Obesity depresses the anti-inflammatory HSP70 pathway, contributing to NAFLD progression

Introduction

The increasing worldwide prevalence of obesity and high fasting glucose concentrations, which are interdependent risk factors for nonalcoholic fatty liver disease (NAFLD), have become a major healthcare problem (1). Additionally, NAFLD encompasses a broad spectrum of successive and progressive pathological conditions that circumscribe hepatic steatosis, its inflammatory variant, nonalcoholic steatohepatitis (NASH), progressive fibrosis, cirrhosis and, eventually, hepatocellular carcinoma (2).

Because obesity is accompanied by an inflammatory status in metabolic tissues (3) and because adipose tissue-derived hormones influence hepatic function, as observed in mouse models of NAFLD (4), the crosstalk between fat and the liver has been the subject of much attention (5, 6). Visceral adipose tissue is inherently pro-inflammatory, although inflammation has also been observed in stressed subcutaneous adipose tissue in a rat model of obesity (7). Important consequences include the release of macrophage cytokines, namely, macrophage chemotactic protein 1 (MCP-1) and tumor necrosis factor-α (TNFα) (8). Moreover, although the complete sequence of events linking obesity to inflammation and insulin resistance has not been fully elucidated, the mechanism by which excess lipids/fuels trigger low-grade inflammation in both adipocyte and liver, which contributes to impaired insulin responsiveness, involves the activation of c-Jun NH2-terminal kinases (JNKs), endoplasmic reticulum (ER) stress, the unfolded protein response (UPR), ceramide, and the low-grade inflammatory response in both humans and rodents (9). In fact, the impaired response of cells from diabetic patients to stress (e.g., high blood glucose and fatty acid levels) is associated with the activation JNK cascades, whereas TNFα-elicited JNK activation is a major constituent of free fatty acid-induced insulin resistance (10).

If obesity operates through JNK activation, which feeds forward to systemic inflammation, then stress (inflammation)-induced expression of the 70-kDa family of heat shock proteins (HSP70) works in the opposite direction. In addition to the now classical molecular chaperone action, the most remarkable intracellular effect of HSP70 is its anti-inflammatory action under stressful situations. Accordingly, HSP70 blocks nuclear factor κB (NF-κB) activation at different levels, for example, by impeding inhibitor of κB (IκB) phosphorylation [see, for instance, Ref. (11)] and by directly binding to IκB kinase gamma (IKKγ) (12). These effects lead to the inhibition of downstream inflammatory signals. These observations are corroborated by the finding that HSP70 assembles with the liver NF-κB/IκB complex in the cytosol, thus hindering further transcription of NF-κB-dependent genes. This situation is the case for TNFα and inducible nitric oxide synthase (NOS2) genes that increase severe inflammation in rats (12). Additionally, stress-induced HSP70 inhibits JNK-dependent signal transduction, leading to programmed cell death in humans (13). The marked anti-inflammatory action of intracellular HSP70 also explains why cyclopentenone prostaglandins (cp-PGs), which are physiological inducers of HSP70 expression, are powerful anti-inflammatory autacoids (14, 15). Nonetheless, one of the most striking observations of obesity states is decreased HSP70 expression in both the skeletal muscle (16) and adipose tissue of obese patients (17), which is consistently associated with insulin resistance. Thus, enhanced HSP70 expression has been convincingly demonstrated to protect against obesity-induced insulin resistance in both humans and animal obesity models, whereas pharmacological (e.g., the hydroxylamine derivative BGP-15, which is now under clinical trial) and physiological (hyperthermic, hot tub) treatments are being tested as possible therapeutic approaches for humans with type 2 diabetes (18).

The interconnection between obesity, inflammation and HSP70 pathways encompasses a much more complex network that operates at the gene regulatory level. The promoter region of the TNFα gene contains a heat shock transcription factor-1 (HSF1) binding site that represses TNFα transcription; thus, the loss of this repressor results in sustained TNFα expression (19). Hence, HSF1 knockout is associated with a chronic increase in TNFα levels and with increased susceptibility to endotoxin challenge, at least in murine models. The opposite regulation of this network has also been reported; TNFα may transiently repress HSF1 activation (20). Furthermore, JNK1 was irrefragably demonstrated to phosphorylate HSF1 in its regulatory domain, causing the suppression of HSF1 transcribing activity in human cells (21), whereas HSP70 prevents apoptosis by inhibiting JNK-dependent pathways.

Taken together, the above observations led us to investigate whether HSP70 pathways could be impaired in the liver and adipose tissue of NAFLD patients, thus allowing JNK-triggered inflammatory signals to reach metabolic tissues. Moreover, systemic oxidative stress markers were analyzed to assess the relation between HSP70-elicited anti-inflammation and NAFLD progression.

Methods

Ninety-five adult patients (75 women/20 men) with a diagnosis of severe (type 3) obesity who underwent bariatric surgery at the Center for Obesity and Metabolic Syndrome of São Lucas Hospital (COM—PUC/RS, Porto Alegre, RS, Brazil) were included in the study. All 95 patients were submitted to intraoperative liver biopsy to confirm the clinical diagnosis of NAFLD (please see below). According to histological criteria, volunteers were divided into the following groups: steatosis (ST), steatohepatitis (SH) and steatohepatitis with fibrosis (SH+F). Patients with evidence of excessive alcohol use (≥10 g day⁻¹) or with other causes of liver disease (e.g., hepatitis B, hepatitis C, or autoimmune liver disease), as well as those patients receiving treatment with PPAR-γ agonists, were excluded. Clinical, plasma and NAFLD scoring studies were performed in all patients (N = 95), whereas 22 of 95 patient samples were randomly assigned among NAFLD groups (at least N = 7 per group) for molecular and immunofluorescence analyses. Ethical questions and patient characteristics, as well as manipulation, histopathology, and sample analyses, including immunoblotting, immunohistochemistry, biochemical evaluations, and statistics, are fully described in the online Supporting Information.

Results

HSP70 vs. JNK status in the adipose tissue of NAFLD patients

Consistent with the proposal that a progressively higher degree of liver inflammation among NAFLD patients could be related to a parallel rise in the inflammatory status in adipose tissue due to a loss of function of HSP70 anti-inflammatory and cytoprotective activity, we investigated HSP70 protein contents in biopsies of visceral adipose tissue from these patients (Figure 1). Notably, total HSP70 (HSP72+HSP73) protein contents in adipose tissue of NAFLD patients inversely correlated with disease progression (ρ = −0.934, P < 0.05) and markedly decreased in the NASH patients compared with the ST groups (P < 0.05), which correlated (ρ = −0.978) with the conspicuous suppression of HSF1 immunocontents in adipose tissue from the same patients, as depicted in Table 2. Furthermore, when the anti-HSP72 antibody clone C92F3A-5 (which specifically recognizes the 72 kDa form of HSP70) was used, we were not able to detect the presence of HSP70 in adipose tissue (data not shown). Paralleling these observations, JNK1 and JNK2 (both total and Thr183/Tyr185-diphosphorylated/activated forms, p-JNKs) were enhanced in adipose tissue from the same patients (P < 0.05). Table 2 summarizes all the correlations between the groups.

Table 2. Correlations between the HSP70, HSF1, JNK1, and p-JNK1 immunocontents in tissue biopsies and either the NAFLD progression or NAFLD fibrosis score

Spearman's rank correlation		p-JNK1	JNK1	HSP70	HSF1§	Score	Progression
(A) Adipose tissue
p-JNK1	Spearman's ρ Coefficient	1.000	0.983**	−0.995	−0.952**	0.713*	0.893**
	σ (2-tailed)	-	0.0014	0.0014	0.0013	0.0011	0.0012
	N	22	22	22	22	22	22
JNK1	Spearman's ρ Coefficient	0.983**	1.000	−0.997*	−0.992**	0.828	0.960**
	σ (2-tailed)	0.0014	-	0.0014	0.0013	0.0011	0.0013
	N	22	22	22	22	22	22
HSP70	Spearman's ρ Coefficient	−0.995	−0.997*	1.000	0.978**	−0.781	−0.934*
	σ (2-tailed)	0.0014	0.0014	-	0.0013	0.0011	0.0013
	N	22	22	22	22	22	22
HSF1	Spearman's ρ Coefficient	−0.952**	−0.992**	0.978 **	1.000	−0.894**	−0.988**
	σ (2-tailed)	0.0013	0.0014	0.0013	-	0.0012	0.0014
	N	22	22	22	22	22	22
Score	Spearman's ρ Coefficient	0.713*	0.828	−0.781	−0.894**	1.000	0.953**
	σ (2-tailed)	0.0010	0.0011	0.0011	0.0012	-	0.0013
	N	22	22	22	22	22	22
Progression	Spearman's ρ Coefficient	0.893**	0.960**	−0.934*	−0.988**	0.957**	1.000
	σ (2-tailed)	0.0011	0.0014	0.0013	0.0014	0.0013	-
	N	22	22	22	22	22	22
(B) Liver
p-JNK1	Spearman's ρ Coefficient	1.000	0.885**	−0.354	−0.968**	0.447*	0.647**
	σ (2-tailed)	-	0.000	0.106	0.001	0.037	0.001
	N	22	22	22	22	22	22
JNK1	Spearman's ρ Coefficient	0.885**	1.000	−0.437*	−0.985*	0.403	0.583**
	σ (2-tailed)	0.000	-	0.042	0.001	0.063	0.004
	N	22	22	22	22	22	22
HSP70	Spearman's ρ Coefficient	−0.354	−0.437*	1.000	0.975**	−0.313	−0.456*
	σ (2-tailed)	0.106	0.042	-	0.001	0.156	0.033
	N	22	22	22	22	22	22
HSF1a	Spearman's ρ Coefficient	−0.968 **	−0.985	0.975**	1.000	−0.343	−0.956 **
	σ (2-tailed)	0.001	0.001	0.001	-	0.843	0.001
	N	22	22	22	22	22	22
Score	Spearman's ρ Coefficient	0.447*	0.403	−0.313	−0.343	1.000	0.795**
	σ (2-tailed)	0.037	0.063	0.156	0.843	-	0.000
	N	22	22	22	22	22	22
Progression	Spearman's ρ Coefficient	0.647**	0.583**	−0.456*	−0.956 **	0.795**	1.000
	σ (2-tailed)	0.001	0.004	0.033	0.001	0.000	-
	N	22	22	22	22	22	22

• Correlations between HSP70, HSF1, JNK1 and p-JNK1 immunocontents in patient biopsies and either NAFLD progression (ST = 1; SH = 2; SH+F = 4) or NAFLD fibrosis severity score were approached. The samples were obtained from 22 of 95 patients enrolled in the study. NAFLD and fibrosis scores were as described in the legend of Table 1. Spearman's rank correlation (followed by Spearman's ρ coefficient) was calculated (± σ). Bi-tailed Student's t test was used to estimate the significance of the calculated coefficients. For the comparisons, *P < 0.05 and **P < 0.01. For details and references, please see the online Supporting Information. A, adipose tissue immunoblots; B, liver immunoblots. §HSF1 contents in liver samples were evaluated only by immunofluorescence.

HSP70 vs. JNK status in the hepatic tissue of NAFLD patients

Liver biopsies from NAFLD patients (N = 22) were also analyzed for HSP70 (HSP72 and HSP73 forms) and JNK (JNK1 and JNK2; both total and diphosphorylated activated forms) protein contents. Similar to that observed in adipose tissue from the same patients, total HSP70 (HSP72+HSP73) protein contents were found to be slightly but significantly decreased as NAFLD progresses (ρ = −0.456, P < 0.05). This finding is associated with a significant reduction in the levels of the inducible form of HSP70 (HSP72), whereas the cognate HSP70 form (HSP73) only slightly decreased (Figure 2). Once again, as observed in adipose tissue, the liver JNK1 and p-JNK1 contents were reciprocally enhanced in relation to HSP70 levels (please see Table 2 for the correlations), thus suggesting that both JNK1 expression and activation were augmented during NAFLD progression. JNK2 was not detected in the liver tissues of these patients.

We analyzed NAFLD liver biopsies for HSP70 and HSF1 contents using immunohistochemistry (IHC) and immunofluorescence (IF). As shown in Figure 3, a progressive decrease in HSP70 immunoreactivity was observed in the livers of the patients as NAFLD progresses from ST to SH and SH+F (Figure 3A-C); this decrease is similar in magnitude to the rate of decrease in the number of HSF1-positive cells (Figure 3D-F). This possible correlation was also tested, with ρ = −0.456 (P < 0.05) for HSP70 and with ρ = −0.956 (P < 0.01) for HSF1 (Table 2). The IF analysis also indicated the co-localization of HSP70 and HSF1 in the slices (Figure 3J-L), thus suggesting a causal effect of decreased HSF1 over HSP72 (the inducible form of HSP70) immunodetection because HSP73 expression is not influenced by HSF1 activation/expression (22). This observation reinforces immunoblot analyses performed with the same samples in which HSP72 was the HSP70 form most affected by NAFLD progression (Figure 2).

The IF analyses of liver biopsies from NAFLD patients (Figure 3) also revealed an inconsistency between the cellular nature of HSP70/HSF1-positive cells and hepatocytes (expected to respond to NAFLD by altering HSP70 expression). In all the examined slices, the co-localization of HSP70 and HSF1 immunostaining (Figure 3J-L) was not on the same cellular structures as were hepatocytes. In fact, hepatocytes were clearly in another histological plane (hepatocytes are the cells whose nuclei are blue-stained by the fluorophore DAPI (Figure 3G-I) and were found to be above the “orange cell” plane of Figure 3J-L). Because of the shape of the structures (small cells) co-localizing HSP70 and HSF1 immunodetection, we conjectured that the cells affected by the progressively decreasing effect of NAFLD over HSP70/HSF1 contents could be Kupffer cells. In fact, as shown in Figure 4, the NAFLD degree-dependent reduction of HSP70 in liver slices (Figure 4A-C) did co-localize with small CD14⁺ cells (Figure 4D-F and Figure 4M-O), which is similar to the decreasing effect observed on the “orange cells” of Figure 3J-L.

When fluorescence intensities measured in liver slices were normalized by the hepatocyte number in the same fields, a sharp decrease in HSP70+ cells that inversely correlated with the progress of hepatic disease was also observed. Additionally, decreased HSP70 contents corresponded with (ρ = −0.975, P < 0.01; Table 2) a similar reduction in HSF1 contents (Figure 5A). Interestingly, in terms of the number of hepatocytes, the ratio of HSP70- to HSF1-positive cells, was unaltered by NAFLD progression (Figure 5C). The same comparison was performed for the fluorescence intensity in HSP70+/CD14+ cells. As depicted in Figure 5B, HSP70+ cell-derived fluorescence significantly decreased with NAFLD progression from ST to SH. No difference in normalized HSP70 contents was observed between the SH and SH+F groups, thus indicating that HSP70 reduction is dependent on the progression of inflammation, regardless of the presence of fibrosis. It is of note, however, that the number of CD14+ cells (normalized to hepatocytes) remained constant, suggesting that the HSP70 reduction in CD14+ cells was not due to a reduced number of small CD14+ cells (Figure 5D).

Although hepatocytes also express CD14, the cells co-localizing HSP70 and CD14+ (Figures 3 and 4) are clearly not hepatocytes. This co-location in the ST patient biopsies is even clearer when a low-magnification image merged with phase contrast micrograph is analyzed, as in Supporting Information Figure S2. Therefore, we believe that NAFLD progression negatively influences HSP70 contents in Kupffer cells, although hepatocytes seem to continue to express some HSP70 (but in CD14-negative cells in our preparations). Table 3 summarizes all the above results.

Table 3. Summary of clinical, plasma, and liver results

	NAFLD progression (Matteoni's ranking)
Parameter investigated	ST (type 1)	SH (type 2)	SH+F (type 4)
Diabetes prevalence (%)	33.3%	74.4% [↑123% vs ST]*	79.3%(ns)*
HOMA-IR	4.69 ± 2.64	9.5 ± 7.85 [↑103% vs ST]*	8.49 ± 4.05(ns)*
HbA1c (%)	5.85±0.78	6.63±1.70 [↑13% vs ST]*	6.24 ± 0.76(ns)*
SAH prevalence (%)	7.4%	60.5% [↑716% vs ST]*	56.0%(ns)*
Total HSP70 levels in the liver (IB)	93.27 ± 11.5	42.75 ± 23.22 [↓54% vs ST]*	59.99 ± 21.68(ns)*
HSP72 levels in the liver (IB)	76.64 ± 34.83	10.00 ± 3.10 [↓87% vs ST]*	10.00 ± 3.87(ns)*
HSP70+ cells in the liver (IF)	49,070 ± 10,722	13,092 ± 1,133 [↓73% vs ST]*	2,440 ± 211†*
HSF1+ cells in the liver (IF)	30,386 ± 9,128	8,233 ± 712 [↓[↓73% vs ST]*	1,486 ± 129†*
CD14+ cells in the liver (IF)	10,883 ± 2,167	12,283 ± 1,842 [not altered vs ST]	8,112 ± 1,298(ns)
HSP70+/CD4+ cells in the liver (IF)	27,534 ± 5,502	13,274 ± 1,725 [↓52% vs ST]*	11,672 ± 2,101(ns)*
JNK1 levels in the liver (IB)	56.18 ± 2.29	77.96 ± 4.57 [↑39% vs ST]*	92.07 ± 10.29†*
Activated p-JNK1 in the liver (IB)	69.00 ± 6.86	85.23 ± 2.86 [↑24% vs ST]*	99.20 ± 1.14†*

• IB: immunoblot values are presented as the means ± s.d. of the data depicted in Figure 2. IF: immunofluorescence intensity normalized to the number of hepatocytes; the data are presented as the means ± s.d. from the data depicted in Figure 5. [↑] and [↓] represent increased and decreased values, respectively, (P < 0.05) with respect to the ST results. *P < 0.05 for the comparisons with the ST group and †P < 0.05 for the comparisons with the SH group by ANOVA. ^(ns) = not significantly different when the SH+F group is compared with the SH group.

Discussion

Although the body fat percentage and plasma lipids may have central roles in NAFLD pathogenesis (2), the present results did not provide a reliable marker for following NAFLD progression from ST to NASH (SH and SH+F). Accordingly, the present data demonstrated that NAFLD may be related to a state of cellular stress and inflammation, whether in adipose tissue and liver or somewhere else, that can be traced in the plasma as alterations in lipoperoxidation. In fact, TBARS measurements, which are a simple but good estimate of lipoperoxidation/malondialdehyde produced throughout the body (systemic oxidative stress), were elevated in NASH plasma samples compared with ST plasma samples, which correlated with the degree of insulin resistance (HOMA-IR) and co-morbidities, such as diabetes and arterial hypertension prevalence (Table 1). In contrast, the plasma levels of SOD, which is an important antioxidant produced in response to oxidative stress, did not correlate with NAFLD progression from its simple ST form toward the more inflammatory SH+F form. Curiously, plasma SOD levels positively correlate with the degree of severity of oxidative stress in metabolic syndrome, with clear augmentation in the livers of NAFLD patients (23).

However, the most striking observation from the present study was the first demonstration that the HSF1-HSP70 axis is progressively suppressed in the adipose tissue and livers of obese patients as NAFLD evolves from ST toward SH+F. Moreover, this suppression strongly correlated (Table 2) with the degree of enhancement of total JNK1 and JNK2 immunocontents in adipose tissue (Figure 1), followed by similar increases in the amounts of Thr183/Tyr185-diphosphorylated activated p-JNK1 and p-JNK2 in the same tissue (Figure 1).

A similar pattern of HSP70 reduction, paralleled by proportionate JNK1 activation, was observed in the livers of NAFLD patients. However, in contrast to adipose tissue, the impairment of the inducible form of HSP70 (HSP72) was much more evident in the liver than in adipose tissue. The presence of JNK2 was not verified in the liver (cf., Figures 1 and 2), consistent with the notion that JNK1, but not JNK2, promotes the development and aggravation of NAFLD, at least in mice (24).

Taken together, the above findings may suggest a causal effect of decreased HSF1 expression over HSP72 for NAFLD progression because HSP72 expression is entirely dependent on induction via the HSF1 pathway [in contrast to that observed in the constitutive HSP73 form, whose expression is HSF1-independent (22)]. Because of HSF1/HSP70 reduction, JNK-based pro-inflammatory pathways may be derepressed, thus allowing for enhanced oxidative stress and worsening of the disease, considering that HSF1 may suppress TNFα expression in the murine model (19). In contrast, TNF-α may impede HSF1 activation (20) and transcribing activity in human cells (21). Hence, any factor that disrupts HSF1 biochemical pathways easily triggers a vicious cycle of inflammation, oxidative stress, tissue damage and so on.

Oxidative stress associated with mitochondrial dysfunction and ER stress are triggering factors for the increased expression and activation of JNK and NF-κB, with the concomitant development of a more pro-inflammatory state in humans (6, 25). In this regard, we have recently shown that increased oxidative and nitrosative stress and reduced antioxidant enzyme activity in rodent model of diabetes are the leading causes of redox imbalance-elicited NF-κB activation, which evolves to animal liver tissue damage in (26) that perpetuates inflammation. In fact, the selective inactivation of both JNK1 and JNK2 in the adipose tissue of mice protects the animals against a diet-induced increase in adiposity and improves insulin sensitivity in both liver and skeletal muscle (27).

In mammalian tissues, the expression of HSP70s is constitutive (HSP73 = HSC70) and stress-inducible (HSP72). Although both HSP70 forms work as molecular chaperones with anti-inflammatory properties, situations that displace cells from homeostasis (e.g., heat shock, heavy metals, and inflammatory signals) quickly trigger HSF1-dependent transcription of much HSP72 to counteract these cellular insults. Therefore, a reduced capacity for stress-induced HSP72 expression is at the center of inflammation and insulin resistance in both the skeletal muscle (16) and adipose tissue of obese patients (17). In fact, HSP72 expression plays a key role in blocking inflammation and in preventing insulin resistance in the context of genetic obesity or high-fat feeding (18).

However, the above considerations raise a central question: what is the mechanism underlying decreased HSF1-mediated HSP70 expression in adipose tissue and liver? Obesity-induced cellular senescence of adipocytes may partly answer this question. We presented this testable hypothesis based on the following observations. A senescent-like state can evolve in fat cells from obese individuals as an adaptation to the metabolic overutilization of fat cells, which resembles cellular aging (28). Moreover, high-fat diet-induced obesity has been shown to induce vascular senescence in mice (29). Fibroblasts from adult segmental progerioid Werner syndrome, which undergo premature senescence, present a strong positive feedback system in which p38-NF-κB pathway overactivation blocks the expression of the RNA-binding factor HuR, which is a potentiating factor for the NAD⁺-dependent protein deacetylase sirtuin-1 (SIRT1), in human cells (30, 31). In turn, SIRT1 enhances HSF1 expression (31), whereas SIRT1 activation prolongs HSF1 binding to the heat shock promoter by maintaining HSF1 in a deacetylated, DNA-binding competent state (32). In contrast, heat shock itself increases the cellular NAD⁺/NADH ratio, leading to the recruitment of SIRT1 to the HSP70 promoter (33). Hence, cellular senescence might occlude HSF1 expression and transcribing activity, consequently repressing a protective anti-inflammatory heat shock response.

The recent findings that hepatocyte senescence predicts NAFLD progression in humans (34) and that hepatocyte senescence is a general marker of human liver cirrhosis (35), which is the next step of NAFLD progression, support the above observations.

Although one cannot discard the possibility that the reduced heat shock response in hepatocytes may account for NAFLD aggravation (36), our results demonstrated that a possible target of decreased HSP70 in the liver could be Kupffer cells (Figures 4 and 5), which may increase hepatic tissue inflammation. In fact, we demonstrated that the HSP70 contents significantly and progressively diminished in small CD14+ cells of NAFLD patients as the disease evolved from ST toward SH+F. Although not completely understood, a hepatic-free radical generation may exacerbate inflammation through Kupffer cell activation. Moreover, an amplified pro-inflammatory response of M1 Kupffer cells may account for the severity of NAFLD and for its progression to NASH, fibrosis, cirrhosis and, eventually, hepatocellular carcinoma (37). Additionally, a growing body of evidence suggests that Kupffer cells critically contribute to NAFLD progression, particularly because abundant or abnormal lipids may confound the recognition of “fatty hepatocytes as dangerous” promoting adverse interactions with these hepatic macrophages (36, 38). The senescence hypothesis is also supported by the observation that experimental Kupffer cell ablation produces a scenario that resembles senescence, with a reduction of the hepatic anti-inflammatory response and a predisposition to steatosis and insulin resistance (39).

To our knowledge, this study is the first to demonstrate a clearly negative correlation between the progression of NAFLD to fibrosis and the expression/activation of the HSF1/HSP72 biochemical pathway in both liver and adipose tissues of NAFLD patients. These alterations were associated with reduced antioxidant enzyme activities and with enhanced JNK1 and JNK2 expression in the adipose tissue and JNK1 expression in the liver. HSP70 suppression also positively correlated with poor maintenance of glucose homeostasis (Table 1), which is notable because weight gain and the baseline HOMA are independent predictors for the development of NAFLD (40).

Finally, these observations may have an important diagnostic value for NAFLD patients because the “HSP70 status” in the circulating blood of human patients is relatively easy to assess (please cf., Ref. (17) and (36)). Clinical and translational confirmatory procedures are currently under evaluation in our laboratories.

Clinical Perspectives