The darkside of leptin

Disclaimer: This post is not related specifically to ketogenic diets or ketosis. 

It is common to see leptin as the "fat-burning" hormone which has antagonizing effects on insulin, the "fat-promoting" hormone. So, in general terms, people tend to see leptin as "good" and insulin as "bad". As I mentioned in my last post, there is no such thing as a bad or good hormone, its all about a correct balance and tissue sensitivity. Before going on discussing leptin, we must review some basics of this molecule. 

Leptin is a 16kDa protein encoded by the ob gene which is produced primarily by adipose tissue so its circulating levels are correlated with adipose tissue mass. Based on its structure and that of its receptor, leptin is a cytokine. It shares many similarities with members of the long chain helical cytokines such as IL-6, IL-1, IL-12, leukemia inhibitory factor (LIF), granulocyte-colony stimulating factor (G-CSF), ciliary neurotrophic factor (CNTF) and oncostatin M (OSM) (1). 

The leptin receptor (OB-R) is related to type I cytokine receptors and form homodimers, both in the absence or presence of ligand. Each receptor binds to one molecule of leptin forming a tetrameric complex composed of two receptors with two leptin molecules. There are different alternatively spliced isoforms of OB-R: OB-Rb mediates leptin's effect in the hypothalamus (weight-regulating effects) but it is also present in several peripheral tissues like endothelial cells, platelets, CD4+ and CD8+ T lymphocytes, CD34+ cells, the yolk sac, the fetal liver, as well as leukemia cells. A short isoform of OB-R is OB-Ra which is found in most tissues and cells like kidney, lung, liver, spleen and macrophages. Other isoform of OB-R is the soluble receptor OB-Re, which bounds leptin in the peripheral circulation. 

We are going to focus in OB-Rb mediated pathway. OB-Rb contains a 302 amino acid cytosolic domain that includes binding motifs associated with the activation of the JAK/STAT signaling pathways. After binding to the receptor, leptin activates STAT-1, -3 and -5. Leptin-OB-Rb interaction has shown to activate the MAPK pathway and induce the expression of supressor of cytokine signaling 3 (SOCS-3), which acts as a negative regulator of its signalling. Leptin increases specifically p38 MAPK phosphorylation and activates the JNK pathway. The main downstream target seems to be NF-kB (2).

Besides its effects in the hypothalamus, leptin, as a pleiotropic protein, has important roles in both immunity and inflammation (reviewed in 3). I will try to summarize what it is known about this relationship.

Leptin and the immune system

Leptin has been shown to promote phagocytosis and induce eicosanoid synthesis, as well as NO and pro-inflammatory cytokines in macrophages and monocytes; increase IFN-gamma induced expression of NO synthase in murine macrophages; induce chemotaxis and the release of ROS in neutrophils and influence proliferation, differentiation, activation and cytotoxicity on natural killer cells (NK). In dendritic cells (DC), leptin upregulates the production of IL-1beta, IL-6, IL-12, TNF-alpha and MIP-1alpha; downregulates IL-10 production and polarizes naive T cells towards Th1 phenotype (4). Moreover, leptin also protects DC from UVB and H2O2-induced apoptosis via NF-kB, bcl-2, bcl-XL and Akt activation (5). 

As studies with db/db mice have shown, leptin is an integral part of the immune system (6). But as a pro-inflammatory cytokine, it has potential downsides.

Leptin and inflammation

Leptin levels have been shown to be elevated during infection and inflammation (7). Exposure to LPS, TNF-a and IL-1 increase its circulating levels and expression in adipose tissue (8). Additionally, leptin increases LPS-stimulated production of TNF-a, IL-6 and IL-12 in murine peritoneal macrophages and human monocytes, as well as TNF-a, IL-1beta and IL-6 from human placenta and adipose tissue (9). This shows a positive feedback between leptin and other pro-inflammatory cytokines. 

Hyperleptinemia has been correlated with cardiovascular disease (10), endothelial dysfunction (11), atherosclerosis (12), arterial hypertension (13), psoriasis (14), multiple sclerosis (15), diabetes (16) and metabolic syndrome (17).

Leptin and cancer

Several in vitro studies have linked leptin signaling to carcinogenesis. For example, leptin and OBR seem to be overexpressed in mammary cancer tissue relative to non-cancer epithelium (18). Leptin has shown to induce growth of breast cancer cells through activation of the JAK/STAT3, ERK1/2 and PI3K pathways, and can mediate angiogenesis by inducing the expression of VEGF (19). Colon cancer cell lines and human colonic tissue also express the OB receptor, and stimulation with leptin induces phosphorylation of p42/44 MAPK and increases proliferation in vitro and in vivo (20). Leptin also induces invasion of colonic cells and formation of lamellipodial structures in human colonic cell lines LS174T and HM7, by activation of RhoA, Cdc42 and Rac1 in a dose-dependent manner (21). Prostate cancer cell lines DU145 and PC-3 treated with leptin show significantly increased expression of VEGF, TGF-beta1 and bFGF, as well as increased cell migration, which was partially inhibited with the addition of MAPK and PI3K inhibitors (22). There is evidence that leptin also increases integrin expression and migration of prostate cancer cells, by activating the OBR1/IRS-1/PI3K/Akt/NF-kB pathway (23). I can go on and on with different types of cancer  which seem to be promoted by leptin but from the examples above a clear trend can be seen (for a review please refer to 24).

Conclusions

Is leptin just a marker of metabolic diseases or has an active role? From the literature we can see that hyperleptinemia is present in many chronic diseases. Hyperleptinemia is a marker of metabolic dysregulation ie. leptin resistance and inflammation. This does not necessarily imply that leptin is the only factor behind these pathologies, but as a pro-inflammatory cytokine, mediates and exacerbates disease together with other cytokines. This should end the myth of leptin as the "good" or "magic" hormone. Unless there is a leptin deficiency issue, there is no point in trying to increase leptin levels for the sake of increasing them, as it could have negative effects*. As with insulin, low leptin is better than high leptin. And by the way, plasma leptin is directly correlated with plasma insulin (25), so an optimal metabolic mielieu is one in which insulin and leptin are kept low. 


* As for instance, exogenous leptin administration for weight loss in the abscence of leptin deficiency (See 26).

Lago R, Gómez R, Lago F, Gómez-Reino J, & Gualillo O (2008). Leptin beyond body weight regulation--current concepts concerning its role in immune function and inflammation. Cellular immunology, 252 (1-2), 139-45 PMID: 18289518

Hypothalamic leptin and insulin signaling pathways

Obesity research is now focusing on dysruption of neuronal pathways involved in energy homeostasis and feeding as one of the main causes of obesity and associated diseases. Insulin and leptin are two key hormones controlling satiety in the hypothalamus (1):

People new to nutrition who find information on low carbohydrate dieting seem to get confused by the fact that insulin stimulates satiety. After all, isn't insulin the evil fat promoting hormone that makes us ultimately fat? Discussing this issue is beyond of this post and interested readers are referred to this review (2). Stephan Guyenet has several posts about satiety and food reward. 

Low carbohydrate diets are known for decreasing plasma insulin levels as well as leptin. Leptin secretion seems to be correlated with insulin secretion, so carbohydrates are the main stimulus for increasing leptin levels (3,4,5). One might think that a very low carbohydrate ketogenic diet (VLCKD) would result in appetite stimulation and low satiety. This leads us to a study published by Park et al (6). 

They used two rodent models: 90% pancreatectomized diabetic (Px) and sham-operated non-diabetic rats (Sham). Rats were infused CSF or 50ug/h of beta hydroxybutyrate (bOHB) (Sigma) into the lateral ventricle for 3h. The hyperinsulinemic euglycemic clamp was performed with a continuous infusion of bOHB. This was the short-term part of the study.

Figure A shows glucose infusion rates and glucose uptake and figure B shows hepatic glucose output at baseline and clamped steady-state. Whole-body glucose infusion rates required to maintain normoglycemia were higher in the bOHB-Px rats, but nearly the same in bOHB-Sham rats, compared to CSF. Glucose uptake was basically the same in all groups, while hepatic glucose output during the clamp was reduced in bOHB-Px rats compared to CSF, but not in Sham rats. 

The long-term study involved Pax rats which were infused CSF or 12ug/h of bOHB into the lateral ventricle for 4 weeks. bOHB levels in Pax-bOHB rats were significately higher compared with CSF, while other metabolic parameters (body weight, epididymal fat pads, food intake, serum leptin, serum insulin) were not significantly different between groups (p<0.05). 

Glucose tolerance was assessed using an OGTT. bOHB rats had lower 30-60min serum glucose levels and lower glucose AUC than the CSF group (A). During the first part of the test (0-40min) insulin AUC was the same in both groups while during the second part (50-120min) was slightly lower for the bOHB group (B).

During the hyperglycemic clamp, peak serum glucose levels were lower on the bOHB group than on the CSF group (since serum glucose increased about 5.5mM from basal levels and these were lower in bOHB rats). Infusion of bOHB increased insulin secretion at 2 min compared to CSF, without differences in second-phase secretion. Insulin AUC was equal in both groups, and glucose infusion rates were higher for bOHB than CSF. Insulin sensitivity at the hyperglycemic clamp state was improved in bOHB rats but not in CSF rats. 

Whole body glucose disposal rates were lower in the bOHB group than in the CSF group without being differences in glucose uptake. As with glucose levels after an overnight-fast, hepatic glucose output was lower in bOHB rats. Because basal serum glucose levels were determined by hepatic glucose production and glucose utilization, and glucose uptake was similar between groups, the differences in glucose infusion rates and glucose uptake could be accounted by insulin-stimulated supression of hepatic glucose production. This reflects increased hepatic insulin sensitivity. As expected, hepatic glucose output during the hyperinsulinemic clamp was lower in bOHB rats, suggesting that ketone infusion was associated with improved hepatic insulin action. 

Now we get to the interesting part: insulin and leptin signaling. Results in the hypothalamus are shown below:

Two samples were used for each group for the immunoblotting assay, and values shown are the mean. As can be seen in the figure, infusion of bOHB potentiated tyrosine phosphorylation of IRS2 and serine phosphorylation of Akt. Expression of GLUT2 and glucokinase were increased. This suggests improved glucose sensing in the hypothalamus.

STAT3 phosphorylation was potentiated more in the bOHB group than in CSF, without differences in AMPK phosphorylation.

Confirming the metabolic parameters evaluated earlier and results in the hypothalamus, insulin signaling was also potentiated in the liver:

Tyrosine phosphorylation of IRS2 and phosphorylation of Akt were increased in the bOHB group. Consistent with these results, PEPCK expression was reduced and GLUT2 and glucokinase expressions were potentiated, as well as phosphorylation of AMPK. 

Summing up

Central infusion of bOHB increased insulin and leptin signaling both in the hypothalamus and the liver, increasing leptin and insulin sensitivity. I think this has huge implications in the treatment of obesity. Insulin and leptin are not deleterious per se. The problems arise when plasma levels reach high enough to start causing metabolic complications and resistance by key tissues to these hormones. 

Additional information

Research on the effects of beta hydroxybutyrate and specifics on food intake regulation with ketogenic diets is scarce. Laeger et al. (7) have reviewed most information available, from which we can get the following conclusions:

- bOHB uptake by the brain is mediated by two forms: difussion and a carrier mediated system by monocarboxylate transporters (MCT) and the sodium coupled MCT 1. 

- Brain bOHB concentration rises concomitantly with increasing plasma ketone body concentrations. The permeability of the BBB for bOHB increases with starvation and high-fat diets, and is reduced with age. 

- Studies examining the effects of bOHB on satiety and the hypothalamus have been done both by central infusion and peripheral infusion.

- bOHB-mediated appetite supression is dependent on the rat strain. In those sensitive, central infusion suppresses food intake and reduces body weight. Conversely, obesity resistant S 5B/P1 rats usually have higher blood ketone concentrations and a higher transport of bOHB across the BBB, leading to chronically higher brain bOHB levels compared with other strains. These rats do not show the same effects as other strains to bOHB infusion. 

- In other rat strains (like Sprague-Dawley), central infusion of bOHB results in loss of bodyweight without affecting blood glucose levels or glycogen content. Other studies have shown that after an initial increase in food intake, there is a long-term supression with an intraventricular infusion of bOHB. Postprandial intermeal interval was also prolonged. 

- Subcutaneous injection of bOHB has shown to reduce food intake under normal physiological conditions. This is also seen in insulin treated hypoglycemic rats receveing an intravenous bOHB infusion.

- In African pygmy goats, hypophagia appears to be related to the amount of bOHB injected intraperitoneally. This leads to a decrease in meal frequency and prolongation of latency to eat without diminishing the meal size. 

- In sham-vagotomized rats, subcutaneous injection of bOHB reduces food intake, but vagotomy of the common hepatic branch eliminated the hypophagic effect. It appears that because most of the common hepatic branch vagal afferent fibers originate in the proximal small intestine and the liver parenchyma barely contain vagal afferents, oxidation of bOHB in enterocytes is signaled to the brain where it is sensed by neurons to affect eating behaviour. This contrasts with earlier hypotheses that speculated that stimulated hepatic bOHB metabolism was recognized by vagal terminals. 

- Attributing reduced food intake to bOHB in studies with ketogenic diets is often difficult because it might include increased amounts of protein, which stimulates satiety.

Park S, Kim da S, & Daily JW (2011). Central infusion of ketone bodies modulates body weight and hepatic insulin sensitivity by modifying hypothalamic leptin and insulin signaling pathways in type 2 diabetic rats. Brain research, 1401, 95-103 PMID: 21652033

Ketones and environmental toxicity

It is a well known fact that ketone bodies protect cells from oxidative stress and apoptosis (1,2,3,4,5). Because toxicity from different environmental toxic substances can be modulated by nutritional status, and specifically, dietary macronutrients, a ketogenic diet could reduce greatly the risk of several associated diseases. 

Soman is an extremely toxic substance which inhibits cholinesterase activity, having profound effects on the CNS. Increased exposure can ultimately lead to death. Langston and Myers (6) tested the influence of diet on soman toxicity in rats. For this purpose, rats were fed four different diets: standard (SD), choline-enriched (CH), glucose-enriched (GL) and a ketogenic diet (KD). The doses used in this study were 0.4-0.5 of the acute 24-h LD50. This dosing regimen was chosen to induce significant cumulative toxicity that would permit characterizing differences in the rate of onset of soman toxicity, the degree of toxicity, and the rate/degree of recovery from soman toxicity as a function of diet composition. Macronutrient composition (as percentage of total calories) was as follows (protein/carbohydrates/fat):

SD: 32.4/54.2/13.4
CH: 32.5/54.2/13.3
GL: 20.2/66.8/12.9
KD: 14.3/0.4/85.3

After 4 weeks, KD-fed rats and CH-fed rats weighted less than SD and GL. But this is not in what I am interested. Lets see how the different diets affected soman mediated toxicity.

From the graph, we can see a clear trend on survival among the different diet groups. 30% of the GL group died after the third administration of soman (0.4 LD50) and the remaining 70% died after the fourth administration of 0.5 LD50. At this cumulative dose (352ug/kg), only 10% of the SD group died, while all of the rats fed a KD and CH survived. When the cumulative dose was greater than 400ug/kg, we can see how the lines representing the CH and SD groups begin to fall. At this point, both groups had an approximate terminal survival value of 55%. In contrast, only one animal in the KD group died after the final soman administration (cumulative dose of 627ug/kg). 90% of KD-fed rats survived. 


Soman administration also reduced the body weight of exposed animals. For simplicity, I will only compare the GL and KD groups. CH and SD rats only lost weight in the last week of exposure, as well as through the recovery period (SD). Time vs. weight graphs are shown below (x represent soman/saline injection days; o represent non-injection days):

GL rats started to lose weight rapidly after soman exposure, compared to vehicle animals. This was evident as early as the second week. Now, lets compare the body weight on soman-exposed animals fed a KD.

KD rats showed little (non significant) weight loss after soman exposure. The only little difference observed was at the maximum cumulative dose.

The last parameter measured by the researchers was avoidance behaviour. Again, I will compare only the GL group with the KD group.

Rats fed an enriched glucose diet had disrupted avoidance performance during the initial 3 days of exposure to 0.4 LD50 of soman. During recovery, avoidance responses were recovered and then were completely supressed (at 0.5 LD50). In contrast, avoidance responses in the KD were only compromised during the final week of exposure. Although the CH group also maintained avoidance responses until the last week, performance in the KD was greater. 

Results were similar for the time spent in the aversive stimulus (AS) (in this case, a scrambled 1.0-mA shock with a frequency of 1.0Hz, 0.5s shock on/0.5 shock off). During the second week of exposure, GL rats failed to respond to the AS and they all died by the end of the week. Once again, KD rats performed the best. AS time was maintained during the entire exposure period. 

Summing up, the authors stated:

"Specifically, all KD animals survived a cumulative 5.0 LD50 dose of soman, whereas all glucose animals died following a cumulative 3.2 LD50 dose of soman. Not only was survival enhanced in KD animals, but there were also minimal differences in body weights compared to dietary controls injected with saline. Furthermore, KD animals exposed to soman exhibited few performance decrements on an avoidance task, and there were fewer instances of behavioral incapacitation in KD animals compared to the other diet groups."

Glucose feeding has shown to exacerbate toxicity from other substances, such as organophosphorus pesticide parathion (PS) (7,8). Interestingly, Slotkin et al. (9) showed that a high-fat ketogenic diet reversed the neurodevelopmental effects of neonatal PS exposure. 

Although the studies mentioned were done in rats, it seems plausible that the effects could be reproduced in humans. Judging by the results, a ketogenic diet could help treating as well as preventing environmentally mediated toxicity. 

Langston JL, & Myers TM (2011). Diet composition modifies the toxicity of repeated soman exposure in rats. Neurotoxicology PMID: 21641933