Liver ENPP1 protein increases with remission of type 2 diabetes after gastric bypass surgery

Clinical data on all patients was collected prior RYGB surgery and included: weight, height, body mass index, glycated haemoglobin (HbA1c), fasting plasma glucose, and fasting plasma insulin. Diagnosis of T2DM was established by either 1) prior documentation of diagnosis and/or receipt of treatment for T2DM or 2) diagnosis based of an oral glucose tolerance test (OGTT) routinely performed in all patients without a known history of diabetes (plasma glucose > 11.1 mmol/l at 2 hours). Impaired glucose tolerance was also diagnosed using the OGTT (plasma glucose > 7.8 and < 11.0 mmol/l at 2 hours). Individuals with T2DM were further classified as: previously unrecognized, diet controlled, needing oral hypoglycaemic drugs or insulin taking. For those individuals who had a second liver biopsy, weight, body mass index, HbA1c, fasting plasma glucose and fasting plasma insulin were measured within 2 months of the second surgery. Remission of T2DM was defined according to recommendations of Buse et al. (HbA1c% <6.5%, fasting plasma glucose was <5.6 mmol/l and no continuing treatment one year after surgery) [24].

All patients undertook a 12 hour fast prior to blood collection for biochemical tests. Fasting was either self-administered or administered during the hospital stay. Clinical biochemistry testing was conducted by Aotea Pathology (Wellington, New Zealand) while testing for insulin was undertaken at Canterbury Health Laboratories (Christchurch, New Zealand).

Homeostasis model assessment (HOMA) was used to estimate insulin resistance [25]. HOMA-IR less than 2.5 has been shown to be representative of a metabolically normal, insulin sensitive population [26].

Total RNA was extracted using TRIzol® (Life Technologies) as per the manufacturer’s instructions. Briefly, 2-3 mg of frozen liver tissue was pulverized under liquid nitrogen and 800 μl of TRIzol® was added to the resultant powder which was then homogenised with an Omni Tissue Homogenizer (Omni International). 200 μl of chloroform was added to separate the mixture into a red phenol-chloroform phase, an interphase and an RNA containing colourless upper aqueous phase. The aqueous phase was collected and total RNA was precipitated with isopropyl alcohol and GlycoBlue (50 μg/mL) (Life Technologies). The resultant RNA pellet was washed with 75% ethanol, air dried and re-suspended in 20 μl DEPC-treated water. RNA was stored at −80°C prior to quality control and use in first strand cDNA synthesis. Quality and quantity of RNA was analyzed on the 2100 Agilent Bioanalyzer, using the Agilent RNA 600 Nano kit and BioSizing software (Agilent Technologies).

Total RNA was reverse transcribed using the SuperScript Vilo cDNA synthesis kit (Life Technologies) according to the manufacturer instructions. 1 μg of RNA was added to a 20 μl reaction mixture and the resulting cDNA was diluted 1:10 and stored at −20°C prior to use in RT-qPCR reactions. EXPRESS qPCR SuperMix (Life Technologies) and TaqMan Gene Expression Assays (ENPP1/PC-1 Hs 01054040_m1- Applied Biosystems) were used for all RT-qPCR reactions which were conducted on an ABI 7300 Real-Time PCR System (Applied Biosystems). Each sample was run in triplicate and each time a threshold cycle (Ct) was obtained using the 7300 Sequence Detection Software 1.3.1. The average of all three replicates for a particular sample was calculated and used in subsequent data analysis. Changes in gene expression were expressed as relative amount normalised to a reference gene and as fold change, both of which were calculated using the RQ method (2^-∆∆Ct). Eukaryotic 18S rRNA (4319413E, Applied Biosystems) was used as the reference gene.

Protein for three of the individuals included in the study was not available for the RYBG time point. Protein was extracted by pulverizing 3-4 mg of frozen liver under liquid nitrogen. The resultant powder was incubated with buffer containing Complete Protease Inhibitor Cocktail (Roche), 30 mM TrisCl, 7 M urea, 2 M thiourea and 4% (w/v) CHAPS) for 45 minutes on ice, with vigorous vortexing every 15 minutes. The mixture was centrifuged at 14000 g for 10 minutes to remove any insoluble material. The supernatant was aliquoted for storage at −80°C. Protein was quantified using Bradford reagent (Bio Rad) on a Bio-Rad Benchmark microplate reader [27].

An E-PAGE 48 well protein electrophoresis system was used to resolve protein samples on pre-cast E-PAGE 48 well 8% gels (20-300 kDa separation range) (Life Technologies) under denaturing conditions. Two separate gels were run, the first with 39 samples from the RYGB only group and the second containing 13 sets of samples from the patients who had repeat operations. Results for the RYGB comparison from the second gel were combined with the first for analysis purposes. SeeBlue Plus2 Pre-stained Standard (Life technologies) and MagicMark XP Western Protein Standard (Life technologies) were used as molecular weights to assess protein migration during electrophoresis and western blotting respectively. Protein was transferred onto Hybond-P PVDF membrane (Amersham) and blocked with 5% skim milk powder in TBS-Tween at 4°C overnight. Membranes were probed for 2 h at room temperature with monoclonal ENPP1/PC-1 (1:500, Clone F-8, Santa Cruz) [28], or monoclonal anti-actin (1:10000, Clone C4, Millipore). Blots were then incubated with alkaline phosphatase-conjugated anti mouse antibody (Life Technologies) for 0.5 h. The chemiluminescent signal was developed using CDP-Star chemiluminesent substrate (Life Technologies) and captured on x-ray film (Kodak). The protein bands on the x-ray film were digitized by the Chemidoc XRS system (Bio Rad) and then analyzed using Quantity One software (Bio Rad). Bands were quantified and expressed as volume; the sum of the intensities of the pixels within a defined volume boundary × pixel area (intensity units × mm ² ). Local background subtraction was used to facilitate background correction. Protein relative abundance was calculated by normalizing the volume of the ENPP1 band to the volume of the Actin band.

Statistical analysis was performed on 2^-ΔCt RT-qPCR data between each study group. Consequently gene expression data was expressed as both 2^-ΔCt in graphical form, whereas the fold change for significant differences was reported in text. Statistical analysis of protein abundance was performed on the value generated from normalisation of band intensity to the loading control.

Normal distribution of all groups to be compared was tested with Kolmogorov-Smirnov test. Non-normally distributed data was log transformed to comply with parametric test assumptions. Pre-post comparisons were carried out using a paired t-test. ANOVA with Bonferroni correction was used to test significance between groups. The Pearson product–moment correlation coefficient was used to test the strength and significance of association between variables. An alpha level of 0.05 was set as the significance threshold. All analysis was performed using Minitab 15. Graphical visualization was performed using GraphPad Prism 5.

We measured levels of ENPP1 mRNA expression and protein abundance in individuals with and without T2DM and again after resolution of insulin resistance and remission of T2DM. Table 1 summarizes the relevant anthropometric data and metabolic characteristics for the study groups around the time of RYGB surgery. All individuals were morbidly obese before surgery (BMI > 40 kg/m²) and there were no statistically significant differences in BMI between any of the groups. The T2DM group was significantly older than the NGT group, likely owing to the chronic and progressive nature of the disease. Mean fasting plasma glucose and HbA1c levels were normal in the NGT group, and slightly raised in the IGT group. However there were no statistically significant differences between the NGT and IGT group in any of the parameters measured at RYGB surgery. The T2DM group had significantly higher (p < 0.001, ANOVA) HbA1c and fasting glucose levels than the NGT group. Although insulin resistance was present in varying severity in all groups, the T2DM group were the most insulin resistant as they had the highest mean fasting plasma insulin concentration and HOMA-IR values, which was only significantly different between the T2DM and the NGT group (p < 0.001, ANOVA).

We found that liver ENPP1 gene expression was −1.5 (95% CI −6.2 to 2.8) fold lower in the T2DM group when compared to the NGT group (Figure 1A, p = 0.04, ANOVA), while there was no difference in ENPP1 mRNA expression between the NGT and IGT group. Figure 1C displays the western blot for ENPP1 protein abundance only for the 39 individuals who did not have a second liver biopsy (See Additional file 1 - Uncropped Western Blots, Figures 1 (ENPP1) & 3 (Actin) for uncropped original blots). To avoid gel bias, sample loading was alternated based on disease state. The molecular weight of ENPP1 in SDS-PAGE is 110–130 kd depending on tissue type. Some proportion of ENPP1 reportedly exists as dimer and the 220kd band visible in Figure 1C is most likely the dimeric form of ENPP1 [29]. The liver ENPP1 protein abundance was −2 fold lower (95% CI, −3.6 to −1.3) in the T2DM group in comparison to the NGT group (Figure 1B), which was also statistically significant (p = 0.04, t-test). To control for the effect of treatment on ENPP1 levels, we performed ANOVA analysis to test whether different treatments for diabetes affected ENPP1 levels. There was no significant differences in ENPP1 mRNA (p = 0.221) or ENPP1 protein (0.362) between previously unrecognized, diet controlled, insulin taking, or oral hypoglycemic taking individuals. Although ENPP1 protein abundance was lower in the IGT group, it did not reach statistical significance. Retrospective power calculations show that while the sample sizes in the NGT and T2DM are sufficient to reject the null hypothesis (α-level 0.05, power 0.83), the small sample size in the IGT group is limiting and another 36 observations would be needed to reach a statistically significant result.

To explore any relationship between liver ENPP1 protein and insulin sensitivity, we calculated a Pearson product–moment correlation between ENPP1 and HOMA-IR. ENPP1 protein abundance negatively correlated with HOMA-IR in the whole cohort (Figure 2), suggesting a positive association between ENPP1 and insulin sensitivity.

One of the novel characteristics of this study is the use of liver tissue biopsied from the same individual before and after significant weight loss and improvement in insulin sensitivity and/or remission of type 2 diabetes. Liver ENPP1 mRNA expression and protein abundance was measured in individuals who had normal glucose tolerance (rsNGT group) or T2DM (rsT2DM group) at RYGB and at a second operation 17 months later. Table 2 lists the metabolic characteristic for the rsNGT and rsT2DM groups at first and second liver biopsy. Out of the 16 individuals who had a second liver biopsy, two had incomplete follow up data. Both the rsNGT and rsT2DM groups had significant decreases in BMI by the second operation. There were no statistically significant differences in BMI between the two groups at either operation or in amount of weight lost, thus removing weight loss as a potential confounding factor. The rsT2DM group had a significant (p < 0.05, paired t-test) decrease in mean fasting plasma glucose and HbA1c levels, representing remission of T2DM. Liver insulin sensitivity (as measured by HOMA-IR) improved dramatically in both the rsNGT and rsT2DM group. The rsT2DM group had the greatest improvement in insulin sensitivity which coincided with remission of diabetes.

Figure 3A shows the mean liver ENPP1 protein abundance for all 13 individuals who had liver protein extracted at RYGB surgery and on average 17 months later, while Figure 3B is the corresponding western blot (see Additional file 1 - Uncropped Western Blots - Figures 2 (ENPP1) and 4 (Actin) for original uncropped images). There were six pairs of liver from the rsNGT group while there were seven pairs of samples from the rsT2DM group. Although the rsNGT group had no great change in liver ENPP1 protein levels after RYGB surgery, the rsT2DM group had a statistically significant (p = 0.01, paired t-test) increase in ENPP1 protein abundance (Figure 3A). The individuals who had resolution of insulin resistance and remission of T2DM following RYGB had a mean 2.7 (95% CI, 1.8 to 4.0) fold increase in liver ENPP1 protein abundance. Thus, improvement in liver insulin sensitivity after RYGB surgery as measured by HOMA-IR was concomitant with increasing levels of ENPP1 protein at the liver in humans. Neither group had significant changes in liver ENPP1 gene expression after RYGB surgery (data not shown), suggesting that post translational regulation is more important in the regulation of ENPP1 protein level with change in insulin resistance and remission of diabetes.