Introduction

Insulin resistance is pathogenic for type 2 diabetes and cardiovascular disease [1]. Unraveling the molecular mechanisms underlying this syndrome is urgently needed.

Ectoenzyme nucleotide pyrophosphate phosphodiesterase 1 (ENPP1) has been proposed as a pathogenic factor for insulin resistance [2]. ENPP1 affects insulin signaling by binding to IR α-subunit and inhibiting receptor β-subunit autophosphorylation [3, 4]. ENPP1 expression is increased in tissues of insulin-resistant individuals [5, 6]. In addition, ENPP1 over-expression causes insulin resistance in rodents. [7, 8]. Finally, several data indicate that gain of function of ENPP1 (as induced by the missense K121Q polymorphism) contribute to insulin resistance and type 2 diabetes [7–9]. Thus, determining the mechanisms whereby ENPP1 is over-expressed may help develop strategies to counteract and possible reverse some forms of insulin resistance. Recent data point to a pivotal role of the ENPP1-3’UTR [10, 11], suggesting the existence of trans-acting proteins that affect ENPP1 mRNA stability. Our aim was to identify these proteins and investigate their role in the modulation of ENPP1 expression and insulin signaling.

Material and methods

Preparation of RNA probes as well as cell culture and solubilization are described in the online appendix methods

RNA electrophoresis mobility shift analysis (REMSA)

Fifty microgram of HEK293 lysates were incubated with the 32P-labeled-RNA probe for 20 min at room temperature (RT) [12]. Following incubation with heparin (5 mg/ml), gel electrophoresis was carried out at RT on 5% non-denaturing PAGE and visualized by autoradiography on Typhoon 8600 (Amersham). For supershift analysis, HEK293 lysates were incubated with 32P-labeled-RNA before adding either heat shock protein 70 (HSP70) specific antibody (SPA-812, Stressgen) or total IgG (Santa Cruz Biotechnology) for 30 min at RT.

Isolation of ENPP1-RNA binding protein

REMSA was carried as described above. Four gels of 15 lanes each were loaded. High molecular weight complexes, located by a 1-h exposure to X-ray film at −80°C, were excised and eluted from the gel [12]. Proteins were pooled, concentrated by acetone precipitation and resolved on 10% SDS-PAGE. The only band present in the gel with an apparent molecular weight of 70 kD was excised, and washed twice with 50% HPLC-grade acetonitrile before subsequent analysis. After proteolytic digestion, peptide composition analysis was performed at the Harvard Microchemistry Facility, Harvard University, (Cambridge, MA USA; http://www.mcb.harvard.edu/microchem/) by microcapillary reverse-phase HPLC nano-electrospray tandem mass spectrometry (μLC/MS/MS) on a Finnigan LCQ DECA XP Plus quadrupole ion trap mass spectrometer.

RNA extraction, cDNA synthesis, and gene expression analysis

Total RNA was isolated from cells using RNAEasy Quick kit (Qiagen). cDNA was generated by reverse transcription with M-MLV Reverse Transcriptase (Promega) and used as template in the subsequent analyses. Gene Expression Assay on Demand Kit Reagents (Applera) were used to quantify relative gene expression levels of ENPP1 and GAPDH on ABI-PRISM 7500 (Applera). Expression levels of ENPP1 were normalized against GAPDH using the comparative C t method, and expressed as percentage of control.

siRNA, cell transfections

Cells were seeded in six-well plates and grown in DMEM/F12 complete medium for 48 h. To down-regulate HSP70 expression, 150 nmol/L of siRNA targeted against HSP70 mRNA (Ambion ID number: 202680 was the only oligonucleotide used in our experiments) were either cotransfected or not with ENPP1 cDNAs by using TransMessanger Transfection Reagent (Qiagen) according to the manufacturer’s instructions.

Cell lines stably over-expressing IR cDNA (HEK293-IR) were generated by co-transfection of the prk5-IR (provided by Dr. Axel Ulrich, Martinsried Germany) and prk-5neo followed by geneticin selection IR expression was evaluated by western blot (WB) as described below. HEK293-IR were transiently transfected with prk7-ENPP1 and/or with pCMVSport6-HSP70 plasmid (ATCC) by using FuGENE6 (Roche).

Western blot analysis

Cell lysates were separated by SDS-PAGE and transferred to nitrocellulose membrane (Amersham Pharmacia Biotech). Blots were probed with following antibodies: anti-HSP70 (SPA-812, Stressgen), anti-IRβ-subunit (C19, Santa Cruz Biotechnology), anti-PY (PY99 HRP, Santa Cruz Biotechnology), anti-ENPP1 (N-20, Santa Cruz Biotechnology) and anti-IRS-1 (A-19, Santa Cruz Biotechnology). Alternatively, cell lysates were immunoprecipitated with anti-PY antibody (4G10, Millipore) and analyzed by WB using IRβ-subunit or IRS-1 antibodies. Immunocomplexes were detected with the ECL Western Blotting System (Amersham Pharmacia Biotech).

ENPP1 mRNA stability

ENPP1 mRNA stability was evaluated by adding Actinomycin D (5 μg/ml) 60 h after HSP70 silencing. RNA extraction was performed at different times as described above, and ENPP1 expressions determined as described.

Insulin stimulation

Insulin (10 nmol/L for 5’ at 37°C) was added to cells and total cell lysates were either immunoprecipitated or not before SDS-PAGE. IRβ-subunit and IRS-1 phosphorylation were evaluated by WB as described.

Results

Identification and characterization of an ENPP1-3’UTR protein complex

REMSA was performed by incubating HEK293 cell extracts with a 395-bp probe corresponding to the ENPP1-3’UTR showing a high degree of conservation between human and mouse genomes (Supplementary Figure 1). RNA–protein complex was indicated by band-shift observed in the presence of the probe (Supplementary Figure 2, lanes 1–2). SDS-PAGE of the eluted shifted-band revealed a single protein in the RNA-protein complex. This was identified by tandem mass spectrometry as the 70 kDa HSP70 encoded by the HSPA1B isoform (Supplementary Figure 3). Addition of increasing amount of HSP70 antibody to the RNA-cell extract mixture induced a gradual loss of the high molecular complex (Supplementary Figure 2, lanes 3-10), confirming the specific interaction between HSP70 and ENPP1-3’UTR.

Effect of HSP70 on ENPP1 expression

We evaluated the effect of HSP70 down-regulation on ENPP1 mRNA stability and expression by transfecting HEK293 cells with HSP70-siRNA. Exposure to the siRNA decreased HSP70 by approximately 80% (Fig. 1a). After transcription inhibition by Actinomycin D, a progressive reduction in ENPP1 mRNA levels was observed (Fig. 1b); the reduction, however, was significantly greater in HSP70 down-regulated than control cells (p < 0.01; Fig. 1b). We also evaluated the effect of HSP70-siRNA on steady state levels of ENPP1 mRNA and protein in cells transfected with ENPP1 cDNA. Both mRNA (Fig. 1c) and protein (Fig. 1d, lanes 1–2) contents were lower following HSP70 down-regulation. This was not observed when ENPP1 cDNA lacked the 3’UTR (Fig. 1d, lanes 3–4). We then evaluated the effect of HSP70 over-expression (i.e. by HSP70 cDNA transfection) on ENPP1 protein content in cells co-transfected with ENPP1 cDNA. ENPP1 protein levels were approximately 30% higher in co-transfected cells as compared to HEK293 transfected with ENPP1 cDNA alone (Fig. 2), thus confirming that HSP70 levels affects ENPP1 expression.

Fig. 1
figure 1

HSP70 down-regulation and ENPP1 expression. a HSP70 siRNA. HEK293 cells were either transfected with HSP70 siRNA (lane 1) or with scrambled siRNA (lane 2). HSP70 expression was evaluated by western blot by using HSP70 SPA-812 specific antibody. b ENPP1 mRNA stability. ENPP1 mRNA content was determined by real time PCR (after transcription inhibition by 5 μg/ml Actinomycin-D) in HEK293 cells either treated (square) or not (circle) with HSP70 siRNA. A significantly greater reduction of mRNA content over time (p = 0.04, by two-way ANOVA) was observed in HSP70 down-regulated cells as compared to control cells. Expression levels of ENPP1 were normalized against GAPDH using the comparative C t method, expressed as percentage of untreated cells at time 0 (Endogenous levels of ENPP1 expression were detectable with Ct values being 26.4 ± 1.3, as compared to those of GAPDH being 15.7 ± 1.9). Data are mean ± SD (n = 4 experiments in each group). c ENPP1 levels in transfected HEK293 cells. ENPP1 mRNA content was determined by real-time PCR in HEK293 cells transfected with ENPP1 cDNA and treated (black bar) or not (white bar) with HSP70 siRNA. The grey bar represents ENPP1 levels in untransfected cells. HSP70 down-regulation caused 80% significant reduction (p = 0.016 by paired sample t-test) of ENPP1 transcript. Expression levels of ENPP1 were normalized against GAPDH using the comparative C t method, expressed as percentage of untreated cells (white bar). Data are means ± SD (n = 3 experiments in each group). d ENPP1 protein content. HEK293 cells transfected with either ENPP1 cDNA or with ENPP1 cDNA lacking 3’UTR (ENPP1 Δ,3’UTR lanes 3–4) were treated (+) or not (−) with HSP70 siRNA. Total cell lysates were then used to measure HSP70, (by using HSP70 SPA-812 specific antibody) and ENPP1 (by using ENPP1 N-20 specific antibody) protein content. HSP70 down-regulation had no effect on ENPP1 protein content when ENPP1 Δ3’UTR cDNA was used

Full size image
Fig. 2
figure 2

HSP70 over-expression and ENPP1 protein content. HEK293 cells were either transfected with ENPP1 cDNA (lane 2) or with both ENPP1 and HSP70 cDNAs (lane 3). Total cell lysates were then used to measure HSP70 (bottom panel) and ENPP1 (upper panel) protein content, by western blot analysis. HSP70 up-regulation (bottom panel, lane 3) increased by approximately 30% ENPP1 protein content (lane 3 vs. lane 2). A representative experiment is shown

Full size image

Effect of HSP70 down-regulation on insulin signaling

When HSP70 expression was down-regulated in HEK293-IR cells, insulin-stimulated IR autophosphorylation was 2.5- to threefold higher than in control cells (Fig. 3a: lanes 2 vs. lane 1; Fig. 3b: second bar vs. first bar). In contrast, in cells transfected with ENPP1 cDNA, insulin-induced IR autophosphorylation was markedly inhibited (Fig. 3a, lanes 3 vs. lane 1; Fig. 3b: third bar vs. first bar). This inhibition was partially abolished by HSP70 down-regulation (Fig. 3a lane 4 vs. lane 3; Fig. 3b: fourth bar vs. third bar), a rescue which was paralleled by a reduction of ENPP1 over-expression (Fig. 3a lanes 3–4). Neither the partial rescue of IR autophosphorylation (Fig. 3a lanes 5–6 and Fig. 3b sixth bar vs. fifth bar) nor the reduced ENPP1 expression (Fig. 3a lanes 5–6) was observed when cells were transfected with ENPP1 cDNA lacking the 3’UTR. In human liver HepG2 cells, HSP70 down-regulation increased insulin-induced IR and IRS-1 phosphorylation (Fig. 3c lane 3 vs. lane 2). As compared to control cells, insulin-induced IRβ-subunit autophosphorylation and IRS-1 phosphorylation were higher in cells treated with HSP70 siRNA (153% and 274% increase, respectively, mean of two different experiments).

Fig. 3
figure 3

HSP70 and insulin signaling. a HEK293-IR cells were either transfected (lanes 3–6) or not (lanes 1–2) with either ENPP1 cDNA (lanes 3–4) or with ENPP1 cDNA lacking the 3’UTR (Δ3’UTR, lanes 5–6) and then treated (lanes 2, 4 and 6) or not (lanes 1, 3 and 5) with HSP70 siRNA. Sixty hrs after silencing, cells were stimulated with 10 nmol/L insulin for 5 min at 37°C and IRβ-subunit autophosphorylation evaluated. IRβ-subunit autophosphorylation was assessed with anti-PY antibody. Blot was then stripped and reprobed with anti IRβ subunit. HSP70 protein content was evaluated with specific SPA-812 antibody. IR and HSP70 bands are from the same blot, but were cropped because of their different motilities. A representative experiment is shown. (b) IRβ-subunit autophosphorylation in each condition was calculated as percentage of that of control untransfected cells (first bar, from the left). IRβ-subunit autophosphorylation was almost 3 times higher in cells treated with HSP70 siRNA (second bar) and 70% lower in cells transfected with ENPP1 cDNA (third bar). As compared to that of cells transfected with ENPP1 cDNA (third bar), IRβ-subunit autophosphorylation was doubled in those co-transfected, with HSP70 siRNA (fourth bar), so that an almost complete rescue of IRβ-subunit autophosphorylation was observed. This rescue was not observed when HSP70 expression was down-regulated in cells transfected with ENPP1 cDNA Δ 3’UTR (lanes 5–6). Data shown are mean ± SEM of 3 experiments. * p < 0.04 vs. control cells. (c) HepG2 cells were either treated (lane 3 or not lane 2) with HSP70 siRNA. Sixty hrs after silencing, cells were stimulated with 10 nmol/L insulin for10 min at 37°C and cell lysates were immunoprecipitated with anti-PY antibody. Blots were probed with both IRS-1 and IRβ-subunit antibodies, respectively. IRβ-subunit and IRS-1 bands are from the same blot, but were cropped because of their different motilities. A representative experiment is shown

Full size image

Discussion

The two major messages of our study are as follows 1) by binding ENPP1-3’UTR, HSP70 stabilizes ENPP1 mRNA and eventually increases ENPP1 transcript and protein levels; 2) through the modulation of ENPP1 expression, HSP70 affects insulin-induced IR and IRS–1 phosphorylation, thus becoming a new potential modulator of insulin resistance. Similar data were observed in two different cell lines, thus suggesting this phenomenon occurs in several insulin target tissues. To the best of our knowledge, this is the first report elucidating a molecular mechanism which modulates ENPP1 expression and, consequently, affects IR signaling.

Most of the molecular mechanisms of insulin resistance so far described are located at the downstream ‘post-receptor level’. However, recent functional, metabolic [5, 13, 14] and genetic studies [9–11, 15–23] have suggested that an additional mechanism of insulin resistance resides at the ‘receptor level’ and is mediated by ENPP1[2]. In this context, we propose HSP70 as a new potential contributor to insulin resistance through up-regulation of ENPP1 expression and subsequent inhibition of IR-autophosphorylation. Of note, HSP70 is known to bind and modulate other 3’UTRs of yeast and human mRNAs [24, 25].Very recently an opposite role of HSP70 on insulin signaling has been reported in transgenic animal models [26]. The proposed mechanism is mediated by HSP70-inhibition of JNK activation, a negative modulator of IRS-1 signaling. These discrepancies are difficult to reconcile and might be due to relevant differences intrinsic to the study models utilized (i.e. in vivo obese rodents vs. in vitro human cells) or variable tissue-specific expression. Further studies are clearly needed to acquire deeper insights on the potential role of HSP70 in both cellular and human insulin resistance.