SIK2 Is Involved in the Negative Modulation of Insulin-Dependent Müller Cell Survival and Implicated in Hyperglycemia-Induced Cell Death

Abstract

Purpose.: To investigate the role of the serine/threonine kinase SIK2, a member of the salt-inducible kinase (SIK) family, in insulin-dependent cell survival and hyperglycemia-induced cell death in Müller glia.

Methods.: Expression studies were performed by RT-PCR, immunostaining, Northern blotting, and immunoblotting. Insulin-dependent changes in SIK2 activity were investigated by in vitro kinase assays in MIO-M1 Müller cell line. Akt activation was studied by immunoblotting and cell death by TUNEL assay. The potential role of SIK2 in insulin signaling was explored by overexpression and sh-RNA knock-down approaches. Effects of hyperglycemia were studied in vitro and in vivo in streptozotocin-injected rats.

Results.: SIK2 expression was detected throughout adult retina, except for the outer nuclear layer. Insulin stimulation of MIO-M1 cells resulted in a rapid 2-fold increase of SIK2 activity, increased insulin receptor substrate 1 (IRS1)-SIK2 interaction, and reduced cell death. pAkt levels following insulin treatment were modulated by SIK2 activity. Under hyperglycemia, increased SIK2 activity/expression was concomitant to decreased Akt activation and enhanced apoptosis; whereas knockdown of SIK2 under normo- and hyperglycemic conditions resulted in a rapid increase in pAkt levels and blunted cell death. SIK2 overexpression under normoglycemia had an opposite effect. SIK2 activity increased significantly within 2 weeks of induction of hyperglycemia in the rat retina.

Conclusions.: Results indicate that SIK2 functions as a negative modulator of the insulin-dependent survival pathway and contributes to hyperglycemia-induced cell death of Müller glia in vitro. Although still hypothetical at this point, our study suggests that SIK2 could serve a similar role during the development of diabetic retinopathy in vivo and that it represents a potential target to control disease progression.

Introduction

Diabetes is characterized by hyperglycemia due to insulin deficiency or impaired insulin signaling. Diabetic retinopathy (DR) is one of the most frequent complications of diabetes and a leading cause of vision loss around the world (World Health Organization, 1995, available in the public domain at http://www.who.int/whr/1995/en/whr95_en.pdf). Although the disease generally is characterized by microangiopathy, earlier dysfunction and cell death in the neural retina are now well established. ^1,2  

Insulin, in addition to its role in glucose homeostasis, is an important survival factor for numerous cell types where the PI3K-Akt axis appears to be the main signaling route. ^3,4 Insulin receptors (IR) are present in retinal neurons and the principal glial cell, the Müller glia. In this tissue, critical node elements of IR signal transduction, insulin receptor substrates (IRS) 1 and 2, were shown to be coexpressed with IR ^5,6 and activated in an insulin-dependent manner. ^7–9 It has been reported that in diabetic retina IR activity is impaired, and with hyperglycemia enhanced retinal cell death is observed. Hyperglycemia results in neuronal cell death via alteration of the neuroprotective effect on insulin-mediated Akt signaling and glycosylation of proteins involved in cell survival. ¹⁰  

Müller cells anatomically and functionally are associated closely with retinal neurons and blood vessels. These cells are critically important in the maintenance of homeostasis of the retinal milieu, such as regulating ion concentrations, and removing neurotransmitters GABA, glycine, and glutamate ^11,12 for the proper functioning of the neurons. Müller cells also are known to express and release trophic factors that are thought to contribute to neuronal survival, and the regulation of blood flow and angiogenesis. ¹³ Thus, Müller glia dysfunction or their death within the context of hyperglycemia has potential to exacerbate neuronal and vascular complications, and thereby contribute to the development of DR. 

In animal models of diabetes, Müller cells become activated at initial stages of the disease, as indicated by glial fibrillary acidic protein (GFAP) expression, and several functions of Müller cells are dysregulated, including changes in proinflammatory cytokine and growth factor expression, as well as impairment in glucose transport and glutamate uptake. ^14–17 IR and IRS1 are colocalized within Müller cells. ¹⁸ In cultured Müller glia, insulin was reported to enhance tyrosine phosphorylation of IRS1, ¹⁹ and silencing of IRS1 enhanced Müller cell apoptosis. ²⁰ Enhanced Müller cell apoptosis and decreased Akt activity have been reported in diabetic animal models. ^21,22 Consistent with these observations, Müller cell apoptosis via downregulation of Akt activity under hyperglycemic conditions has been observed in vitro. ²³  

The role of serine phosphorylation of IRS proteins is well established in attenuation or fine-tuning of IR signaling and insulin resistance during diabetes. ²⁴ Thus, an increasing number of serine/threonine kinases are gaining attention as modulators of the cellular response to insulin stimulation, such as SIK2, a member of the Salt Inducible Kinase (SIK) family. SIK2 was identified first in mouse adipose tissue, where it was shown to phosphorylate IRS1 on S789. ²⁵ SIK2 expression and activity were upregulated in parallel to an elevation in S789 (S794 in human) phosphorylation of IRS1 in insulin-resistant rats, ²⁶ and in the adipose tissue of diabetic mice. ²⁵ Although the physiologic consequence of this phosphorylation remains to be resolved, it was suggested that SIK2 might be involved in the development of type 2 diabetes. ²⁵  

In our study we reported widespread expression of SIK2 in adult rat retina and cultured Müller glia. Subsequently, using the MIO-M1 cell line we observed that insulin leads to a rapid and transient activation of SIK2. Overexpression and downregulation studies in vitro suggest that SIK2 is involved in negative modulation of the insulin-dependent Müller cell survival pathway and that it contributes to hyperglycemia-induced cell death. Furthermore, we observed enhanced SIK2 activity in retinas of streptozotocin-induced diabetic rats within 2 weeks. 

Methods

Experimental Animals and Tissue Collection

The use and handling of the animals conformed to the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research and were approved by the local Ethics Committee. Animals anesthetized by CO₂ inhalation were killed by cervical dislocation. For RNA and protein extraction, retinas were dissected from enucleated eyes and flash frozen. Retinal outer nuclear layer (ONL) and inner nuclear layer (INL) slices were prepared by Vibratome sectioning. ²⁷ For immunohistochemical studies, eyecups were fixed in 4% paraformaldehyde (PFA) in PBS overnight at 4°C, washed thoroughly with fresh PBS, impregnated sequentially with PBS containing 10%, 20%, 30% sucrose, and embedded in OCT compound (Tissue-Tek, Torrance, CA). Cryosections were stored at −20°C until use. 

To generate diabetic animals, 8 male albino Wistar rats (190–250 g), were injected intravenously with 60 mg/kg streptozotocin (STZ) solubilized in 10 mM citrate buffer (1.8 mM citric acid and 8 mM sodium citrate, pH 4.5); control rats (n = 6) received 10 mM citrate buffer alone. ²³ Animals were sacrificed after 2 weeks. Fasting blood glucose levels of the animals were between 75 and 100 mg/dL before the injection. On the day of sacrifice, animals with fasting blood glucose levels higher than 200 mg/dL were considered as diabetic. Dissected retinas were frozen rapidly in liquid nitrogen and stored at −80°C until use. 

Cell Culture

MIO-M1 Müller glia cell line ²⁸ was kindly provided by Astrid Limb (University College London, Institute of Ophthalmology). Cells were maintained in DMEM-glutamax (Invitrogen, Carlsbad, CA) containing 5.5 mM glucose supplemented with 10% fetal bovine serum (Biochrom, Berlin, Germany) and 0.1% penicillin/streptomycin (Invitrogen) under 5% CO₂. Cells were treated with 0.05% trypsin (Invitrogen) in PBS, resuspended, and divided into three when the plates reached confluence. For hyperglycemia studies, subconfluent cells were grown for 48 hours in Dulbecco's modified Eagle's medium (DMEM) with 25 mM glucose, and supplemented with 1 g/L transferrin, 96.6 μg/mL putresceine, and 300 nM sodium selenate. Media were changed every 24 hours to maintain the glucose concentration. 

For insulin treatment, subconfluent MIO-M1 cell cultures were serum starved overnight before treatment with 100 pM insulin (Sigma, St. Louis, MO) for the indicated times. At the end of incubation periods, cells were washed with ice-cold PBS containing protease and phosphatase inhibitor cocktails (Roche, Mannheim, Germany), scraped, and collected by centrifugation. Primary cultures of purified Müller glia were prepared from PN10 rat retina as described previously. ²⁹ Cells were allowed to grow to confluence in DMEM–10% fetal calf serum for RNA isolation. 

RNA Isolation and RT-PCR

RNA isolation and RT-PCR reactions were performed as described previously. ^30,31 Briefly, RNA from dissected retinas was extracted by acid guanidium isothiocyanide-phenol-choloform method. ³¹ Subsequent to DNase treatment, 1 μg of total RNA was reverse transcribed at 42°C for 2 hours with 200 units M-MLV reverse transcriptase (Promega, Madison, WI) in the presence of 1 unit/μL RNasin in the buffer supplied by the manufacturer. Amplification cycles consisted of 30 seconds of template melting at 95°C, annealing at 55°C, and extension at 72°C, and the amplification products were analyzed on 1% agarose gels. 

For quantitative PCR (qPCR) studies, reaction mixtures containing the primers for SIK2 and the reference gene β-actin were prepared in SYBR Premix Ex Taq (Takara, Otsu, Japan). Following a 2-minute initiation step at 50°C and denaturation at 95°C for 10 seconds, 35 cycles of amplification were done where each cycle included a denaturation step at 95°C for 5 seconds, an annealing step at 55°C for 10 seconds, and an extension step at 72°C for 10 seconds. Subsequent to the melting curve construction the data were analyzed by LinRegPCR software 7.2. ³²  

In all cases for amplification the primer pairs 5′-TTGCTGAACAAACAGTTGCC-3′ and 5′-TCAAGCAGACAGCCATTCAC-3′ for SIK2, and 5′-AAGATCAAGATCATTGCTCCTC-3′ and 5′-GGGTGTAACGCAACTAAGTC-3′ for β-actin were used. Reactions where the reverse transcriptase was omitted constituted negative controls. The results represent data from three different RNA samples and three reverse transcription reactions from each experiment. 

Northern Blot Analysis

Northern blots of 15 μg total retinal RNA were performed as described previously. ³⁰ Sense and antisense SIK2 probes were generated in vitro using 207 base pair (bp) RT-PCR fragment in the presence of digoxigenin (DIG)-labeled UTP using an RNA labeling kit (DIG RNA Labeling Kit; Roche). The membranes were washed for 30 minutes at the final stringency of ×0.1 SSC, 0.5% SDS at 60°C. Northern blots from two independent RNA preparations were performed. 

Immunostaining

For immunohistochemistry, 12 μm cryosections were rehydrated in PBS, permeabilized with 0.1% Triton X-100 in PBS for 5 minutes at room temperature, and preincubated in PBS containing 5% donkey serum, 1% glycine, 0.1% BSA, 0.1% Tween-20 for 1 hour. For SIK2 staining, sections were incubated with two independent affinity purified polyclonal antibodies, made either in rabbit hosts against the peptide spanning amino acid residues 600 to 650 of the human protein (Novus Biologicals, Cambridge, UK) or against recombinant human protein raised in mouse (Biolegend, San Diego, CA). In coimmunostaining studies, mouse monoclonal anti-CRALBP antibody (Abcam, Cambridge, UK) was used with the SIK2 antibody obtained from Novus Biologicals. Sections were incubated with the primary antibodies at 1:250 dilution overnight at 4°C. Subsequent to extensive washing with PBS, the sections were incubated with secondary antibodies, Alexa Fluor 555 conjugated antirabbit IgG (Invitrogen), and/or Alexa Fluor 488 conjugated antimouse IgG at 1:1000 dilution; diaminophenylindolamine (DAPI) also was included during this treatment. Samples were examined using a SP5-AOBS confocal microscope equipped with LAF software (LAS AF; Leica, Wetzlar, Germany). Images were optimized for color, brightness, and contrast, and double-labeled images overlaid by using Adobe Photoshop 6.0 (Adobe Systems, Inc., New York, NY). All control and test images were digitally enhanced in an identical manner. 

For immunostaining, cultured MIO-M1 cells were fixed in 4% PFA for 10 minutes, blocked with 1% donkey serum, 0.3% Triton X-100 in PBS for 1 hour. Subsequently, cells were incubated with anti-SIK2 antibodies as described above. Cells were examined on an Axio Observer Z1 inverted fluorescent microscope equipped with AxioVision Rel. 4.6 SP1 analysis system (Carl Zeiss AG, Jena, Germany). 

In Vitro Kinase Assay

To evaluate kinase activity in vitro, SIK2 was immunoprecipitated using antiserum, kindly donated by Hiroshi Takemori (National Institute of Biomedical Innovation, Osaka, Japan), as described by Horike et al. ²⁵ Briefly, protein A-agarose beads (Roche) were incubated with the serum preadsorbed with the insoluble fraction of Escherichia coli lysate at 4°C for 4 hours. Subsequent to extensive washing with PBS, a 50% suspension was prepared in lysis buffer containing 50 mM Tris-HCl (pH 8.0), 150 mM NaCl, 1% NP40 with protease, and phosphatase inhibitor cocktails (Roche). Lysates prepared from 10⁷ cells in the same buffer were incubated with the 1/100 of protein A-agarose slurry. The beads were collected, washed, and resuspended in the kinase buffer composed of 50 mM Tris-Cl, pH 7.4, 1 mM DTT, 10 mM MgCl₂, 10 mM MnCl_2. ³³ Half the samples were analyzed by Western blotting to assess the input. To the remaining samples, 1 μCi (γ ³² P)-ATP (3000 Ci/mmol; Isotop, Budapest, Hungary) and 500 ng of purified recombinant GST-IRS1 were added. The kinase reactions were allowed to proceed at 30°C for 1 hour, then terminated by the addition of the SDS sample buffer, and boiled at 95°C for 5 minutes. The beads were removed by centrifugation, the supernatants were run on 8% SDS-PAGE. The gels were fixed, dried, and exposed to x-ray film (Amersham, Buckinghamshire, UK) for varying times at −80°C. 

IRS1 Immunoprecipitation

Protein A–agarose beads (Roche), washed in 20 mM HEPES (pH 7.4), 150 mM NaCl, 0.1% Triton X-100, and 10% glycerol, were incubated with 8 mg of anti-IRS1 rabbit (Santa Cruz Biotechnology, Santa Cruz, CA) antibody for 2 hours at room temperature. Subsequently the beads were incubated with cell lysates, 500 μg of total protein, prepared in 50 mM Tris-Cl (pH 8.0), 300 mM NaCl, 5 mM EDTA, 5 mM EGTA, 2 mM DTT, and 0.5% Triton X-100 supplemented with protease and phosphatase inhibitor cocktail (Roche), for 4 hours at 4°C. The beads were washed with ice-cold HEPES buffer with protease/phosphatase inhibitor cocktail, adsorbed proteins were eluted by incubation at 75°C for 5 minutes in SDS sample buffer and subjected to Western analysis. 

Western Blot

The samples fractionated on polyacrylamide gels were electroblotted to polyvinyl difluoride membranes (Roche) as described previously. ²⁹ Following blocking with 1% BSA, 5% skimmed milk powder in 150 mM NaCl, 20 mM Tris-HCl, pH 8.0, 0.1% Tween 20 for 1 hour at room temperature, the membranes were incubated with one of the primary antibodies. Polyclonal SIK2 antibody (rabbit; Novus Biologicals) was used at 1:2500 dilution, anti-GFP (rabbit; Abcam) at 1:1000, anti-pAkt, rabbit or pan-Akt, rabbit (Cell Signaling Technologies, Danvers, MA) at 1:2000 and 1:1000 dilutions, respectively; pERK, ERK2, IRS1, phosphotyrosine antibodies were obtained from Santa Cruz Biotechnology and used at 1:1000 dilution. Subsequently, the membranes were washed and incubated with appropriate HRP-conjugated secondary antibodies. The bands were visualized using Lumi-light Western blotting substrate (Roche) for 1 minute, and images were captured using Stella gel imaging system (Raytest, Straubenhardt, Germany). 

Vector Constructs and Transfections

The human SIK2 coding sequence (Genbank ID: NM_015191.1) was obtained by RT-PCR using the primer pair AGATCTCGAGATGGTCATGGCGGATGGCCCGAG/AGAGGGCCCTAATTCACCAGGACATACCCGT, where the melting temperature was 61°C. The 2780 bp PCR product was subcloned to XhoI and ApaI restriction sites in pEGFP-C3 (Clontech, Mountain View, CA) or pcDNA4-HisMax vector (Invitrogen). A total of 5 × 10⁵ cells was transfected with 20 μg/mL of plasmid DNA using FugeneHD (Roche) according to the manufacturer's instructions. At 24 hours after transfection, cells were washed three times with PBS and insulin treatments were performed. The same vector was used to transfect HEK293 cells with the same protocol as above. The GFP-SIK2 fusion protein was purified by immunoprecipitation using anti-GFP antibody (Abcam) and used in antibody neutralization in immunostaining studies. 

An IRS1 fragment containing the SIK2 target residue (Genbank ID: NM_012969.1) was obtained by RT-PCR using the primer pair 5′-AGAGGATCCGGTGGTAAGCTCTTGCCTTGC-3′/5′-AGAGAATTCCTAGTTGGTCTGTGCAGCTGTGT-3′, cloned to BamH1 and EcoRI sites of pGEX-2TK-P (Amersham). The GST-IRS1 fusion protein was expressed in E. coli BL21, purified using glutathione-Sepharose 4B microspin columns (Amersham) following the manufacturer's instructions. 

TUNEL Assay

Approximately 3 × 10⁴ serum-starved MIO-M1 cells were treated with 100 pM insulin or vehicle (25 mM HEPES) for 24 hours. Apoptotic cell death was examined by TUNEL assay using the In Situ Cell Death kit (Roche), as per manufacturer's instructions. To visualize the nuclei, cells were incubated with DAPI for 5 minutes. The samples from three independent experiments were evaluated by fluorescent microscopy. 

SIK2 Gene Silencing

Human SIK2 sh-RNA lentiviral particles, a pool of three target-specific constructs, and scrambled sh-RNA were purchased from Santa Cruz Biotechnology. MIO-M1 cells were subjected to lentiviral infection as instructed by the manufacturer and propagated in the presence of puromycin (1 mg/mL; Sigma). After 2 weeks, colonies were isolated and ones showing SIK2 downregulation were stored at −80°C until further use. 

Statistical Analysis

Experimental groups were compared statistically using the Mann-Whitney test (one-tailed) or one-way ANOVA. Means with P < 0.05 were considered statistically significant. 

Results

SIK2 Is Widely Expressed in Adult Rat Retina

SIK2 expression in adult rat retina was studied by Northern and Western blot analyses. In Northern blots, the antisense probe revealed the presence of two transcripts of 7.5 and 8 kb, while no bands were detected with the sense probe (Fig. 1A). This finding is supported by the isolation of two overlapping SIK2 cDNAs encoding for two proteins that differ by 12 amino acids at the carboxyl terminus and within their respective 3′ UTRs, that were reported by our group previously (Uysal A. IOVS 2005;46:ARVO E-Abstract 3151). Two polypeptides with approximate apparent molecular weights of 120 and 100 kDa could be detected by anti-SIK2 antibody on Western blots (Fig. 1B). Prior incubation of the primary antibody with recombinant SIK2 abolished the signal. Thus, two different SIK2 isoforms resulting from alternatively spliced messages are present in the rat retina. 

Figure 1

 

Expression of SIK2 in rat retina and Müller cells. (A) Northern blot prepared with total RNA isolated from adult rat retina was probed with in vitro generated DIG-labeled anti-sense (lane 1) and sense (lane 2) SIK2 transcripts. (B) Western blot analysis of retinal lysates prepared from adult retina was done using anti-SIK2 and HRP-conjugated secondary antibodies (lane 1); in controls primary antibody was preincubated with recombinant SIK2 (lane 2). (C) RNA isolated from ONL, INL, entire retina (WR), and cultured Müller glia were subjected to RT-PCR analysis using SIK2 specific primers. 100 bp ladder was the size marker (M). (D) Retinal sections from adult animals were stained with anti-SIK2 (red) and anti-CRALBP (green) antibodies. The last panel represents staining with anti-SIK2 antibody preincubated with affinity purified recombinant GFP-SIK2 protein. White arrowheads point to cell bodies that are SIK2- and CRALBP-positive, yellow arrows indicate Müller cell extensions in ONL. Scale bar: 50 μm for all panels. (E) Subconfluent culture of MIO-M1 cells were stained with anti-SIK2 and DAPI. Scale bar: 10 μm.

SIK2 expression in retinal layers and in Müller glia was investigated by RT-PCR. An expected 207 bp SIK2 amplification product was obtained from the ONL and the INL as well as from cultured primary Müller cells generated from PN10 rat retina (Fig. 1C). In all cases the identity of the amplification products was verified by sequencing; in control samples lacking reverse transcriptase no amplification products were detected (data not shown). Immunochemical staining of SIK2, irrespective of the antibody used, demonstrated strong labeling in cell bodies within the INL and ganglion cell layer (GCL). Although immunoreactivity was not detectable in the ONL cell bodies, it clearly was seen enveloping some cells at this level (arrow, Fig. 1D). In INL there were strongly SIK2-positive cells that were devoid of CRALBP signal, suggesting some neuronal cells express SIK2 (Fig. 1D). Double label immunostaining with CRALBP and SIK2 antibodies showed coincidental expression in Müller cell bodies, albeit the intensity of SIK2 staining was less than in neurons (Fig. 1D). Immunocytochemical staining of MIO-M1 cells by anti-SIK2 antibody revealed predominantly cytoplasmic labeling (Fig. 1E). Preincubation of the primary antibody with recombinant SIK2 abolished staining of the retinal layers (Fig. 1D) and the cells (data not shown). 

SIK2 Activity and SIK2/IRS1 Interaction Are Modulated by Insulin

To investigate whether SIK2 activity was modulated in an insulin-dependent manner, SIK2 was immunopurified from MIO-M1 cells treated with insulin for 0 to 120 minutes and analyzed by in vitro kinase assays using GST-IRS1 as substrate. Densitometric readings of GST-IRS1 bands from the autoradiograms of samples resolved on SDS-PAGE were normalized to SIK2 input as evaluated by Western blots. The results indicated that SIK2 activity was doubled within 5 minutes after insulin addition, and in 10 minutes started to return to preinduction levels (Fig. 2A). In contrast, SIK2 protein levels remained constant during the first hour of insulin stimulation, but increased 1.6-fold at 120 minutes after induction (Fig. 2B), suggesting that SIK2 activity, rather than changes in SIK2 expression are affected by insulin stimulation, at least during the initial response phase. 

Figure 2

 

Insulin-dependent changes in SIK2 activity, Akt phosphorylation, SIK2 coimmunoprecipitation with IRS. Following overnight serum starvation cells were treated with 100 pM insulin for 0 to 120 minutes. (A) SIK2 immunoprecipitated from the lysates was used in kinase assays in the presence of (γ ³² P)-ATP, using purified GST-IRS1 as the substrate. Reaction products were fractionated on PAGE and visualized by autoradiography, SIK2 levels were assessed by Western blot analysis using anti-SIK2 antibody in matching samples. (B) To follow possible changes in SIK2 protein levels during the course of insulin treatment, Western blotting using anti-SIK2 antibody was carried out; β-Actin is the loading control. (C) Immunoprecipitates obtained from the lysates with anti-IRS1 antibody were analyzed by Western blotting with anti-SIK2 or anti-IRS1 antibodies. (D) The lysates were subjected to Western blot analysis with anti-pAkt(S473), anti-pAkt(T308), and anti-panAkt. The histograms represent in each case mean values of three independent biological samples after normalization of band intensities of GST-IRS1 to that of SIK2 bands in (A), SIK2 to β-actin in (B), and pAkt with that of pan-Akt in (D). *P < 0.05 compared to untreated samples.

SIK2-induced phosphorylation of IRS1 at S789 has been proposed to down-regulate insulin signaling in rat adipocytes. ²⁵ Although our in vitro kinase studies confirmed that SIK2 is able to phosphorylate IRS1 (Fig. 2A), due to the lack of an antibody that recognizes the human counterpart (S794 in human), we were unable to explore whether this residue was the cellular target. However, coimmunoprecipitation assays indicated that these two proteins interact in Müller cells and this interaction appeared to be enhanced with 10 minutes of insulin treatment (Fig. 2C). These observations lend support to the possibility that SIK2 is an upstream kinase of IRS1 in vivo. 

We next investigated the insulin-dependent Akt activation profile in the MIO-M1 cell line by Western blotting using antibodies specific to Akt phosphorylated at S473 and T308, the residues implicated in full activation of Akt. Relative phosphorylation levels at these sites did not show significant changes in the first 5 minutes of insulin treatment, but thereafter increased steadily, to reach a maximum by 60 minutes, representing a 4.5-fold increase over the basal level. At 120 minutes after induction, although pAkt levels declined significantly, they remained 3-fold higher than in untreated samples (Fig. 2D). In these same cells, pERK levels did not show statistically significant changes in the course of insulin treatment (data not shown). 

The activation profiles obtained revealed that SIK2 activation precedes that of Akt and argued for SIK2 being upstream of Akt in insulin signaling in the context of MIO-M1 cells. 

SIK2 Negatively Regulates Insulin-Dependent Akt Activation in MIO-M1 Cells

The potential involvement of SIK2 in insulin-dependent Akt activation in glial cells was investigated further by overexpression and sh-RNA–mediated gene knockdown. In overexpression experiments, MIO-M1 cells either were left untreated or treated with insulin for 60 minutes for maximal Akt phosphorylation to occur. Insulin treatment of mock-transfected cells resulted in approximately a 4-fold increase in pAkt levels (Fig. 3A), consistent with our previous results in untransfected cells. However, overexpression of SIK2 in parallel cultures resulted in pAkt levels showing only a 1.5-fold increase after 60 minutes following insulin treatment; that is, a 60% decrease compared to untransfected control cells (Fig. 3A, lanes 2 and 4). In contrast, when SIK2 was downregulated by sh-RNA, a 50% decrease in SIK2 levels resulted in a 4-fold upregulation of Akt phosphorylation within 5 minutes of insulin exposure as compared to cells treated with scrambled sh-RNA (Fig. 3B). 

Figure 3

 

Effect of SIK2 overexpression or silencing on insulin-dependent Akt activation. (A) MIO-M1 cells were transfected transiently with GFP-SIK2 or with empty vector. (B) Knockdown of SIK2 gene was performed by infecting cells with lentiviral particles carrying sh-SIK2, control samples received scrambled (Src) sh-RNA. The cells either were untreated or treated with 100 pM insulin for the indicated times. pAkt levels were evaluated by antipAkt(S473); anti-panAkt and anti-β-actin used as the loading, anti-GFP or anti-SIK2 as the transfection control. Results represent data from three independent biological samples, and the histograms represent relative mean pAkt(S473) band intensities normalized to that of pan-Akt. *P < 0.05 compared to the untreated samples.

Earlier reports indicate that insulin-dependent Müller cell survival is mediated via the IRS1/Akt pathway, but that the ERK pathway is not involved. ^20,34,35 In line with these results, insulin treatment for 24 hours led to a 20-fold decrease in the percentage of apoptotic MIO-M1 cells compared to untreated cultures, as evaluated by TUNEL assay (Fig. 4A). Insulin addition did not stimulate proliferation (data not shown). However, in MIO-M1 cells overexpressing SIK2, insulin treatment only reduced the apoptotic loss by 5-fold (Fig. 4B). In other words, elevated SIK2 expression actually hindered the antiapoptotic action of insulin by a factor of four. The rate of cell death in mock-transfected cells was similar to that of untransfected cells (Fig. 4B). 

Figure 4

 

Effect of SIK2 overexpression on insulin-dependent cell survival. After serum starvation overnight, cells either were treated with 100 pM insulin for 24 hours or left untreated. Apoptotic cells were assessed by TUNEL assay, nuclei were stained with DAPI. (A) The effect of insulin on cell death of untransfected MIO-M1 was evaluated; the graph represents mean values of percent apoptotic cells (±SE). (B) MIO-M1 cells overexpressing SIK2 or cells carrying the vector alone were used, the histograms show the mean ratio of per cent apoptotic cells (±SE) in the presence and absence of insulin. In all cases 3 independent experiments were carried out, *P < 0.05 for treated versus untreated cells in (A), and for SIK2 overexpressing samples compared to the mock-transfected ones in (B). Scale bar: 50 μm.

Taken together, these results are consistent with SIK2 being a negative modulator of insulin-dependent Akt activation and enhancing apoptosis in Müller cells. 

Chronic Hyperglycemia Attenuates Insulin-Induced Akt Signaling and Leads to Enhanced SIK2 Activity and Expression

Several lines of evidence have suggested that chronic hyperglycemia enhances apoptosis via Akt inactivation in a wide range of cell types, including Müller cells. ^23,36,37 Thus, we next explored SIK2 involvement in hyperglycemia-induced Akt inactivation and apoptosis in MIO-M1 cells. 

Our initial experiments indicated that in cells grown under chronic hyperglycemia the basal levels of pAkt(S473) were decreased 40%, and the presence of insulin was not sufficient to restore Akt activation to normoglycemic levels (Fig. 5A). In contrast with Akt, SIK2 activity was doubled and its transcript levels were increased 50% when the cells were exposed to high glucose (Figs. 5B, 5C). Again the addition of insulin did not affect these parameters. 

Figure 5

 

Effect of hyperglycemia on Akt phosphorylation, SIK2 activity, and transcript levels. MIO-M1 cells cultured in the presence of 5.5 mM or 25 mM glucose for 2 days were treated with 100 pM insulin, and compared to the untreated samples. (A) Lysates were immunoblotted with anti-pAkt(S473) and anti–pan-Akt antibodies. (B) SIK2 immunoprecipitated from the lysates was used in kinase assays. GST-IRS1 kinasing was visualized by autoradiography following PAGE. SIK2 levels in matching samples were evaluated by Western blotting using anti-SIK2 antibody. (C) Relative SIK2 transcript levels were evaluated by qRT-PCR using primers specific to SIK2 and β-actin. Data represent the mean ± SE of three independent experiments where in pAkt levels normalized to that of Akt, GST-IRS1 band intensities to the input SIK2. *P < 0.05.

To investigate further the observed inverse correlation between Akt and SIK2 activation under chronic hyperglycemia, cells were transfected with GFP-SIK2 or GFP alone as control, with a transfection efficiency of 35%. In mock-transfected cells exposed to hyperglycemic conditions, pAkt levels were reduced by 25% and apoptosis was increased 2-fold (Figs. 6A, 7). SIK2-overexpressing cells cultured in normal glucose concentrations showed a 45% decline in Akt phosphorylation and apoptosis was enhanced 2.5-fold. On the other hand, SIK2 silencing under normo- and hyperglycemic conditions resulted in a 2.5-fold increase in pAkt levels (Fig. 6B). Importantly, the reduced Akt phosphorylation seen under hyperglycemia was reversed by SIK2 silencing. Therefore, we proposed that, under diabetic conditions, increased SIK2 activity and/or protein levels contribute to hyperglycemia-mediated impairment of Akt activation. 

Figure 6

 

Effect of SIK2 overexpression or silencing on pAkt levels of Müller cells grown under normal or hyperglycemic conditions. (A) Cells were transfected with GFP-SIK2 or empty GFP vector. (B) Knockdown of SIK2 gene was done by infecting cells with lentiviral particles carrying sh-SIK2 or scrambled sh-RNA. The cells either were grown under 5.5 mM or 25 mM glucose conditions for 2 days. pAkt levels were evaluated by anti-pAkt(S473) antibody. Anti–pan-Akt antibody was used as loading, anti-GFP for transfection and anti-SIK2 for knockdown control. Relative Akt phosphorylation levels were plotted on the graph after pAkt(S473) band intensities were normalized to that of corresponding pan-Akt. Data represent the mean ± SE of three independent experiments. *P < 0.05 compared to the cells grown in 5.5 mM glucose.

Figure 7

 

Effect of SIK2 overexpression on MIO-M1 cell apoptosis. MIO-M1 cells overexpressing SIK2 were grown under hyperglycemic conditions for 2 days. Mock transfected cells grown under normal glucose conditions constituted the control. Apoptotic cells were assessed by TUNEL assay; nuclei were stained with DAPI. The graph represents apoptotic cell numbers in percent of total cell numbers. *P < 0.05 compared to the mock-transfected cells grown in 5.5 mM glucose.

To assess whether SIK2 activity and expression could be modulated by hyperglycemia in vivo, we used the streptozotocin-induced diabetic rat model and focused on a very early stage, 2 weeks after treatment, where cell death ^35,38–40 and metabolic changes ^16,41,42 are limited. On the day of sacrifice the fasted blood glucose levels of the experimental animals were roughly three times that of controls (Fig. 8A). While we observed no change in steady state SIK2 transcript levels by qRT-PCR (Fig. 8B), SIK2 activity was significantly increased in hyperglycemic animals as assessed by an in vitro kinase assay using IRS1 as substrate (Fig. 8C). However, no differences in immunostaining with anti-SIK2, nor in TUNEL-positive cell numbers, were evident between the groups at this time point (data not shown). 

Figure 8

 

SIK2 activity and transcript level change in STZ injected rat retinae. Male albino Wistar rats received streptozotocin (diabetic group), or vehicle alone (control group) intravenously were sacrificed after 2 weeks. (A) Average weight and fasted blood glucose level of rats before and after the injections were measured. (B) Relative SIK2 transcript levels were evaluated by qRT-PCR using primers specific to SIK2 and β-actin. Data represent the mean ± SE of three independent experiments. (C) SIK2 immunoprecipitated from the retina lysates were used in kinase assays. Reaction products were fractionated on PAGE visualized by autoradiography (top). SIK2 levels in matching samples were evaluated by Western blot analysis using anti-SIK2 antibody as a control (bottom).

Discussion

Current data suggest that serine/threonine phosphorylation of IRS proteins has a central role in the regulation of insulin signaling, and perturbations at this level may be critical in the development of diabetes and insulin resistance. ⁴³ Studies indicate that the neurotrophic effects of insulin on retinal neurons involve IRS1/2 phosphorylation and activation of the PI3K/Akt axis. ^9,18,44–46 In Müller cells, Akt activation via IRS1 mediates insulin-induced survival responses, ^18,19 but detailed information on the regulation of the pathway is lacking. 

In our study we showed expression of the recently identified IRS1 upstream serine kinase SIK2²⁵ in neurons and Müller cells of rat retina. Our subsequent in vitro studies revealed that insulin treatment of Müller glia results in rapid and transient activation of SIK2. Coimmunoprecipitation assays showed, for the first time to the best of our knowledge, that endogenous SIK2 and IRS1 interact directly, and this interaction is enhanced by insulin stimulation. These findings lend support to the hypothesis that IRS1 is an in vivo SIK2 substrate in insulin signaling in Müller cells. The temporal profiles of SIK2 and Akt activation in Müller glia indicate that SIK2 functions upstream of Akt in this pathway. sh-RNA–mediated SIK2 knockdown studies, in which insulin-dependent Akt activation occurred earlier corroborated this idea and suggested that rapid insulin-induced SIK2 activity delays full Akt activity under normal conditions. In line with these results, pAkt levels remained close to baseline at all time points in Müller cells overexpressing SIK2, and insulin-induced cell survival was reduced significantly. Based on these observations we proposed that SIK2 acts as a negative modulator of the insulin-induced IRS1/Akt-mediated Müller glial survival pathway. The physiologic significance of an early and transient activation of a negative modulator of the pathway is not clear at this time. We speculated that this modification is a component of an IRS1 “phosphorylation code” and might be instrumental for the integration of different signaling events. For instance, SIK2-dependent phosphorylation might be priming IRS1 for a final inhibitory readout from stress, nutrient sensing and inflammatory cytokine cascades. ⁴⁷  

It has been reported previously that SIK2 activation also is required for Akt phosphorylation in ovarian cancer cell lines during cell cycle progression. ⁴⁸ However, in hepatocytes and adipocytes where insulin mainly regulates gluconeogenesis and lipogenesis, respectively, SIK2 was proposed to be an Akt substrate. ^49,50 These differences may stem from alternative roles played by SIK2 in Müller cells and the core insulin-responsive tissues, or variations in the time-frames that have been examined. 

Increased Müller cell apoptosis and decreased Akt activity have been reported in diabetic animal models, ^21,22 and Müller glia apoptosis via Akt downregulation under hyperglycemic conditions has been shown in vitro. ²³ We observed an attenuation of basal Akt activation in Müller cells cultured under chronic hyperglycemia, and this could not be reversed by addition of exogenous insulin, suggesting development of insulin resistance. However, the same hyperglycemic conditions, irrespective of insulin addition, resulted in an upregulation of SIK2 activity and expression. We also showed that pAkt levels could be restored to control levels by SIK2 gene silencing during hyperglycemia, and that overexpression of SIK2 in normoglycemic cells decreases Akt activation and increases cell death. These observations provide evidence that modulation of SIK2 activity may be an important component during the development of insulin resistance and Müller cell apoptosis. However, additional experiments will be necessary to show whether SIK2 has a similar role in vivo, especially under diabetic conditions. 

It is widely accepted that IRS1 serine phosphorylation after prolonged exposure to high glucose, coupled with the consequent failure of tyrosyl-phosphorylation or enhanced degradation of IRS1, results in attenuation of insulin signaling and so-called insulin resistance. ^51–59 SIK2 has been implicated in this process via phosphorylation of IRS1 on Ser789 in adipocytes of diabetic animals. ²⁵ Although we were not able to test IRS1 phosphorylation at this residue due to the lack of an available antibody, we observed a decrease in IRS1 protein levels and its tyrosyl phosphorylation in Müller cells under hyperglycemia (data not shown). Thus, it is possible that increased SIK2 activity/levels during hyperglycemia enhance IRS1 serine phosphorylation, which may interfere with its tyrosine phosphorylation and/or modulate its stability, thereby preventing Akt activation. It also is possible that the downregulation of the PI3K-Akt axis may be due to altered scaffolding on IRS proteins. In a recent study, IRS1 was shown to regulate specific apoptotic markers in rat Müller cells exposed to hyperglycemic conditions. ²⁰ Currently, we are investigating whether SIK2-mediated IRS1 phosphorylation also modulates the expression of pro- and antiapoptotic genes. In Müller glial cells cultured under hyperglycemic conditions, TNFα was reported to inhibit insulin signaling and induce apoptosis by phosphorylating IRS1 on Ser307. ³⁷ Therefore, it is conceivable that in these cells SIK2 may contribute directly to reduced survival, together with TNFα and other factors. Our results obtained with STZ-injected animals indicated that SIK2 activity is increased during early diabetes, at a time when hyperglycemic conditions are installed, but before they result in retinal damage or cell death. Whether this early dysregulation can be correlated with downregulation of Akt and later increased Müller cell death in vivo, or during the etiology of DR, awaits further investigation. 

We have shown that SIK2 negatively regulates the ability of insulin to promote cell survival through Akt phosphorylation in Müller cells grown under normal conditions and chronic hyperglycemia. Because SIK2 also is expressed in neurons, especially retinal ganglion cells, it is conceivable this kinase also may participate in the regulation of Akt-mediated survival of retinal neurons. Knockdown studies implicated SIK2 in TORC1/CREB-dependent survival of cortical neurons during ischemia. ⁶⁰ It would be interesting to investigate the contribution of the SIK2/TORC1/CREB-mediated pathway to neuronal as well as Müller glia survival in addition to the SIK2/IRS1/Akt axis that was studied here. Though the current findings were derived from cultured Müller cells, it is tempting to speculate that inappropriate SIK2 activation may contribute to retinal dysfunction in diabetes in vivo. Further experiments will demonstrate whether such a link exists, and whether this kinase might represent a new potential target molecule for the treatment and prevention of diabetic retinopathy. 

Acknowledgments

The authors thank David Hicks, PhD (Université de Strasbourg, France), for valuable guidance in immunohistochemistry and advising on this manuscript, Bahar Sahin, PhD (Bogazici University, Istanbul), for excellent technical assistance on confocal imaging, and Arzu Temizyürek for animal care. Authors also thank Bekir Cinar, PhD (Cedars-Sinai Medical Center, Los Angeles, CA), and Arzu Celik, PhD (Bogazici University, Istanbul), for critically reading the manuscript. 

Supported by Bogazici University Research Projects Grants 08B104D and 13B010D and Turkish Scientific and Research Council Grants 103T063 and 108T646. 

Disclosure: G. Küser-Abalı, None; F. Ozcan, None; A. Ugurlu, None; A. Uysal, None; S.H. Fuss, None; K. Bugra-Bilge, None 

Footnotes

 Supported by Bogazici University Research Projects (08B104D; 13B010D); Turkish Scientific and Research Council (103T063; 108T646).