Impact of Adenoviral Transduction With SREBP1c or AMPK on Pancreatic Islet Gene Expression Profile: Analysis With Oligonucleotide Microarrays

Collagenase (type V) was obtained from Sigma (Poole, Dorset, U.K.). Culture medium, fetal bovine serum (FBS), and glutamine were obtained from Gibco (Glasgow, U.K.). Histopaque solutions and antibiotics were from Sigma.

Male Wistar rats (200–225 g) were sacrificed by cervical dislocation. Islets were isolated by pancreatic distension as described previously (6) and purified using Histopaque gradient solutions (10 ml of 1.119 g/l, 6 ml of 1.083 g/l, 6 ml of 1.077 g/l). After centrifugation for 20 min at 1,000g, islets were removed from the top layer and washed once with Dulbecco’s modified Eagle’s medium (DMEM). Isolated islets were cultured in suspension for 16 h in DMEM containing 30% (vol/vol) FBS, 11 mmol/l glucose, 2 mmol/l glutamine, 100 IU/ml penicillin, and 100 mg/ml streptomycin and were incubated at 37°C with 95% air and 5% CO₂.

Adenoviruses encoding constitutively active SREBP1c (amino acids 1–403, wild-type sequence) (3), constitutively active AMPK (16), or only enhanced green fluorescent protein (eGFP; “Null” virus) were amplified as described (6). SREBP1c and AMPK adenoviruses also express eGFP, under a distinct CMV promoter. Virus particles were purified on cesium chloride gradients before infection at a multiplicity of infection (MOI) of 30 (SREBP1c) or 100 (AMPK) viral particles per cell (6). Islets were infected with adenoviruses the day after isolation and cultured for 24 h in DMEM containing 10% FBS and 11 mmol/l glucose and were then cultured for a further 32 h in DMEM containing 3 mmol/l glucose before RNA extraction. Triplicate experiments were performed for each condition.

Total RNA (at least 5 μg/sample) was extracted from rat islets (1,000 islets/condition) using TRIzol reagent (Invitrogen Life Technologies, Paisley, U.K.) according to Affymetrix recommendations (Affymetrix, Santa Clara, CA). Processing of total RNA (5 μg) was performed at the Bioinformatics Facility of the University of Wales College of Medicine at Cardiff (http://www.cf.ac.uk/biosi/research/molbiol/Arrayer/index.html), as described (AffymetrixGenechip Expression Analysis Manual; Affymetrix).

Quality of total RNA was first controlled using an AgilentChip (Agilent Technologies, West Lothian, U.K.). Double-stranded cDNA was synthesized using a T7-oligo(dt) primer. The resultant cDNA was purified using Phase Lock Gel phenol/chloroform extraction and precipitation with ethanol. Biotin-labeled cRNA was synthesized using an RNA Transcript labeling kit (Enzo BioarrayHigh Yield; Enzo Life Sciences, Farmingdale, NY). In vitro transcription reactions were carried out at 37°C for 5 h, and the labeled cRNA obtained purified using RNeasy columns (Qiagen). The cRNA was fragmented in fragmentation buffer for 35 min at 94°C. Fragmented cRNA (10–11 μg/probe array) was used to hybridize to rat 230A GeneChip arrays (Affymetrix) at 45°C for 24 h in a hybridization oven with constant rotation (60 rpm). The chips were then washed and stained using Affymetrix fluidics stations. Staining was performed using streptavidin phycoerythrin conjugate (SAPE; Molecular Probes, Eugenes, OR), followed by the addition of biotinylated antibody to streptavidin (Vector Laboratories, Burlingame, CA). Probe arrays were scanned using fluorometric scanners. The scanned images were inspected and analyzed using established quality control measures.

MAS 5.0 software (Affymetrix) was used to obtain an expression signal and a present/absent/medium call status for every probe set on each of the hybridized chips. Probe sets with an absent/medium call in two of the three replicates in at least one of the two states (Treatment and Control) were removed from the subsequent analysis.

Log₂-transformation of the data were performed (19) to allow the normalization of intensities and before statistical Student’s t test (two tailed; assuming unequal variances). Positive false discovery rates (pFDR) were calculated as described (20).

One microgram of total RNA, prepared as described above, was reverse-transcribed. First-strand cDNA synthesis was performed as described (21) using oligo (15)-dT primers and reverse transcriptase (500 units Moloney-murine leukemia virus; Promega) in a total volume of 50 μl. PCR reactions were performed in a total volume of 50 μl, comprising 5 μl cDNA product, 0.2 mmol/l of each nucleotide triphosphate, 20 pmol of each primer, and 0.875 units Taq polymerase (Expand High Fidelity; Roche). Oligonucleotide primer sequences and PCR conditions used in this study are available upon request. PCR was performed at 95°C for 2 min, followed by determined cycles for each target gene at 95°C for 1 min, 1 min at the melting temperature (determined for each target gene; not shown), and 72°C for 1 min. The last cycle was followed by a final extension step at 72°C for 10 min.

PCR products were separated on 1% (wt/vol) agarose gels containing ethidium bromide (0.5 μg/ml) and quantitated by digital imaging (ImageQuant; Molecular Dynamics).

To identify SREBP1c-regulated genes in isolated pancreatic islets, we used adenoviruses to express a recombinant, truncated, and constitutively active form of SREBP1c, comprising the nuclear NH₂-terminal domain (amino acids 1–403) (1). As described in detail elsewhere (6), infection with this virus leads to the expression of SREBP1c in 20–30% of cells, the majority (>70%) being insulin-positive β-cells at the islet periphery.

After transduction with a SREBP1c-encoding or null (eGFP-encoding) adenoviruses and culture for 32 h at 3 mmol/l glucose, 8,748 of a total of 15,923 probe sets were retained for analysis. In comparison to null adenovirus-infected islets, SREBP1c affected the expression of 2,992 genes significantly (P < 0.05); of these, 1,238 genes were altered highly significantly (P < 0.01). Of the latter group, the vast majority (1,180, or 95.3%) were upregulated. By contrast, overexpression of a constitutively active form of AMPK catalytic subunit (amino acids 1–312, T¹⁷²D, corresponding to the identical NH₂-terminal domains of AMPKα1 and α2) (22) altered the expression of 2,531 genes significantly (P < 0.05) and 752 genes highly significantly (P < 0.01). Of these, 702 (93.4%) were upregulated.

To identify genes that may be direct targets for SREBP1c in the islet, as opposed to genes whose expression results from the indirect effects of lipid accumulation in response to transduction with SREBP1 (3,6), we next applied three further criteria to increase the stringency of the search. Specifically, genes were included only if their expression 1) changed highly significantly (P < 0.01) in response to SREBP1c overexpression but did not change significantly and in the same direction in response to both SREBP1c and AMPK overexpression (since AMPK overexpression also inhibits insulin secretion and glucose-stimulated increases in intracellular ATP [16,22] this allowed correction for the effects of SREBP1c on these parameters); 2) was altered by at least 2.0-fold (increased or decreased) by either SREBP1c or AMPK; and 3) passed a positive false discovery rate test (P < 0.05). This led to the identification of 206 genes that were regulated by SREBP1c (Table 1); of these, 195 (94.6%) were increased. Within this group, six genes were upregulated by SREBP1c and downregulated by AMPK. Of the SREBP1c-regulated genes identified, 82 were identified functionally (Table 1) using Pathway Assist (Stratagene) software.

Genes affected highly significantly by either activated SREBP1c or AMPK are illustrated in Fig. 1. Of those affected by SREBP1c, the majority were involved in either metabolism or metabolic signaling, with smaller numbers involved in growth or apoptosis. Interestingly, 7% of SREBP1c-stimulated genes were involved in the immune response, an observation that may reflect either the presence of immune cells with the isolated islets or the activation of these genes in endocrine cells, or both. Genes involved in the neural regulation of insulin secretion were among those strongly affected by SREBP1c, including receptors for the inhibitory neuropeptides somatostatin (sstr1) and galanin (Table 1). These and other changes identified in the microarrays were also verified by semiquantitative RT-PCR analysis (underlined in Tables 1 and 2).

A high-stringency analysis, as described above, revealed that of 48 genes affected uniquely by AMPK overexpression, 33 (70%) were highly significantly upregulated. As shown in Table 2 and Fig. 1B, a greater proportion of AMPK-affected genes were involved in cellular functions, including adhesion and growth. It should be stressed, however, that under the conditions used here (3 mmol/l glucose) AMPK is significantly activated in β-cells (14–18,22) such that further increases in AMPK activity as a result of adenoviral transduction may be relatively small (22). Future studies will be required to explore the effects of changes in AMPK activity over the physiological range of glucose concentrations (17).

As observed in analyses of livers from transgenic mice overexpressing SREBP1a or SREBP2 (10), expression of SREBP1c in isolated rat islets led to changes in the expression of a large number of genes (>1,200, representing 14% of all genes called present in the islet preparations used). It seems likely that many of these are changed as a result of the nonspecific effects of lipid accumulation and the consequent metabolic changes (10). Thus, SREBP1c expression in either clonal β-cells (3) or primary rat islets (6) leads to a decrease in ATP levels and to an inhibition of insulin secretion. To control for these changes, we used as a control condition islets overexpressing AMPK. While the effects of AMPK expression on lipid synthesis are likely to be directly opposed to those of SREBP1c (12), the expression of AMPK also leads to a reduction in glucose oxidation and GSIS (16–18,22). Thus, by this combinatorial approach, we sought to identify genes likely to be under the direct control of SREBP1c.

In contrast to the changes observed after SREBP1c expression (see below), the effects of AMPK overexpression were relatively modest with the levels of only three genes (prostaglandin E synthase, lipoprotein lipase, and peptide YY precursor) changing by >3.0-fold. Further studies will be required to determine whether the induction of PPY, a neuropeptide present at very early stages in islet development (23), may reflect dedifferentiation of the islet cell population.

The findings reported here in respect to the effects of SREBP1c overexpression are in broad agreement with those reported for the livers of transgenic SREBP-expressing mice, including genes involved in cholesterol and fatty acid biosynthesis (Table 1). However, other, unsuspected SREBP1c-regulated genes were also identified in islets. These include the glycine transporter; solute transporter family 6, member 6 (NM_017206; Table 1); and receptors for the inhibitory neuropeptides somatostatin (receptor 1, sstr1, NM_012719) and galanin (receptor 2, NM_019172), which were not detectably expressed in the liver (10). Because sstr1 (24) is localized largely to β-cells within the islet, upregulation of this receptor may render islets more susceptible to inhibition by endogenous somatostatin. By contrast, type 2 galanin receptors (NM_019172) are usually linked to phospholipase C activation and the activation of insulin release (25).

We did not, however, observe any significant effect of SREBP1c overexpression on the levels of mRNAs encoding a number of key β-cell transcription factors such as pdx-1 or Nkx6.1, or on the key “glucose sensing” genes glut2/slc2a2 or glucokinase, in line with previous studies in islets (6) but in contrast to INS-1 cells (4). Moreover, we detected no significant changes in the expression of acetyl-CoA carboxylase I, consistent with only very small changes observed previously observed by RT-PCR (6). Finally, no large changes in the expression of mitochondrial enzymes were detected, although UCP2 mRNA levels were increased 2.0-fold, albeit at the limit of significance (P = 0.025) (6).

We also detected changes in the expression of a number of proapoptotic genes (6). These included GADD45β, the p75-like apoptosis-inducing death domain protein, PLAIDD, and prostaglandin D2 synthase. On the other hand, expression of T-cell death-associated gene and BAX inhibitor-1 (Table 1) were decreased in SREBP1c-infected islets. Further exploration of the impact of SREBP1c on islet cell apoptosis may thus be warranted given reports of decreases in β-cell mass in type 2 diabetes (26).

This study was supported by grants from the Wellcome Trust (Project 062321; Programme 067081/Z/02/Z, Prize Studentship to E.M.), the Biotechnological and Biological Research Council, the Juvenile Diabetes Research Foundation International, and the Medical Research Council (U.K.). G.A.R. and I.L. are grateful to the Wellcome Trust for Research Leave and Advanced Fellowships, respectively, and F.D. is a recipient of the European Union for a Marie Curie Fellowship.

We thank Rebecca Rowe for excellent technical assistance.