Fasting intervention and its clinical effects on the human host and microbiome

Experimental trials in organisms ranging from yeast to humans have shown that various forms of reducing food intake (caloric restriction) appear to increase both overall and healthy lifespan, delaying the onset of disease and slowing the progression of biomarkers of aging. The gut microbiota is considered one of the key environmental factors strongly contributing to the regulation of host health. Perturbations in the composition and activity of the gut microbiome are thought to be involved in the emergence of multiple diseases. Indeed, many studies investigating gut microbiota have been performed and have shown strong associations between specific microorganisms and metabolic diseases including overweight, obesity, and type 2 diabetes mellitus as well as specific gastrointestinal disorders, neurodegenerative diseases, and even cancer. Dietary interventions known to reduce inflammation and improve metabolic health are potentiated by prior fasting. Inversely, birth weight differential host oxidative phosphorylation response to fasting implies epigenetic control of some of its effector pathways. There is substantial evidence for the efficacy of fasting in improving insulin signaling and blood glucose control, and in reducing inflammation, conditions for which, additionally, the gut microbiota has been identified as a site of both risk and protective factors. Accordingly, human gut microbiota, both in symbiont and pathobiont roles, have been proposed to impact and mediate some health benefits of fasting and could potentially affect many of these diseases. While results from small‐N studies diverge, fasting consistently enriches widely recognized anti‐inflammatory gut commensals such as Faecalibacterium and other short‐chain fatty acid producers, which likely mediates some of its health effects through immune system and barrier function impact.

Experimental trials in organisms ranging from yeast to humans have shown that various forms of reducing food intake (caloric restriction) appear to increase both overall and healthy lifespan, delaying the onset of disease and slowing the progression of biomarkers of aging. The gut microbiota is considered one of the key environmental factors strongly contributing to the regulation of host health. Perturbations in the composition and activity of the gut microbiome are thought to be involved in the emergence of multiple diseases. Indeed, many studies investigating gut microbiota have been performed and have shown strong associations between specific microorganisms and metabolic diseases including overweight, obesity, and type 2 diabetes mellitus as well as specific gastrointestinal disorders, neurodegenerative diseases, and even cancer. Dietary interventions known to reduce inflammation and improve metabolic health are potentiated by prior fasting. Inversely, birth weight differential host oxidative phosphorylation response to fasting implies epigenetic control of some of its effector pathways. There is substantial evidence for the efficacy of fasting in improving insulin signaling and blood glucose control, and in reducing inflammation, conditions for which, additionally, the gut microbiota has been identified as a site of both risk and protective factors. Accordingly, human gut microbiota, both in symbiont and pathobiont roles, have been proposed to impact and mediate some health benefits of fasting and could potentially affect many of these diseases. While results from small-N studies diverge, fasting consistently enriches widely recognized anti-inflammatory gut commensals such as Faecalibacterium and other short-chain fatty acid producers, which likely mediates some of its health effects through immune system and barrier function impact. strategies, while always recognized as relevant, is a key target for translational and personalized omics-informed medicine.
In the wild, animal populations will increase to the level set by limiting factors; frequently, those result in food scarcity within habitats. It can be expected that before various human technologies of civilization, food production, and storage, our evolutionary history will have included recurring periods of food scarcity or even starvation, including during cold or dry seasons, as a given [37]. Accordingly, we should expect animal biology to be profoundly adapted to these pressures [17,38,39], and in some form, gene expression programs for adjusting to changes in nutrient availability are ubiquitous already in microbial life. A system under external constraints can rely on those for some of its regulation, meaning there is little fitness gain from evolving or retaining costly internal regulatory mechanisms, though high variability of habitat conditions will again promote some regulatory sophistication. As such, to the extent recurring starvation to a greater or lesser extent has been a fact of a form of life, we should expect systems to cope with it-functions for acquiring, consuming, and storing nutrients in excess to weather deprivation when it comes-to lack internal counterbalances in the organism. An analogy can be made to cancer-freed from those restrictions imposed to impede proliferation in the uninjured adult, uncontrolled growth will result (and accordingly, restricted diets are an active area of research and discussion in cancer treatment [34,40,41]).

Calorie restriction as controlled starvation
The discovery nearly a century ago that various animals kept at substantially restricted caloric intake experienced extended lifespans and improved health spurred further inquiry into potential mechanisms [21,[42][43][44][45][46]. Many follow the same principle-the more metabolic turnover, the more molecular and tissue-level wear and tear an organism experiences, including through reactive oxygen species and other potentially toxic side products of its metabolism, driving the accumulation of damage at different scales underlying both disease risk and more general processes of aging, with protective mechanisms in turn activated under functioning homeostasis [1,38,47]. Even where damage is repaired, such reparation in turn involves increased metabolic turnover and ultimately contributes to the same process, as do compensatory phenomena where some processes escalate to account for deficits in others. It has been further suggested that calorie restriction (CR) induces a more efficient use of limited resources in homeostasis maintenance through molecularlevel adaptations and conservation of stem cell reserves [48]. While true CR has been difficult to leverage in the same way in the human setting, it has informed dietary advice and offers an arena for understanding downstream mechanistic bases for health consequences of varying metabolic activity directly or indirectly [18,38]. Some trials have been undertaken on human CR and have revealed improvement in aging-related biomarkers [49], and observed typically long-lived human populations have also been documented as maintaining lowcalorie diets [50]. Studies comparing very old and less old adults revealed a complex pattern of gut microbiome differences, where the very old harbor, for example, more Akkermansia bacteria, but with inconsistent results for short-chain fatty acid (SCFA) producers [51]. Interpreting these studies faces the complexity that signatures of healthy aging will load onto a survivor bias difficult to disentangle from any microbiome derailment with age, with both factors likely playing parts.

Fasting as a human historical practice
Many human cultures practice some form of a fasting tradition, often serving functions of ritual purification or ordeal, and considered within their system to promote the health of the body and mind [30,[52][53][54]. Passed down, these practices have inspired complementary or alternative health approaches that have gained more public awareness in modernity, with widespread belief in their utility for a variety of aspects of wellbeing [9,52], a progression they also share with practices stemming from ancient food preservation such as fermentation where microbes or microbial metabolic products play a role. An exciting task of modern molecular medicine is to evaluate what effects this may have on the commensal gut microbiome, where so far little research has been done. However, a recent study showed increased gut community diversity and enrichment in taxa, including from the Roseburia, Lachnospira, Ruminococcus, Streptococcus, and Faecalibacterium genera, under a fermented food intervention [55].

Fasting in context of host and microbiome multicellularity
Reflecting an evolutionary strategy of increasing sophistication, specialization, and regulation, complex multicellular life forms can be thought of as (largely clonal within a body) ecosystems of independent, yet interdependent cells differentiated into tissues and organs. From this perspective of organisms as communities, the integral presence and role of diverse mobile cells in this system are unsurprising. Circulating blood and/or immune cells form distinct and branching subpopulations moving around the human body, especially inhabiting interface surfaces, and frequently perform intracommunity functions such as defense against foreign or rogue cells. To allow selective uptake through such surfaces, mucosal barriers involve a concerted expression of proteins and polysaccharides by adjacent cells forming these tissues and maintaining the integrity of this barrier while still permitting necessary transport, thus requiring careful gene regulation reactive to a variety of signals. Both microbial metabolites and microbial surface components activate a variety of signaling pathways that can either strengthen or weaken this barrier [56] alongside the impact on host cell proliferation [57] and through bacterial degradation of the mucin layer above the epithelium. While the long-standing claim that host-associated bacteria by far outnumber hostderived cells has been updated to be roughly 1:1, that still makes for around 30 trillion bacterial cells in each of us [58]. Far from being sterile, our skin as well as all external and internal mucosal surfaces are home to a diverse community of bacteria and other microbes, acquired and accumulated from birth (and before) and in a constant state of compositional flux reflecting processes of motility, translocation, introduction, competition, extinction, and differential capacity in making use of available nutrients [59,60]. Work to date on the impact of fasting and fasting-like [61,62] interventions broadly point to a scenario where nutrient restriction initially drives a shift from adaptive to innate immune reliance [2,63]. Immune cell populations shrink in the gut and other secondary lymphoid organs but expand, also through migration, in the bone marrow [64][65][66] while also shifting transcriptional programs. This process has been described as optimization and reorganization for greater sustainability [64] of the adaptive immune repertoire and forms a credible mechanism for observed improvements in autoimmune diseases such as both RA and multiple sclerosis (MS) by CR, ketogenic diet, or intermittent fasting (IF; in MS); by periodic fasting (RA); and by fasting-mimicking diets (RA and MS alike). [62] Refeeding again allows an expansion of the cellular immune arsenal, reverting many of the changes to this repertoire as different subpopulations grow more numerous again [2].

Box 1 Some forms of fasting and related practices
• Periodic fasting: Food intake is low or zero (water fast) for a period of several consecutive (commonly [5][6][7][8][9][10][11][12][13][14] days. The Buchinger protocol and certain forms of religious fasting are examples of this and may be undertaken at an interval from monthly to one or a few times yearly.
• Intermittent fasting: Maintaining an ongoing fasting pattern at all times, including: • Weekly fasting (e.g., 5:2 or alternate-day fasting): Periodic fasting for a shorter interval, restricting to at most ∼500 calories for 2 days per week with unrestricted eating otherwise (5:2), with similar restrictions on every other day (alternate), and so on.
• Calorie restriction: General term for a nutritionally complete but calorically restricted diet. "Grazing" eating patterns of frequent but small meals would not trigger fasting responses but may still constitute CR.
At least following the introduction of solid food, these microbiomes show stability over time [67], with communities from the same sample donor being more similar over time than those from other donors, in both of which bacterial taxa are represented and in their relative abundances [60]. This stability should be seen in context to substantial variability between individuals [68], with large-scale community structure falling within certain broad patterns each seen in samples from donors from different cultures [69], and also substantial variability within the gut. For the latter, different taxa tend to dominate the lumen of the small intestines (e.g., Lactobacillaceae and Enterobacteriaceae species) versus the colon (e.g., Bacteroidaceae and Prevotellaceae species), whereas relatively few species can colonize the mucosa and the epithelial structures [70,71]. Stool, while often the only accessible sample matrix, thus integrates and flattens this spatial (and temporal, reflecting intestinal transit time) organization of the gut microbiome, providing a simplified and sometimes biased view [72]. The origin of microbiome stability over time is not completely known but will reflect the action of the host by way of individual patterns of immune tolerance versus vigilance. Animals born and raised truly sterile reveal immune abnormalities (partly reverted by subsequent colonization by microbiota) reflecting underdevelopment of gut-associated lymphoid tissue and morphological differences compared to control animals, resulting in impaired antibody production and both innate and adaptive immune response to later bacterial infection being compromised [73][74][75]. This thus reflects a process of immune training in wildtype development where the presence of bacteria as (controlled or uncontrolled) foreign guests and invaders primes the immune system cells to respond to their recognition by specific increased or decreased inflammation, with particular developmental windows where this learning process most easily occurs. In this manner, host immune and microbial cell populations interact bidirectionally [70,[76][77][78][79][80], with these patterns of favored coexistence versus antagonism also seen within the microbial part of the resulting ecosystem alone. Among interesting patterns seen are recurring anticorrelation between, on one hand, a module of anti-inflammatory commensals like Faecalibacterium prausnitzii and Eubacterium rectale, and on the other hand inflammation-associated bacteria like Ruminococcus gnavus, further supporting the role of microbial immunomodulation in the formation of these networks [68]. Bacteria coordinate, especially when forming biofilms as is the case also in the gut [81], forming trophic webs where metabolic capacities are distributed between cohabiting species to form complete pathways; signal to each other through metabolite secretion; and use specialized peptides and small molecules to fight each other, exerting lethality or imposing growth disadvantage on their rivals while their allies are protected through resistance gene systems [82,83]. Finally, to a great extent, the microbiome is affected by host-external influences where the nutrients supplied to our internal soil gardens reflect the changing diet of the host [84][85][86], as well as any medications, whether antibiotic or otherwise [87], that the host is taking.

Microbial metabolites as potential fasting mediators
The totality of host and microbial cells has been termed a holo-organism or holobiont [88][89][90]. With microbes especially in the intestines performing many functions of nutrient processing into better host-accessible forms, this symbiosis, which is seen in the internal (micro-)environment, mediates, modulates, and moderates influences of the external (macro-)environment, for example where a dietary factor such as insoluble fiber is differentially produced into SCFAs (primarily acetate, butyrate, and propionate), which in turn has protective effects on the mucosal barrier. The gene functional capacities of the microbiome, reflective of collective gene repertoire and reactivity, vary substantially between individuals, meaning the resulting nutrient availability to the host will differ from person to person under the same diet, leading to individual dietary responses (including, e.g., post-prandial glucose in circulation [91,92]), suggesting the potential for personalized dietary interventions reflecting gut microbiome composition. Several central products of bacterial metabolism in the gut not only respond to host diet but also to the ability of a particular microbiome to mediate that response and have a complex impact on the host. Particular attention has been paid to SCFAs acting as preferred nutrients for particular host cell types, activating (usually anti-inflammatory) immune gene programs (e.g., in macrophages [93]), triggering a variety of host signaling cascades (several passing through chromatin modification and so acting through epigenetic effects) or affecting the permeability of mucosal membranes to reduce the degree to which bacteria may enter the systemic circulation and end up within remote tissues, where they can cause further (including low-grade and chronic) inflammation, which in turn triggers further responses from the host. One important mechanism possibly enabling further feedback loops is that SCFAs can induce the production of antimicrobial peptides including various C-lectins, defensins, and cathelicidins in cells in the intestinal lining [94].
As noted in the systematic overview of findings below, both fasting and refeeding are linked to the expansion of different species known to produce SCFAs and we have reported the same being seen for propionate on the level of direct readouts of gene functions integrated across species [2]. Another relevant potential mechanism, supported by animal studies [95][96][97] but as yet relatively unexplored in human studies, is for fasting to impact host health and homeostasis (e.g., with regard to blood pressure [96] or diabetic sequelae [95]) through altering the pool of secondary bile acids produced by the gut microbiota in the course of enterohepatic circulation [98][99][100][101][102]. Increased permeability of the gut will release a heterogeneous class of compounds into circulation that, among other functions, activates an immune response, termed alarmins in this capacity, with some being directly antimicrobial and several useful as gut permeability markers [103]. While such a release may serve to raise vigilance against extraintestinal infection through concordantly translocated bacteria, the resulting state of low-grade inflammation appears to form a contributing risk factor for a diverse set of health conditions [104][105][106][107][108], representing a body in a state of chronically elevated alertness to the perceived foreign threat. A Mediterranean diet (which substantially overlaps with the Dietary Approaches to Stop Hypertension diet) has been linked to reduced inflammation and improved metabolic health compared to a Western diet [109][110][111][112], also in our work [2]. Several nutrient categories, including host-indigestible fibers, form good substrates for SCFA production by a capable gut microbiome, and they are accordingly central to prebiotic practices [113], where dietary supplements of particular nutrients aim to increase levels of their bacterial metabolism products. Together with probiotic [114] supplementation of live bacteria and the postbiotic [115] supplementation of products of external microbial fermentation, such biotic functional food approaches (found, as noted above, often as a consequence of premodern food storage techniques) can be contrasted with the antibiotics that revolutionized 20th-century medicine, in that while the latter seem to induce a microbiome pattern also associated with low-grade inflammation [87], the former seem in various regards able to prevent or reverse such changes.

Other fasting targets in the human host
Aside from microbiome-mediated modes of action, forms of fasting have several other possible ways to exert systemic impact alone or in interaction with the former. CR tends to involve a shift towards ketone metabolism [116,117], which in turn triggers complex changes in gene expression and epigenetic modifications [21,118]. Several forms of epigenetic modification, including DNA methylation, histone methylation, and histone acetylation, have been demonstrated to follow from forms of CR in human or animal models across a range of tissues, including but not limited to the liver, pancreas, and adipose tissue. Epigenetic markers of metabolic disease within these tissues are known and linked to genes where expression is normalized upon CR but large-scale data are as yet sparse on how often epigenetic normalization also drives this change, though with substantial indications, for example, for the sirtuin family of histone deacetylases [119,120] and also in human impacts through DNA methylation on leptin and adiponectin expression [121] and downregulation of genes involved in oxidative phosphorylation [122], though in a manner complexly depending on birth weight still in the adult. Epigenetic signatures of aging have been increasingly identified and refined, and animal studies on CR have shown a slower accumulation of age-related epigenetic markers [123]. With nutrient gathering being perhaps the most central activity required to stay alive, it has been conducive to fitness to build and maintain ways by which hungry animals will seek out food and will be effective in doing so. The intestines as a site of digestion are the site to sense fullness or conclude hunger, and by that complex set of pathways comprising the gut-brain axis [124][125][126], involving direct nervous connections as well as immune, endocrine, and circulating metabolite signaling, activity elsewhere in the body will shift at all levels. Accordingly, we expect and observe sweeping and wide-ranging effects of reduced nutrition, and as an emergency signal of sorts, for these pathways to have been less optimized for carefully avoiding off-target or secondary effects on the organism. Mood strongly reflects eating and hunger, directly and profoundly involving central nervous system reward mechanisms and triggering instinctive behaviors. These mechanisms in turn can drive a feedback loop over time and dominate habitual behaviors, with these processes likely underpinning some of the anecdotal utility of fasting as a psychological tool. Fasting has been shown to reduce levels of leptin, a satiety signal, beyond its regulation from adiposity itself [127][128][129], and there are indications that prolonged CR may revert resistance to leptin, which has been linked to overeating [130]. Slightly weaker evidence supports CR increasing adiponectin levels [128]. At least one trial showed increases in hunger suppressing GLP-1 and peptide YY levels following seasonal fasting [131]. Further studies into the impact of different forms of fasting on these signaling systems are needed.
Another mechanism that should not be underestimated either on shorter or longer time scales is the turnover of tissue-bound energy storage. As circulating blood glucose is essential to maintain at all times for the processes required for even vegetative survival, a calorie deficit must swiftly and decisively be countered by catabolic processes, recycling especially adipose and muscle tissue for energy. If the body cannot eat elsewhere, it must eat itself; analogous to how excess in turn is stored through anabolic processes gradually increasing the mass of those same tissues in different proportions. The autophagic process eventually takes place [38,53,132], releasing various compounds into circulation as anatomy begins to reshape. The steps needed for this to occur may serve the purposes of recycling and clearing out damaged cellular components or misfolded proteins [133], and as such, affect a broad range of processes across many tissues and cell types. It is further important to consider the context wherein fasting and catabolism are often followed by refeeding and anabolism, which in turn involve activation of regulatory programs and compensatory processes throughout the organism.

Fasting in the context of modern life
Given the way human bodies reflect adaptations to a (diverse, rather than monolithic) premodern past of recurring food scarcity, the changed conditions beginning with settled agricultural or pastoral lifestyles and proceeding to life in the industrialized modern world, at least in the global north, results in a major homeostatic challenge. For many people alive today, high-fat, high-salt, high-carbohydrate food, if monotonous, is available ad libitum. With bodies evolved to recognize conditions of plenty as an opportunity to build up energy stores for the inevitable winter, dry season, or period of poor fortune, overeating and the resulting weight gain are epidemic and resistant to most interventions [37,38,53]. Many modern processed foods are lower in pre-and postbiotic components than traditional fare, hampering the ability of microbiota to exert beneficial aspects, with this state also possibly linked, alongside changes in hygiene and sterility of our habitats, to a loss of co-evolved commensal microbiota, further compounding increased susceptibility to slowly progressing systemic loss of homeostasis and to disease. Comparisons of affluent populations from the global north to some remaining indigenous populations maintaining premodern practices, while heavily confounded by many other factors, indicate an accordingly elevated risk of many diseases in the former. As social and economic transitions take place in the developing world, changes in diet and lifestyle accompany them. For this process, predictions were made and epidemiologically confirmed of a gradual increase in the same diseases, which can be projected onto a global shift in sources of mortality over the upcoming decades [134,135]. Yet another intriguing mechanism for the potential impact of restricted feeding (especially IF) on homeostasis occurs through circadian rhythms. Most life is adapted to a day-night cycle and a review of available literature [136,137] reveals evidence that time-restricted feeding can amplify fluctuations over that cycle in the composition and activity of the gut microbiome, interacting with host intestinal sensors, in turn driving signal cascades. Essentially, adaptations for within-day cyclicity may exist similar to seasonal adaptations, such that maintaining homeostasis is more challenging in its absence, and possible to further through its induction. Disruptions of the circadian rhythms exhibited and also mediated by the microbiota [138] and the immune system [10] are associated with an increased risk of cardiovascular and metabolic illness as well as cancer [138]. Microbial SCFA production plays a role in maintaining this homeostasis, forming another mechanism by which fasting may reduce disease risks. Moreover, especially IF may help strengthen or stabilize impaired cyclicity so that beneficial processes active during the nonfeeding phase become more prevalent again [139], also with time-restricted feeding such that endogenous cyclicity is optimally aligned with nutrient availability from food intake [140]. Genes connected with the circadian clock also affect the regulation of production of insulin, thyroid hormones, and glucocorticoid, among others, with varying degrees of support for an impact of restricted feeding (for insulin, summarized below), but with as yet more research needed to establish under which circumstances these effects are beneficial [141]; but this again suggests a mechanism by way of which restricted feeding contributes to increased longevity [140].

Open questions: Locating the site of fasting action
One important largely unresolved question is to what extent the spatial heterogeneity of both host gene expression and microbiome composition/activity along the gastrointestinal tract [70][71][72] are relevant for mechanisms of fasting. Available restricted diet trial data in humans are largely always based on stool samples representing a rough summary of the gut as a whole, making spatially resolved changes upon dietary intervention thus far largely the domain of a very small number of animal studies, though with the expectation of substantial differences between responses in proximal and distal gut. Fourteen days of CR in a mouse model did show differentiated activation of host genes as well as differential microbial metabolite responses along the gut [142]. Novel sampling approaches for the human setting may here eventually complement what otherwise becomes a reliance on animal models with concomitant limitations.

Fasting and impact on the gut microbiota
With the above context in mind, the present review aims to summarize the main trends of the state of the art of fasting specifically (broadly defined as restricted feeding, see Box 1 for an overview of common forms and Fig. 2 for an overview of fasting mechanisms) as a tool for health improvement, with a particular focus on its interlocking impact on the human gut microbiome and immune cell populations, and with a particular focus on high-throughput ("-omics") studies that can support computational systems analysis. A summary of key findings (aiming primarily to identify recurring patterns as indicators of phenomena robust enough to be reproduced) is given, alongside some discussion of clinical relevance, translational potential, and remaining major knowledge gaps.
We summarize here the state of the art of human microbiome studies on the impact of fasting or CR diets. For a description of inclusion criteria and the comparison approach resulting in the articles shown in Table 1 and the results shown in Table  S1 and Fig. 3, please see Supplementary Methods.
Overall, the literature on microbiome changes under any kind of fasting diet skews towards relatively small studies, and while most report significant changes in host health and metabolism, particularly weight loss, most are poorly powered to conclude specific changes robustly and exhaustively in microbiome composition. In those study designs where a fasting period is followed by a maintenance or refeeding period, as a rule, microbiome changes largely revert, suggesting changes are transient. Studies with no significant microbiome impact (Louis et al. [143], Cignarella et al. [144], and Heinsen et al. [145]) under these criteria were omitted from further discussion.

Effect of periodic fasting
Lilja and coworkers [146] investigated the gut microbiome and host gene expression before and after a Buchinger fast using 16S sequencing, revealing an overall change in microbiome structure yet with little that replicates in other available studies.
Maifeld and coworkers [2] carried out Buchinger fasting in metabolic syndrome patients followed by 3-month refeeding on a Mediterranean diet, assessed through shotgun and 16S sequencing of the gut microbiome, linking findings to changes in immune cell subpopulation proportions while, uniquely, adjusting the analysis for changing medication in many subjects throughout the follow-up period. Individuals in whom blood pressure control either improved or was maintained at a reduced medication dose were considered responders and differed already at base-line from nonresponders in immune profile and some microbiome features, including propionate production capacity. This deficit was normalized during fasting itself, whereas a relative lack of other SCFA-producing commensals, especially butyrateproducing Faecalibacterium but also Coprococcus and Roseburia, normalized during refeeding, particularly in responders. No effects on alpha diversity were seen, and most microbiome alterations had reverted to baseline by 3 months follow-up.
Mesnage and coworkers [147] assessed a Buchinger fast and subsequent 3-month refeeding in healthy volunteers, assessing the gut microbiome with 16S sequencing and additionally investigating serum metabolite levels. In many ways, changes during fasting resembled those seen in responders in the work by Maifeld et al., with initial depletion of SCFA producers (e.g., Faecalibacterium, Coprococcus, Roseburia, as well  as Oscillibacter) then reversed during refeeding; and the opposite pattern was seen for potentially opportunistic taxa (including Bacteroides, Alistipes, Intestinimonas, and Anaerotruncus).
Remely and coworkers [148] report on a combined small-sample pilot trial where subjects undergo a Buchinger fast followed by refeeding with additional treatment with probiotics. Increased alpha diversity (reflecting more different taxa present in each ecosystem as opposed to bloom or monoculture) and carriage of the probiotic genera involved are the best-supported results on the microbiome level. This study also reports an increase in SCFAproducing mucus-associated Akkermansia, and Faecalibacterium prausnitzii at the species but not the genus level.

Effect of IF
Stanislawski et al. [149] report results of the DRIFT2 trial, which is a 12-month weight loss intervention in overweight subjects. Microbiome analysis through 16S sequencing was done for the first 3 months, with subjects randomized to CR or 4:3 IF. Multivariate compositional change of the gut microbiome is reported alongside increased alpha diversity, though no specific microbiome changes replicate in any other considered study.
Su and coworkers [150] followed an old and a young cohort undergoing yearly Ramadan IF, with an additional longer follow-up in the older cohort, analyzing the microbiome through 16S sequencing. Multivariate analysis revealed overall gut remodeling in both groups, with support for increased alpha diversity only in the younger group, and with an increase in Faecalibacterium being the main signal shared by both.
Ali and coworkers [151] similarly assessed microbiome changes throughout Ramadan IF in two different cohorts from two different countries using 16S sequencing of stool samples. No effect on alpha diversity was seen, and changes in other regards under the fast differ substantially between the subcohorts, with some signatures resembling that seen in Buchinger fasting, and some resembling refeeding.
Ozkul and coworkers [152] also investigated microbiome alterations in a small sample of volunteers undergoing Ramadan fasting using 16S sequencing. A clear shift towards gut eubiosis as previously described in the literature was seen, revealing elevated alpha diversity as well as elevated levels of SCFA producers including Faecalibacterium, Roseburia, Eubacterium, and Akkermansia. The bestknown species within the latter, A. muciniphila, is a mucin degrader strongly associated with metabolic health in a variety of studies [153].
Guo and coworkers [154] investigated gut microbiome changes after 8 weeks of IF using 16S sequencing. Increases in SCFA producers including Roseburia and Butyricoccus were seen, alongside microbiome changes towards eubiosis on other taxonomic levels as well. Butyricoccus was depleted in inflammatory bowel disease patients and demonstrated protection in a rat colitis model [155], leading to suggestions for its probiotic use.

Effect of CRs
Alemán and coworkers [156] report from one of several overall restricted daily calorie diets, assessing gut microbiome composition through 16S sequencing under a weight loss intervention. The very limited sample number may underlie the largely negative findings in the microbiome space (while the intervention as such was effective).
Frost and coworkers [157] placed a small number of type 2 diabetic obese subjects on a low calorie diet followed by a food reintroduction period, investigating gut microbiome composition using 16S sequencing. From the reported results, alpha diversity, as a marker of eubiosis, increased, along with depletion of some pathobiont taxa. Most microbiome changes had reverted by the time of follow-up though sustained weight loss remained visible.
Gutiérrez-Repiso and coworkers [158] in one study compared bariatric surgery, the Mediterranean diet, and a CR diet to understand the possible scope of CR intervention in humans. While microbiome changes occurred in each study arm, signatures largely did not overlap, indicating that the mode of dietary intervention rather than the weight loss itself is what is most salient in accompanying gut microbiome alterations. The ketogenic CR arm is what most resembles other studies included in this review and has some overlap with taxa seen altered elsewhere (especially Roseburia, Parabacteroides, and Alistipes) in either fasting or refeeding stages, though the resulting heterogeneity suggests specifics of intervention and the starting point may represent different aspects of an overall more complex process of nutrition-induced microbiome change. All three diets lowered blood sugar, though significance was achieved only in the bariatric surgery arm, in line with improved insulin sensitivity accompanying the microbiome changes.

Consensus findings on fasting impact on the gut microbiome
The most frequently found microbiome impact of fasting interventions, whether periodic or intermittent, is an enrichment of Faecalibacterium (resolved further as F. prausnitzii), well known to produce anti-inflammatory SCFA from dietary fiber and for being protective against both metabolic and inflammatory disease. Where the study design allows distinguishing of a fasting phase from a refeeding phase, this enrichment takes place during refeeding, sometimes following an initial suppression during fasting itself. Roseburia, Butyricoccus, and Coprococcus, also genera populated by core gut SCFA producers, display similar patterns, with evidence for depletion during fasting and enrichment during refeeding.
Several studies report increase in gut Alistipes abundance either in fasting or refeeding state, with a single counterexample of depletion-associated refeeding in a cohort of healthy volunteers. All in all, this suggests a taxon associated with the health improvements of fasting but with more work needed to clarify its specific role. Similar patterns are seen for the Anaerotruncus and Intestinimonas genera. Enrichment of Bifidobacterium, a common probiotic, has also been seen across multiple studies with some ambiguity as to the role of fasting and refeeding stages (and note also its direct use as a probiotic in one of the included studies), and an analogous case holds for Parabacteroides. It should be noted that in one of two studies concluding the increase of Bifidobacterium, this taxon is also present as a probiotic, so less likely to represent a fasting effect, though its support from another study suggests a protective role.
Inversely, Bacteroides is seen enriched during fasting itself in three studies, then depleted during refeeding; its specific impact seems robust nonetheless. As the central Bacteroidetes genus it is core to the Bacteroidetes-Firmicutes phylum ratio often proposed as a biomarker of gut eubiosis and reflected in the high-level summary of gut microbiome composition patterns as enterotypes. Several studies report depletion in the Lachnospiraceae taxa. Eubacterium is enriched in two but depleted in one intermittent (Ramadan) fasting intervention. Other previously discussed commensal and pathobiont associations reported, such as fasting-associated depletion of Streptococcus or Collinsella or increase in Akkermansia, Eggerthella, or Lactobacillus, are seen to be replicated but not consistently. Gut alpha diversity, well recognized as a marker of health and homeostasis, increases in several long-term moderate CR and IF scenarios but is generally not seen altered during intensive fasting. In particular, there is no indication of any loss of gut diversity from depriving the gut microbiome of nutrition on a shorter time scale, which might otherwise have been expected.
In summary, there is a good foundation for concluding that a variety of fasting interventions result in the enrichment of various anti-inflammatory core commensals, especially SCFA producers, consistently though with variability, which may reflect a form of fasting, time of follow-up, or state at baseline.

Conclusions
Fasting and refeeding, intermittent and ongoing CR are associated with eventual shifts in the gut ecosystem away from pathobionts and towards major anti-inflammatory commensal taxa. Such shifts seem transient, with the microbiome returning to near baseline within months of cessation of the intervention, but are frequently accompanied by longer-lasting changes to the metabolism and overall health induced when they are visible. There clearly seem to be grounds to adapt fastingtype diets in a variety of health indications, particularly components of the metabolic syndrome and its sequelae, as well as other immune-and inflammation-mediated diseases, especially combined with other modalities such that it may complement and strengthen homeostasis. Interventions showing fasting-induced improvement to a disease entity thus also indicate a potential gut microbiome protective effect or pathological mechanism involvement and can guide the design of further trials to test this. Such fasting-induced microbiome shift typically involves enrichment of bacteria such as F. prausnitzii that produce antiinflammatory SCFA. Indeed, the impact of high fiber on immune homeostasis showed such effects, suggesting possible ways to "prime" an individual to be more receptive to a wider scope of interventions.
It is also clear that while there exists a broad trend of interventions studied so far-Buchinger fasting, restriction of daily calories to less than 800 (very low calorie diets [159]), and intermittent (including religious) fasting-the specifics of these matter and any comparison of results from studies using different interventions of this kind will diverge reflecting factors of the interventions other than the calories ingested. To what extent do healthy versus metabolically ill individuals respond differently? And to what extent is this-as suggested by our work also in line with other recent studies suggesting individually variable and predictable glycemic responses to different diets-predictable from microbiome baseline, so that it represents a target for personalized nutrition? To address this, it is motivated to run larger-scale harmonized interventions where the individual baseline varies sufficiently within the cohorts.
Going forward, observing emerging standards for systematic data collection [160] and accounting for confounders [161], both in study design and analysis, are likely to further solidify our insights into the potential microbiome-mediated health benefits of fasting. This may guide it to a central role in future interventions, as expected from its anchoring within the scope of our human evolutionary past.