mice fed either normal (ND) or high fat (HFD) diet gained weight at a similar rate to the C57Bl/6J control cohort (), and their weights were not statistically significantly different at the time of metabolic studies. Consistent with this finding, NMR analyses of lean () and adipose () mass demonstrated no genotype-specific differences between CD14-ko
mice from controls. Similarly, serum cholesterol, triglyceride, and non-esterified fatty acid (NEFA) in CD14-ko
mice were not significantly different from wild type mice (). As total energy expenditure and altered circadian rhythms can contribute to metabolic variation in whole-body knockout mice, all mice studied were individually housed in activity cages 
. Total spontaneous ambulatory movements and their circadian rhythmicity were indistinguishable from age- matched controls ().
Body weight, adiposity, and activity in wild-type and CD14-ko mice.
Lipid Profile and Fasting Insulin of wild-type and CD14-ko mice at 52 weeks of age following 41 weeks of dietary treatment.
Despite the lack of genotype-specific changes in overall body composition in CD14-ko mice, glucose tolerance tests (GTT) uncovered glucose intolerance. This phenotype was seen in 26-week-old animals (), and became more pronounced with age, as seen in 52 week-old old mice (). Glucose tolerance was impaired in response to a HFD in wild-type animals, and was further impaired in CD14-ko mice (). We investigated whether the glucose intolerance seen in these mice could be attributed to insufficient insulin secretion or peripheral insulin resistance. Fasting insulin levels were not significantly different between genotypes (), although they increased by HFD in both genotypes. CD14-ko mice did not have defective insulin secretion, as the time-weighted average insulin values following an i.p. bolus of glucose (2 g/kg body weight) in these mice exceeded that in wildtype controls (3213±344 ng-min/mL in CD14-ko vs 1187±82 ng-min/mL in control mice; p<0.005 respectively). Thus, glucose intolerance in CD14-ko mice does not appear to be due to impaired insulin secretion.
Effects of disruption of the LPS signaling pathway on glucose tolerance.
However, insulin tolerance tests (ITT) did demonstrate genotype-related alterations (). Specifically, while no significant differences were observed during the first 15 min post insulin injection, the rebound phase of the ITT, which reflects the counter regulatory response to hypoglycemia, was significantly enhanced in CD14-ko
compared to control mice (). This effect became more pronounced with age (), and was most pronounced in 52 week-old-mice fed a HFD () where, despite similar declines in the first 30 min after insulin injection, blood glucose levels of CD14-ko
mice were significantly higher than controls after 60 min, and significantly higher than values seen before insulin injection after 90–120 min post-injection (142±8% and 128±4% of initial glucose values at 120 min in 52-week-old WT and CD14-ko
mice on HFD; P values<0.0001 for CD14-ko
vs WT, n
Acute counter-regulatory pathways to hypoglycemia rely on adrenal-derived catecholamines to inhibit insulin secretion and peripheral glucose utilization, while promoting hepatic glycogenolysis and stimulating lipolysis 
. To determine whether adrenal catecholamines may explain the altered rebound phase in the ITT in CD14-ko
mice, we measured norepinephrine at baseline, at 45 min post-injection to coincide with the inflection in circulating glucose, and at 90 min post-injection when basal glucose levels are re-established. At both 45 and 90 minutes post-injection, norepinephrine levels in CD14-ko
mice exceeded the values seen in control mice (). Moreover, increased tonic catecholamine secretion could explain basal glucose intolerance in these mice, as the level of norepinephrine found in urine collected over a 24 hr period was significantly higher in CD14-ko
mice compared to controls (). To further test the hypothesis that a hyperactive adrenal response to hypoglycemia contributes to impaired glucose homeostasis in CD14-ko
mice, mice were treated with the β-adrenergic receptor antagonist propranolol prior to insulin administration. Pre-treatment with propranolol mitigated the excess glucose production in response to insulin-induced hypoglycemia in CD14-ko
Effects of disruption of the LPS signaling pathway on counter-regulatory response to hypoglycemia.
To search for additional evidence for an effect of CD14 depletion on adrenal function that could underlie the observed alteration in sympathoadrenal response to hypoglycemia seen in CD14-ko
mice, we analyzed the morphology of their adrenal glands. Adrenal glands from 5 mice from each genotype were excised, fixed and serially sectioned at 8 um intervals (). Every 10th section was photographed (), and the areas of the whole adrenal and of the medulla in each section were measured. These were summed to obtain a direct estimate of total adrenal and adrenal medullary volume (). This morphometric method was advantageous in that it eliminated variation due to extraneous tissue associated with the gland during dissection. While the total adrenal volume of CD14-ko
mice, assessed by the sum of area of all sections, was not different from WT mice, (), the associated adrenal medullary volume was significantly larger (). Strain-associated variation in adrenal volume has been documented 
. However, the mice used in these studies were backcrossed for more than 12 generations, making it unlikely that strain-related factors account for the differences observed.
Effects of disruption of the LPS signaling pathway on adrenal gland morphology.
The results presented here suggest that mice lacking a functional LPS signaling pathway become glucose intolerant due to a tonic effect of innate immune signaling to suppress the sympathoadrenal axis, suppressing the production of norepinephrine by the medulla or sympathetic nerve terminals, which produce the majority of norepinephrine in the mouse 
. The importance of the sympathoadrenal axis on both glucose homeostasis and innate immune signaling is well documented. Insulin sensitivity and glucose uptake are normalized in ob/ob mice by adrenalectomy 
. In humans, adrenal hyperactivity in Cushing's disease is marked by hyperglycemia and adrenal insufficiency in Addison's disease is marked by hypoglycemia. Thus, enhanced adrenal tone leads to hyperglycemia, through the activation of gluconeogenesis by glucocorticoids, as well as by the sustained counter-regulatory mechanisms mediated by norepinephrine.
Adrenal tone also plays an important role in innate immune signaling. Adrenalectomy results in dramatic hypersensitivity to LPS 
, possibly due at least in part to impairment in catecholamine-mediated protection from LPS-induced hypoglycemia 
. Also, LPS administration leads to a rapid induction of glucocorticoid production, which is impaired by TLR4 ablation 
, and which is thought to provide a counter regulatory response to mitigate exaggerated inflammatory responses to infection 
. Because of the strong suppressive effect of catecholamines on LPS-induced inflammatory and metabolic responses, a counter-regulatory input in which CD14/TLR4 signaling might exert a tonic suppression of adrenergic tone might be necessary to maintain adequate innate immune responsiveness. This may be the basis of the phenotype reported in this manuscript, where absence of CD14, which is expected to impair TLR4 signaling, causes enhanced medullary adrenal output.
Several studies have reported that deletion of TLR4 ameliorates insulin resistance produced by a high fat diet 
, raising the question of why in our studies CD14-ko
mice display enhanced insulin resistance. The amelioration of insulin resistance in TLR4-deficient mice has been interpreted to reflect a direct activation of TLR4 by saturated fatty acids present in the diet. However, the lack of direct interaction between the extracellular domain of TLR4 and saturated fatty acids suggests alternative possibilities 
. One such possibility may be that some high fat diets may enhance the levels or activity of natural TLR4 ligands such as LPS, for example through changes in gut permeability, microbiota composition, or metabolic endotoxemia 
. Thus, high-fat diets could produce insulin resistance through at least two concurrent mechanisms; first, by increasing LPS levels or sensitivity to LPS, and second, through direct impairment of insulin signaling pathways by excess free fatty acids 
, which may be enhanced by tonic TLR4 activity. Depending on the relative predominance of these mechanisms, ablation of TLR4 signaling would result in greater or lesser amelioration of HFD induced insulin resistance. This model could explain the results presented here where the absence of CD14 in extensively backcrossed C57Bl6 mice raised and fed in strict pathogen-free conditions does not ameliorate HFD-induced insulin resistance.
This model would also predict that the effects of fatty acids to induce inflammatory responsiveness in macrophages would be negligible compared to those induced by LPS. To test this directly, we compared the induction of IL6 protein secretion by primary macrophages in response to fatty acids and LPS. As expected, secretion of IL6 by peritoneal macrophages was potently stimulated by LPS (), and this effect was absent in macrophages obtained from CD14-ko
mice. However, significant IL6 secretion in response to palmitic acid, oleic acid or both could not be detected in this assay, indicating that the magnitude of the effect is very small compared to that of LPS. These fatty acids were functional, as they induced a potent increase in oxygen consumption when added to cultured myocytes, stemming from their entry into the beta-oxidation pathway (). Also, at high concentrations, palmitic acid had toxic effects on macrophages, reflected by a gradual decline in ATP levels over time (). These results confirm that LPS activation of TLR4 is much more potent that its possible activation by fatty acids, consistent to that seen by others 
. Under our experimental conditions the impairment of glucose homeostasis due to lack of CD14/TLR4 signaling overrides the small beneficial effects that might ensue from the mitigation of non-LPS induced pro-inflammatory signaling.
Effect of free fatty acids on peritoneal macrophages.
In human populations, high TLR expression is associated with greater inflammatory responses 
. Conversely, TLR4 polymorphisms (Asp299Gly) that confer decreased responsiveness to LPS 
increase susceptibility to infection 
. These polymorphisms are strongly associated with a decreased susceptibility to atherosclerosis 
, consistent with the possibility that the decrease in chronic inflammation due to hyporesponsive alleles suppresses atherogenesis. Interestingly, these same polymorphisms are associated with increased insulin levels, decreased insulin sensitivity and family history of diabetes 
. In large cross sectional studies the association of these TLR4 alleles with decreased insulin sensitivity persisted when corrected for body fat 
. Whether these polymorphisms in humans are associated with altered adrenal tone is not known, but they are consistent with our findings of impaired glucose tolerance in animals deficient in innate immune inflammatory signaling. Future studies designed to address the potential role of the LPS signaling pathway in modulating adrenal tone in humans will be required to establish the significance of the results presented here to human metabolic disease.