The gut microbiota regulates the metabolic response to fasting

Metabolic adaptation to fasting is an essential mechanism developed by mammals in order to survive. The transition from the fed to the fasted state is tightly regulated. This metabolic shift includes reducing glucose oxidation and storage, and increasing the supply of free fatty acids (FFA) and ketone bodies (KB) to peripheral tissues. Glucose is spared for obligate glucose-consuming cells (such as some neurons, erythrocytes, kidney cells) by FFA's effects on membrane glucose transporters in peripheral tissues, upregulation of lipolytic enzymes and downregulation of glycolytic enzymes. Overall, if extended, fasting will produce a state of peripheral insulin resistance, which implies that skeletal muscle will become desensitized to insulin's effect, thereby reducing glucose transport into cells. However, this scenario differs from that observed during pathological insulin resistance, in which skeletal muscle, liver and adipose tissue are insulin resistant, so there is no control of glucose and FFA levels, leading to glucotoxicity and lipotoxicity.

Healthy humans should be able to fast without problems (metabolic flexibility). If everything is working as supposed to, there should be no problem when switching to a predominantely lipolytic/ketogenic metabolism. How hard the transition to a fasting state is may be a marker of the functioning of intermediate metabolism. 

Crawford et al. (1) tested a basic hypothesis, based on previous findings:

• Germ-free (GF) mice are leaner than conventionally raised (CONV-R) mice, even when GF mice consume more food (2).
• Transplantation of the gut microbiota from obese mice to GF recipients causes a greater increase in adiposity than does a microbiota from lean mice (3).

Overall, there seems to be an alteration in the composition and the microbiome of gut bacteria from obese mice, which increases energy harvest from available food. This is an aberration from its normal function, which is to provide an adequate amount of energy from otherwise indigestable nutrients. If true, then we might expect that this mechanism is beneficial to the host in periods of nutrient deprivation. If not, there would be no obvious evolutionary explanation for this symbiotic relationship. Accordingly,  after withdrawal of nutrients, GF mice die more rapidly that CONV-R mice, even when the rate of body weight loss is the same (4). Because adaptation to fasting involves a shift towards ketogenesis, the gut microbiota might regulate this process.

The gut microbiota regulates ketone body metabolism during fasting

Compared to CONV-D*, GF mice showed 37% lower levels of serum betahydroxybutyrate (BHB) in the fasted state, with no differences in the fed state. Moreover, levels of insulin, glucose, FFA and triglycerides where unchanged. CONV-D mice had increased hepatic triglyceride stores compared to GF mice, difference which was enhanced dramatically with fasting. As expression of Pnpla2 was increased similarly in both CONV-D and GF mice, fasting-induced fatty acid mobilization was not impaired by the absence of gut microbiota. PPARa expression was higher in CONV-D than in GF mice, and the fasting-induced ketogenesis was impaired in CONV-D PPARa-/- mice. 

The liver produces BHB for utilization in peripheral tissues. It lacks the key enzyme 3-oxoacid CoA transferase, so is incapable of oxidizing ketone bodies. When fasted, GF mice had 50% less concentration of liver BHB compared to CONV-D mice. Expression of Fgf21 and Hmgcs2, both targets of PPARa which stimulate ketogenesis, was lower in fasted GF mice compared to CONV-D mice. 

These results suggest that the gut microbiota stimulates ketogenesis during fasting by a PPARa-dependent mechanism. Additionally, it promotes hepatic triacylglycerol synthesis and storage. 

There are two possible ways of generating acetyl CoA in the liver during a fast: from acetate produced in the gut and from oxidation of fatty acids from adipose tissue. In GF mice, the only source is the latter. Cecal acetate levels were very low in fed GF mice and 20-fold greater in CONV-D mice. During fasting, these levels were reduced in CONV-D, but remained significantly higher compared to GF mice, which showed no reduction. 

One unexpected result was the change in the microbiota composition induced by fasting. It was found that there was a significant increase in the proportion of Bacteroidetes and a significant reduction in the proportion of Firmicutes. Previous studies have found a correlation between a reduction in Bacteroidetes and an increase in Firmicutes with obesity (the high Firmicutes/low Bacteroidetes hypothesis) (5, 6). 

The gut microbiota influences myocardial metabolism

Analysis of the myocardial transcriptome of CONV-D and GF mice revealed an enrichment in the ketone body metabolic pathway in both groups, compared to their PPARa -/- counterparts. Expression of genes involved in fatty acid and ketone body metabolism were increased with fasting in both groups, but Oxct1 gene expression was higher in CONV-D mice. Conversely, an increased Glut1 expression was only observed in GF mice. 

The rate of glucose oxidation was significantly increased in isolated working hearts of GF mice, without alteration of fatty acid oxidation. In the absence of BHB, glucose utilization was also significantly greater. Glycogen levels were reduced in the myocardium of fasted GF mice compared to CONV-D mice, and there were no significant differences in heart rate, cardiac hydraulic work, mitochondrial morphology or number, or mitochondrial respiration. 

The shift towards a ketogenic metabolism in the myocardium is one hallmark of adaptation to fasting, as BHB is more energy efficient than glucose. So in periods of deprivation of nutrients, the myocardium maintains its normal functioning by using BHB instead of relying in glucose. This adaptation is impaired in GF mice.

The gut microbiota affects myocardial mass

Heart weight was reduced in fasted GF mice compared to CONV-D mice. Training elicits an hypertrophic response in the heart. However, this response was blunted in the absence of gut microbiota, as the hearts of trained GF mice remained smaller than the hearts of trained CONV-D mice. This correlated with alterations in a subset of pathways, which included ketone body metabolism. Administration of a ketogenic diet rescued heart weight in GF mice and shifted the myocardial transcriptome toward ketone body metabolism.

These results suggest that the gut microbiota is an important component for cardiovascular health, and that ketone bodies represent an essential substrate for the heart. 

Summing up

This is one of the most interesting studies that I have read lately. It provides a new template for the relationship between metabolism and gut microbiota, and shows the importance of gut bacteria for the normal response to fasting. I have summarized the findings of the study in the following figure (my interpretation):

The gut microbiota regulates ketogenesis during fasting. Fasting induces an increase in the proportion of Bacteroidetes and a reduction in the proportion of Firmicutes. These changes promote the production of acetate, which serves as substrate for hepatic acetyl CoA synthesis. The gut microbiota also stimulates hepatic triglyceride stores, providing another source of energy during fasting. The increase in acetyl CoA levels stimulates ketogenesis by a PPARa-dependent mechanism, increasing serum BHB levels. The elevated concentration of BHB levels supplied to the heart promotes the shift towards a ketone body-based metabolism, and inhibits glucose oxidation. Myocardial ketone body metabolism maintains myocardial mass and the normal hypertrophic response to exercise.

* CONV-D (conventionalized) mice were transplanted with distal gut microbiota from CARB-fed CONV-R lean mice. 

Crawford PA, Crowley JR, Sambandam N, Muegge BD, Costello EK, Hamady M, Knight R, & Gordon JI (2009). Regulation of myocardial ketone body metabolism by the gut microbiota during nutrient deprivation. Proceedings of the National Academy of Sciences of the United States of America, 106 (27), 11276-81 PMID: 19549860

Fecal bacteriotherapy

Proper gut microbiota establishment begins in the moment we are born and is shaped by lifestyle and environmental factors in subsequent years. In some cases, the degree of dysbiosis is so severe that there is not turning back and practical dietary/lifestyle recommendations are useless.

Fecal bacteriotherapy

The logic behind this intervention is simple: it tries to "reset" the gut microbiota. It has shown promising results in intestinal bowel disease (IBD) and resistant Clostridium difficile infections. The following protocol is taken from Silverman et al (1). My intent is to facilitate information, not to encourage the realization of this protocol without medical supervision. Interested persons should consult with their doctors before doing any procedure of this nature. Donors and recipients should be examined carefully before the intervention. The complete set of tests can be consulted in the mentioned study.

Pretreatment

Recipients are initiated on maintenance therapy with oral Saccharomyces boulardii (probiotic), 500mg orally, twice per day. Metronidazole (500mg/3 times per day, PO) or vancomycin (125mg/4 times per day, PO) are also used. Both are antibiotics normally used against C.difficile infections.

Equipment

- 1 bottle of normal saline (200mL)

- 2 standard 2 quart enema bag kits (available at drug stores)

- 3 standard kitchend blenders (1L capacity) with markings for volume

Procedure

- Vancomycin/metronidazole should be stopped 24-48 hours before procedure.

- S.boulardii should be continued during the transplant and 60 days afterwards.

1.     Add 50mL of stool (volume occupied by solid stool) from the healthy donor immediately prior to administration (< 30 minutes) to 200mL of normal saline in the blender.

2.     Mix until getting a "milkshake" consistency.

3.     Pour mixture (approximately 250mL) into the enema bag.

4.     Administer enema to the recipient following the kit instructions. The patient should hold the infusate as long as possible and lie still as long as possible on his/her left side to prevent the urge of defecation. The procedure should be ideally performed after the first bowel movement.

5.     If diarrhea recurrs within 1 hour, the procedure may be immediately repeated.

Modifications and perspectives

This procedure was made to treat C.difficile infections. Accordingly, the antibiotics and the probiotic used aimed to eliminate C.difficile from the gut. However, there are certain modifications which can be useful for treating severe dysbiosis. First, broad-spectrum antibiotics can be used to wash out most bacterial species and reduce colonization resistance. In addition, utilization of probiotics such as Bifidobacteria or Lactobacilli during and after the treatment should help preventing colonization by enteropathogenic species. Why Bifidobacteria? The use of broad-spectrum antibiotics increases the risk for colonization of enteropathogens. Bifidobacteria competes and prevents colonization by these pathogens directly and indirectly, via production of antibacterial molecules (2). In addition, dysbiosis is characterized by low levels and expression of Foxp3+ Tregs, which compromises immune tolerance and promotes inflammation. Oral administration of B.infantis has been shown to increase expression of Foxp3+ and IL-10 in peripheral blood and to drive maturation of dendritic cells towards a regulatory phenotype (3), and certain strains of Bifidobacteria are capable of modulating the plasticity of Th17/Treg populations in human PBMCs (4). On the other hand, Lactobacilli has also shown protective properties (specially against vaginal infections) (5) and competes with enteropathogens for adhesion on intestinal epithelial cells (6). Importantly, the effects over Treg induction and T cell differentiation differ between strains from the same species. I should address this issue in future posts. One thing that is not emphasized in the above protocol is the importance of diet for maintaining a correct microbiota. This, in my opinion, is key to success.

It is worth noting that because of the nature of the procedure, the microbiota of recipient subjects is altered and reduced, but not completely eliminated such as seen with studies in fecal transplantation. The utilization of fecal transplantation in humans is promising and should result in better outcomes. Indeed, positive preliminary results from the FATLOSE trial (7, 8) have been recently published in which patients with metabolic syndrome improved insulin resistance and lipid profiles after feces infusion from healthy donors. The positive results seem to be correlated with increases in colonic butyrate concentrations. These results fit nicely with the ones found previously with fecal transplantation in obese mice.

Turnbaugh et al (9) found astonishing differences in the microbiome of obese mice, compared to lean mice (greater abundance of Firmicutes). Metagenomic analysis revealed that the obese microbiome is enriched for EGT (environmental gene tags) encoding many enzymes invoved in the break down of otherwise indigestable dietary polysaccharides. These included KEGG pathways for starch/sucrose metabolism, galactose metabolism and butanoate metabolism. Increased concentrations of butyrate and acetate were also observed, as the fact that obese mice were able to harvest more energy compared to lean mice (assessed by less energy remaining in feces by bomb calorimetry,). Despite equal amount of food consumed in both groups, colonization of lean mice with obese microbiota led to an increase in bodyfat percentage of approximately 47% after two weeks. The potential for fecal bacteriotherapy in the treatment of several diseases has been observed in different animal models of inflammatory and autoimmune diseases.

Thus, it seems possible that future therapies for obesity, metabolic syndrome and other inflammatory/autoimmune conditions will aim to modulation of the gut microbiota.

Silverman MS, Davis I, & Pillai DR (2010). Success of self-administered home fecal transplantation for chronic Clostridium difficile infection. Clinical gastroenterology and hepatology : the official clinical practice journal of the American Gastroenterological Association, 8 (5), 471-3 PMID: 20117243

Glucose restriction and TSC

Recently, zooko asked me about my opinion on a recent study just published (1).

Background

TSC1 and TSC2 are a pair of tumor supressor genes, which relevance lies in the inhibition of mTORC1 activity. mTOR (the mammalian target of rapamycin) is a master regulator of cell proliferation, cell growth, cell motility, cell survival, protein synthesis and transcription. Because of this, dysregulation of the mTOR pathway is seen in many cancers (2).

mTOR forms two complexes, mTORC1 and mTORC2 ("rapamycin-insensitive"), which respond to different stimuli. TSC2 has a GAP (GTPase activating protein) domain that stimulates the GTPase activity of Rheb. GDP-Rheb is inactive, while GTP-Rheb is active. By this mechanism, TSC2 accelerates the hydrolysis of GTP, inactivating Rheb. Active Rheb is a potent activator of mTORC1. The interplay between these proteins is shown below (3):

In response to growth factors, Akt phosphorylates TSC2 directly on four or five residues (Ser939, Ser981, Ser1130, Ser1132 and Thr1462). Phosphorylation of TSC2 by Akt impairs its ability to inhibit Rheb, thereby blocking the inhibitory effect of Rheb on mTORC1. Other mechanisms proposed to explain contradictory experimental results include the action of PRAS40 and binding of 14-3-3 to phosphorylated TSC2.

The most conserved pathway for Akt activation is the PI3K/Akt pathway. Insulin/IGF-1 binding to the insulin receptor produces phosphorylation of its cytosolic domain, promoting the binding of IRS (insulin receptor substrate) by its PTB (phosphotyrosine binding) domain. This promotes the association (by SH2 domains in the p85 regulatory subunit) and activation of PI3K. PI3K phosphorylates phosphatidylinositol-4,5-biphosphate (PtdIns(4,5)P2), producing PtdIns (3,4,5)P3. PtdIns (3,4,5)P3 binds to the PH domain of Akt and promotes its translocation to the plasma membrane. PI3K-dependent kinase 1 (PDK1) then phosphorylates Akt on Thr308 and PDK2 phosphorylates Ser473. Both phosphorylations activate Akt. Phosphorylated Akt, as previously mentioned, phosphorylates and inactivates TSC2 and PRAS40 promoting mTORC1 activation. The activation of the PI3K/Akt pathway has many downstream cellular events promoting cell survival and proliferation, which include inactivation of several proapoptotic factors (BAD, procaspase-9 and Forkhead transcription factors), activation of antiapoptotic factors (CREB), activation of IKK, inactivation of p53, among other. The net result is promoting proliferation and cell survival, hallmarks of cellular malignancy development and progression.

Discussion of the study

The basis for the utilization of glucose restriction for treating TSC related tumors can be easily inferred from the above explanation. By restricting glucose, insulin signaling is reduced, so is activation of mTORC1. mTORC1 also upregulates HIF1a, promoting aerobic glycolysis and lactate production (4). Thus, glucose restriction should promote apoptosis, specially in TSC1/TSC2-null cells.

What the authors basically did was treating Tsc2-/- mice with a carbohydrate-free diet (CF) and a Western Diet, alone or combined with 2-deoxyglucose (2-DG). Preliminary in vitro studies showed that Tsc2-/- cells were sensitive to glucose restriction and 2-DG in an additive manner. Contrary to what was expected, in vivo experiments showed that tumor size and growth rate were highest in the CF group and 2-DG supressed tumor growth independently of diet. These results also contradicted the observed standard uptake values (SUV) during the FDG-PET scan (presented as the maximum SUV within each tumor). Theoretically, as these tumors are sensitive to glucose deprivation, there should be a correlation between glucose uptake (measured by uptake values of FDG) and tumor size (increased tumor size should show increased SUV). However, there was no correlation between these parameters, as the CF+2-DG group showed the minimum mean SUV but the largest tumor size. What can be fueling tumor growth? Ketone bodies? An in vitro assay showed that addition of either acetate or beta-hydroxybutyrate to Tsc2-/- cells increased cell confluence and reduced the number of non-viable cells (assessed by trypan blue), compared to glucose alone. To further complicate things, ketonemia was not developed in CF mice, but beta-hydroxybutyrate levels were higher with the Western +2-DG diet. Testing the effects of fatty acids in vitro showed that palmitic acid induced necrosis and oleic acid induced proliferation. This correlated with the histologic analysis of CF mice. Addition of rapamycin reduced cell-size, in contrast with 2-DG, which decreased proliferation. Finally, there was increased activation of mTORC1 (measured by phospho-S6) and low levels of phosphorylated Akt (secondary to feedback inhibition) in all groups, with no differences between groups.

Interpretation

First, the results confirm the potent anti-tumor activity of 2-DG. Second, the CF group failed to establish ketosis, and the Western group had increased levels of beta-hydroxybutyrate, as well as reduced tumor size. This (despite the observed growth-promoting properties of acetate and beta-hydroxybutyrate in vitro, see below) can be interpreted as an inhibitory effect of ketonemia on cancer growth. The comparison of glucose and beta-hydroxybutyrate levels is shown below:

The diet which resulted in lower glucose levels and higher ketone bodies was associated with reduced tumor size, and the diet which produced greater glucose levels and lower ketone bodies was associated with increased tumor size. The results observed in vitro with Tsc2-/- cells and ketone bodies suggest that in this cell line, it is necessary an additional anti-glycolytic factor to control tumor growth (2-DG), because these (and other) cancer cells seem to be capable of metabolizing ketone bodies (5, 6). This underscores the importance of the phenotype of the tumor being treated, an important factor that is not taken into account by some "low-carb" advocates who think that restricting dietary glucose will magically cure all cancers.

Another important factor to take into consideration is that mice were not calorie restricted, and more importantly, that the CF diet was high in protein. Glutamine is a major substrate that can fuel cancer cells (7, 8). On a more general level, aminoacids are potent stimulators of mTORC1 (9, 10). This is specially relevant for this model, because excess aminoacids can by-pass the inhibition of the PI3K/Akt pathway by promoting Rheb co-localization with mTORC1 (11), activating mTORC1 in the abscence of TSC2 (12). Ketone bodies increased ATP levels in Tsc2-/-in vitro. This reduces AMPK activity. AMPK inhibits mTORC1 activity by TSC2 dependent and independent mechanisms (possibly by phosphorylation of Raptor) (12). 2-DG also increases intracellular AMP levels (activating AMPK), which would explain the benefits of its utilization observed in this model (13). Supporting the role of AMPK as a target for cancer treatment, the combination of metformin and 2-DG seems to be more toxic to cancer cells than either by itself (14). Interestingly, AMPK activity was not changed in response to 2-DG in this model, which suggests that there are other mechanisms mediating the anti-proliferative effect of 2-DG.

Summing up

Cancer is a very complex disease which treatment has to be personalized depending on the phenotype. With the increase knowledge in cancer molecular biology and genetics, therapies should be designed depending on specific markers evaluated. This complexity explains why not all cancers can be treated just by restricting glucose and making such statement is ludicrous. Besides calorie, glucose and protein restriction, compounds such as 2-DG and metformin show promising effects for controlling most types of cancer.

Jiang X, Kenerson HL, & Yeung RS (2011). Glucose deprivation in Tuberous Sclerosis Complex-related tumors. Cell & bioscience, 1 (1) PMID: 22018000