysis, we were also able to conclude how manipulations of the microbiome alter pain thresholds in naïve animals and persistent pain intensity and duration in cutaneous and visceral pain models. This review summarizes by identifying existing gaps in the literature and providing recommendations for how to best plan, implement, and interpret data collected in preclinical microbiome pain experiments.

1. Introduction

The microbiome, or collection of microorganisms that exist in a given environment, has become exponentially popular in biomedical research and the public press over the past 2 decades.⁶⁶ With technical advances in genetic sequencing, local disruptions in microbial populations—a state known as dysbiosis—have been demonstrated in many diseases that are associated with sensory symptoms. For example, gut dysbiosis is associated with both irritable bowel syndrome (IBS¹²⁸) and inflammatory bowel disease (IBD¹²⁷), 2 disorders characterized by persistent pain, and skin dysbiosis is a hallmark of atopic dermatitis, a condition characterized by persistent itch.⁸¹ However, a question unanswered by these observational studies is which came first?—disease pathophysiology or microbiome dysbiosis? Animal models provide an unparalleled platform in which these and other mechanistic questions can be answered. Germ-free animals (ie, animals that have no microorganisms living on or in them) can be bred and raised in gnotobiotic facilities, and unlike in human patients who frequently have negative perceptions of the process, fecal microbiota transplant (FMT) studies are routinely performed in rodents.¹⁰⁷ Experiments using these approaches allow for definitive conclusions regarding if and how select microbes contribute to a disease state or symptom.

Pain is a symptom of many chronic diseases that may be directly or indirectly caused by changes in microbial communities. For example, we know that many bacteria produce neuroactive metabolites or are themselves capable of directly activating nociceptors, the class of peripheral sensory neurons that encode and relay noxious stimuli (reviewed in the study by Staurengo-Ferrari et al.¹⁵⁵). Dysbiosis may also contribute to pain by altering activity of the innate or adaptive immune systems⁸⁹ or by changing host metabolism through endocrine signaling.¹³⁶ Despite this knowledge and the increasing popularity of microbiome research, the pain field currently lacks a comprehensive, systematic assessment of the preclinical, microbiome-related literature. Review of this work will either provide answers to or allow for the identification of gaps in our understanding of how microbes contribute to pain, a fundamental, evolutionarily conserved sensation. In this review, we report current methodological trends in preclinical microbiome articles, summarize the behavioral effects of microbiome manipulations performed in naïve rodents and pain models, and finally, provide recommendations for best practices and future experimental questions in this field.

2. Methods

This assessment was conducted to review all original research articles that published pain behavior test results collected in animals that had a microbiome manipulation. For this analysis, we searched PubMed for all articles published in or before 2023 that contained the word “pain” and 1 or more of the following terms: “microbiome,” “probiotic,” “FMT.” This search was completed on February 7, 2024, and resulted in 1336 articles (Fig. 1).

A first round of screening eliminated articles that met our a priori exclusion criteria: 12 articles were not written in English, 2 articles were retracted, and 527 articles were not primary research articles (of note, 481 of these articles were review articles). In a second round of screening, articles were reviewed for topic relevance. To be included, studies had to include 1 or more animal pain behavior test in conjunction with a microbiome manipulation. Although they can be used to assess pain-related changes in mood or cognition, the following tests were not considered bona fide pain assessments in our analysis: open field, sucrose preference, tail suspension, forced swim, elevated plus maze, elevated zero maze, and novel object recognition. Articles that exclusively used prebiotics (ie, food ingredients that can promote the growth or activity of select bacteria; N = 9) were excluded. We needed to set strict criteria on the types of manipulations that would be included in this study given that many orally ingested drugs—including those not used for their microbiome-related activity—can change bacterial populations¹⁸¹ in a similar manner to prebiotics. To prevent analysis of all studies in which an animal orally received a drug, we only included studies in which germ-free animals, FMT, antibiotics (ie, compounds that kill or inhibit the growth of bacteria) or probiotics (ie, compounds that contain live bacteria) were used. We also excluded experiments in which the primary objective was measuring morphine analgesia or tolerance^75,87,111 in the absence of an additional injury to the animal. Finally, we excluded 3 articles in which administration of a potentially noxious bacterium was the primary pain model.^56,125,137 Articles meeting all inclusion criteria were read by at least 2 members of our laboratory. Individual experiments within the articles were coded for the following: species used, injury model (or naïve state), pain measurement, microbiome manipulation, and effect of microbiome manipulation on pain measure. All results are reported in Supplemental Table 1, https://links.lww.com/PAIN/C113. The following information was also coded for all probiotic and antibiotic studies: compound dosage (concentration and administration time frame), timing of administration (before, after, or concurrent with injury), and administration route (Supplemental Tables 4 and 5, https://links.lww.com/PAIN/C113). Probiotic manufacturers and antibiotic classes were also reported. The following was reported for FMT studies (Supplemental Table 6, https://links.lww.com/PAIN/C113): administration route, state of donor fecal material (fresh from colon, freshly defecated, frozen), and state of recipient (germ-free, pseudo–germ-free, naïve).

3. Results

3.1. Overview of preclinical microbiome pain studies

A systematic review of the preclinical microbiome pain literature yielded 303 individual experiments across 97 articles (Supplemental Table 1, https://links.lww.com/PAIN/C113). The first article in this field was published in February of 2006; in this article, Verdú et al.¹⁶¹ used colorectal distension (CRD) to determine whether administration of Lactobacillus paracasei (now known as Lacticaseibacillus paracasei) relieved antibiotic-induced visceral pain in mice. Over the subsequent 17 years, there has been a gradual increase in the number of articles annually published on this topic (Fig. 2A).

3.1.1. What animals are being used in preclinical microbiome pain studies?

To date, no preclinical microbiome pain study has been completed in animals that are not rodents. Instead, 58% of experiments have used mice and 42% of experiments have used rats (Fig. 2B). Although not the main topic of this review, we would be remiss to not briefly highlight some of the key species differences that exist between human and rodent microbiomes because these may impact translatability of results obtained in the latter (for a detailed discussion of this topic, we recommend the review written by Nguyen et al.¹¹⁸). Mice and humans maintain relatively similar gastrointestinal tract anatomy and physiology, and as such, their core gut microbiomes are relatively similar in terms of the phyla and genera present therein.⁸³ The relative abundance of each phylum and bacterial species, however, varies between organisms. Lifestyle factors may contribute to these differences. For example, the life history of laboratory rodents is intentionally limited and well-controlled to minimize variability in experimental outcomes. Most laboratory rodents are born and raised in pathogen-free facilities, housed in cages filled with autoclaved bedding and otherwise limited enrichment, and maintained on the same defined diet throughout their life. This is in stark contrast to the average human lifestyle which is much more variable and filled with regular exposure to environmental challenges including pathogens. Although understanding the scientific basis for maintaining rodent life homogeneity, we recently argued that increasing diversity in rodent lifestyle (eg, using outbred rodent strains, increasing complexity of housing conditions) may increase the translatability of basic pain science findings¹⁴³; the same may also be true with regards to microbiome-related animal studies. Because human microbiomes will never be as controlled as those in laboratory rodents, studies that are purely aimed at translation—but not those intending to prove causality—may prove more successful if completed in subjects who maintain a more diverse microbiome. This was recently exemplified in so-called wildling mice (ie, C57BL/6 laboratory embryos transplanted into and ultimately born from wild dams).¹⁴⁰ Wildling mice maintain microbiomes that are more similar to wild mice than laboratory raised mice. Notably, wildlings also exhibited immune responses that mirrored those observed in humans during a failed, phase 1, clinical trial; had these same responses been observed in laboratory raised animals, the experimental therapeutic would likely have never proceeded to clinical trials.¹⁴⁰ Acute pain and injury-related pain behaviors have not yet been measured in wildling mice but should a priority for the future.

Diversity in rodent microbiome studies is further confounded by the fact that mice are coprophagic (ie, cohoused rodents ingest cagemate feces). A recent study determined that the gut microbiome of mice that were prevented from participating in this practice were more similar to the human gut microbiome than those of mice that were permitted to engage in this natural behavior.¹⁵ The recolonization that happens as a result of coprophagia not only decreases gut microbial diversity in an individual animal but also gradually shifts the microbiome of all cohoused animals to the point that cagemates essentially become technical replicates of one another. In an effort to identify strategies that could minimize these cage effects—and thus potentially allow cagemates to be considered biological replicates—Russell et al. recently demonstrated that mouse gut microbiome richness and diversity was increased in cages that had reduced housing density.¹⁴² In conclusion, the use of mice and rats in microbiome-related studies is founded on a strong biological basis, but preclinical scientists should be aware of the caveats that exist when using rodents for translationally relevant studies.

3.1.2. How is pain being assessed in preclinical microbiome studies?

Reflexive somatic pain behaviors, or behaviors that result from stimulating the surface of an animal’s body, are the most frequently reported measures in preclinical microbiome pain studies. Specifically, 39% of studies incorporated somatic mechanical behaviors, 17% of studies incorporated somatic heat behaviors, and 2% of studies incorporated somatic cold behaviors (Fig. 2C). These trends closely align with those observed in all preclinical pain studies; evoked somatic mechanical measures like von Frey withdrawal thresholds are the most popular assessment performed by basic pain scientists.¹⁴³ Frequent use of these measures is also technically appropriate given the high prevalence of somatic injury models in preclinical microbiome studies (discussed below). Relative to the rest of preclinical pain science, microbiome-related pain studies more frequently incorporate visceral pain measures, in particular, abdominal reflex to colorectal distension. In microbiome-specific studies, visceral pain measurements were performed in 35% of experiments, but in all articles published in the journal Pain between 2016 and 2020, visceral pain measurements were only performed in 2% of all experiments.¹⁴³ This is not entirely surprising given that the initial preclinical pain microbiome studies were performed using models of IBS or IBD. Finally, similar to the larger preclinical pain field,¹⁴³ ongoing pain measures like facial grimace and attending behaviors are infrequently used in microbiome-specific studies (Fig. 2C). An interesting consideration for future studies is the incorporation of machine learning tools that can identify novel behavior patterns that result following microbiome manipulation.¹⁶

3.2. Review of microbiome manipulation techniques in animal pain studies

Experiments included in this review manipulated rodent microbiomes in 4 possible ways: probiotics were administered in 47% of experiments, antibiotics were administered in 25% of experiments, FMT was performed in 23% of experiments, and 4% of experiments assessed pain behaviors in germ-free rodents (Fig. 3). Specifics on the implementation of these techniques are described in the following sections.

3.2.1. Trends in probiotic usage

Probiotics are defined as live microorganisms that, when administered in adequate amounts, provide health benefits to the host.⁶⁰ Humans routinely consume probiotics through normal diet (eg, breastmilk, yogurt, fermented foods) and may additionally supplement their intake with commercially available bacteria. The global commercial probiotic industry is estimated to exceed $100 billion USD by 2029¹²⁹ and is often regulated like other food or dietary supplements that do not contain living organisms.¹⁵¹ In other words, unlike pharmaceutical products that are intended to treat a certain disease and thus subject to strict regulatory testing, probiotics are frequently marketed as general health supplements, which allows for generalization between manufacturing lots. Therefore, the exact identification and number of live/dead bacteria contained within a given product may vary even when coming from the same distributor. In a similar manner, genetic sequencing is not performed on every lot of product, and thus, loss or gain of genes relevant to the functional effects of the probiotic may unknowingly occur over time. Regardless, probiotics have become increasingly popular tools for modulating gastrointestinal symptoms, including pain.

To date, probiotics have been used in 143 preclinical pain experiments (Fig. 3, Supplemental Table 4, https://links.lww.com/PAIN/C113). Across these 143 experiments, there was not one in which probiotic administration increased visceral or cutaneous sensitivity. Members of the genus formerly known as Lactobacillus were the most frequently used probiotics appearing in 52% of experiments. Lactobacilli are Gram-positive facultative anaerobes that have been popular dietary supplements ever since Stamen Grigorov first identified L. bulgaricus as the microbe required for the fermentation of yogurt in 1905.⁵⁵ In more than half of the experiments that employed this genus, L. reuteri or L. rhamnosus species were used. Potential mechanisms through which these bacterial species can alter gut physiology and ultimately reduce pain sensation are detailed elsewhere,^114,145 but it should be noted that in studies performed to date, the administration of L. rhamnosus more frequently leads to analgesia as compared with the administration of L. reuteri. In the current analysis, L. reuteri administration decreased pain in 33% of experiments (7 of 21) and had no effect in the remaining 67% of studies, whereas L. rhamnosus decreased pain in 89% of experiments (16 of 18) and had no effect in the remaining 11% of studies. One potential explanation of this efficacy discrepancy may be the fact that L. rhamnosus is not a native commensal of the murine gastrointestinal tract, whereas L. reuteri is.¹⁷² Thus, introduction of a novel bacterium may have exaggerated functional effects relative to the results obtained when levels of a bacterium that already exists within a microbial community are increased. The second most popularly used probiotic is Bifidobacterium (used in 18 of 143 experiments). Like Lactobacilli, Bifidobacteria are Gram-positive anaerobes that produce lactic acid during sugar fermentation. B. breve and B. lactis were the most frequently used species of Bifidobacteria. Mechanisms through which these specific bacterial species alter sensory signaling are not clearly resolved.

In addition to the administration of single bacterial species, probiotics were often administered in combination. Some combinations like VSL#3 are well-defined commercially available formulations. VSL#3 is classified as a medical food by the US Food and Drug Administration. Despite this, lot-specific differences—which may contribute to the functional effects observed following compound administration—have been reported for this formulation.¹³ One of the caveats of using probiotic mixtures in preclinical studies is the inability to distinguish which bacterium or bacteria are critical for the ultimate effect observed in vivo. This may not be the most important aspect of a preclinical experiment but may be relevant for translational success, particularly in situations where non-native bacteria are being introduced to an existing microbiome.

Two other factors included in this analysis of probiotic use were treatment duration and administration time frame (ie, were probiotics administered before, after, or concurrent with pain model). The current analysis found that probiotic delivery ranged from 4 to 126 days with an average of duration of 25 days (3.6 weeks). Initiation of probiotic treatment relative to injury/pain model induction also varied across experiments with 62 experiments starting administration after injury, 33 experiments starting administration before injury, and 3 experiments starting administration concurrent with the injury. Variations in probiotic induction timing can answer different experimental questions regarding probiotic prevention (before injury) or treatment (after injury) of pain; both questions are clinically important. All other variables aside, in the current analysis, probiotic initiation before and after injury resulted in similar levels of analgesia; 42% of experiments that initiated probiotic treatment before injury and 40% of experiments that initiated probiotic treatment after injury resulted in decreased pain. In conclusion, probiotic therapies are commonly used in preclinical pain studies and, to date, have only been reported as decreasing pain or having no effect on somatic/visceral sensitivity levels. Suggestions for future use include determining whether and what overlapping mechanisms individual species and strains use to induce analgesia across disparate pain conditions.

3.2.2. Trends in antibiotic usage

Since their initial discovery in the early 20th century, antibiotics have changed the course of medicine, making once deadly bacterial infections largely survivable. In recent decades, antibiotics have been increasingly used as a research tool, especially in the rapidly expanding field of microbiome science. In the current analysis, 75 experiments used antibiotics to manipulate the microbiome (Fig. 3, Supplemental Table 5, https://links.lww.com/PAIN/C113). Antibiotics were always administered orally; 58% of experiments dissolved antibiotics in drinking water, 31% administered the drugs through oral gavage, and 11% used a combination of both techniques. Limitations regarding oral antibiotic administration include the following: (1) the time course over which the effects of oral antibiotic administration persist are not the same as the time course for intravenous administration effects,⁷⁷ and (2) if administering drugs through drinking water, exact dose consumption will likely vary from night to night. In addition to drug administration route, the timing of drug initiation and length of time over which antibiotics were administered varied greatly. In the present analysis, the time course of antibiotic treatment ranged from 4 to 90 days, with a mean treatment duration of 18 days or 2.6 weeks. In the United States, the average outpatient antibiotic prescription duration is 10 days, a time frame that is both shorter than that used in most preclinical experiments and longer than that typically advised by the Infectious Diseases Society of America.⁷⁹ Apart from the benefits that chronic antibiotic use can provide in very specific disease states (eg, prevent pneumococcal sepsis in children with sickle cell disease⁵¹), excessive or prolonged antibiotic use is strongly discouraged so as to limit the spread of antibiotic-resistant microbes. Related to this, the persistence and functional effects of antibiotic-resistant microbes in rodents has not been explored in the pain field. Research into this topic may provide critical insight into the microbiological basis of the chronic pain that frequently occurs following sepsis or admission to the intensive care unit where antibiotic prescription is routine.¹¹

Initiation of antibiotic treatment relative to injury/pain model induction also varied across experiments. In 43% of experiments, antibiotic treatment started before injury, treatment started after injury in 13% of studies, treatment was started concurrent with injury in 7% of experiments, and in 23% of experiments, the antibiotic(s) was the injury. In clinical practice, few chronic pain conditions are predictable, save pain that may develop following surgery or chemotherapy treatment. Thus, experiments in which antibiotics are prescribed following pain model induction may provide more valuable insight into whether and how the microbiome can be manipulated toward a therapeutic end.

Although all antibiotics ultimately result in bacterial death, the mechanism through which this is achieved differs between antibiotic classes. There are 3 main mechanisms through which antibiotics can eventually lead to bacterial death: antibiotics either cause lysis through inhibition of bacterial cell well synthesis, or they slow reproduction and growth through protein synthesis inhibition or DNA replication interference. In the current analysis, 54% of antibiotics inhibited protein synthesis, 45% were active against the cell wall, and 0.9% interfered with DNA replication. Further subclassification revealed that 9 classes of antibiotics were represented across the 75 experiments; aminoglycosides (29%), beta-lactams (22%), glycopeptides (19%), and nitroimidazoles (19%) were the most frequently used drug classes.

Another key distinction between antibiotics is their oral bioavailability, or their ability to be absorbed in the gastrointestinal tract. If attempting to manipulate only the gut microbiome, this is a critically important factor to consider in experimental design. For instance, if the intent of antibiotic treatment is to induce/measure gut-specific changes in microbiome, the experimenter should choose a nonabsorbable antibiotic to restrict activity to the gastrointestinal tract. If absorbable antibiotics are used, experimenters may be unknowingly altering microbiomes in or on other body surfaces. In a recent test of this hypothesis, oral administration of absorbable antibiotics (doxycycline, cephalexin, trimethoprim/sulfamethoxazole) in humans was found to not only alter gut microbial populations but also induce skin dysbiosis.⁷¹ Similar effects were also observed in mice following oral administration of a cefazolin/enrofloxacin mix.⁸⁰ Given that so many preclinical microbiome pain studies incorporate somatic pain measures, one must wonder if antibiotic-induced behavioral changes can be exclusively ascribed to changes in the gut microbiome, or if changes in skin microbiome could also contribute to these results. Notably, contributions of skin microbes to somatosensation have received virtually no attention outside of itch that accompanies atopic dermatitis. Once rigorous methods are established that allow for independent manipulation of the rodent skin microbiome in the absence of gut microbiome manipulation, investigations should prioritize this research question.

In total, 13 unique antibiotics were used across experiments. To further analyze administration practice, daily antibiotic doses were converted to mg/kg units assuming average mouse/rat body weight and daily water intake. Extremely wide dose ranges were recorded for all antibiotics that were used in more than one study. For example, daily oral vancomycin doses ranged from 5 to 100 mg/kg in rodents. Given that the average daily oral human dose of this drug is 12.5 mg/kg, many studies are thus using a clinically irrelevant dose of vancomycin that may have unintended physiological consequences. Also noted in our analysis was oral administration of select antibiotics in rodents that are typically administered either topically or intramuscularly in humans. For example, bacitracin, a compound that is widely used by humans in ointment form to prevent topical infection, was orally administered to rodents in several experiments.

Finally, in the current analysis, antibiotics were administered either on their own (20% of experiments) or in combination with 1 or more additional antibiotics (80% of experiments). Notably, 49% of the antibiotic cocktail studies used the specific combination of vancomycin, metronidazole, ampicillin, and neomycin. Use of this specific cocktail to completely eliminate colonic commensals in mouse was first described 2 decades ago by Rakoff-Nahoum et al.,¹³⁴ and now, it is commonly used to induce a pseudo–germ-free state in rodents before FMT. In summary, current trends in preclinical antibiotic use include administering drug cocktails at higher doses and for longer periods relative to clinical prescriptions. In the future, it will be important to determine whether and how individual antibiotics, prescribed in clinically relevant doses, affect acute pain thresholds and chronic pain sensitivity in rodent models.

3.2.3. Trends in fecal microbiota transplant usage

Fecal microbiota/material transplant is a method through which the gut microbiota of one organism is introduced to the GI tract of another organism. Clinically, FMT is used to prevent recurrent Clostridioides difficile (formerly known as Clostridium difficile) infection¹³²; as of this publication, there are 2 FDA-approved biologics available for this indication. Select randomized control trials and case studies have also demonstrated FMT efficacy in alleviating the pain experienced by individuals with IBS, IBD, and fibromyalgia.^22,44,120 Significantly more research into stool donor and recipient compatibility is needed before FMT will be universally prescribed for these indications. In preclinical pain research, FMT has been used in both a parallel (ie, determining if transplant of a so-called healthy gut microbiome into an injured animal alleviates pain) and reciprocal or causative fashion (ie, determining whether transplant of a microbiome from an injured animal induces pain in a non-injured animal). In the current analysis, 71 experiments used FMT paradigms (Fig. 3) and manipulations of the causative variety indicated a role of the gut microbiome in the cutaneous or visceral hypersensitivity associated with models of spared nerve injury (SNI), chronic restraint stress,¹⁶⁸ water avoidance stress,¹⁰⁹ dinitrobenzene sulfonic acid (DNBS) colitis,¹⁰⁰ dextran sodium sulfate (DSS) colitis,⁴¹ and ovariectomy.¹⁶⁵

Preclinical FMT paradigms can vary widely, and thus, we analyzed several design and implementation parameters that may contribute to transplant efficacy (for further review on this topic, we recommend the guidelines written by Secombe et al.¹⁴⁴). First, we assessed the frequency with which intra- or interspecies FMT were performed. The most common transfers involved rodent donors and recipients of the same species (73%); human to rodent transfers were the second most popular application (21%), and very few studies administered FMT from one rodent species to another (6%).

The microbial status of the FMT recipient’s gastrointestinal tract also varied between the studies. In more than half of the experiments included in this analysis (52%), FMT was delivered to pseudo–germ-free animals; bona fide germ-free animals were used in 25% of experiments, naïve animals (ie, no prior microbiome manipulation) were used in 18% of experiments, and in 4% of experiments, the methods failed to state the condition of the FMT recipient. Although considered by many to be the gold standard for FMT experiments, germ-free animals that completely lack microbes in or on their bodies, are associated with several practical and biological limitations that should be considered when planning this type of experiment. First, germ-free animals are not accessible to all investigators based on the required specialized housing.¹⁵³ Second, although germ-free animals provide what is essentially a “blank slate” for colonization, they lack microbiome-induced training of the immune system which can lead to impaired immune functions in adulthood that are clinically irrelevant.^23,153,156 These immune system discrepancies can be reversed when colonized later in life with microbes from the same species, but if germ-free mice receive a human microbiota transplant for example, atypical immune responses will ensue.²⁷ In addition to this limitation, FMT from humans more successfully colonizes germ-free rats than germ-free mice suggesting that rats may be a better model organism when assessing the functional consequences of FMT from a human patient population.¹⁶⁷ Notably, in this current analysis there were only 2 experiments in which rats were the recipient of human FMT.^176,183

A low-cost alternative to germ-free animals is pseudo–germ-free animals which are created by treating otherwise naïve mice or rats with antibiotics. The most popular way to induce pseudo–germ-free status is to administer a four-drug cocktail consisting of ampicillin, vancomycin, neomycin, and metronidazole.^134,144 This drug cocktail can be administered orally and results in the elimination of Gram-positive, Gram-negative, and anaerobic bacteria. Despite the ease and relatively inexpensive cost associated with the generation of pseudo–germ-free mice, experimenters should remember the following caveats regarding their use: (1) antibiotic effects on the microbiome can persist for months after treatment cessation,³⁴ (2) antibiotic use can promote the emergence of antibiotic-resistant bacteria,⁶⁴ (3) the antibiotic cocktail described above does not completely sterilize the gut if administered for only 3 to 4 days, but does lead to fungal blooms if administered for >7 days,¹⁵⁹ and (4) antibiotics can increase gut permeability by disrupting the intestinal tight junction barrier.⁴⁷ Related to this last point, antibiotic-induced changes in rodent intestinal physiology may further enhance the effects of FMT from some patient populations that are also associated with intestinal inflammation or damage (eg, ulcerative colitis⁷⁸). Lastly, it is important to note that there is debate in the field as to whether antibiotic treatment is even necessary for colonization of donor bacteria; the mixed literature on this topic may result from species differences in donor microbiome.^49,70 Overall, neither germ-free nor pseudo–germ-free mice provide the perfect environment in which to complete FMT studies. Thus, experimenters must carefully consider the caveats of each model and decide if those limit interpretation of their results when designing FMT experiments.

Other important considerations in FMT study design include method of fecal preparation, route of administration, and duration/frequency of FMT. All of these factors should be explicitly described in article methods sections but are oftentimes not.¹⁴⁴ In the current analysis, freshly defecated feces were used in 48% of experiments, frozen fecal material was used in 41% of experiments, cagemate coprophagia was the transfer method in 6% of experiments, and in 4% of studies, the source of the FMT was not described. Use of frozen fecal material allows for flexibility of experimental procedures, and previous analyses found no difference in the richness or diversity of recipient microbiotas following FMT with fresh or frozen stool.⁴⁰ This suggests that both frozen and fresh samples can successfully colonize recipient animals. However, caution should be taken if using frozen samples as the number of freeze–thaw cycles, and cryoprotection mechanism can affect the diversity of microbial populations.¹⁴⁶ Regardless of storage properties, experimenters should also report if samples are being prepared in an aerobic or anaerobic fashion because this can affect the quality and viability of the sample.^12,21 In the current analysis, only 4 experiments used cohousing (ie, coprophagia) to transfer fecal microbiota between cagemates.^101,133 Although this is a low-cost and hands-off method for examining the effects of FMT, this technique does not allow for precise control over amount and timing of FMT in recipients.

Cagemate transfer withstanding, the frequency of FMT in the current analysis varied widely and included experiments in which only a single transplant was performed (35%) or in which multiple transplants were performed (65%). The majority (94%) of experiments performed the FMT through oral gavage as opposed to 4% of experiments that used intracolonic infusion. In repeated FMT experiments, the average FMT frequency was 8 transfers over an 8-day period. However, additional experiments ranged from 3 to 24 days in length. Although transplant frequency may be dictated by experimental question or behavior testing duration, repeated transplants may prove to be more robust given that previous studies demonstrated that a single microbiome transplant may not allow for sufficient engraftment of human microbiota of human donor material in mice.⁶¹ Finally, the FMT initiation time frame was analyzed relative to injury model induction. Similar to what was observed with probiotic and antibiotic treatment, the current analysis found a relatively even split between experiments that started FMT before (27 experiments) or after (24 experiments) injury model induction; in a third group of experiments, FMT was considered the injury model itself (18 experiments).

The final factor considered in this analysis is the inclusion/specificity of the FMT control. Given that >75% of FMT experiments described herein were completed in germ-free or pseudo–germ-free mice, it seems most appropriate for FMT controls to include live microbes because that is what is being delivered to the experimental group. For example, when determining if the fecal microbiota of DSS-induced colitis mice was sufficient to drive visceral hypersensitivity in pseudo–germ-free animals, Esquerre et al.⁴¹ compared behavioral responses from animals that received FMT from DSS-treated mice with animals that received FMT from vehicle-treated mice; in other words, neither experimental nor control animals were pseudo–germ-free following FMT because both had been colonized, albeit with different microbiomes. This biologically appropriate control was used in 55% of the FMT studies included in this analysis. In the remaining 45% of studies, controls were either not described in the methods section or consisted of a sterile saline infusion. Given that the presence of bacteria, regardless of genus or species, can have drastic effects on gastrointestinal physiology, it is critical to include a biological control that will allow for the appropriate interpretation of results. In other words, this control will allow one to answer “Is my effect simply due to the presence of bacteria in the GI tract or specific to the environment from whence those bacteria were collected?”

In conclusion, FMT is an effective strategy for studying the causality of the microbiota in driving or alleviating acute and chronic pain conditions. When developing FMT paradigms, experimenters need to carefully consider the manner in which donor fecal material is collected and processed, the microbial status of the recipient animal, the timing, frequency, and delivery method of FMT, and what specific control is being used to appropriately interpret data. These and other factors related to rigor and reproducibility were recently highlighted in the Guidelines for Reporting on Animal Fecal Transplantation.¹⁴⁴

3.3. Does manipulating the microbiome affect baseline pain sensitivity in animals?

After detailing the methods through which the microbiome is being manipulated in preclinical pain studies, we next examined what effect microbiome manipulation has on acute pain sensitivity in otherwise naïve rodents. To this end, we analyzed the results of all experiments that were performed in naïve mice and rats (Supplemental Table 2, https://links.lww.com/PAIN/C113). In total, 106 behavior assessments were made in injury-free rodents; 42% of experiments incorporated probiotic manipulation, 29% of experiments manipulated the microbiome with antibiotics, 24% of experiments used FMT paradigms, and 5% of experiments compared pain thresholds between germ-free mice and conventionally raised controls (Fig. 4). Below, we discuss the results of experiments that measured visceral or somatic pain in non-injured animals and conclude what effects the microbiome has on acute pain sensitivity.

Behavior measurements performed in germ-free animals are the most comprehensive way to assess if microbes in or on the body affect acute pain thresholds. To date, 3 studies have measured visceral pain behaviors in germ-free mice; in 2 of these studies, germ-free mice had elevated visceromotor responses (VMR) to colorectal distension (CRD) relative to conventionally colonized controls,^101,130 and in the third study, no difference was observed between CRD responses from germ-free and conventional controls.¹⁶⁰ What variables differed between these experiments and thus may account for the result discrepancies? In the null result study, germ-free and conventionally raised female Swiss webster mice were used.¹⁶⁰ Of note, male germ-free mice of the same strain and age exhibited visceral hypersensitivity relative to conventional controls in a second study performed by the same laboratory.¹⁰¹ In the third study, visceral hypersensitivity was noted in both male and female germ-free C57BL/6 mice relative to conventional controls. Thus, complete absence of the microbiome from birth may have both sex-specific and strain-specific effects on mechanisms underlying visceral pain.

Fewer studies have used germ-free animals to assess the role of the microbiome in cutaneous sensitivity. To date, only 2 experiments have performed reflexive somatic pain behavior tests in germ-free mice. Both studies assessed mechanical sensitivity using von Frey filaments; one study reported decreased hindpaw mechanical withdrawal thresholds in germ-free mice (ie, germ-free mice exhibited mechanical allodynia⁷⁴), and the other reported no difference in facial withdrawal thresholds between germ-free mice and conventionally colonized controls.¹⁵⁸ Given that we do not believe that von Frey filament application ever induces a noxious sensation in uninjured animals,¹³⁸ this means that the contribution of the microbiome to cutaneous pain sensation, regardless of modality, has never been examined in germ-free rodents.

As a complement to studies performed in germ-free animals, antibiotic depletion of microbes can provide insight into how bacteria acutely modulate pain sensitivity while ignoring the developmental effects that are inherent to germ-free models. To date, 8 studies have administered antibiotics to naïve rodents, then measured visceral pain in response to mechanical distension of the colon or intracolonic instillation of noxious chemicals shortly thereafter. In 5 studies (63%), antibiotic treatment increased visceral pain behaviors,^{41,48,100,161,164} in 2 studies (25%), antibiotic treatment decreased visceral pain behaviors,^4,62 and in the final study, antibiotic treatment had no effect on visceral pain.¹⁶² Antibiotic cocktails were used in all but one study; in the only experiment to use a single antibiotic, minocycline administration (50 mg/kg) had no effect on abdominal von Frey thresholds.¹⁶² There was no consensus on cocktail components that were associated with pain. In fact, 2 groups independently observed that the combination of oral bacitracin and neomycin increased responses to colorectal distension^48,161 whereas a third group observed that this same drug combination decreased spontaneous pain behaviors exhibited following intracolonic installation of capsaicin.⁴ In the second and final study that observed antibiotic-induced visceral hyposensitivity, cocktail components included ampicillin, vancomycin, and metronidazole—drugs that were also included in cocktails that induced visceral hypersensitivity⁴¹—as well as ciprofloxacin and imiperem.⁶² It is possible that the addition of ciprofloxacin and imiperem, a fluoroquinolone and carbapenem (synthetic β-lactam), respectively, could eliminate pain-inducing bacteria that persist following treatment with the 3-drug cocktail and thus lead to visceral hypoalgesia. This hypothesis could be empirically tested with 16S ribosomal RNA sequencing. Finally, it should be noted that a ninth study also measured visceral sensitivity following antibiotic use but after a particularly extended drug washout period. O’Mahony et al.¹¹⁹ administered vancomycin to rat pups on postnatal days 4-13 then observed drug concentration-dependent visceral hypersensitivity when CRD was performed 9 weeks later. The extended time course for this drug effect may be attributed to the administration time frame (ie, in early life during circuit development) or be a generalizable quality of vancomycin and related drugs.

More studies (N = 21) have examined the effects of antibiotics on cutaneous sensitivity of naïve rodents. In 10 of these experiments, antibiotic administration increased somatic sensitivity; in the remaining 11 studies, antibiotic-treated animal behaviors was identical to that observed in controls. Notably, antibiotic treatment never decreased somatic pain in naïve mice. The nociceptive effects of antibiotic treatment were not limited to one specific sensory modality because these drugs increased both heat^87,168,178 and mechanical^{63,74,164,168,173,178} sensitivity in addition to inducing facial grimace.¹⁶² Similar to observations made in visceral studies, antibiotics were almost always administered in cocktails and identical drug combinations were reported to both increase and have no effect on cutaneous pain, sometimes by the same laboratory. The popular combination of vancomycin, metronidazole, ampicillin, and neomycin was used in 8 experiments and had no effect on pain behaviors in 7 of these reports^35,41,104; in one report, this cocktail decreased mouse hindpaw mechanical thresholds.⁷⁴ Interestingly, hindpaw heat and mechanical sensitivity increased in all experiments wherein animals received the combination of neomycin, ampicillin, and metronidazole without vancomycin.^63,168,178 Future sequencing analyses would allow for the comparison of bacteria that persist following treatment with these 2 different cocktails and ultimately lead to the identification of bacteria that can directly or indirectly modulate nociceptive somatosensory circuits.

In summary, complete absence or depletion of the microbiome in naïve rodents increased sensitivity in the majority of experiments performed to date (18 of 34 experiments); germ-free mice never exhibited hyposensitivity relative to conventionally colonized controls, and antibiotic treatment only decreased sensitivity in 7% of studies (2 of 29 experiments). From this analysis, we can conclude that in the absence of injury, the collective activity of commensal microbes more-often-than-not contributes to an antinociceptive tone across internal and external surfaces of the rodent body. Microbe absence or depletion frequently results in increased sensitivity to exogenous stimulus application; the cellular and molecular basis of this heightened sensitivity should be the topic of future investigations.

In a reciprocal fashion, we also analyzed studies in which probiotics were administered to naïve mice. Probiotic administration decreased cutaneous or visceral sensitivity in 29% of experiments (13 of 45) and had no effect on pain measures in the remaining 71% of studies (32 of 45). Notably, probiotic administration never induced cutaneous or visceral hypersensitivity in naïve rodents. In attempting to identify factors that differentiated between analgesic and null effect studies, we determined the following: (1) probiotic treatment was only analgesic in mechanical tests and never in thermal tests, (2) all probiotics that decreased sensitivity also had no effect when tested in additional experiments save for Bifidobacterium infantis¹⁰⁶ and Lactobacillus acidophilus,¹⁴¹ which decreased visceral sensitivity in the single experiments in which they were used. The modality specificity of probiotic analgesia would be a notable finding if it stands following testing with additional bacteria and behavior tests. In the current set of experiments, all probiotics were delivered orally and analgesia was observed most frequently during mechanical stimulation of the colorectum,^{37,38,73,106,126,141,161} an anatomical site where probiotics could directly modulate activity of the nociceptors that relay distension signals to the central nervous system. After 8 weeks of treatment, probiotics also decreased hindpaw mechanical sensitivity in 2 experiments that were performed by the same team.^8,32 Given that hindpaw afferents are not being directly acted upon by the orally administered probiotics, this behavior effect may be explained by changes in immune activity, or more likely, plasticity in ascending or descending sensory circuits. If the latter is true, it would thus be surprising to not observe generalized decreases in mechanical, thermal, and chemical sensation across all surfaces of the body; future experiments should directly investigate the reasons for probiotic modality specificity. In summary, probiotics more-often-than-not have no effect on acute pain thresholds in uninjured rodents, likely because there is limited functional effect when increasing the number of microbes that already exist in a homeostatic environment.

3.4. Does manipulating the microbiome affect pain that arises in injury models?

After assessing how the microbiome contributes to pain in naïve animals, we next reviewed all experiments performed in injured rodents (Fig. 5). In total, 197 experiments measured pain in injured rodents after a microbiome manipulation (Supplemental Table 3, https://links.lww.com/PAIN/C113); neuropathic pain models were used in 28% of experiments, cutaneous inflammatory injuries were used in 19% of experiments, stress was used to induce pain in 18% of experiments, visceral pain models were used in 16% of experiments, diet was used to induce pain in 4% of experiments, and the remaining 15% of studies used other pain models. Below, we discuss what effect microbiome manipulation had on the pain that resulted in each model class.

3.4.1. Role of the microbiome in gastrointestinal (GI) pain models

Intestinal pain was a logical starting point for preclinical microbiome studies given that the gut microbiota resides therein. Indeed, relative to healthy controls, humans with chronic visceral pain conditions, including IBS,¹²⁸ ulcerative colitis, and Crohn’s disease,¹²⁷ have well-described gut dysbiosis. Thus, unsurprisingly, microbiome manipulations, including probiotics^67,112 and FMT,^39,45 have been touted as potential therapies for these conditions, whereas antibiotics have received more attention as a potential risk factor for the development of these diseases.^{46,82,108,117} Rodent colitis models attempting to mirror patient physiology also report gut dysbiosis (eg, dysbiosis observed in DSS-⁴¹ and DNBS-induced colitis¹⁰⁰) and, as such, have been the focus of many preclinical microbiome pain experiments. To date, 32 experiments have measured the effects of various microbiome manipulations in animal models of visceral hypersensitivity. An additional 9 experiments used FMT to determine if the microbiota of patients or animals with visceral hypersensitivity could drive pain in naïve recipients in the absence of disease pathophysiology. In all but one⁵⁸ of the non-FMT experiments, visceral hypersensitivity was induced through chemical means. Dextran sodium sulfate (DSS) was the most popular chemical used to insight pathology (7 of 32 experiments)^41,52,162; additional methods are listed in Supplemental Table 3, https://links.lww.com/PAIN/C113. Fittingly, colorectal distension (CRD) was the most popular method used to assess visceral hypersensitivity (used in 22 of the 32 experiments).

Antibiotics were used in 12 experiments with mixed results. Supporting a potential role of antibiotics in visceral pain development, use of the ampicillin, neomycin, vancomycin, and metronidazole cocktail increased visceral hypersensitivity—as measured by CRD but notably, not abdominal von Frey—in the DSS colitis model.⁴¹ This same cocktail had no effect in any of the other visceral pain models in which it was tested.⁴¹ When administered to adult animals, the combination of bacitracin and neomycin had no effect in some visceral pain models but decreased pain in other models.^3,5 Finally, administration of minocycline or rifaximin in isolation decreased visceral pain in all models tested.^162,179 Given that minocycline is also widely used as a microglia inhibitor,¹¹³ one must wonder if the analgesic effects of this drug can be exclusively ascribed to changes in the microbiota make-up. Regardless, in rodents with ongoing visceral pain, antibiotic administration fails to produce the level of analgesia observed when the same drugs are administered in somatic pain models. Probiotics successfully alleviated visceral hypersensitivity in all 17 experiments in which they were used, regardless of genus or species.^{10,30,31,52,58,72,76,90,97,99,115,131,139,141} This impressive statistic may be a confound of positive date publication biases or may indicate the extent to which the host gut microbiome is altered in these experimental models. In other words, if chemical perturbation of the gastrointestinal tract alters microbial communities to a great enough extent, (re)introduction of “healthy” bacteria may be sufficient to provide analgesia in that context. Given that these same chemical perturbations are not the actual cause of IBS or IBD, it is unclear if this high level of probiotic preclinical success will translate to patient populations.

In this analysis, FMT was used as both a treatment for visceral hypersensitivity and as a means of inducing visceral pain in otherwise naïve recipients. In 3 experiments, fecal microbiota from control animals was transplanted into rodents that had dinitrobenzene sulfonic acid (DNBS)–induced or trinitrobenzene sulfonic acid (TNBS)–induced colitis^100,103 or postinfectious IBD¹⁰; this treatment alleviated pain in all models tested. Reciprocally, FMT from patients with IBS induced pain in recipient rodents in 4 separate experiments.^{50,122,176,183} Visceral hypersensitivity was also reported in mice following FMT from an infant with colic.⁴² Finally, FMT from rodents with DSS- or DNBS-induced colitis was found to induce visceral hypersensitivity—but not somatic hypersensitivity—in naïve recipients.^41,100 Collectively, these findings suggest that alterations in local signaling between microbes, immune cells, and extrinsic sensory neurons can alter ascending nociceptive circuits that relay visceral pain signals to the central nervous system. Tremendous therapeutic potential lies in better understanding of the complex, multidirectional signaling cascades that exist in the gastrointestinal tract.

3.4.2. Role of the microbiome in non-GI inflammatory pain models

Following injury, appropriate inflammatory responses are crucial for tissue healing and eventual pain resolution.¹²³ Given the intimate, coevolutionary relationship between the microbiota and innate and adaptive immune systems (reviewed by Zheng et al.¹⁸²), it is perhaps unsurprising that preclinical pain researchers have examined the role of the microbiome in cutaneous pain models that result from immune system activation. To date, 38 studies have investigated the role of the microbiome in non-GI inflammatory pain models. Arthritis models were used in 23 studies, cutaneous inflammogen injection was used in 12 experiments, and postsurgical pain models were used in 2 experiments.

Arthritis is generally defined as inflammation or swelling of one or more joints. Although there are many types of arthritis, preclinical pain research has primarily investigated what role the microbiome plays in osteoarthritis (22 experiments) and gout (1 experiment)-associated pain. Osteoarthritis (OA), the most common form of arthritis, is associated with cartilage break down within a joint and resulting changes the underlying bone. In the current analysis, OA was induced through several methods, including anterior cruciate ligament transection, partial medial meniscectomy, and injection of complete Freund’s adjuvant (CFA) or monoiodoacetate into the knee joint. Probiotic administration was the only microbiome manipulation performed in OA models. In 18 of the 22 OA experiments, probiotic administration decreased mechanical or heat hypersensitivity.^{26,68,69,84,94–96,121,148} In the remaining 4 experiments, Bacillus subtilis and Limosilactobacillus reuteri (formerly known as Lactobacillus reuteri) administration had no effect on pain that developed in the partial medial meniscectomy model.¹²¹ In the single gout experiment included in this analysis, germ-free mice did not develop the monosodium urate–induced mechanical allodynia that was observed in conventionally colonized controls.¹⁶³ In summary, although a causative role of the microbiome has not yet been demonstrated in any arthritis model, select probiotic treatments may alleviate pain that arises in osteoarthritis. Future studies should test the extent to which the microbiome is required for pain in various arthritis models, including models of rheumatoid arthritis, which may involve additional neuroimmune interactions not captured in OA models.

The most frequently used inflammatory pain models in basic pain research are those in which a known inflammatory agent is directly injected into the hindpaw.¹⁴³ To date, 12 experiments have investigated the role of the microbiome in the following types of injections; CFA (6 experiments), formalin (2 experiments), carrageenan (2 experiments), and lipopolysaccharide (LPS; 2 experiments). Compared with conventionally colonized controls, germ-free mice exhibited fewer pain-like behaviors when injected with formalin and developed less mechanical allodynia when injected with carrageenan or LPS.⁹ Colonization of germ-free mice with fecal material from conventional controls reversed this protective effect, thus suggesting that a healthy microbiome is required for the development of LPS-associated mechanical allodynia.⁹ In a similar manner, antibiotic treatment decreased ongoing pain behaviors¹²⁴ and mechanical allodynia¹⁶⁹ in rats that received formalin or LPS injection, respectively. Similar to OA models, hindpaw inflammation model pain was not relieved by B. subtilis or L reuteri treatment⁶⁵; no other probiotics were tested in these models. Finally, FMT experiments demonstrated that transfer of a naïve gut microbiome into CFA-treated mice partially alleviated the mechanical allodynia and heat hyperalgesia associated with this model.¹⁷⁵

Chronic postsurgical pain is estimated to impact between 10% and 30% of surgical patients 1-year postoperation.²⁰ Given that surgery, by definition, involves manipulation of the skin, which contains vibrant bacterial communities, it is likely that cutaneous microbiomes play a role in postoperative pain development, even when appropriate sterilization is used.⁹⁸ To date, the role of the skin microbiome in postoperative pain has not been examined in animal models. However, manipulation of the gut microbiome has been performed in 2 preclinical experiments that modeled postoperative pain. In both studies, FMT was performed following surgery to alter the microbiome of pseudo–germ-free mice. One study transferred feces from human subjects with congenital insensitivity to pain with anhidrosis (CIPA) to mice with acute incisional pain and observed a decrease in mechanical allodynia.¹⁷⁸ In the second study, FMT from rats exhibiting postoperative pain had no effect on pain behaviors in germ-free recipients.⁹¹ With only 2 relatively niche articles published on postsurgery pain and the microbiome, there are clear gaps to be filled in this research area.

3.4.3. Role of the microbiome in neuropathic pain models

Chronic neuropathic pain affects approximately 10% of the general population^57,152 and results from metabolic, traumatic, or drug-induced injury to peripheral or central nervous system structures. Neuropathic pain models are incredibly popular in preclinical pain research—more than 40% of experiments published in the journal Pain over the past 5 years used a model of this kind¹⁴³—and these models are also becoming more popular in preclinical microbiome pain studies. In the current analysis, 28% of the experiments performed in injured animals incorporated a neuropathic injury model. Of those experiments, 49% used models that involved chemotherapy treatment, 47% used traumatic nerve injury models, and 4% used diabetic neuropathy models.

Chemotherapy-related pain has become an increasingly prevalent condition as the numbers of cancer diagnoses and successfully treated patients have risen.²⁸ Chemotherapy-related pain, which often results from direct damage to peripheral nerves, can be a dose-limiting side effect of treatment, and thus, many efforts have been directed to developing novel preventative and analgesic therapies for this symptom. In preclinical studies, rodents are treated with the same chemotherapeutic agents used in patients, resulting in relatively high model validity. In the current analysis, 25 experiments used chemotherapy agents; 58% of experiments administered paclitaxel and 42% of experiments administered oxaliplatin. Unsurprisingly, 16S rRNA and rDNA sequencing have revealed that, even in the absence of a tumor, paclitaxel¹⁵⁰ and oxaliplatin⁹² treatment alter the composition of the rodent gut microbiome. Thus, manipulation of the microbiome may have significant therapeutic potential for this condition. Antibiotics—specifically the combination of ampicillin, neomycin, vancomycin, and metronidazole—were administered in 11 preclinical chemotherapy studies and alleviated mechanical or cold allodynia in 6 of these experiments.^104,135,149 In the remaining 5 experiments, the same drug cocktail had no effect on chemotherapy-related hypersensitivity.^104,150 Probiotics, specifically the SLAB51 formulation, was used in 2 experiments and decreased chemotherapy-related mechanical allodynia in both assessments.²⁹ Finally, in the 13 chemotherapy-related experiments that used FMT, it was determined that (1) an intact microbiome is required for the development of chemotherapy-related pain¹⁴⁹ (2) the microbiome of select strains of mice,¹²⁶ confer protection from chemotherapy-related pain,¹³⁵ and perhaps most notably, (3) FMT from vehicle treated or naïve animals into chemotherapy-treated animals can alleviate mechanical and cold allodynia associated with this injury.^105,150 Interestingly, no study has determined if FMT from chemotherapy-treated animals is capable of inducing pain in animals that have not received the same drug. If true, this would reveal an entirely new environment (ie, gastrointestinal tract as opposed to peripheral nerve landscape) and set of therapeutic targets that could be developed for chemotherapy-related pain management. Finally, it should be noted that 1 article investigated the role of the microbiome in cancer-related pain in the absence of chemotherapy treatment. In a single experiment, administration of L. rhamnosus alleviated the mechanical allodynia that developed following tumor implantation in the tibia.¹⁷⁴ Future microbiome investigations should prioritize experiments in which a tumor and chemotherapeutics are administered to animals to better capture the complex neuroimmune interactions that are affected by both experimental variables.

Traumatic nerve injury models are incredibly popular amongst basic pain scientists; approximately 70% of preclinical neuropathic pain articles published in Pain over the past 5 years used a traumatic nerve injury model.¹⁴³ In the preclinical microbiome pain literature, traumatic nerve injury models were used less frequently; chronic constriction injury (CCI) was used in 16 studies, spared nerve injury (SNI) was performed in 5 studies, and spinal never transection was used in 3 studies. Despite significant anatomical separation between the injured peripheral nerve and gut, 16S rRNA or rDNA sequencing revealed that the gut microbiome of animals who underwent CCI,²⁵ SNI,¹⁹ or spinal nerve transection⁸⁶ differs from sham control animals. These differences may result from systemic changes in immune activity following injury but, regardless, suggest that microbiome manipulation may exert some therapeutic potential in animals with nerve injury. Administration of the ampicillin, neomycin, metronidazole, and vancomycin cocktail effectively decreased heat and mechanical allodynia in 6 of the 8 experiments in which it was tested^35,86,159 and had no effect in the remaining 2 experiments.¹⁰⁴ Probiotic formulations containing at least 1 species of Lactobacilli alleviated traumatic nerve injury–related hypersensitivity in all 6 experiments in which they were used^24,86,147; isolated use of L. reuteri or Bifidobacterium BL5b had no effect on nerve injury–related heat or mechanical hypersensitivity.⁶⁵ Finally, like in chemotherapy models, FMT was the only manipulation that increased pain associated with nerve injury. Specifically, pseudo–germ-free recipient animals that received FMT from animals that exhibited anhedonia following SNI developed pain following transplant.¹⁷⁰ In conclusion, it will be important to uncover the mechanisms through which traumatic nerve injury alters the gut or skin microbiome so that new therapies may be developed to help those suffering from nerve injury pain, particularly those individuals who are also suffering from comorbid mood disorders.

3.4.4. Role of the microbiome in diabetes and diet-induced pain models

Although typically grouped with neuropathic pain models, discussion of preclinical diabetes models—especially those mirroring type 2 diabetes—may have more in common with the diet-induced pain models included in the current analysis. Despite having obvious associations with gastrointestinal chemosensation, diet-induced pain models were only used in 7 experiments and diabetes-related pain was the topic of only 5 preclinical microbiome pain studies. Western or high-fat diets were used in 4 experiments,^8,17,32 a high FODMAP (fermentable oligosaccharides, disaccharides, monosaccharides, and polyols) diet was used in 1 experiment,¹⁸³ and alcohol consumption was used in the final 2 diet experiments.⁵⁴ Animals were maintained on these diets for 2 to 14 weeks before microbiome manipulation. Notably, in the experiments with longer diet manipulation, animals started exhibiting changes in skin innervation and primary afferent gene expression, 2 hallmarks of diabetic neuropathy.¹⁷ Regardless of the length of diet manipulation, administration of L rhamnosus or the De Simone Formulation, which includes 4 other species of Lactobacilli. successfully alleviated diet-induced mechanical allodynia in all models tested.^8,32,54 Antibiotic administration also successfully reversed diet-induced cutaneous and visceral mechanical hypersensitivity^17,183 but failed to alleviate heat hyperalgesia.¹⁷

All bona fide diabetic neuropathy experiments were performed by the same group and used the streptozotocin model of type 1 diabetes,¹⁰⁴ a model that selectively destroys pancreatic β cells. In this model, treatment with the ampicillin, neomycin, metronidazole, and vancomycin cocktail alleviated diabetes-related mechanical allodynia but had no effect on heat hypersensitivity.¹⁰⁴ To date, no probiotics have been tested in diabetic neuropathy models. Finally, FMT experiments from this group demonstrated that an intact “healthy” gut microbiome is required for the development of streptozotocin-related pain.¹⁰⁴ Moving forward, diabetic neuropathy models, particularly those that involve diet manipulations, should be a focus of the preclinical microbiome pain field given the high prevalence of both diabetes and diabetic neuropathy around the world and the logical associations between microbiome placement and diabetes pathophysiology.⁵⁹

3.4.5. Role of the microbiome in migraine models

Despite being the second leading cause of disability world-wide,¹ migraine research does not receive proportional attention from preclinical pain researchers.¹⁴³ Even fewer articles have questioned if the microbiome contributes to migraine headache despite the high prevalence of gastrointestinal symptoms in migraine (eg, nausea, vomiting) and the new migraine therapies that target calcitonin gene-related peptide (CGRP), a critical signaling molecule in normal gut physiology.⁶ In this analysis, migraine was the primary focus of only 2 articles; 17 total experiments were carried out across the 2 articles.^74,158 In both articles, migraine was induced through injection of nitroglycerin, a nitric oxide donor. Three experiments injected nitroglycerin into germ-free mice. Relative to conventionally colonized mice, germ-free mice exhibited increased periorbital mechanical allodynia¹⁵⁸ but similar levels of hindpaw mechanical sensitivity and grooming following nitroglycerin injection.⁷⁴ In a parallel fashion, antibiotic administration also increased periorbital mechanical sensitivity¹⁵⁸ but had no effect on hindpaw mechanical sensitivity⁷⁴ in mice that received nitroglycerin injections. Thus, the effects of microbiome depletion or absence are anatomically restricted to the facial region in migraine models. This is notable given that migraine pain is thought to result from the activation of sensory neurons in the trigeminal nerve. Unlike paw-innervating afferents that have a higher likelihood of communicating with gut-innervating afferent cell bodies in the dorsal root ganglia or central terminals in the spinal cord, trigeminal afferent fibers and cell bodies are anatomically separated from the spinal afferent system. Thus, the effects of microbiome depletion in migraine models likely result from compounding changes that specifically occur in the trigeminal system. Probiotics, specifically the VSL#3 formulation, were administered in only 1 experiment and had no effect on orofacial mechanical sensitivity.¹⁵⁸ Finally, FMT was used to determine if the gut contents of migraineurs induce migraine-like behaviors in germ-free recipients. Indeed, mice that received FMT from individuals with migraine, but not those who received FMT from healthy control humans, developed hindpaw mechanical allodynia and exhibited increased grooming behaviors.⁷⁴ Thus, from the limited studies completed on this topic, microbiome dysbiosis or depletion appears to increase the likelihood of migraine-like symptom development. Future experiments should determine whether disrupted CGRP signaling is a critical factor that drives both migraine headache and microbiome dysbiosis.

3.4.6. Role of the microbiome in ovariectomy models

As life expectancy continues to increase around the world, a greater number of women are experiencing menopause and developing postmenopausal symptoms, including pain hypersensitivity, mood fluctuations, osteoporosis, and cardiovascular disease.³³ Ovariectomy is a surgical procedure done in nonhuman species wherein the ovaries are removed from the body, thus eliminating the primary estrogen source in females and creating a state that mimics human menopausal and/or postmenopausal conditions.¹⁵⁴ In this analysis, 3 articles investigated how microbiome manipulations alter ovariectomy-related pain.^88,93,165 In total, 4 experiments used probiotics to manipulate the microbiome of ovariectomized animals. Administration of different Lactobacilli^88,93 and the VSL#3 formulation⁹³ decreased the mechanical hypersensitivity that develops following ovariectomy. Fecal microbiota transplant was performed in 1 of the 3 articles. In these experiments, transfer of control animal feces into ovariectomized mice decreased heat hyperalgesia and mechanical hypersensitivity, and in a reciprocal fashion, FMT from ovariectomized mice induced heat and mechanical hypersensitivity in control recipients.¹⁶⁵ However, it should be noted that these changes in pain-like behaviors did not occur until at least 5 weeks following the end of FMT, so it is unclear whether this was the direct result of FMT or a confounded consequence of aging. Future studies completed in ovariectomized models should consider how experimental manipulations alter not only the gut microbiome but also the vaginal microbiome, which is becoming an increasingly popular target for the treatment of gynecological diseases.¹⁰²

3.4.7. Role of the microbiome in stress-induced pain models

Stress is a widely accepted risk factor for IBS¹⁴ and an aggravator of IBD symptoms.¹⁵⁷ Therefore, stress is commonly used in animal models to induce IBS and IBD symptoms, including visceral hypersensitivity. In the current analysis, 35 experiments used stress models to study the effects of microbiome manipulation on pain; neonatal limited bedding or maternal separation was used in 15 experiments, restraint stress was used in 11 experiments, and water avoidance stress was used in 9 experiments. All but 5 experiments used probiotics to manipulate the gut microbiome, and in 83% of these experiments, probiotics alleviated stress-induced visceral hypersensitivity.^{2,7,31,36,43,53,85,110,115,116,177,180} No obvious explanation could be derived for the experiments in which probiotics had no effect on pain^43,52,85,110; the same probiotics that had no effect in certain stress-induced pain models had analgesic effects in other models and probiotics alleviated pain in all stress models with relatively similar efficacy. Antibiotics were used in 4 experiments and successfully alleviated stress-induced mechanical allodynia and spontaneous pain each time they were employed.^5,53,171 In the final experiment, FMT from naïve controls alleviated pain when transplanted into rats that had water avoidance stress-induced visceral hypersensitivity.¹⁰⁹ These results parallel reciprocal FMT procedures performed in naïve rodents wherein FMT from animals that underwent water avoidance stress¹⁰⁹ or chronic restraint stress¹⁶⁸ induced mechanical and heat hypersensitivity in naïve controls.

3.4.8. Role of the microbiome in “other” pain models

Few additional pain models have been used in the preclinical microbiome literature. Despite being the leading cause of disability worldwide,¹ low back pain is the topic of too few preclinical pain studies, including only 1 microbiome article in this analysis. This article performed lumbar disc herniation, then observed that Lacticaseibacillus paracasei (formerly known as Lactobacillus paracasei) administration decreased the mechanical and heat hypersensitivity that developed in this model.¹⁶⁶ A final article examined the extent to which the gut microbiome contributes to pain in interstitial cystitis/bladder pain syndrome (IC/BPS), a debilitating chronic pain condition categorized by generalized pelvic pain and increased urinary frequency and urgency. Fecal microbiota transplant from IC/BPS patients exacerbated visceral hypersensitivity in mice that lacked acyloxyacyl hydrolase (AOAH), a gene associated with spontaneous bladder pain generation, and in a reciprocal fashion, FMT from wild-type mice decreased bladder and generalized pelvic pain in AOAH mice.¹³³

4. Discussion

This is an exciting time to be in preclinical microbiome pain research. Relative to other subfields of mammalian somatosensation, we have a very limited understanding of how microbes communicate with the sensory nervous system in the uninjured state and even less knowledge regarding how these interactions change in disease states or following injury. For example, although viruses and fungi are known to inhabit the same mammalian tissues as bacteria and have similarly complex interactions with host nervous and immune systems, exceedingly few articles have manipulated the virome or mycobiome, then measured acute or chronic pain behaviors in laboratory animals (only 1 article in our current assessment manipulated the mycobiome).¹⁸ Given continuous technical advancements, declining costs of genetic sequencing, and increasing access to human patient microbiome samples, we optimistically predict that the microbiome will yield important therapeutic targets for chronic pain management in the near future. This review aims to expedite that process by summarizing critical gaps in the literature and providing best practice suggestions for those just entering the preclinical microbiome space.

4.1. Suggestions for planning, performing, and interpreting preclinical microbiome pain experiments

Below, we list experimental suggestions for future preclinical pain microbiome experiments:

• (1) Include multiple cages of rodents in each experimental group. Coprophagia limits the microbial diversity between cagemates. Thus, to ensure that your experimental result is not simply an artifact driven by limited sample size, include multiple cages of male and female rodents in each treatment or control group.
• (2) Use clinically relevant doses and administration routes when treating rodents with antibiotics. To date, preclinical antibiotic use has failed to mirror patient prescribing practices despite the similarity between rodent and human microbiomes. In future studies, antibiotic dosing, prescription length, and administration route should more accurately mimic clinical practice.
• (3) Report all details for each microbiome manipulation. In completing this analysis, we read too many methods sections that failed to describe experimental manipulations with a level of detail that would have allowed us to repeat the experiment. This information is not only critical for reproducibility’s sake but also required for appropriate interpretation of results. We believe that “more is more” when it comes to reporting experimental methods.
• (4) Include appropriate controls, particularly in FMT paradigms. In the current analysis, almost half of the FMT experiments transferred resuspended fecal material into experimental animals and sterile saline into control animals. The introduction of bacteria into germ-free or pseudo–germ-free animals can have dramatic consequences on immune and neuronal function. Thus, we argue that the better control in FMT experiments would be transplant of resuspended fecal material from control animals/humans.
• (5) Use a battery of pain behavior tests to fully assess all aspects of the rodent pain experience. Until it is clear if a given microbiota-based therapy is more likely to affect reflexive–nociceptive pain behaviors or affective–motivational pain behaviors, both should be included in experimental design to increase the likelihood of translational success.

4.2. Current knowledge and future research opportunities

General conclusions that can be made from this current analysis include the following:

• (1) Manipulation of the microbiome affects both visceral and cutaneous sensitivity.
• (2) In naïve rodents, symbiotic microbes contribute to body-wide antinociceptive tone. When microbial populations are disrupted with antibiotics, animals typically become hypersensitive to exogenous stimuli.
• (3) Antibiotics have the opposite effect in injured rodents, particularly in somatic pain models. Antibiotic administration more-often-than-not reduces pain in experimental injury models.
• (4) To date, probiotic administration has never made pain worse in naïve or injured rodents; probiotics have only been found to decrease pain or have no effect on pain behaviors.
• (5) FMT is effectively being leveraged in animal models to determine the extent to which gastrointestinal contents drive pain in the absence of disease pathophysiology.

Given this understanding, we believe the following are some of the most urgent experimental questions that need to be answered by preclinical microbiome pain scientists:

• (1) Do cutaneous mechanical, thermal, and chemical pain thresholds differ in germ-free and conventionally colonized mice?
• (2) How do non-gut microbiomes contribute to pain sensation? For example, to what extent does the skin microbiome contribute to cutaneous pain?
• (3) How do individual antibiotics alter pain sensitivity? The field’s overreliance on the ampicillin, vancomycin, neomycin, and metronidazole cocktail has prevented drug-specific conclusions from being made.
• (4) Does the analgesic efficacy of different probiotics result through similar mechanisms? Or are there species-specific molecular and cellular changes that occur when each bacterium is administered?

[Tratto da: www.journals.lww.com ]