Exploring the role of the microbiota member Bifidobacterium in modulating immune-linked diseases

The gut-associated microbiota is essential for multiple physiological processes, including immune development. Acquisition of our initial pioneer microbial communities, including the dominant early life genus Bifidobacterium, occurs at a critical period of immune maturation and programming. Bifidobacteria are resident microbiota members throughout our lifetime and have been shown to modulate specific immune cells and pathways. Notably, reductions in this genus have been associated with several diseases, including inflammatory bowel disease. In this review, we provide an overview of bifidobacteria profiles throughout life and how different strains of bifidobacteria have been implicated in immune modulation in disease states. The focus will be examining preclinical models and outcomes from clinical trials on immune-linked chronic conditions. Finally, we highlight some of the important unresolved questions in relation to Bifidobacterium-mediated immune modulation and implications for future directions, trials, and development of new therapies.


Introduction
The human gastrointestinal (GI) tract is home to a complex ecosystem of microbes, including bacteria, fungi and viruses, which play a critical role in host health [1,2]. Owing to the ability of these bacteria to interact with the host directly, through physical interactions with the intestinal mucosa, and indirectly, via production of metabolites that can enter the blood stream, there is significant interest in understanding how these bacteria affect our physiology, particularly with respect to immune development and modulation. For many years, there has been a commercial and scientific interest in using beneficial bacteria, such as 'probiotics', to positively modulate host health. Probiotics are defined as 'live microorganisms that, when administered in adequate amounts, confer a health benefit on the host' [3] and are, for the most part, consisting of strains from the genus Lactobacillus and Bifidobacterium. Bifidobacteria have been used for many years as supplements to promote host wellbeing, as their presence, including the high levels observed in infants and stable levels in adults, is associated with a 'healthy' state. These bacteria are particularly effective at protecting against infectious diseases [4][5][6][7] and modulating immune responses [7,8]. This review discusses Bifidobacterium across the life course, and focuses on species and specific strains that have been studied in the context of immune modulation and treatment of disease.

Bifidobacterium across the life course
Bifidobacteria are Gram-positive, heterofermentative, anaerobic bacteria with a distinctive bifid (i.e. Y) shape after which they are named. Originally isolated from the faeces of breast-fed infants by Tissier in 1899, members of the genus Bifidobacterium are commonly found in the GI tract of mammals. They have also been isolated from birds, social insects such as honey bees [9,10], and more recently from water kefir [11][12][13]. There are currently 55 recognised (sub)species of Bifidobacterium [14]. Recently, the genomes of representative strains of these taxa have been sequenced allowing greater resolution when classifying potential new strains of bifidobacteria [14][15][16]. An analysis of 317 core genes, across all 67 representative genomes of Bifidobacteriaceae [including representative strains of the 55 (sub) species of Bifidobacterium], classified Bifidobacterium into seven phylogenetic clusters: Bifidobacterium longum; Bifidobacterium adolescentis; Bifidobacterium pseudolongum; Bifidobacterium boum; Bifidobacterium asteroides; Bifidobacterium pullorum; and Bifidobacterium bifidum [14]. Bifidobacterium genomes range from 1.63 Mb (Bifidobacterium commune R-52791) to 3.25 Mb (Bifidobacterium biavatii DSM 23 969) and have a high G + C content ranging from 65.53% (Bifidobacterium choerinum LMG 10 510) to 52.29% (Bifidobacterium aquikefiri LMG 28 769). The analysis of the pan genome of Bifidobacterium revealed that 38% of all truly unique genes are involved in carbohydrate metabolism, highlighting the importance of this function in the genus [14,16]. Moreover, Bifidobacterium possesses a large arsenal of genes encoding glycosyl hydrolases (GHs), with 3989 genes predicted to have this function in the 55 Bifidobacterium genomes. The highest number of GH genes was identified in isolates from humans and primates, reflecting the diverse range of dietary carbohydrates consumed by these hosts [14].
In humans, Bifidobacterium resides within the GI tract, from birth to old age, which has recently been reviewed by Arboleya et al. [17]. Briefly, bifidobacteria colonise the new-born gut within the first days and weeks after birth, and they represent the most abundant bacterial family ranging from 40 to 80% of the total gut microbiota [18,19]. There is also evidence to suggest that bifidobacteria could begin colonisation of the GI tract in utero [20,21]; however, this remains controversial as direct proof for microbial colonisation, and the mechanisms by which bacteria pass from the mother to the foetus remain to be elucidated. Current studies indicate that bifidobacteria are transmitted vertically from the mother's vagina, GI tract, or breast milk. This is supported by findings by Duranti et al. [22], who used a novel internal transcribed spacer (ITS) approach trialled previously [23]. Duranti et al. found genomically identical bifidobacteria strains in faecal and milk samples from 24 mother-infant pairs. These findings provide initial insights as to why vaginal delivery provides a higher abundance of Bifidobacterium in infants, over a caesarean section (C-section) delivery [24,25]. Following birth, breast milk may provide a secondary delivery route for further bifidobacteria [22,26] and additionally drives proliferation of bifidobacteria due to its unique nutritional milieu of human milk oligosaccharides (HMOs), proteins, and lipids [27][28][29]. Notably, a reduced abundance of Bifidobacterium in infants is highly correlated to chronic diseases, including asthma and obesity [30].
As the infant begins to consume solid foods (∼6 months onwards), overall bacterial diversity increases in response to an expanding nutritional environment, and the abundance of bifidobacteria decreases quite rapidly to 30-40% [17,31], and continues to fall gradually during childhood and adolescence. This can be an unstable time period, and Bifidobacterium levels can be influenced by puberty, nutrition, and antibiotic use [32][33][34]. As we reach adulthood, bifidobacterial populations stabilise between 0 and 18%. A further decline is then seen as we enter the elderly phase of life [35], which interestingly also correlates to a decrease in immune function, so-called immunosenescence. Exactly when or why this happens is still unclear, but higher bifidobacteria levels in the elderly are correlated with health and longevity [36,37].
Notably, bifidobacteria levels across the life course align with key stages in immune maturation ( Figure 1) and are associated with improved host well-being. However, we are at a relatively early stage in understanding the specific mechanisms whereby Bifidobacterium influence this critical homeostatic development and programming, including impact on specific immune populations and signalling pathways. Current studies have focused more on immune-linked diseases, in both patients and preclinical in vivo disease models, and thus, this review discusses the role of bifidobacteria in modulating different immune populations and intervention studies in disease cohorts.

Effects of bifidobacteria on the immune system
Data from mouse models and clinical trials indicate that bifidobacteria may have beneficial effects for treating and preventing immune-linked diseases, including gut-associated and systemic conditions. However, we still do not fully understand the mechanisms employed by bifidobacteria to exert their immunomodulatory effects [38]. Studies to date indicate that bifidobacteria have a complex role, having both pro-and anti-inflammatory effects, promoting anti-pathogen immune responses, and modulating immunity in the context of auto-immune or immune-mediated diseases. A significant complication in evaluating these responses lies in the fact that many distinct species and strains of bifidobacteria have been tested, and additionally many of these studies include combination testing with other species or phylum. Furthermore, the cell type, species of animal, model used, and human cohort supplemented also affect immune responses generated [39]. Currently, most mechanistic studies have focused on inflammatory bowel disease (IBD) (Figure 2), allergy, and infection models, reporting bifidobacterial-associated modulation of specific immune cells and their outputs. There are also some limited reports highlighting immune receptor-ligand interactions and downstream signalling events Although further studies are required to test this hypothesis, this figure illustrates the potential correlation between bifidobacteria and immune cell maturation in early life. The distinct bifid shape (white Y) represents Bifidobacterium as a percentage of the total microbiota. DNA analysis indicates that Bifidobacterium spp. may cross the placenta, but whether Bifidobacterium spp. begins colonisation before birth has not been evidenced, and thus indicated with a question mark. After birth Bifidobacterium spp. quickly colonises the infant gut and represents the most abundant bacteria by 2-3 weeks and remains prominent at 40-80% of the total microbiota until solid food is introduced ∼6 months of age. At this age, bifidobacterial populations begin to decrease through childhood and adolescence. It stabilises as we enter adulthood ∼0-18% where it remains for most of our adult life. A further reduction in Bifidobacterium levels is then observed as we enter the elderly stages of life. Interestingly, a similar trend is seen with the immune system. Studies have shown that Bifidobacterium spp. has an important role in stimulating the immune system. These interactions could potentially occur as early as in utero and within the critical early life window after birth linking to the high levels of bifidobacteria also observed at this time period. Illustrated are total number of T cells (top), shown as naive (T N ) or mature (T M ), and cell maturation (bottom) for NK cells, B cells, neutrophils (Neutro), and macrophages (Mw). At birth, a reported 75% of T cells are naive, with 25% mature, indicating potential in utero priming. Post-natal immune development is complex, and beyond the scope of this review, however, there is a trend with respect to immune cell maturation; B cells and macrophages are mature by 1 year of age and neutrophils fully mature by 5 years of age. The exception is NK cells that do not mature until 17 years of age, but previous studies have shown that these innate immune cells can be influenced by Bifidobacterium spp. Further investigation is required to provide mechanistic correlation, but we hypothesise that bifidobacteria may potentially modulate foetal immune development at the very first stages of life. and links to specific bifidobacteria molecules [40,41], such as pili and exopolysaccharide (EPS), on immune responses [5,42,43]. However, it is apparent that the bifidobacteria-immune field requires a greater number of investigations detailing key mechanistic targets and pathways in different immune compartments and immune cell types.

T cells
From an adaptive immune development perspective, the ratio of T-cell subsets, including T helper1 (Th1), Th2, Th17, and T regulatory cells (Tregs), is key for maintaining homeostasis, while also promoting inflammatory responses in response to appropriate external antigenic stimuli [44]. Notably, irregularities in T-cell responses at different life stages are associated with allergic and chronic inflammatory diseases [45]. Exacerbated Th1 or Th17 responses have been linked to auto-immune disease [46], whereas uncontrolled Th2 responses or reduced Treg responses are associated with allergic reactions [47]. A lack of Tregs is often also found in patients with IBD [48]. Notably, several studies have reported that different strains of bifidobacteria can modulate T-cell responses in immune-driven diseases. In a murine model for chronic allergic asthma, Bifidobacterium breve M16-V was shown to increase Treg cell responses (defined as CD4 + FoxP3 + cells) and additionally increase the anti-inflammatory cytokine IL-10 in lung tissue [49]. This was also found to have similar effects as budesonide (i.e. glucocorticoid) treatment. In an ovalbumin-induced food allergy mouse model, the same strain of B. breve M16-V (in combination with non-digestible oligosaccharides) was shown to normalise aberrant Th2 responses including a decrease in IL-5 and an increase in IFN-γ, which correlated with a reduction in allergic symptoms [50]. In an IBD-like model, B. breve NutRes 204 ameliorated dextran sodium sulfate (DSS)-induced colitis. This was linked to increases in Tregs and decreases in Th17 (CD4 + IL-17 + ) cell subsets in Peyer's patches of DSS-treated mice and concurrent differential expression of Th1 cells, Th2, and Treg-associated cytokines [51]. Zuo et al. reported an increase in mesenteric lymph node (MLN) Tregs (i.e. CD4 + FoxP3 + cells) in healthy Balb/c mice, and a reduction in Th1-associated cytokines, including IFN-γ and TNF-α. An increase in Treg-associated FoxP3 and anti-inflammatory cytokines IL-10 and TGF-β expression in the MLNs during trinitrobenzene sulfonic acid (TNBS) colitis was also observed [52]. The importance of IL-10 in modulating T-cell responses was further demonstrated in an interesting study where a recombinant B. longum NCC2705, producing human IL-10 from a plasmid, was shown to ameliorate colitis in mice by increasing Treg cells and decreasing Th17 cells [53]. However, the use of genetically modified Bifidobacterium in humans is a significant regulatory issue, and thus, more in-depth preclinical trials are required to identify the efficacy of these strains and inform regulators.

Dendritic cells
A potential mechanism whereby bifidobacteria induce T cells may be through dendritic cells (DCs), via antigen presentation and stimulation of antigen-specific T cells. Jeon et al. [54] observed that CD103 + DCs isolated from the lamina propria (LP), and stimulated with B. breve YAKULT strain, and co-cultured with naive splenic CD4 + T cells, lead to IL-10 production and expression of cMaf, Ahr, and Il21, markers of type 1 regulatory T cells. Moreover, this effect was abolished in CD103 + DCs from Il10 −/− , Tlr2 −/− , and Myd88 −/− mice. Konieczna et al. [55] determined that B. longum subsp. infantis 35 624 increased numbers of CD103 + retinaldehyde dehydrogenase (RALDH) + DCs within the LP of mice and that this was associated with decreased Th1 and Th17 cells within the LP, and improved colitis outcomes. Furthermore, in a mouse allergic response to bovine β-lactoglubulin model, B. longum BBMN68 induced both CD4 + CD25 + Fox3 + Treg cells in the MLNs and CD103 + DCs in Peyer's patches, which restored the Th1/Th2 balance. Additionally, ex vitro experiments indicated that DCs from B. longum BBMN68 fed mice also increased TGF-β, IL-10, IFN-γ secretion, and reduced IL-4 secretion from CD4 + T cells, further highlighting the indirect role, via DCs, that bifidobacteria have on T-cell cytokine secretion [56].

Epithelial cells
As bifidobacteria reside within the GI tract, intestinal epithelial cells (IECs) represent a key immune cell type for bifidobacteria-associated modulation. IECs are fundamental for maintaining barrier function during homeostatic conditions, and many different species and strains of Bifidobacterium, or their metabolic products, have been shown to increase epithelial cell integrity in vitro and in vivo [57][58][59]. In the context of disease, IBD patients, who also have reduced bifidobacteria levels [60], display what is called pathological cell shedding. This is characterised by redistribution of tight junction (TJ) proteins, such as Zonula occludens-1 (ZO-1) and E-cadherin, and increased apoptosis of the epithelial cells at the villus tip, resulting in excessive cell shedding into the lumen [61,62]. In a mouse model of epithelial cell shedding, B. breve UCC2003 was shown to reduce the number of apoptotic IECs and corresponding apoptosis signalling molecules. This was mediated via the bifidobacterial EPS capsule and host immune-associated adaptor protein MyD88 [40]. In an IBD-like experimental model, administration of B. longum subsp. longum 7952, but not B. longum subsp. longum 372, enhanced expression of TJ proteins in the epithelial layer, which was associated with reduced development of DSS-induced symptoms [63]. This was further highlighted by Hsieh et al. [57] who also showed that only some species and strains of Bifidobacterium prevented TNF-α-induced disruption of the epithelial barrier, and promoted tight junctions which, in vitro, was attributed to TLR2, but is yet to be defined in vivo. These studies IBD is characterised by a damaged or 'leaky' IEC barrier and chronic inflammation. A weakened barrier, in tandem with a reduced mucus layer (depicted by light green layer over IECs), enables translocation of luminal microbes into the underlying lamina propria which triggers NF-κB and release of pro-inflammatory cytokines from IECs and immune cells such as macrophages (Mw) and DCs. Cytokines such as IL-6, IL-23, and TNF-α activate T H cells; CD is marked by an increase in T H 1 cells, whereas UC is characterised by an increase in T H 2 cells. In both diseases, there is a reduction in Treg cells, linked to increased IL-12 secretion. Bifidobacterium has been shown to reduce levels of key IBD-related pro-inflammatory cytokines such TNF-α, IFN-γ, and IL-1β, and increase the production of IBD protective cytokines TGFβ and IL-10 in vitro and in vivo, and mucus production in vitro. Furthermore, Bifidobacterium has been shown to induce Treg cells and reduce restore the T H 1/T H 2 cell balance in murine models. emphasise the importance of the species and even the strain of Bifidobacterium that is used. In a necrotising enterocolitis (NEC, which is also linked to epithelial barrier disruption) mouse model, B. breve was shown to up-regulate TJ proteins Claudin 4 and Occludin [51], and a non-specified species of Bifidobacterium increased ZO-1 in a rat NEC model [8]. Studies have also shown that the effects of bifidobacteria are only exerted, or are increased, when live bifidobacteria are used. Grimm et al. [64] showed that the beneficial effects of B. bifidum S17 in DSS colitis were seen from only live and not UV-killed bacteria, and Hsieh et al. [57] showed that only live B. bifidum had a restorative effect on a TJ impaired Caco-2 cell monolayer. They found that acetate and formate were produced more by B. bifidum than by B. adolescentis. Whether it is necessary for bacteria to be 'alive' to be effective remains a matter of debate, but differences in structure and components could hold key findings for future therapeutic development.

Other cells types
Currently, there are limited studies examining the role of bifidobacteria with other immune populations. Kawahara et al. [65] reported that supplementation with B. longum MM-2 was linked to increases in natural killer (NK) cell activity, potentially via an increase in NK cell-activating cytokines such as IL-18, and correlated with anti-influenza virus responses. In an obesity-associated inflammation model, B. pseudocatenulatum CECT 7765 reduced B-cell (CD19 + ) and pro-inflammatory macrophages (F4/80 + CD11c − CD206 + ), as well as increasing Treg responses, which correlated with reduced body weight gain and improved glucose tolerance [66]. Recently, B. breve pre-treatment was shown to significantly decrease the total inflammatory cell number, including decreasing the relative number of eosinophils and neutrophils in a murine airway inflammation model [49].
Overall, these studies indicate that bifidobacteria may have beneficial effects on inflammatory and immunedriven diseases via regulation of specific immune cells and cellular networks, including cytokines (details on Bifidobacterium-associated cytokine modulation are shown in Table 1). The implications that bifidobacteria are important modulators of immune responses during disease, both locally and systemically, therefore make them attractive therapeutic targets. Bifidobacteria possess many proteinaceous factors, such as EPS and sortasedependent pili, that modulate immune responses. This includes the presence, on some strains, of a surface-associated EPS, which has been shown in both B. breve UCC2003 and B. longum subsp. infantis 35624 to modulate innate immune cells, such as neutrophils, macrophages, and peripheral monocytes [5,43]. An EPS deletion mutant in B. breve UCC2003 induced more pro-inflammatory cytokine secretion from splenocytes and also increased the number of Ly6G + neutrophils, F4/80 + macrophages, DX5 + /CD3 + NK cells, and CD19 + B cells in the spleen of treated mice compared with mice treated with the wild-type strain [5]. Similarly, an EPS deletion mutant of B. longum subsp. infantis 35624 stimulated more IL-12p70, IL-17, and IFN-γ from peripheral blood mononuclear cells than the wild-type strain. B. bifidum PRL2010 expresses sortase-dependent pili, which when heterologously expressed in a Lactobacillus lactis strain induced a higher TNF-α and IL-10 response compared with the non-piliated L. lactis strain in a U937 macrophage cell line [42]. A similar response was seen in a murine TNBS colitis model when mice were pretreated with B. bifidum PRL2010 [79]. Despite these insights, further studies to elucidate these, and other mechanisms used by bifidobacteria to regulate the immune system, are required. This could include expanding studies to cover exploration of other immune-linked conditions (e.g. inflammatory arthritis), important cell types, specific signalling pathways, and bifidobacteria components or metabolites, and is critical for designing new bacteriotherapies or 'probiotics' (Box 1). This may offer a more targeted or personalised approach for patients, as there does not appear to be a one-strain-fits-all scenario.
Bifidobacteria supplementation in patientsevidence from clinical trials Disturbances in the microbiota are linked to an ever-growing number of immune-linked disease states including IBD, atopic allergy, arthritis, and obesity [80]. Therefore, there is a significant interest in treating these diseases through microbial or 'probiotic' supplementation of patients, including with Bifidobacterium. Many clinical trials use combinations of Lactobacillus and Bifidobacterium; however, for this review, we will discuss only studies where Bifidobacterium (single or multiple species) were administered as the sole bacteria and/or in combination with a prebiotic (Table 2).

Inflammatory bowel diseases
IBD encompasses both Crohn's disease (CD) and ulcerative colitis (UC). Both diseases are characterised by chronic intestinal inflammation; UC inflammation is continuous from the rectum to the proximal colon, CD inflammation is patchy and discontinuous, and frequently occurs in the distal ileum or colon. The incidence of Box 1.
Areas for exploration in Bifidobacterium-immune interactions, and potential experimental tools/approaches that could be used to uncover key mechanisms involved 16S rRNA (metataxonomic profiling), whole genome sequencing (WGS), global RNA sequencing (RNASeq), knockout (KO).

Key questions
• Does bifidobacteria modulate immune responses directly or indirectly (i.e. via wider microbiota modulation)?
• What are the specific strains and species that regulate immune modulation?
• What are the specific components and metabolites that mediate beneficial effects?
• Does bifidobacteria modulate diverse immune cell populations?
• What cell-associated receptors and downstream signalling events are involved in pro-and anti-inflammatory events?
• Does Bifidobacterium modulate immune development across the life course, from in utero to old age?
• How does bifidobacteria modulate dysregulated immune-linked conditions? Is it via similar pathways as observed in homeostasis?
Experimental approaches/tools • Mono-colonised or defined gnotobiotic models • Novel cell models to study cross-talk • In-depth genomic (e.g. WGS) and phenotypic characterisation on key strains and combination studies • Comparative WGS analysis and transcriptional profiling (e.g. RNASeq) of Bifidobacterium and utilisation/development of molecular tools to test key mutants • Profiling immune populations with flow cytometry and transcriptomics and use of cellspecific mouse KO models • Use of network analysis and systems biology to define the specific pathways involved • Human cohort studies and use of life stage-specific (e.g. neonatal) in vivo models and immune readouts • Characterise responses in homeostasis (i.e. 'healthy') and correlate to clinically relevant disease models and patient/volunteer cohorts Abbreviations: RDBPCT, randomised; double-blind; placebo-controlled trial; RCT, randomised clinical trial; UC, ulcerative colitis; CD, Crohn's disease; Bif, Bifidobacterium supplemented; CAIA, clinical activity index assessment; GOS, galactooligosaccharide; scGOS, short-chain galactooligosaccharides; lcFOS, long-chain fructooligosaccharides; 5-ASA, 5-aminosalicylic acid; PPI, protein pump inhibitor; CRP, C-reactive protein.
both these diseases is increasing in Western Europe and North America, and represents a significant burden on health services [92,93]. The aetiology of IBD is multifactorial, but it is widely accepted that the microbiota plays a key role in disease pathology. Patients with IBD have decreased microbial diversity, and many studies have shown a decrease in Bifidobacterium levels in both CD and UC patients during active disease [60,[94][95][96] For a recent review on the topic, see Buttó & Haller [97].
Owing to the anti-inflammatory properties exhibited by many strains of Bifidobacterium, in conjunction with reduced levels of bifidobacteria in IBD, there have been several studies testing this bacteria as a treatment for IBD; one published trial for CD and six for UC (Table 2). However, a limited number of species (B. longum subsp. longum, B. breve, and B. longum subsp. infantis) have been used in these trials. Additionally, the treatment duration, number of patients, and disease makers studied in each trial vary greatly, and thus, comparison between trials is difficult. Despite these differences, the limited number of clinical trials shows some promise for using Bifidobacterium in the treatment of IBD. A pilot study in UC patients, which used a prebiotic (a fructo-oligosaccharide/inulin mix; Synergy 1) in conjunction with B. longum subsp. longum strain isolated from a healthy rectum, showed promising results despite low numbers of patients in the trial [73]. After 28-day treatment, patients in the treatment group had reduced TNF transcripts, a key cytokine in UC, and reduced clinical symptoms. A follow-up trial on patients with active CD, using the same probiotic/prebiotic mix, showed a reduction in CD activity index and histology score in patients receiving the synbiotic compared with the controls [82]. However, due to the short duration of these studies, it is not clear if this strain is effective in the induction or maintenance of remission, and whether a longer-term study would prove continued efficacy. In another short-term study, B. longum subsp. infantis 35 624 administered to UC patients for 6 weeks resulted in a significant reduction in C-reactive protein and a non-statistically significant reduction in IL-6 when compared with the baseline [70]. While this study indicated a decrease in inflammatory markers, no clinical outcomes were measured and therefore it is not possible to conclude that this strain is effective in the treatment of UC. More recently, a trial where patients with mild-to-moderate UC (UCDAI 3-9) were supplemented with B. longum subsp. longum 536 resulted in a significant decrease in disease activity following 8-week supplementation, whereas a significant decrease was not seen in the placebo group [84]. Taken together, these trials suggest that bifidobacteria may be a promising therapy for the treatment of IBD; however, the limitations of the studies must be considered. Many trails did not test whether the strain administered had colonised patients making it difficult to directly attribute an effect to the probiotic, or indeed, if the strain modulated the wider microbiota, as no microbiota profiling (i.e. 16S rRNA or shotgun metagenomics) was performed. In all trials, Bifidobacterium supplementation was additional to standard treatment therapies (e.g. immunosuppressants/ aminosalicylates or steroids); therefore, the efficacy of bifidobacterial treatment alone is unclear. Furthermore, all trials reviewed had a low number of participants (<100) over a short duration, and larger clinical trials are needed to clarify the efficacy of bifidobacteria in treating IBD. The differences between strains studied, intervention time, frequency and concentration of dose, and the addition of synbiotic and clinical outcomes measured mean that studies are difficult to compare. Finally, two Cochrane reviews, focused on clinical trials testing the use of probiotics in the induction or remission of UC or CD, highlighted a lack of well-designed trials in this area. Furthermore, the authors could not make any conclusion about the efficacy of probiotics in the treatment of UC or CD [98,99]. Thus, a more robust standardised approach to clinical trials with Bifidobacterium species (and other probiotics) would benefit future studies.

Irritable bowel syndrome
Another GI disorder that has been the focus of treatment with species of Bifidobacterium is irritable bowel syndrome (IBS). The pathophysiology and cause of IBS is not fully understood; however; there is an immune component, as IBS patients have higher serum cytokine levels [100]. B. longum subsp. infantis 35 624 has been studied in two double-blind, placebo-controlled clinical trials [89,90]. In both studies, the bifidobacteriasupplemented group had reduction in symptoms, and in one trial, a reduction in cytokine production by peripheral blood mononuclear cells (PBMCs) was reported in vitro [89]. These data suggest that at least in some conditions, bifidobacteria could be useful in the management of IBS.

Atopic eczema and asthma
The intestinal microbiota is important in early life immune development, and disturbances via antibiotics usage, formula feeding, or C-section are proposed to contribute to extra-intestinal disease, such as asthma and atopic eczema [101]. Studies have shown that infants who develop atopy have a lower Bifidobacterium to Clostridium difficile ratio [102]. Several trials have tested the use of probiotics as an intervention for infants with eczema and asthma. In an intervention study, Van Der Aa et al. [88] found that supplementation of B. breve M-16V, plus a prebiotic, to infants less than 7 months old, who were positive for atopic eczema, resulted in less children on asthma medication 1-year post-treatment. The three studies of eczema carried out in infants under 24 months, who had developed atopic eczema, and had a variety of study designs, used the SCORAD (Scoring atopic dermatitis), allowing for some comparison between studies [85][86][87]. An early study focused on 3-6-month-old infants who had developed eczema during breast-feeding and had never been exposed to infant formula [85]. Children were provided with Bifidobacterium animalis subsp. lactis Bb-12 supplemented exclusively with hydrolysed whey formula for 2 months, resulting in a reduction in SCORAD from 16 to 0 vs. 13.4 in the supplement group. Another study supplementing B. animalis subsp. lactis Bi-07 to infants diagnosed with eczema resulted in a correlation between an increase in Bifidobacterium spp. in the infant microbiota and a decreased SCORAD index, but this could not be directly attributed to probiotic consumption [86]. While these two studies suggest that supplementation with bifidobacteria could help reduce the symptoms of atopic eczema, another larger, longer-term clinical trial showed no benefit of supplementation with B. animalis subsp. lactis CNCM I-3446, highlighting that not all clinical interventions with bifidobacteria are successful [87].

Necrotising enterocolitis
NEC primarily occurs in premature, and low-birth-weight infants, and can result in death. These infants have an underdeveloped intestinal immune system and are given broad-spectrum antibiotics prophylactically to prevent infection. Colonisation with opportunistic pathogens may contribute to the pathogenesis of NEC, which is characterised by an exacerbated inflammatory cascade [103]. A recent study, where preterm infants were supplemented with B. breve M-16V, showed a significant reduction in NEC ≥ Stage II, highlighting a role for Bifidobacterium in this disease [104]. The mechanism of a bifidobacteria-protective effect in NEC is not clear, but one study in a rat NEC model showed that B. bifidum OLB6378 modulated mucosal immunity by reducing Il6 and Tnfa expression, and improving TJ protein distribution in the ileum [107]. However, bifidobacteria are also known to inhibit pathogen colonisation [4] and thus may directly modulate the microbiota and inhibit NEC; however, more studies are required to clarify this. A recent large-scale study supplementing preterm infants with B. breve BBG-001 suggested that supplementation does not prevent NEC or late-onset sepsis in the study group [106]; however, the outcomes of this study remain controversial [107,108].

Recommendations for future Bifidobacterium intervention trials in human patients
Clinical study design recommendations • Profile colonisation ability of strain(s) • Stratify responders vs. non-responders and cross-talk capabilities • Determine the impact of supplementation on wider microbiota (e.g. 16S or shotgun) • Define clear primary standardised clinical readouts • Define clear immune markers associated with disease as secondary readouts, using markers from preclinical models • Define clear microbiota and immune baselines for patients before intervention • Longitudinal sampling throughout intervention • Define cohort to be tested based on preclinical model data (e.g. paediatric vs. adult) The above clinical trials have identified some positive roles for the treatment of immune-driven diseases with Bifidobacterium therapy; however, other studies have shown no benefit. Furthermore, there is a current lack of understanding, with respect to the underlying immune-modulatory factors involved in improving clinical outcomes. Currently, there is also a lack of bifidobacterial supplementation studies aimed at positively modulating other immune-linked conditions, such as arthritis and psoriasis. Further identifying the mechanisms by which bifidobacteria modulate the immune system in humans, building on in-depth mechanistic animal studies, will allow for better screening of new potential therapeutic strains. In IBD, with the highest number of trials, there is scope for better standardisation of secondary outcomes to allow for better comparison between independent studies (Box 2).

Conclusion and future perspectives
The studies, to date, have shown that Bifidobacterium are resident within the GI tract across our lifespan, and are associated with immune well-being. Notably, reductions in bifidobacterial populations are associated with various immune-linked conditions, and studies using in vivo models and clinical trials indicate strategies that use Bifidobacterium may beneficially modulate immune responses to improve clinical symptoms. However, we are still somewhat removed from understanding how different strains of bifidobacteria specifically modulate immune responses (Box 1), and how we link this to comprehensive and well-planned clinical trials (Box 2). These studies are critical if we are to perform more personalised interventions in patients with immune-linked diseases, with the aim of improving clinical outcomes and providing cost-effective and potentially non-toxic therapies.

Summary
• Bifidobacterium spp. are present in the human gastrointestinal tract from birth and throughout the life course, and their presence is associated with health.
• Reduction in bifidobacterial abundance occurs in multiple inflammatory diseases.
• Bifidobacteria can modulate T-cell responses to reduce inflammation.
• Bifidobacteria may modulate T cells indirectly through dendritic cells to reduce inflammation.
• The overall mechanisms of bifidobacterial-associated immune modulation are currently incompletely understood.
• Bifidobacteria supplementation to treat inflammatory diseases shows promise, but more studies are required.