Per-arnt-sim (PAS) domain-containing protein kinase is downregulated in human islets in type 2 diabetes and regulates glucagon secretion
- First online:
We assessed whether per-arnt-sim (PAS) domain-containing protein kinase (PASK) is involved in the regulation of glucagon secretion.
mRNA levels were measured in islets by quantitative PCR and in pancreatic beta cells obtained by laser capture microdissection. Glucose tolerance, plasma hormone levels and islet hormone secretion were analysed in C57BL/6 Pask homozygote knockout mice (Pask −/−) and control littermates. Alpha-TC1-9 cells, human islets or cultured E13.5 rat pancreatic epithelia were transduced with anti-Pask or control small interfering RNAs, or with adenoviruses encoding enhanced green fluorescent protein or PASK.
PASK expression was significantly lower in islets from human type 2 diabetic than control participants. PASK mRNA was present in alpha and beta cells from mouse islets. In Pask−/− mice, fasted blood glucose and plasma glucagon levels were 25 ± 5% and 50 ± 8% (mean ± SE) higher, respectively, than in control mice. At inhibitory glucose concentrations (10 mmol/l), islets from Pask−/− mice secreted 2.04 ± 0.2-fold (p < 0.01) more glucagon and 2.63 ± 0.3-fold (p < 0.01) less insulin than wild-type islets. Glucose failed to inhibit glucagon secretion from PASK-depleted alpha-TC1-9 cells, whereas PASK overexpression inhibited glucagon secretion from these cells and human islets. Extracellular insulin (20 nmol/l) inhibited glucagon secretion from control and PASK-deficient alpha-TC1-9 cells. PASK-depleted alpha-TC1-9 cells and pancreatic embryonic explants displayed increased expression of the preproglucagon (Gcg) and AMP-activated protein kinase (AMPK)-alpha2 (Prkaa2) genes, implying a possible role for AMPK-alpha2 downstream of PASK in the control of glucagon gene expression and release.
PASK is involved in the regulation of glucagon secretion by glucose and may be a useful target for the treatment of type 2 diabetes.
KeywordsΑlpha cells Glucagon secretion Human islets of Langerhans Knockout mouse PASK
AMP-activated protein kinase
PAS domain-containing protein kinase
- Pask −/−
Pask homozygote knockout mice
Small interfering RNA
Elevated glucose concentrations (>3.5 mmol/l) normally suppress the release of glucagon from pancreatic alpha cells; dysregulation of this process is a feature of type 1 and type 2 diabetes [1, 2]. Both direct  and indirect  effects of glucose on the release of glucagon have been described. Whereas the latter appear to involve the release of secretory products, including insulin [3, 5], gamma-aminobutyric acid [6–8] or zinc ions [4, 9] from neighbouring beta cells, as well as somatostatin from delta cells [10, 11], the signalling events involved in the direct sensing of glucose are more controversial . These may involve enhanced metabolism of glucose and the closure of ATP-sensitive K+ channels, followed by limited membrane depolarisation [13, 14]. AMP-activated protein kinase (AMPK) appears also to be involved in the regulation of glucagon release .
Per-arnt-sim (PAS) domain-containing protein kinases (PASKs) are related to AMPK and are common in prokaryotes. However, there is currently only one known mammalian counterpart [16, 17]. We  and others [19, 20] have shown that PASK is important for energy sensing and maintenance of normal cellular energy balance in mammalian systems.
In pancreatic beta cells, PASK activity is regulated by glucose and is involved in the regulation of glucose-induced preproinsulin and pancreatic duodenum homeobox-1 (Pdx1) gene expression [18, 20]. Expression of the Pask gene in rodent islets and beta cell lines [18, 20] is also glucose-sensitive. Recently, PASK has been implicated in regulation of lipogenic gene expression  and might, therefore, influence glucose signalling through lipid intermediates as proposed for glucose-induced insulin secretion .
Pask homozygote knockout mice (Pask−/−), which are globally inactivated for Pask, have previously been reported to display lower plasma insulin levels than control littermates , but normal glucose tolerance [19, 22], reflecting enhanced insulin sensitivity . Insulin secretion from Pask−/− islets has variously been shown to be not different  or lower  than in control islets. It has also been previously reported that total insulin content and/or beta cell mass were not altered in Pask−/− mice [19, 22]. However, data from these earlier studies are difficult to interpret for a number of reasons. In the first instance, total islet insulin was not always measured, making it difficult to evaluate the insulin secretory capacity of Pask−/− islets. Moreover quantification of beta cell mass using pancreatic sections is prone to substantial variability . Due to these limitations [19, 22], the role of PASK in the regulation of pancreatic hormone release and glucose homeostasis remains to be clarified.
A recent report  showed that inhibition of insulin gene expression by palmitate was reversed by Pask overexpression in MIN6 beta cells. The loss of PASK, and hence the loss of regulation by Pdx1 and Mafa gene expression , was proposed to be a mechanism by which insulin gene expression in the beta cell might be lost on exposure to palmitate . Thus, PASK appears to exert a protective effect in mature beta cells. Moreover, aberrant PASK expression and/or function may play a significant role in the development of diabetes and, interestingly, Pask −/− mice develop glucose intolerance on a high-fat diet .
Although glucose homeostasis reflects the release of multiple islet hormones in addition to insulin, there are currently no published data on the role of PASK in other islet cell types. The present study demonstrates that PASK is involved in the regulation of glucagon secretion from pancreatic alpha cells. We show that PASK expression is decreased in the islets of human type 2 diabetic patients and that Pask is at least as strongly expressed in highly purified mouse alpha cells as in beta cells. Silencing or ablation of Pask in clonal alpha cells or islets, respectively, drastically blunted the inhibition of glucagon secretion by glucose. Whereas the insulin content of PASK-deficient islets was dramatically reduced, the acute regulation of insulin secretion by glucose was normal. Thus, PASK regulates hormone release reciprocally from pancreatic alpha and beta cells. Given that dysregulation of insulin and glucagon secretion are characteristics of type 2 diabetes, we propose that PASK may be a potential drug target to modulate glucagon release in vivo.
Adenoviruses encoding for human PASK have been previously described . All general chemicals and tissue culture reagents were purchased from Sigma (Poole, UK) or Invitrogen (Paisley, UK), unless otherwise stated in the text.
Isolation and culture of islets
Studies on human islets were conducted with local ethics committee approval at all sites (Charing Cross Research Ethics Committee Ref. 07/H0711/114). Human islets were isolated as previously described in Oxford, UK , or in Pisa, Italy , and maintained in medium containing 11 mmol/l glucose for 10 days to allow the loss of exocrine tissue . Human type 2 diabetic donors were selected according to established criteria [23, 26]. Mouse pancreatic islets were isolated and cultured as described in . Alpha-TC1-9 cells (passage 35–45; American Type Culture Collection, Manassas, VA, USA) were cultured as previously described .
Laser capture microdissection and microarray analysis
These procedures were performed as described in Electronic supplementary material (ESM) Methods.
We prepared 300 ng of total RNA with a kit (mRNA-Seq 8; Illumina, Little Chesterford, UK). For clustering and sequencing we used Illumina cluster generation and sequencing kits v4. 9pM were loaded to the flowcell (one sample per lane) and sequenced on a sequencer (GAIIx) with RTA software version 2.6 (Little Chesterford, UK). On average we obtained 28 million reads per lane, passing Illumina’s quality filtering. Short reads (35 bp) were aligned to the mouse genome (Ensembl v54) using a mapping station (Genomatix)  and to a database of known and potential splice junctions using Bowtie . Up to three mismatches and no insertions or deletions were allowed, and, on average, more than 85% of the reads in a sample could be mapped to the genome. Gene expression was measured as proposed by Mortazavi et al.  and reads per kilobase per million (RPKM) values were computed for every gene.
Mice Pask−/− mice (kindly provided by R. Wenger, Institute of Physiology and ZIHP, Zurich, Switzerland)  were back-crossed for ten generations with C57BL/6 mice prior to use. Mice were housed with two to five animals per cage in a pathogen-free facility on a 12 h light–dark cycle with free access to standard mouse chow diet, unless otherwise stated. All in vivo procedures were performed in the Imperial College Central Biomedical Service in accordance with the Principles of Laboratory Care and the UK Home Office (Animals Scientific Procedures Act, 1986), and approved by the local ethics committee. Genotyping was performed as previously described .
Intraperitoneal glucose tolerance test
An intraperitoneal glucose tolerance test was performed on 8-week-old mice as described in , at 09:00 hours on each experimental day.
Measurement of total pancreatic insulin and glucagon
Pancreases were excised from 8-week-old mice and suspended in ice-cold acid–ethanol (75.0% ethanol–23.4% molecular grade water–1.5% HCl–0.1% Triton X-100, vol./vol.) prior to disruption by sonication (microsonicator; Misonix, Farmingdale, NY, USA) at 4°C. Total protein was measured by Bradford’s assay , and insulin and glucagon content were measured by radioimmunoassay (Linco, Watford, UK).
Measurement of plasma glucagon
At 8 weeks of age mice were starved overnight prior to being killed by cervical dislocation. Blood (200 μl) was immediately removed by cardiac puncture. Plasma was collected using high-speed centrifugation (2,000 g, 5 min) in heparin-coated tubes (Microvette; Sarstedt, Leicester, UK) and plasma glucagon assessed by radioimmunoassay (Linco, Watford, UK).
Manipulation of PASK content in alpha-TC1-9 cells and cultured islets of Langerhans
Alpha-TC1-9 cells and islets were infected with adenoviruses encoding for PASK or enhanced green fluorescent protein at a multiplicity of infection of 100 and cultured for 48 h prior to use. Alpha-TC1-9 cells were transfected with anti-Pask small interfering RNA (siRNA) or control siRNA (1 nmol/l) . Protein content was assessed by western (immuno-)blot analysis.
Culture of E13.5 rat pancreatic epithelia
E13.5 rat pancreatic epithelia were isolated and cultured as previously described . Epithelia were cultured in the presence of anti-Pask siRNA or control siRNA  (1 nmol/l) for 10 days prior to RNA isolation for real-time quantitative PCR analysis.
Real-time quantitative PCR analysis
Highly purified primary mouse islet beta and alpha cells were obtained by fluorescence-activated cell sorting of transgenic mice expressing the fluorescent protein Venus selectively in the alpha cell under the preproglucagon promoter [34, 35]. RNA was extracted from alpha-TC1-9 cells and cultured rat E13.5 explants using Trizol (Invitrogen) according to the manufacturer’s guidelines. Total RNA was subjected to DNAse treatment (Ambion, Warrington, UK), followed by cDNA conversion (high-capacity cDNA conversion kit; Applied Biosystems, Warrington, UK) and real-time quantitative PCR using SYBR green (Applied Biosystems) in a 7500 Real-Time PCR System (ABI, Warrington, UK).
Western (immuno-)blot analysis
Western (immuno-)blot analysis was performed as previously described .
Data are the means ± SE for the number of observations indicated. Statistical significance and differences between means were assessed by Student’s t test with Bonferroni correction for multiple analyses.
PASK gene expression is regulated by glucose in human pancreatic islets of Langerhans and is lowered in type 2 diabetes
mRNA expression data obtained using whole islets may be confounded by variable degrees of contamination with other pancreatic cell types, as well as by other factors. Therefore, we sought to further support the above results by using laser capture microdissection and array analysis to quantify mRNA levels selectively in beta cell-enriched samples . A clear tendency for decreased PASK expression in islets in type 2 diabetic pancreases vs controls was also observed using this approach (ESM Fig. 1). In contrast, consistent changes in levels of the AMPK-alpha1 and -alpha2 subunit (encoded by PRKAA1 and PRKAA2, respectively) were not observed with array data on samples obtained by laser capture microdissection. However, with quantitative PCR done on the same samples with enough RNA (10 vs 10), a small but statistically significant decrease in PRKAA1 expression was apparent (18%; p < 0.019), using RPL32 as an internal control.
Pask is expressed in murine pancreatic beta and alpha cells
In the mouse, Pask mRNA has previously been shown to be most abundant in testes as well as in haemopoietic tissues including thymus and spleen . Levels in other tissues, including islets, are much lower  (G. A. Rutter and G. Sun, unpublished observations). Indeed, the expression of Pask in adult mouse islets and its induction by glucose have recently been questioned, since expression in these cells of the lacZ gene, expressed in mice null for Pask alleles under the Pask promoter, was barely detectable by beta-Gal staining . By contrast, we have previously detected Pask mRNA in mouse islets and MIN6 cells , data recently confirmed by others . Examined here in islets from mice on a mixed C57BL/6/sv129 background cultured at 11 mmol/l glucose, massive parallel sequencing (RNAseq) confirmed the presence of Pask mRNA at low but detectable levels, lying in the lower 30th percentile of all mRNAs at approximately 0.2 ± 0.3% (n = 3 female mice) of β-actin mRNA levels (ESM Fig. 2).
To determine in which cell types Pask was expressed in mouse islets, we next compared the expression of Pask mRNA in highly purified primary mouse alpha and beta cells (Fig. 1c). These were obtained by fluorescence-activated cell sorting of transgenic mice selectively expressing the fluorescent protein, Venus, in the alpha cell [34, 35] and by real-time quantitative PCR analysis. This approach confirmed the presence of Pask mRNA in both cell types and indicated that levels in purified alpha cells (1.45 ± 0.46% of beta-actin mRNA) tended to be higher than in beta cells (0.54 ± 0.044%; n = 3 preparations; p = 0.07 by two-tailed Student’s t test).
Pask−/− mice display normal glucose tolerance but impaired plasma glucagon concentration
Islets of Langerhans from Pask−/− mice display impaired glucose-regulated glucagon secretion
The expression of mouse preproinsulin 2 (Ins2) and Pdx1 genes was strongly impaired in Pask −/− mouse islets (Fig. 3e), as previously reported [18, 20]. Moreover, the insulin content of whole Pask −/− mouse pancreases was 1.98 ± 0.3-fold lower than that of wild-type pancreases (67 ± 15 vs 133 ± 12 ng/mg protein for Pask −/− and wild-type mice, respectively; n = 3 mice for each genotype). While glucagon protein content in Pask −/− islets (Fig. 3c) was unaffected, preproglucagon (Gcg) gene expression was slightly but significantly increased, as was Prkaa2 mRNA expression (Fig. 3e).
Glucagon release is activated in the absence of Pask in alpha-TC1-9 cells
Glucagon release is inhibited by PASK overexpression in alpha-TC1-9 cells and human islets of Langerhans
Similarly, overexpression of PASK in human islets of Langerhans (Fig. 5b) inhibited glucagon secretion at all glucose concentrations tested, while the stimulatory effects of KCl were largely maintained. Interestingly, there was still an apparent effect of glucose on glucagon secretion in human islets overexpressing PASK (Fig. 5b), consistent with a maintained effect on glucagon release mediated by beta cell-derived factors.
Silencing of Pask gene expression in alpha-TC1-9 cells and E13.5 rat pancreatic epithelial explants causes increased AMPK-alpha2 and glucagon gene expression
To determine whether the dysregulation of glucagon gene expression and release in Pask −/− mice was due to a role of the enzyme in the development of pancreatic alpha and beta cells, we next examined the impact of Pask silencing in developing pancreatic epithelia. E13.5 rat pancreatic epithelial explants , in which Pask gene expression was silenced with an anti-Pask siRNA, developed endocrine buds to an extent indistinguishable from that observed in explants treated with a control (scrambled) siRNA (ESM Fig. 3). However, explants in which Pask was silenced displayed similar increases in Gcg and Prkaa2 gene expression (ESM Fig. 3) to those observed in alpha-TC1-9 cells (Fig. 6). Explants overexpressing Pask did not survive the 10 day culture period.
PASK is expressed in pancreatic islets
An important finding of the present study is that Pask mRNA is clearly present in mouse islets maintained at permissive glucose concentrations (11 mmol/l), in contrast to a recent suggestion . These results and those from another recent publication  support our own previous findings . We would stress, however, that the level of gene expression, at least at the mRNA level, was relatively low compared with tissues in which Pask is strongly expressed (notably in the testes) , an observation that probably underlies the failure to detect significant expression using alternative approaches .
PASK regulates insulin and glucagon secretion
It has previously been shown that islet architecture and beta cell mass in Pask −/− mice are intact [19, 22]. Thus, it was hypothesised in earlier studies that the observed decrease in insulin release may have been due to defective glucose-signalling in the beta cell . In contrast, our data show that glucose-regulated insulin secretion was intact in Pask −/− islets (Fig. 3b), but that the insulin content of the islet (Fig. 3d) and pancreas (see Results section) were lowered by Pask deletion. This apparent discrepancy between the present and earlier [19, 22] studies may reflect differences in the genetic background, age and sex of the mice used. We would, however, note that Pask −/− beta cell mass was previously assessed by immunohistochemical analysis of fixed pancreatic sections, which may be prone to error.
Nevertheless, it was previously reported that plasma insulin content was compromised in Pask −/− mice , a result consistent with the present findings. In our hands, glucagon release from Pask −/− mouse islets (Fig. 3a) and human islets in which PASK was overexpressed (Fig. 5b) still displayed some degree of glucose-responsiveness, even though this was markedly decreased compared with the control condition. We hypothesise that this ‘residual’ regulation of release may be due, at least in part, to the release of insulin or other factors (Zn2+, gamma-aminobutyric acid, etc.) from neighbouring pancreatic beta cells and is consistent with the maintained effect of exogenous insulin on glucagon release from alpha-TC1-9 cells (Fig. 4b).
Close inspection of the data of Fig. 3a vs Fig. 4 reveals that whereas loss of Pask enhanced glucagon secretion from islets only at high (inhibitory) glucose levels (Fig. 3a), stimulation of glucagon release was apparent at all glucose concentrations examined in alpha-TC1-9 cells (Fig. 4). The mechanisms responsible for this difference are unclear. However, we note that while insulin mRNA was present at low but detectable levels in control alpha-TC-1-9 cells (Ct values ≥ 29; data not shown), silencing of Pask led to decreased insulin gene expression such that this was below the limit of detection. Hence an action of Pask silencing on endogenous insulin secretion from alpha-TC-1-9 cells, and thus loss of tonic inhibition of glucagon secretion by this hormone, may contribute to the above mechanisms. Further studies are needed to resolve this question.
Interestingly, the increase in plasma glucagon levels measured after 16 h fasting in 8-week-old Pask −/− mice was similar to that observed in alpha cell-specific insulin receptor knockout mice of the same age . We have previously shown that PASK activity is not regulated by insulin, but that PASK may regulate insulin gene expression in pancreatic beta cells . Thus, we propose that the regulation of insulin content by Pask is one of the mechanisms through which PASK may indirectly regulate glucagon release.
A role for AMPK alpha2 downstream of PASK?
Our gene-expression studies revealed that expression of the catalytic subunit of another glucose-responsive fuel gauge, Prkaa2 (encoding AMPK-alpha2), is upregulated when Pask gene expression is compromised (Fig. 6). This is an interesting observation, since AMPK activity has been shown to be glucose-responsive in pancreatic beta cells  and to regulate insulin release [37, 38]; in addition, we have recently shown that activation of AMPK in the alpha cell stimulates glucagon secretion . However, as also observed here, Prkaa2 mRNA levels are normally much lower (more than tenfold) than Prkaa1 in purified mouse beta and alpha cells , suggesting that the inappropriate activation of this isoform and an increase in AMPK activity in the nucleus (from which AMPK alpha1 is excluded in mature pancreatic endocrine cells)  may impact on the transcription of key genes, including Gcg, to reprogram the alpha cell. Interestingly, induction of Prkaa2 expression was also observed in E13.5 pancreatic epithelia  in which Pask was silenced (Fig. 6c), indicating that defects during pancreatic development may contribute to the dysregulation of glucagon release in Pask −/− mice.
PASK may have a role in the pathophysiology of type 2 diabetes
The observation that pancreatic islets from patients with type 2 diabetes display lower PASK mRNA levels than islets from non-diabetic individuals (Fig. 1b) suggests that lowered PASK activity may contribute to decreased insulin release and to enhanced glucagon secretion in this condition . PASK may therefore represent a potential therapeutic target, the activation of which might favourably affect the secretion of both hormones in patients with type 2 diabetes.
The authors thank S. Vakhshouri for technical assistance, and F. Semplici and M. K. Loder for scientific input during manuscript preparation. This work was supported by a Juvenile Diabetes Research Foundation post-doctoral fellowship and an European Foundation for the Study of Diabetes Albert Renold travelling fellowship to G. da Silva Xavier, and by grants to G. A. Rutter from the Wellcome Trust (081958/2/07/2) and the Medical Research Council (G0401641). H. Farhan, H. Kim and S. Caxaria were supported by Imperial College London BSc and MSc project funds. The Diabetes Research and Wellness Foundation supported P. Johnson and S. Hughes.
Duality of interest
The authors declare that there is no duality of interest associated with this manuscript.
This article is distributed under the terms of the Creative Commons Attribution Noncommercial License which permits any noncommercial use, distribution, and reproduction in any medium, provided the original author(s) and source are credited.