Last week I reviewed four recent papers on the impact of gut microbial commensals and pathogens on immune function, focusing on regulatory T cell (Treg) generation and on the role of effector Th17 T cells (Th17s) on disease. See the post here: http://www.sugarconebiotech.com/?p=6.
The other day another paper on the role of dietary fatty acids in the regulation of immune responses appeared (see http://www.nature.com/nm/journal/vaop/ncurrent/full/nm.3444.html). Honestly, the results presented in this new paper are sufficiently distinct from the prior two papers in Nature that a reappraisal makes sense, as there are clearly a whole host of unresolved questions in this body of work. Also, we will touch briefly on an Immunity paper, also just out.
The driver for all of these studies is the extensive observations on the ability of particular fatty acids to modulate the immune response. There are similar observations on the role of fatty acids in regulating metabolism – that work will not be discussed here.
In particular, the investigators are trying to understand if, and how, short-chain fatty acids (SCFAs) produced by gut bacteria, such as butyrate, propionate and acetate, can modulate the immune system. The stakes are high, as there is early clinical work aimed at manipulating the microbiome in order to treat diseases, notably gut diseases such as inflammatory bowel disease (IBD) and severe diarrhea. Also there are the massive supplement and wellness industries already selling such SCFAs, without much understanding of the science.
Suffice to say we are entering the high fiber diet metaverse, cautiously.
Here is a quick recap of the earlier papers, taken one at a time and in a little more detail. I’ll highlight some similarities and differences. The first two studies were published in Nature on 19 December 2013. The key findings are summarized below. Note that this and the other studies discussed here are done in pure strain inbred mice.
- commensal microbes support Treg development
- specifically, large bowel (colonic) production of the SCFA butyrate induces Tregs
- butyrate functions by inhibiting HDAC IIa
- HDAC inhibition allows more extensive acetylation of FoxP3 and other genes
- expression of FoxP3 and other genes drives Treg differentiation
- butyrate blocks the development of IDB in a T-cell dependent mouse colitis model
The first study was led by Hiroshi Ohno from the RIKEN Center in Kanagawa, Japan. The colitis experiment was shown in that paper. Now onto some of the other details – there are 4 figures in the text and 22 supplemental figures so we can’t cover everything.
The Tregs under discussion in this paper are peripherally generated (i.e. not thymic in origin) and are identified in flow cytometry (FACS) experiments as FoxP3+/neuropilin-/Helios-. About half the high-fiber diet (HFD) induced Tregs were activated and therefore CD103+. Critically, Ohno and colleagues show that this Treg population in restricted to the local (colonic lamina propria) environment. There are plenty of Tregs in the lymphoid organs – mesenteric lymph nodes and spleen – but the percentage of FoxP3+/neuropilin-/Helios- cellsdoes not change in these organs in response to the HFD. The investigators then demonstrate that the microbes responsible for fermenting the HFD into beneficial SCFAs are bacteria of the class Clostridiales and that the potentially beneficial SCFAs include the very potent butyrate and the less potent propionate. Acetate had no effect.
Now this is where is starts to get a little complicated. Having demonstrated that the gut SCFAs induce Treg differentiation only in the colonic lamina propria, the authors next show that T cells isolated from the mouse spleen can be differentiated into Tregs using stimulation conditions that include TGFbeta and butyrate plus stimulatory antibodies to CD3 and CD28. This might indicate that there is some barrier that prevents butyrate reaching sites other than the gut wall, and that this accounts for the local aspect of the Treg response to butyrate. However, while most of these SCFAs are passively and actively taken up by intestinal cells, SCFAs can also be detected in circulation. The absorption of SCFAs can be saturating, i.e. above some concentration SCFA uptake into the circulation is maxed out. These observations suggest that there is a requirement for locally high concentrations of SCFAs in order to induce Treg differentiation – this would explain why these induced Tregs were only found in colonic tissue and not in spleen or LN. I can’t find the relevant concentration data nor is there any dose response data – this is disconcerting. They also report that butyrate can drive naive T cells to a Treg phenotype irrespective of pro-inflammatory Th1, Th2 or Th17 inducing conditions. This is a very dramatic result but suffers from the same lack of critical experimental detail.
The observations made using butyrate stimulation of T cell are followed up in vivo using SCFA supplemented diets. As noted earlier the MOA here is the antagonism of the deacetylase HDAC IIa, thereby allowing increased acetylation and activation of the FoxP3 gene. Finally, using an activated CD4+ T cell transfer colitis model (a model in which Tregs are specifically excluded from the transferred cells) the investigators show protection from disease when the mice are fed diets containing butyrate.
OK, we still have no idea how this is mediated, but the observation is in line with other papers that have examined to ability of butyrate to control colitis (its a large body of work). So, we are not criticizing this whole story, but just suggesting that more experimental detail would be useful, especially in a world where one can buy butyrate capsules or arrange for butyrate enema treatment. A more general critique is offered at the end of this post.
The second Nature paper is by Alexander Rudensky and colleagues from Memorial Sloan-Kettering in NY and reaches broadly similar conclusions as the Ohno paper. Their naive T cell culture conditions are a little different, substituting dendritic cells and Il-2 for the anti-CD28 antibody signal, and they do show dose response data. The latter results indicate a sharp rise in Treg induction above 30uM butyrate. To manipulate the system in vivo these investigators used antibiotics to clear the SCFA producing bacteria. Thus the model is rigged to show an increase above an artificially low background. That’s OK, but let us just be clear about it.
Using sodium butyrate in the drinking water, the investigators induced Treg differentiation in the peripheral LN and spleen. The serum concentration achieved with the drinking water regimen was 500pg/ul which is ~ 4.5uM, a physiological concentration in wildtype, untreated mice. In other words, they brought the SCFA level back to normal, and that induced T regs. In order to induce Tregs in the colonic lamina propria they used either butyrate-enriched food, or a butyrate enema.
The conclusion of all that work is that systemic exposure (drinking water) can induce Tregs in the periphery (spleen, LN) but that local exposure (food, enema) is required to induce Tregs in the gut. Note that this latter conclusion echoes the Ohno paper. Turning to propionate and acetate, they next show that propionate in the drinking water can induce peripheral Tregs in the spleen and LN, and that both propionate and acetate can induce local Tregs in the colon. However, these latter cells are possibly thymic-derived, not extrathymic Tregs, as shown by independence from the CNSI gene (required for extrathymic Treg differentiation). The other option is that these cells are preferentially recruited from the circulation. Note that the acetate result is at odds with this prior paper.
So this has now gotten very complicated, with three abundant starch-derived fatty acids being endowed with both unique and overlapping abilities to induce different types of Tregs in different geographies. Just to make this even more complicated, the Rudensky team next shows that this biology is not just T cell specific, but that butyrate can also endow dendritic cells (DCs) with the ability to preferentially induce Tregs. Lets not go into detail except to say that this effect on DCs did not depend on GPR109a, the niacin and butyrate specific G-protein-coupled receptor (GPCR). More on GPCRs later. The rest of the story – HDAC inhibition and FoxP3 induction – is familiar from the Ohno paper (and many others, the HDAC mechanism is pretty well known).
Finally, I mentioned at the top the Nature Medicine paper that triggered this reappraisal (http://www.nature.com/nm/journal/vaop/ncurrent/full/nm.3444.html). Using a mouse model of house dust mite (HDM) antigen allergic asthma, Benjamin Marsland and colleagues from the University of Lausanne, Switzerland, demonstrate that the susceptibility to and severity of HDM-induced asthma was worsened on a low fiber diet and improved on a high fiber diet. They traced the change in asthmatic response to a change in the gut microbiota supported by the different diets, bringing us again to the SCFAs produced. Higher concentrations of the usual suspects (butyrate, propionate, acetate) were produced when the microbiome was dominated by the Phylum Bacteroidetes, to which the class Clostridiales belongs. Note that the differences in the composition of the microbiome on the low and high fat diets were not significant. This is an issue we will revisit.
The asthma paper is strikingly different from the two Treg papers. In this paper the focus in on propionate, not butyrate, and on systemic effects, not local effects. The premise is that the dietary changes impact the bone marrow, not the local lung tissue. Indeed SCFAs could not be detected in the lung. Propionate treatment reduced Th2 immune responses to HDM in a manner that was dependent on GPR41, a SCFA receptor. This receptor is expressed at high levels in the colon where it mediates a variety of responses to SCFAs, however, in this paper the impact of propionate was traced to the CD11bhisubset of DC in the lung-training LN. How this impacts the allergic-asthmatic response is hypothesized to be (and I’m quoting the paper here) “after inflammation, the lung DC compartment is replenished with inflammatory monocyte-derived DCs that have been exposed to SCFAs in the bone marrow and circulation, leading to a maturation profile that is ineffective at driving Th2 cell responses.”
Therefore the authors conclude that they have elucidated a “gut-lung axis for the formation of the airway microbiota” and therefore I suppose, lung immune responses.
What we have here are three reductionist tales, necessary to help us understand the rules of the system but perhaps not sufficient in themselves to draw sweeping biological and pathological conclusions. It’s very clear from the disparate results obtained that we are still working out the rules. Also, one cautionary note, a recent study in human subjects showed that plasma butyrate concentration remained very close to 2uM under a variety of high fiber meal and fasting conditions, suggesting that this SCFA may not be as variable in concentration as is seen in mice (http://jn.nutrition.org/content/140/11/1932.full).
A somewhat more straightforward study just published last week in Immunity (http://download.cell.com/immunity/pdf/PIIS1074761313005645.pdf?intermediate=true). Vadivel Ganapathy and colleagues from the Georgia Regents University in Augusta show that GPR109 signaling is required to maintain IL-10 dependent Treg activity in the colon, and they trace this function to DC and macrophage responses. Butyrate (or niacin, the nominate ligand for Niacr1 aka GPR109a) treatment of DCs and macrophages induced a phenotype that supported Treg differentiation. Note that this result contradicts the Rudensky paper, in which GPR109a was ruled out as the causative receptor, at least on DCs.
GPR109a gene-deficient mice were then shown to be more susceptible to colitis and inflammation induced colon carcinogenesis, and this effect was shown to be dependent on both the hematopoietic compartment and colonic tissue cells. This final study is satisfying, as now we are seeing pharmacological manipulation of a defined receptor, albeit with a molecule (niacin) that has a pretty checkered history as a therapeutic.
Where does this leave us? I think the take home message is that these systems are very complex, and by trying to simplify them we have the benefit of gaining some insight but the risk of over-interpretation. The human microbiome is incredibly variable, over time and between individuals. The fact that we are seeing different results from manipulation of highly inbred strains of mice on very carefully defined diets should give us pause, especially when some studies can’t statistically distinguish between components of the microbiota they are describing. However, at the very least these studies support the belief that high fiber diets that producing lots of butyrate and propionate should be beneficial, and we have identified some targetable GPCRs, which should drive further research. Finally, we are perhaps a step or two closer to understanding how to manipulate Treg cell populations in human disease. This last goal, the ability to regulate immune responses via regulatory T cell modulation, has proved to be an elusive one so far.