In the present study, we observed that insulin resistance in healthy, non-obese, non-diabetic subjects was accompanied by an increased activation of the JNK pathway with a corresponding increase in IRS-1 serine phosphorylation in skeletal muscle tissue, the site responsible for the majority of insulin-mediated glucose disposal 
. These indices of stress kinase activation were accompanied by reduced post-receptor activation of the insulin signaling pathway at the level of IRS-1 and Akt. Consistent with a possible causal role of lipids in JNK activation, we observed increased levels of intramyocellular lipids, and higher total and abdominal adipose stores in the insulin resistant subjects, despite their normal body weight.
Our principal findings were activation of JNK and dramatic increase in phosphorylation of IRS-1 on SER312
in the insulin resistant non-obese subjects. JNK has previously been implicated only in obesity-mediated insulin resistance 
. Increased serine phosphorylation of IRS-1 has been demonstrated in insulin resistant, non-obese subjects previously, but only in a carefully selected population of first degree offspring of diabetic parents 
. In those studies, however, no specific serine kinase was implicated. While our data do not allow us to rule out the contribution of other cellular serine kinases on the altered phosphorylation state of muscle IRS-1, it is highly plausible that JNK can serve as a causal mechanism for insulin resistance in this population.
We observed that the intracellular triglyceride content of soleus and tibialis anterior muscles were twice as high in the insulin resistant group in our study, indicating the potential for lipotoxic effects in muscle cells of these subjects. When this condition is modeled in cultured hepatocytes by incubation with saturated fatty acids, a direct activation of the JNK pathway and serine phosphorylation of IRS-1 results 
. JNK activation in this model is directly attributable, at least in part, to oxidative stress resulting from increased lipid oxidation 
. Thus, a large component of the toxic impact of cellular lipids may occur as a result of the burden of high rates of lipid oxidation on mitochondria, due to the increased generation of reactive oxygen species 
, which, in addition to activating JNK, may also impact insulin signaling via the p38MAPK and NF-κB pathways 
. We could not, however, find any evidence for p38MAPK or NF-κB activation in the present study. The influence of oxidative stress on insulin signaling is likely complex, however, and is not yet fully understood. It is known that angiotensin II can both contribute to oxidative stress-induced insulin resistance 
, and directly result in serine phosphorylation of IRS-1 
, but the specific kinases involved and the role of this hormone on human models of insulin resistance have yet to be determined. Beyond oxidative stress, high rates of lipid oxidation can also lead to incomplete lipid oxidation, with a resultant accumulation of β-oxidation intermediates 
. These β-oxidation intermediates, such as acyl-CoAs or acyl-carnitines, have been linked to insulin resistance 
, although their potential impact on insulin signaling is unclear.
It was not possible to determine the amount of lipid oxidation occurring in muscle, or to assess markers for oxidative or ER stress in the present study, although assessing these parameters along with additional components of JNK signaling will be important avenues for subsequent studies. Therefore, we can not conclude whether serine kinase activation is associated with the process of metabolizing excess lipids, or as a result of a direct effect of intracellular lipids intermediates on stress kinase pathways. Protein kinase C (PKC) enzymes, for example, can be directly activated by lipid intermediates such as long chain fatty acyl-CoAs, diacylglycerols, and ceramides 
are associated with insulin resistance in humans 
. While the MRS technique we employed quantifies the relatively benign triglyceride component of the intracellular lipid pool, IMCL measures have been used as surrogate indices of these intermediates. Due to constraints in sample size, we were unable to determine the amount of distinct lipid intermediate species or to assess activity of PKC enzymes. Further work will be needed to explore the potential role for PKC activation in this population.
It can not be ruled out that JNK activation in this population also results from the effects of increased lipid stores in adipose depots throughout the body. JNK can be activated via receptors for adipose-derived cytokines such as TNF-α and IL-1 
, and cytokines such as these have been implicated in translating the negative effects of expanded adipose stores into impaired insulin signaling in muscle. Insulin resistant subjects in the present study had significantly increased total fat stores and greater fat volume in the abdominal region. Still, the total amount of adipose tissue in these subjects is significantly less than that observed in obese subjects, making it unclear whether a negative impact of adiposity on muscle insulin action can be observed at this range of adiposity. There are little data to suggest that potential adipokine mediators of muscle insulin resistance are increased in non-obese subjects. In the present study we found some evidence that, even in this absence of obesity, increased fat stores have the potential to produce a detrimental effect on muscle insulin action.
Circulating levels of TNF-α are increased in obese states. Although the difference in serum TNF-α between the RES and SEN groups did not reach statistical significance, serum TNF-α was correlated both with total fat and visceral fat across subjects. While TNF-α can produce activation of the JNK pathway, it has been demonstrated that disruption of insulin signaling in cultured muscle cells by TNF-α is mediated by p38 MAPK 
. Neither activation of p38 MAPK, nor the TNF-α sensitive IKKβ were observed in our resistant subjects when compared to the sensitive subjects. Thus, circulating TNF-α is not likely to constitute a primary determinant of JNK activity and insulin signaling in this non-obese population. The correlations between TNF-α and adiposity, as well as muscle IKKβ, suggest that even in non-obese subjects, small increases in total and visceral fat mass, although relatively small compared to values for obese subjects, are still sufficient to produce increased circulating levels of TNF-α. The lack of an association between these variables and insulin action suggest that a threshold may be required for TNF-α to activate the NF-κB and other pathways sufficiently to negatively influence insulin signaling, and that this threshold is not readily met in non-obese subjects.
The roles of other potential adipose-derived cytokines in this population remain to be determined, and can not be excluded as contributors to JNK activation or insulin resistance in this population. Likewise, non-adipose derived inflammatory factors could contribute to muscle stress kinase activation. Homocysteine has been identified as an activator of JNK 
. However, homocysteine levels were not different between these groups, and have been shown to not associate with insulin resistance in this non-obese population 
While the causes of JNK activation and impaired insulin signaling in this population remain uncertain, it is clear that the insulin resistant subjects in our study represent a less dramatic version of an obese phenotype with an overabundance of adipose and muscle lipid storage, and elevated muscle JNK activation. An obese phenotype in non-obese subjects with metabolic disturbances has been described previously 
. These data, as with our data on serum lipids and cardiovascular risk in this same population 
, suggests that the threshold at which lipid deposition may impact metabolic health may be relatively low, and well within the range of what would be considered healthy weights. Our finding of increased IMCL levels in the insulin resistant subjects is also is in agreement with other reports demonstrating in association of increased IMCL levels with insulin resistance in non-obese populations 
, suggesting that this local accumulation of lipids in muscle is an important component of the development of insulin resistance in the absence of overall obesity.
It has been suggested, based on studies of first degree offspring of diabetic patients, that IMCL accumulation occurs in lean subjects as a result of impaired mitochondrial content and capacity 
. In contrast to studies in that unique population 
, we found no evidence for a reduced mitochondrial content as assessed by SDH, suggesting a substantial difference between these populations. While SDH content is a very crude estimate of mitochondrial content, the variables employed in the present study to measure total, abdominal and visceral fat, as well as IMCL values, were all correlated across subjects, as has previously been observed for subjects with lower total fat stores 
. Our data is therefore suggestive of an accumulation of fat across multiple depots and tissue beds driving a phenotype similar to obesity, rather than a mechanism promoting tissue-specific lipid deposition, that results in insulin resistance. Whether the accumulation of lipids that does occur in muscle provides a disproportionate negative impact on insulin signaling in cells of that tissue, as compared to other fat stores, remains to be determined. Further studies are required to determine whether reduced mitochondrial content is indeed a significant contributor to muscle lipid accumulation and insulin resistance, and whether the first degree offspring of diabetic parents present with a unique phenotype across the spectrum of insulin resistance. The present study included a subgroup of subjects with a parental history of diabetes. While the possibility exists that this introduced heterogeneity into the determinants of insulin resistance, our study was designed to assess mechanisms of insulin resistance in the general population, which includes subjects both with and without the potential for genetic influences on insulin resistance.
Although some studies of non-obese subjects have employed a BMI of 25 as the cutoff point for normal weight, our results were not influenced by the inclusion of subjects with BMI values up to 27. When only the data from those individuals with BMI values ≤25 were analyzed, the results were identical; insulin resistance was associated with increased JNK activation reduced insulin signaling, and elevated intramyocellular lipids, abdominal and visceral fat. Likewise, there was no difference in the trends of the data when men and women were analyzed separately, except for the trend for insulin resistant men to have higher amounts of visceral fat than insulin sensitive men, a trend not observed for women. However, the sample size involved precludes drawing any conclusions from these data, and further studies would be required to assess any gender differences in visceral fat accumulation, or the potential for visceral fat depots to influence insulin action in this population.
The group analysis aspect of the study design facilitated the identification of characteristics that differentiated insulin resistant from insulin sensitive subjects in an otherwise healthy population. Pre-screening for insulin-mediated glucose disposal with subsequent enrollment of only the most insulin sensitive and resistant subjects allowed for in depth study of a sufficient number of subjects with substantial insulin resistance, considering that there is no accepted clinical threshold to identify insulin resistance. This design did not, however, allow us to run uni- and multivariate analyses in an attempt to identify independent determinants of insulin resistance of JNK activation. Further cross sectional studies are needed to segregate the various contributors to JNK activation and insulin resistance in this population.
In summary, we observed that insulin resistance in the non-obese population is associated with an activation of the JNK pathway with increased serine phosphorylation of IRS-1. Implicated in this disruption of cellular insulin action is the accumulation of lipids within skeletal muscle, and the greater degree of overall adiposity that was observed in the insulin resistant subjects. Additional work is required to study additional stress kinase pathways in greater depth, and to assess the extent that these factors individually contribute to insulin resistance across the non-obese population.