Rapid Changes of mRNA-binding Protein Levels following Glucose and 3-Isobutyl-1-methylxanthine Stimulation of Insulinoma INS-1 Cells ^,

Abstract

Glucose and cAMP-inducing agents such as 3-isobutyl-1-methylxanthine (IBMX) rapidly change the expression profile of insulin-producing pancreatic β-cells mostly through post-transcriptional mechanisms. A thorough analysis of these changes, however, has not yet been performed. By combining two-dimensional differential gel electrophoresis and mass spectrometry, we identified 165 spots, corresponding to 78 proteins, whose levels significantly change after stimulation of the β-cell model INS-1 cells with 25 mm glucose + 1 mm IBMX for 2 h. Changes in the expression of selected proteins were verified by one- and two-dimensional immunoblotting. Most of the identified proteins are novel targets of rapid regulation in β-cells. The transcription inhibitor actinomycin D failed to block changes in two-thirds of the spots, supporting their post-transcriptional regulation. More spots changed in response to IBMX than to glucose alone conceivably because of phosphorylation. Fourteen mRNA- binding proteins responded to stimulation, thus representing the most prominent class of rapidly regulated proteins. Bioinformatics analysis indicated that the mRNA 5′- and 3′-untranslated regions of 22 regulated proteins contain potential binding sites for polypyrimidine tract-binding protein 1, which promotes mRNA stability and translation in stimulated β-cells. Overall our findings support the idea that mRNA-binding proteins play a major role in rapid adaptive changes in insulin-producing cells following their stimulation with glucose and cAMP-elevating agents.

Pancreatic β-cells store insulin within secretory granules (SGs).1 Hyperglycemia triggers the fusion of SGs with the plasma membrane and the extracellular release of insulin, which in turn lowers glycemia by promoting glucose uptake into cells. In addition to insulin secretion, glucose promotes the biogenesis of SGs by enhancing the synthesis of their components, including preproinsulin, prohormone convertases 1/3 (PC1/3) (1, 2) and 2 (PC2) (2), chromogranin A (3), and ICA512 (4). This rapid up-regulation of SG biogenesis is largely attributed to post-transcriptional mechanisms because it is insensitive to the transcription blocker actinomycin D (AmD) (5–9). These post-transcriptional mechanisms include reduced degradation of mRNAs encoding SG components (10–13), recruitment of these mRNAs from the cytosol to the endoplasmic reticulum (14), and increased translation (15–18). Elevation of cAMP levels also increases stability and translation of mRNAs encoding insulin SG proteins (12, 17). A key factor for glucose/cAMP-mediated up-regulation of mRNA stability and translation is polypyrimidine tract-binding protein 1 (PTBP1), formerly known as PTB1 or heterogeneous nuclear ribonucleoprotein I (hnRNP I) (11, 12, 19, 20). PTBP1 was identified in 1989 based on its ability to bind polypyrimidine tracts of pre-mRNAs and has multiple functions (21). In the nucleus it regulates pre-mRNA splicing (22–25) and poly(A) site cleavage (26). In the cytoplasm, it has been shown to regulate cap-independent translation through the internal ribosome entry site (27–29), mRNA localization (30, 31) and the stability of mRNAs for CD154 (32, 33), inducible nitric-oxide synthase (34), insulin, and other SG proteins (11, 12, 19, 35). Alternative splicing of the Ptbp1 transcript generates different isoforms (36), the largest of which measures 59 kDa (20) and contains four RNA recognition domains.

In this study we examined the proteomic changes occurring shortly after stimulation of INS-1 cells, an in vitro model of β-cells, with glucose and/or the cAMP-elevating agent 3-isobutyl-1-methylxanthine (IBMX). To this aim, protein samples were analyzed by mass spectrometry following their separation by fluorescent two-dimensional (2-D) DIGE (37–39). This method facilitates quantitative comparisons of samples by labeling proteins prior to 2-D electrophoresis with dyes that differ in fluorescence spectra, such as Cy3 and Cy5.

2-D DIGE routinely allows the separation of >2,000 and >1,600 spots in the range of pH 4–7 and pH 6–9, respectively, for a total number of >3,000 distinct spots. For comparison, 800–2,500 spots over the wider range of pH 3–10 are typically resolved in proteomics studies on isolated islets that rely on non-fluorescent dyes for protein staining (40–42). 2-D DIGE is also preferable to other procedures such as silver staining because of the greater linearity, sensitivity (0.025 ng), and wide dynamic range of the fluorescence signal (39). Using this approach, we identified mRNA-binding proteins as a major class of molecules whose expression pattern rapidly changes in response to glucose and IBMX stimulation.

MATERIALS AND METHODS

Cell Culture and Stimulation of INS-1 Cells

Rat insulinoma INS-1 cells were grown as described previously (43). Cells in 75-cm² flasks were preincubated in resting medium (15 mm HEPES, pH 7.4, 5 mm KCl, 120 mm NaCl, 24 mm NaHCO₃, 1 mm MgCl₂, 2 mm CaCl₂, 0 mm glucose, 1 mg/ml ovalbumin) for 1 h before stimulation for 2 h by addition of fresh medium containing 25 mm glucose and/or 1 mm IBMX (Sigma). Transcription was blocked using 5 μg/ml actinomycin D (AppliChem, Darmstadt, Germany), which was added to both the resting and stimulating media as indicated.

Transfection of INS-1 Cells

The cDNA of rat PTBP1 in INS-1 cells was cloned into pcDNA3.1 (Invitrogen) as described previously (12). INS-1 cells were transiently transfected with cDNA vectors using a Laboratory PulseAgile Electroporation System (Model PA-3000, Cyto Pulse Sciences, Inc., Glen Burnie, MD). Cells were harvested by trypsinization of a 175-cm² confluent flask (sufficient for ∼4 transfections) followed by centrifugation (500 × g, 5 min, room temperature). The cell pellet was resuspended in 250 μl of Cytoporation Medium (Cyto Pulse Sciences, Inc.). For overexpression of PTBP1, 4 μg of pcDNA3.1-PTBP1 (PTBP1-V5) and pcDNA3.1 (empty control vector) were added before cells were transferred into an electroporation cuvette (Eppendorf AG, Hamburg, Germany; 4-mm gap). The electroporation program was performed twice in an interval of 1.5 min (560 V; 0.2-μs pulse width; 0.2-s pulse interval four times). After electroporation, 600 μl of INS-1 cell medium (43) was added, and cells were seeded in two to three wells of a 6-well plate covered with sterile coverslips. Seventy-two hours after transfection, 10 mm sodium butyrate (Sigma) was added to fresh medium, and cells were harvested 24 h later.

RNA Interference (RNAi)

Knockdown of rat Ptbp1 mRNA in INS-1 cells by RNAi was performed using the pGENEClip U1 Hairpin vector (Promega, Madison, WI). The hairpin 5′-TCTCGTCATGGAAGAGTGTAAATTTCAAGAGAATTTACACTCTTCCATGACCT-3′ was designed as described previously (12). Four micrograms of the pGENEClip-PTBP1 or the scrambled pGENEClip control (5′-TCTCGTGAAATAGAGTGTAGGAATTCAAGAGATTCCTACACTCTATTTCACCT-3′) vectors were independently electroporated in INS-1 cells as described above.

Cell Extract Preparation

For 2-D gel electrophoresis (2-DE), cells were washed poststimulation three times with ice-cold PBS (Dulbecco's PBS, 1× without Ca²⁺ and Mg²⁺) (PAA Laboratories GmbH, Pasching, Austria) and twice with ice-cold sucrose buffer (250 mm sucrose, 10 mm Tris-HCl, pH 7.5). Cells were scraped in sucrose buffer, pelleted by centrifugation (500 × g, 5 min, 4 °C), and then lysed in 7 m urea, 2 m thiourea, 4% (w/v) CHAPS, 30 mm Tris-HCl, pH 9.3, 1× protease inhibitor mixture (GE Healthcare). Cell lysates were further passed through needles of decreasing sizes (18, 22, and 25 gauge). Interfering DNA was broken by sonication (5 × 10 s, 2-min break). The lysates were then centrifuged at 15,800 × g for 10 min at 4 °C to remove insoluble material. Protein concentration was measured using the RC DC protein assay (Bio-Rad). Proteins were used for 2-DE or shock frozen in liquid nitrogen and stored at −80 °C.

For other applications, cells were washed with ice-cold Dulbecco's PBS, scraped in PBS, pelleted by centrifugation (500 × g, 5 min, 4 °C) and then lysed in 20 mm Tris-HCl, pH 8.0, 140 mm NaCl, 1 mm EDTA, 1% (v/v) Triton X-100, 1× protease inhibitor mixture P8340 (Sigma), 1× phosphatase inhibitor mixture set II (Merck). Cell lysates were incubated for 10–30 min on ice and then centrifuged at 23,800 × g for 10 min at 4 °C to remove insoluble material. Protein concentration was measured using the BCA protein assay kit (Pierce). Proteins were frozen and stored at −80 °C or directly used for SDS gel electrophoresis after adding 6× SDS sample buffer (350 mm Tris-HCl, pH 6.8, 36% (v/v) glycerol, 12% (w/v) SDS, 600 mm DTT, 0.12% (w/v) bromphenol blue).

2-D DIGE

The minimal labeling of protein samples with CyDye DIGE Fluors Cy2, Cy3, and Cy5 (Ettan™ DIGE, GE Healthcare) was performed according to the manufacturer's instructions. Briefly, 50 μg of protein extracts from resting and stimulated cells was labeled with 200 pmol of Cy3 and Cy5, respectively. All experiments were independently repeated six times using different samples. To ensure the same labeling efficiency, three of six experiments were labeled in the opposite manner. Fifty micrograms of protein resulting from the mixture of resting and stimulated cell extracts in an equal ratio was labeled in parallel with 200 pmol of Cy2 as an internal control for normalization (38, 39, 44). After labeling, the proteins were reduced with 50 mm DTT, and the differentially labeled samples were mixed and supplemented with 0.5% (v/v) ampholytes (Bio-Lyte 3/10 ampholyte, 40% (Bio-Rad) for pH 4–7; IPG Buffer pH 6–11 (GE Healthcare) for pH 6–9). To better resolve protein spots and minimize their chance of co-migration, we used the longest available IPG strips (24 cm) with overlapping narrow range pH (pH 4–7 and 6–9) (45). For separation on Immobiline™ DryStrip Gels pH 4–7 (IPG strips; 24 cm, GE Healthcare) labeled samples were brought to a final volume of 450 μl with rehydration buffer (7 m urea, 2 m thiourea, 4% (w/v) CHAPS, 50 mm DTT) and passively loaded during rehydration of the strips. For separation on IPG strips pH 6–9, labeled samples were instead brought to a final volume of 80 μl with rehydration buffer and “cup-loaded” on strips already rehydrated in 7 m urea, 2 m thiourea, 2% (w/v) CHAPS, 5% (v/v) glycerol, 0.5% (v/v) IPG Buffer pH 6–11, 12 μl/ml DeStreak (GE Healthcare)). First dimension IEF was performed using the Ettan IPGphor IEF II™ unit (GE Healthcare). Focusing conditions were as follows: pH 4–7: (i) 0.5 h, linear 0–150 V; (ii) 1.5 h, 150 V; (iii) 1 h, 250 V; (iv) 4 h, linear 250–1,000 V; (v) 1.5 h, linear 1,000–5,000 V; (vi) 2 h, linear 5,000–10,000 V; and (vii) 8 h, 10,000 V; pH 6–9: (i) 4 h, 150 V; (ii) 3 h, 300 V; (iii) 6 h, linear 300–10,000 V; and (iv) 7 h, 10,000 V at 20 °C. According to the pH, a maximum of 50 μA (pH 4–7) or 30 μA (pH 6–9) was applied to the IPG strip. After IEF, IPG strips were first soaked with equilibration buffer (50 mm Tris-HCl, pH 8.8, 6 m urea, 30% (v/v) glycerol, 2% (w/v) SDS) supplemented with 20 mg/ml DTT for 20 min and then in new equilibration buffer supplemented with 25 mg/ml iodoacetamide for additional 20 min. The equilibrated IPG strips were transferred on the top of 10% polyacrylamide gels for the second dimension, which was run at 1 watt/gel at 20 °C using an Ettan DALTsix electrophoresis unit (GE Healthcare).

Image Analysis

DIGE-labeled gels were scanned with a Typhoon 9410 Variable Mode Imager (GE Healthcare) using excitation/emission wavelengths specific for Cy2 (488/520 nm), Cy3 (532/580 nm), and Cy5 (633/670 nm). For statistical analysis, DeCyder Differential Analysis Software with the Biological Variation Analysis (BVA) module version 5.0 (GE Healthcare) was used. Spots were automatically detected, matched, and normalized to the internal standard labeled with Cy2. Afterwards spots were manually checked to guarantee correct matching across the gels. The 18 images from the corresponding six gels for each condition and pH range were used to calculate the average ratio. Only spots that were detected in at least four of six gels and showed a threshold limit of 1.5-fold difference and a Student's t test of 99% (p ≤ 0.01) were regarded to differ significantly. For each comparison of condition sets a separate BVA setup was used.

Protein Identification by MS

For MS, 1–1.5 mg of protein extracts was separated by 2-DE, and proteins were stained with colloidal Coomassie Brilliant Blue G-250 (Bio-Rad) as described by Kang et al. (46). Protein spots were excised from gels, processed, and digested with 50–100 ng of trypsin (Promega), and peptides were extracted as described previously (47). MALDI-TOF spectra were obtained using an Ultraflex MALDI-TOF/TOF mass spectrometer (Bruker Daltonics, Bremen, Germany) in the reflectron mode and α-cyano-4-hydroxycinnamic acid as matrix. MALDI-TOF/TOF measurements were carried out in the LIFT mode. Peptide mass fingerprint spectra were internally calibrated with trypsin autolysis peaks. All spectra were processed, and peak lists were generated using flexAnalysis software (version 2.2) and the following parameters: signal-to-noise threshold of 5 and exclusion of contaminant ion masses as given in supplemental Table 1. Peptide mass mapping and fragment ion analysis were performed using the on-line available Mascot version 2.2 (Matrix Sciences Ltd., London, UK) (48). The following search criteria were used: (i) taxonomy, Rattus norvegicus; (ii) enzyme specificity, trypsin; (iii) mass accuracy, 50 ppm and 0.5 Da for peptide mass fingerprinting and fragment ion analysis, respectively; (iv) fixed and variable modifications, cysteine carbamidomethylation and methionine oxidation, respectively; (v) maximum of one missed cleavage site; and (vi) databases, National Center for Biotechnology Information (NCBI) version 20080221 (6,122,577 sequences; 2,096,230,148 residues; Rattus: 68,243 sequences; February 27–29, 2008) and NCBI version 20080229 (6,251,073 sequences; 2,135,462,495 residues; Rattus: 68246 sequences; March 3–4, 2008) (as indicated). Proteins were considered as identified if the peptide mass fingerprint exhibited a significant Mascot score (score > 61; p < 0.05).

Western Blot

Protein extracts separated by 2-DE or one-dimensional SDS-PAGE were transferred onto nitrocellulose membrane. Proteins were immunoblotted with the following antibodies: mouse monoclonal anti-PTBP1 (Zymed Laboratories Inc., South San Francisco, CA); rabbit polyclonal anti-phospho-PTBP1 (12); rabbit polyclonal anti-hnRNP K (H-300) and rabbit polyclonal lamin A/C (H-110) (Santa Cruz Biotechnology Inc., Santa Cruz, CA); mouse monoclonal anti-hnRNP A1 (ab50949) and rabbit polyclonal anti-hnRNP A3 (ab10685) (Abcam, Cambridge, UK); mouse monoclonal anti-chromogranin A (CGA) (BD Biosciences); rabbit polyclonal anti-PC1/3, anti-PC2, and anti-carboxypeptidase E (CPE) (Millipore, Billerica, MA); mouse monoclonal anti-PAI-RBP1 and mouse monoclonal anti-PCBP2 (M07) (Abnova Corp., Taipei, Taiwan); mouse monoclonal anti-ICA512 (49); mouse monoclonal anti-γ-tubulin (Sigma); polyclonal rabbit anti-N-CBF-A (CArG binding factor-A) (a gift from T. Leanderson); and polyclonal rabbit anti-Staufen2 (a gift from M. Kiebler). Blots were incubated with horseradish peroxidase-conjugated goat anti-mouse or goat anti-rabbit IgG (Bio-Rad), developed with SuperSignal West Pico Chemiluminescent Substrate or SuperSignal West Femto Maximum Sensitivity Substrate (Pierce) according to the manufacturer's instructions, and visualized using a LAS-3000 imaging system (Fuji Film Co. Ltd., Tokyo, Japan). Alternatively the ECL Plex Western Blotting System (GE Healthcare) containing secondary antibodies conjugated with CyDyes was used for protein detection. The secondary antibody signal was detected by scanning the membrane with a Typhoon 9410 Variable Mode Imager.

Computational Prediction of PTBP1 Binding Sites

Collection of Sequences—

Regulated proteins of rat INS-1 cells were mapped to their NCBI Entrez Gene identifiers. Human and mouse orthologs were obtained using the InParanoid database (50). NCBI Entrez Nucleotide was then queried for available mRNA sequences associated with the rat, mouse, and human genes. 5′- and 3′-untranslated regions (UTRs) as defined in Entrez Nucleotide were separated, and the 3′-UTR poly(A) tail was removed. On average, seven 5′-UTR and eight 3′-UTR sequences were retrieved per protein and species. The average length of the 5′- and 3′-UTR (without poly(A) tail) was 144 and 654 nucleotides, respectively.

Conservation—

To identify conserved regions, a multiple sequence alignment of rat, mouse, and human sequences was constructed for each UTR of a regulated protein using the multiple alignment program MAFFT with the L-INS-i iterative refinement parameter (51). A consensus sequence was calculated from residues that were at least 50% conserved in the multiple sequence alignment. For each conserved position of the alignment, the number of species showing the conserved amino acid was recorded. Positions conserved in only one species were excluded from further analyses.

Search for PTBP1 Binding Sites—

Each UTR was screened for the presence of the motif CYYYYCYYYYYG, corresponding to the consensus for PTBP1 binding according to Tillmar et al. (19). Hits were counted whenever a match was found considering zero, one, or two mismatching nucleotides. Overlapping hits were counted separately, allowing a continuous binding region to give rise to several hits.

Significance of Detected Binding Sites—

To assess the significance of hits matching the PTBP1 binding motif, we calculated the p value for each binding site found in a UTR. To this end, each UTR sequence was shuffled 100,000 times and subsequently searched for the binding motif. Shuffling ensured that the nucleotide composition was preserved. The frequency of observing at least the same number of hits in the random sequences as in the original sequence was recorded. The p value was calculated as this frequency divided by 100,000. It resembles the probability that the same number of hits observed occurs by chance. Hits with a p value <0.01 were considered to be significant.

Secondary Structure Assessment—

Because structural studies show that PTBP1 RNA recognition motif domains bind single-stranded RNA (52), we predicted the secondary structure of consensus UTR sequences using RNAfold (53), which calculates the minimum free energy structure of a given RNA sequence. We mapped binding site hits onto the predicted secondary structure. A binding site was counted as single-stranded if more than half of its nucleotides were non-paired in the minimum free energy structure. Significant hits were manually checked for their secondary structure.

RESULTS

Analysis of Spots That Rapidly Change Expression upon Stimulation—

To identify the potentially large set of proteins regulated shortly after stimulation of INS-1 cells with glucose and/or IBMX, we separated and analyzed protein samples by 2-D DIGE. Among the four different conditions that were tested (25 mm glucose, 1 mm IBMX, 25 mm glucose + 1 mm IBMX, and 25 mm glucose + 1 mm IBMX + 5 μg/ml AmD), we detected 2,224 ± 182 and 1,694 ± 99 spots within the ranges of pH 4–7 and 6–9, respectively (supplemental Table 2). Only spots that were detected in at least four of six gels and showed an average increase or decrease ≥1.5-fold and a ≤0.01 t test were considered to differ significantly.

First we compared cell extracts of INS-1 cells incubated for 2 h in either resting (0 mm glucose) or stimulating (25 mm glucose + 1 mm IBMX) buffer. Upon stimulation, 33 spots increased, and 42 decreased (Fig. 1, A and B). Next we compared the profiles of cells kept at rest or stimulated with either 25 mm glucose or 1 mm IBMX. In cells stimulated with 25 mm glucose, eight spots increased, and five decreased (supplemental Fig. 1, A and D), whereas upon stimulation with 1 mm IBMX, 40 spots increased, and 37 decreased (supplemental Fig. 1, B and E). Four (two up-regulated and two down-regulated) of the 13 glucose-responsive spots did not change in response to IBMX. Of the remaining nine spots, six were up-regulated and three were down-regulated in IBMX-treated cells. Thus, the number of regulated spots (33 + 42 = 75) in INS-1 cells co-stimulated with glucose and IBMX was lower than the sum of the spots independently regulated by glucose and IBMX (13 + 78 − 9 = 82). This discrepancy is even greater considering that only 39 of these 82 spots (47.6%) also changed upon co-stimulation with glucose and IBMX, thus indicating that synergism of glucose and cAMP is only partial.

Fig. 1.

2-DE proteomic profile of INS-1 cells after stimulation with 25 mm glucose and 1 mm IBMX for 2 h. The 2-DE proteomic profiles of INS-1 were compared using the DIGE technology (A and B). Shown are representative gray images of stimulated cell extracts of six independent experiments. Only spots detected in ≥4 gels and having an average ≥1.5-fold increase or decrease and a ≤0.01 t test were regarded to differ significantly. Using this threshold, 18 (pH 4–7) and 15 (pH 6–9) spots were significantly increased (red spots) upon stimulation with 25 mm glucose + 1 mm IBMX, whereas 24 (pH 4–7) and 18 (pH 6–9) spots were significantly decreased (green spots). Representative images of preparative gels containing ∼1.5 mg of loaded protein extracts stained with colloidal Coomassie Brilliant Blue G-250 are shown (C and D). IEF was performed on pH 4–7 (A and C) and 6–9 (B and D) IPG strips. The second dimension was separated by 10% SDS-PAGE.

To detect changes among proteins with low molecular weight, samples were also separated on 15% polyacrylamide gels. Only four spots with low molecular weight increased in cells stimulated with 25 mm glucose + 1 mm IBMX, whereas five spots decreased (data not shown).

To determine whether the rapid proteomic changes observed in response to stimulation with 25 mm glucose + 1 mm IBMX depend on post-transcriptional mechanisms, we compared the DIGE profiles of cells co-stimulated in the presence or absence of 5 μg/ml AmD (supplemental Fig. 1, C and F). In the presence of AmD, 20 of 33 (61%) spots still increased, and 28 of 42 spots (67%) decreased (Fig. 2A, overlapped area of green and red rectangles), indicating that their regulation occurs because of post-transcriptional mechanisms. Notably AmD treatment of co-stimulated cells correlated with an increase in 18 spots and a decrease in 21 spots that were unresponsive to glucose and/or IBMX stimulation alone (Fig. 2A, non-overlapped area of red rectangle).

Fig. 2.

Functional annotation of identified proteins. A, number of rapidly up- (↑) or down-regulated (↓) spots by stimulation with 25 mm glucose (Glc) + 1 mm IBMX (green rectangle), 25 mm glucose (blue rectangle), 1 mm IBMX (yellow rectangle), and 25 mm glucose + 1 mm IBMX + 5 μg/ml AmD (red rectangle). The number of spots regulated by multiple conditions is represented in overlapped areas. B and C, number of identified genes that were differentially expressed in one or more of the four stimulation conditions separated according to their molecular function (B) or biological process (C) annotation in the PANTHER database. Fold enrichments were calculated by comparing the number of occurrences of a term in proteins found in this study with the number of occurrences of that term in the PANTHER database. To assess significance of the enrichment, p values were calculated using the hypergeometric distribution. The enrichment and its significance are shown below the diagram. The significance is indicated as *** for p ≤ 0.001, ** for 0.001 < p ≤ 0.01, and * for 0.01 < p ≤ 0.05 (for more detailed information see supplemental Table 7).

Identification of Rapidly Regulated Proteins by Mass Spectrometry—

MALDI-TOF-MS analysis was performed to identify the spots regulated upon stimulation with 25 mm glucose + 1 mm IBMX in the presence or absence of AmD (Fig. 1, C and D, supplemental Fig. 2, and supplemental Table 3). We identified 63 (84%) of the 75 spots whose levels changed in the absence of AmD (see Table I) and 70 (72%) of the 97 spots whose expression changed in the presence of AmD. Many of the unidentified spots had a basic isoelectric point, making difficult their separation and isolation when preparative protein samples were loaded on the IPG strips.

Table I.

Summary of proteins with increased and decreased levels after stimulation with 25 mm glucose and 1 mm IBMX

Proteins were identified by MALDI-TOF-MS. Glc, glucose; ND, spots were not detected in these conditions; —, spots were not significantly changed in these conditions; FUSE, far upstream element.

Spot no.a	Accession no. (gi)	Protein	Expression changeb				Gene symbol
			Glc + IBMX	Glc	IBMX	Glc + IBMX + AmD
pH 4–7, increased
1	16923998	Heterogeneous nuclear ribonucleoprotein K	3.57	—	3.22	3.07	Hnrpk
	51592098	Staufen, RNA-binding protein, homolog 2 isoform LS					Stau2
	11560133	Tubulin, α1					Tuba1
2	16758782	Lamin B1	3.29	—	3.48	1.98	Lmnb1
3	56799436	Chromobox homolog 3	3.22	—	—	2.12	Cbx3
4	8393696	Stathmin 1	2.94	—	—	—	Stmn1
5	16923998	Heterogeneous nuclear ribonucleoprotein K	2.60	—	—	2.28	Hnrpk
	149040328	Internexin, α					Inexa
6	16758782	Lamin B1	2.56	—	3.44	5.81	Lmnb1
8	149040328	Internexin, α	2.25	1.73	2.10	1.81	Inexa
9	62078893	Ubiquitin-activating enzyme E1	2.01	1.69	1.66	1.91	Ube1
10	157816973	Protein phosphatase 1, regulatory (inhibitor) subunit 8	1.90	1.50	3.88	2.00	Ppp1r8
11	16923998	Heterogeneous nuclear ribonucleoprotein K	1.82	—	—	1.58	Hnrpk
12	13592133	β-Actin	1.71	1.72	1.88	—	Actb
	14010837	NSFL1 (p97) cofactor (p47)					Nsfl1c
13	56971386	Gars protein (glycyl-tRNA synthetase)	1.67	—	—	1.76	Gars
	1346413	Lamin A					Lmna
14	16923998	Heterogeneous nuclear ribonucleoprotein K	1.63	—	—	—	Hnrpk
15	13929082	Pyridoxal (pyridoxine, vitamin B6) kinase	1.61	—	—	2.19	Pdxk
16	1346413	Lamin A	1.57	—	—	1.65	Lmna
17	16923998	Heterogeneous nuclear ribonucleoprotein K	1.56	—	—	—	Hnrpk
18	149053241	Rabaptin, RAB GTPase binding effector protein 1	1.50	—	—	1.63	Rabep1
pH 4–7, decreased
19	157786744	Dihydropyrimidinase-related protein 2	−1.51	—	—	−1.93	Dpysl2
20	71361625	Adaptin ear-binding clathrin-associated protein	−1.52	−1.55	−2.22	—	Necap1
21	16923998	Heterogeneous nuclear ribonucleoprotein K	−1.54	—	—	−1.94	Hnrpk
22	149040328	Internexin, α	−1.58	—	—	−1.52	Inexa
24	74095899	PRP19/PSO4 pre-mRNA processing factor 19 homolog	−1.60	—	—	—	Prpf19
25	16923998	Heterogeneous nuclear ribonucleoprotein K	−1.65	—	—	−2.15	Hnrpk
26	16923998	Heterogeneous nuclear ribonucleoprotein K	−1.66	—	—	−2.47	Hnrpk
27	58865398	Leucine aminopeptidase 3	−1.66	−1.52	—	−1.80	Lap3
28	8393855	Nucleoporin 54 kDa	−1.75	—	−1.97	−1.74	Nup54
29	149040328	Internexin, α	−1.78	—	—	—	Inexa
30	58865550	N-myc downstream regulated gene 1	−1.79	—	—	—	Ndrg1
31	1346413	Lamin A	−1.80	—	−1.97	−2.57	Lmna
32	20302113	Stress-induced-phosphoprotein 1	−1.83	—	−1.95	−1.76	Stip1
33	11693176	Acidic ribosomal phosphoprotein P0	−1.92	—	−2.52	−2.75	Arbp
34	203941	Vitamin D-binding protein precursor	−2.08	—	—	—	Gc
	27465535	Tubulin, β5					Tubb2b
35	8393696	Stathmin 1	−2.23	—	—	—	Stmn1
36	13592133	β-Actin	−2.31	−1.97	−2.93	−3.02	Actb
	14010837	NSFL1 (p97) cofactor (p47)					Nsfl1c
38	56799436	Chromobox homolog 3	−2.72	—	−2.62	−2.62	Cbx3
40	56799436	Chromobox homolog 3	−3.38	—	−3.91	−3.23	Cbx3
pH 6–9, increased
43	206205	M2 pyruvate kinase	2.90	—	—	1.93	Pkm2
44	40538742	ATP synthase, H+-transporting, mitochondrial F1 complex, α subunit, isoform 1	2.57	1.99	—	2.09	Atp5a1
45	62825891	Phosphofructokinase, muscle	2.42	—	3.44	2.94	Pfkm
46	13592065	Ribosomal protein S6 kinase polypeptide 1	2.38	—	1.94	—	Rps6ka1
47	13592065	Ribosomal protein S6 kinase polypeptide 1	2.26	—	1.89	1.62	Rps6ka1
48	112984344	Adaptor-related protein complex AP-1, μ subunit 1	2.09	2.78	—	—	Ap1m1
49	140971918	Heterogeneous nuclear ribonucleoprotein A/B	2.02	—	2.33	1.73	Hnrpab
	157816973	Protein phosphatase 1, regulatory (inhibitor) subunit 8					Ppp1r8
50	34327779	Heterogeneous nuclear ribonucleoprotein A3 isoform c	1.93	—	3.48	—	Hnrpa3
51	52783155	Plasminogen activator inhibitor 1 RNA-binding protein	1.82	—	—	—	Serbp1
52	40538860	Aconitase 2, mitochondrial	1.78	—	—	2.44	Aco2
53	6981602	Syntaxin-binding protein 1	1.70	—	—	—	Stxbp1
54	8394162	Aconitase 1	1.65	—	1.57	—	Aco1
55	162287306	ATP-citrate lyase isoform 1	1.56	—	1.98	—	Acly
56	8393418	Glyceraldehyde-3-phosphate dehydrogenase	1.55	—	2.16	—	Gapdh
	34327779	Heterogeneous nuclear ribonucleoprotein A3 isoform c					Hnrpa3
57	13592065	Ribosomal protein S6 kinase polypeptide 1	1.55	—	—	—	Rps6ka1
pH 6–9, decreased
59	66911068	Pcbp2 protein	−1.51	—	—	—	Pcbp2
61	1346413	Lamin A	−1.56	—	—	—	Lmna
62	1346413	Lamin A	−1.60	—	—	−2.63	Lmna
	76253725	Chaperonin subunit 6a (ζ)					Cct6a
63	1346413	Lamin A	−1.71	—	−1.72	−1.70	Lmna
64	52783155	Plasminogen activator inhibitor 1 RNA-binding protein	−1.79	—	—	−2.35	Serbp1
65	13592065	Ribosomal protein S6 kinase polypeptide 1	−1.81	—	−1.94	−2.40	Rps6ka1
66	52783155	Plasminogen activator inhibitor 1 RNA-binding protein	−1.91	—	—	−2.30	Serbp1
68	157816973	Protein phosphatase 1, regulatory (inhibitor) subunit 8	−1.97	—	−2.18	—	Ppp1r8
	6978487	Aldolase A					Aldoa
	81294202	Psmc6 protein					Psmc6
69	52783155	Plasminogen activator inhibitor 1 RNA-binding protein	−2.00	—	—	−2.20	Serbp1
71	162287306	ATP-citrate lyase isoform 1	−2.14	—	−2.12	−2.54	Acly
73	19424312	KH-type splicing regulatory protein (far upstream element (FUSE)-binding protein 2)	−2.39	−1.86	−2.50	−4.28	Khsrp
74	1346413	Lamin A	−4.85	—	−2.87	−3.08	Lmna
75	1346413	Lamin A	−5.15	—	−5.82	−5.44	Lmna
pH 4–7, increased, 15% SDS-PAGE
76c	8393696	Stathmin 1	10.42	ND	ND	ND	Stmn1
80c	8393696	Stathmin 1	4.28	ND	ND	ND	Stmn1
82c	112983968	Sorting nexin 3	2.87	ND	ND	ND	Snx3
pH 4–7, decreased, 15% SDS-PAGE
77c	40018580	Hypothetical protein LOC308869	−1.50	ND	ND	ND	C11orf59
79c	228542	Myosin: subunit = regulatory light chain	−4.46	ND	ND	ND	Mylc2b
81c	8393696	Stathmin 1	−1.77	ND	ND	ND	Stmn1
pH 6–9, increased, 15% SDS-PAGE
83c	149037907	DnaJ (Hsp40) homolog, subfamily B, member 1 (predicted), isoform CRA_b	1.92	ND	ND	ND	Dnajb1
83c	38328245	Hnrpa1 protein					Hnrpa1

^aNumbering of spots is according to the 2-D gels (10% SDS-PAGE) shown in Fig. 1. Spots 7, 23, 37, 39, 41, 42, 58, 60, 67, 70, 72, 78^c could not be identified.

^bAverage ratio of changed expression between resting and stimulated (25 mm glucose + 1 mm IBMX) INS-1 cells calculated using the BVA module version 5.0 (p ≤ 0.01).

^cProteins identified from regulated spots as detected by 2-D DIGE using 15% SDS-PAGE (data not shown).

Proteins recognized by MS belonged to 22 molecular function categories in the PANTHER database (54) (see Fig. 2, B and C, and supplemental Tables 4–7 for detailed information). Interestingly the largest class of regulated proteins, corresponding to 26.9%, was represented by nucleic acid-binding proteins (Fig. 2B), including the following 14 mRNA-binding proteins: CUGBP1, hnRNP A1, hnRNP A3, hnRNP A/B (CBF-A), hnRNP E2 (PCBP2), hnRNP K, hnRNP L, hnRNP H1, KH-type splicing regulatory protein, PAI-RBP1, PRP19/PSO4 pre-mRNA processing factor 19, RNA binding motif protein 8A, Srp20 (splicing factor arginine/serine-rich 3), and Staufen2. According to their biological process annotation in the PANTHER database, 27 (34.6%) of the regulated proteins are involved in nucleoside, nucleotide, and nucleic acid metabolism, whereas 19 (24.4%) are implicated in protein metabolism and modification (Fig. 2C). A drawback of 2-DE is the lack of detection of low abundance proteins and the poor resolution of membrane and large hydrophobic proteins (55). Thus, we cannot rule out that the identification of nucleic acid-binding proteins as a major class of regulated factors is because of their preferential detection by the selected method relative to other protein classes. Changes in 10 mRNA- binding proteins (CUGBP1, hnRNP A/B (CBF-A), hnRNP K, hnRNP L, hnRNP H1, KH-type splicing regulatory protein, PAI-RBP1, RNA binding motif protein 8A, Srp20, and Staufen2) were not inhibited by AmD. The changed expression patterns of hnRNP K, PAI-RBP1, PCBP2, and lamin A, another protein identified in our screen, were verified by immunoblotting of 2-D gels (Fig. 3). The immunoreactive pattern of hnRNP K and PAI-RBP1 correlated well with the one detected by 2-D DIGE (equivalent spots are indicated by the same numbers in Figs. 1 and 3). In the case of PCBP2 and lamin A, the immunoblotting pattern showed a shift similar to the one visualized by 2-D DIGE, but an accurate match of the spots detected with the two procedures was not possible. Additionally analysis by one-dimensional gel electrophoresis showed increased levels or an upward shift suggestive of phosphorylation in the case of hnRNP A1, hnRNP A3, the 52-kDa isoform of Staufen2, and lamin A (Fig. 4).

Fig. 3.

Validation of 2-D DIGE results by Western blotting on 2-D gels. Selected rapidly regulated proteins were identified from extracts of cells kept in resting (0 mm glucose) or stimulating (25 mm glucose + 1 mm IBMX) buffer for 2 h. Immunoprobing was carried out using anti-hnRNP K (pH 4–7; A and B), anti-PAI-RBP1 (pH 6–9; C and D), anti-PCBP2 (pH 6–9; E and F), and anti-lamin A/C (pH 4–7 and 6–9; G and H) antibodies. PTBP1 in extracts from resting (I) or stimulated (J) cells was detected using the ECL Plex system in combination with a mouse monoclonal antibody directed against PTBP1 (green channel) and an affinity-purified rabbit antibody against PTBP1 phosphorylated on serine 16 (red channel). The numbering of spots detected in A, B, C, and D is according to the 2-D gels (10% SDS-PAGE) shown in Fig. 1. Changes in spot pattern are indicated with arrows if no direct match to the 2-D DIGE was possible.

Fig. 4.

mRNA-binding proteins are regulated upon stimulation. Western blots for several mRNA-binding proteins in extracts of cells kept in resting (0 mm glucose (Glc)) or stimulating (25 mm glucose + 1 mm IBMX) buffer for 2 h are shown. Immunoprobing was carried out with anti-Staufen2 (Stau2) (A), anti-hnRNP A1 (B), anti-hnRNP A3 (C), anti-PCBP2 (D), anti-CBF-A (E), or anti-lamin A/C (F) antibodies. For quantification, three lanes were probed for each protein under each condition. Equal loading was monitored by immunoblotting for γ-tubulin. G, quantification of the immunoblots shown in A–F. The level of each detected protein in resting cells was equaled to 100%. Each bar shows quantification from three independent experiments normalized to γ-tubulin (*, p ≤ 0.05; **, p ≤ 0.01). Error bars represent standard deviation of three independent experiments.

We have shown previously that stimulation of INS-1 cells with IBMX induces the cAMP-dependent protein kinase A-dependent phosphorylation of PTBP1 on serine 16 (12). Having a predicted pI of 9.17, PTBP1 migrates beyond the range of proper separation by the pH 6–9 IPG strips. Thus, its absence in the list of regulated spots identified by 2-D DIGE/MS was not unexpected. Its change upon stimulation, however, was visualized by 2-DE immunoblotting with an antibody that specifically binds to PTBP1 phosphorylated on serine 16 (Fig. 3, I and J).

Insulin, ICA512, PC1/3, PC2, and chromogranin A are SG components whose increased expression in cells stimulated with glucose/IBMX is regulated by PTBP1 (11, 12) but that were not identified in our screen. These proteins, however, are either too small to be detected (insulin) or too hydrophobic to be properly separated (all other cases) by conventional 2-DE. Nonetheless 2-DE immunoblotting confirmed that in cells stimulated with glucose + IBMX, the levels of chromogranin A (Fig. 5, A and B), PC1 (Fig. 5, C and D), and PC2 (Fig. 5, E and F) increased, whereas the transmembrane fragment of ICA512 (ICA512-TMF) (Fig. 5, G and H) and carboxypeptidase E (CPE) (Fig. 5, I and J) decreased either because of proteolytic cleavage (4, 56) or secretion (3), respectively.

Fig. 5.

SG proteins are regulated upon stimulation. Western blotting on 2-D gels (pH 4–7) for some SG proteins in extracts of cells kept in resting (0 mm glucose) or stimulating (25 mm glucose + 1 mm IBMX) buffer for 2 h is shown. Immunoprobing was carried out using anti-chromogranin A (CGA) (A and B), anti-PC1 (C and D), anti-PC2 (E and F), anti-ICA512 (G and H), and anti-carboxypeptidase E (CPE) (I and J) antibodies. Changes in spot pattern are indicated with arrows or numbers.

Bioinformatics Prediction of PTBP1 Binding Motifs in Regulated Proteins—

Next we investigated whether mRNAs encoding proteins regulated by glucose and IBMX contain potential PTBP1 binding sites in their 5′- and 3′-UTRs. Conservation of these putative binding sites among rat, mouse, and human (Table II) or just between rat and either mouse or human (Table III) was considered a necessary criterion for significance. Thirteen of the identified proteins included fully conserved potential PTBP1 sites in mRNA UTRs from all three species. By lowering the conservation criterion to include regions conserved in at least two species, we found putative PTBP1 binding sites in mRNA UTRs of 22 proteins in our list (Table III), including PAI-RBP1, PCBP2, and Staufen2. The changed expression of PAI-RBP1 and Staufen2, but not PCBP2, was most likely because of post-translational mechanisms because of its insensitivity to AmD. Eighteen of these regulated proteins, including Staufen2 but not PAI-RBP1, were identified as a single spot by 2-D DIGE, suggesting that their changed level was not the result of phosphorylation. In most cases, at least one of the putative PTBP1 binding sites in each of these mRNA UTRs was predicted to be located within single-stranded regions by secondary structure analyses (supplemental Fig. 3). This is interesting in view of the prevailing data suggesting that PTBP1 preferentially binds single-stranded RNA; albeit this opinion has recently been challenged (57). For insulin, ICA512, PC1/3, PC2, and chromogranin A, which are regulated by PTBP1 through its binding to their 3′- and/or 5′-UTRs (11, 19),2 our method confirms potential PTBP1 binding sites fully conserved among rat, mouse, and human with at least one potential binding site in a single-stranded region (data not shown).

Table II.

Predicted conserved (human, rat, and mouse) PTBP1 binding sites in the mRNA UTRs from regulated proteins

Hits are sorted by p value, p < 0.01. Stm1 is listed twice in the table because of a different number of mismatches detected in the PTBP1 consensus sequence and consequently different p values. NECAP, adaptin ear-binding clathrin-associated protein.

Entrez Gene symbol	Protein	Untranslated region	Number of mismatchesa	Number of hits	p valueb	Number of hits in single strandc
Aldoa	Aldolase A	3′	2	12	0.0001	4
Dpysl2	Dihydropyrimidinase-related protein 2	3′	2	35	0.0006	8
C11orf59	Chromosome 11 open reading frame 59	3′	2	14	0.0007	0
Lmna	Lamin A	3′	2	22	0.0013	7
Pkm2	M2 pyruvate kinase	3′	2	15	0.0017	6
Mapk1	Mitogen-activated protein kinase 1	5′	1	3	0.0026	1
Dnm2	Dynamin 2	3′	1	5	0.0029	1
Stmn1	Stathmin 1	3′	1	3	0.003	1
Stub1	STIP1 homology and U-box-containing protein 1	3′	1	4	0.0033	0
Sgta	Small glutamine-rich tetratricopeptide repeat (TPR)-containing, α	3′	1	7	0.0041	2
Stxbp1	Syntaxin-binding protein 1	3′	2	28	0.0047	15
Stmn1	Stathmin 1	3′	2	8	0.0063	4
Ndrg1	N-myc downstream regulated 1	3′	1	4	0.0098	1
Necap1	NECAP endocytosis-associated 1	3′	1	4	0.0099	2

^aNumber of mismatches between 0 and 2.

^bThe p value is the probability of obtaining at least the same number of hits in a random sequence of the same composition.

^cThe secondary structure of RNA was predicted using RNAfold (53).

Table III.

Predicted conserved (minimum of two species) PTBP1 binding sites in the mRNA UTRs from regulated proteins

Hits are sorted by p value, p < 0.01. Some genes are listed several times in the table because of a different number of mismatches detected in the PTBP1 consensus sequence and consequently different p values. NECAP, adaptin ear-binding clathrin-associated protein.

Entrez Gene symbol	Protein	Untranslated region	Number of mismatchesa	Number of hits	p valueb	Number of hits in single strandc
Stau2	Staufen2	3′	1	9	0	4
Lmna	Lamin A	3′	2	33	0	12
Dpysl2	Dihydropyrimidinase-related protein 2	3′	2	42	0	10
C11orf59	Chromosome 11 open reading frame 59	3′	2	23	0	0
Aldoa	Aldolase A	3′	2	12	0.0001	4
Tubb2b	Tubulin β chain 15	3′	2	13	0.0001	3
Stau2	Staufen2	3′	2	21	0.0001	5
Stxbp1	Syntaxin-binding protein 1	3′	2	36	0.0001	21
Pkm2	M2 pyruvate kinase	3′	2	18	0.0002	6
Stau2	Staufen2	3′	0	2	0.0005	1
Sgta	Small glutamine-rich tetratricopeptide repeat (TPR)-containing, α	3′	1	9	0.0006	4
Serbp1	Plasminogen activator inhibitor 1 RNA-binding protein	5′	1	3	0.0007	3
Snx3	Sorting nexin 3	3′	1	5	0.0007	2
Glul	Glutamate-ammonia ligase (glutamine synthase)	3′	2	19	0.0014	4
Dpysl2	Dihydropyrimidinase-related protein 2	5′	2	9	0.0018	1
Inexa	α-Internexin	3′	2	21	0.0026	6
Stmn1	Stathmin 1	3′	1	3	0.003	1
Ndrg1	N-myc downstream regulated 1	3′	2	17	0.003	3
Necap1	NECAP endocytosis-associated 1	3′	1	5	0.0032	1
Tubb2b	Tubulin β chain 15	3′	1	3	0.0037	0
Smc3	Structural maintenance of chromosomes protein 3	5′	2	4	0.0038	0
Atp6v1a	ATPase, H+-transporting, lysosomal V1 subunit A	3′	2	25	0.0039	6
Pkm2	M2 pyruvate kinase	3′	1	4	0.0063	1
Stmn1	Stathmin 1	3′	2	8	0.0063	4
Pcbp2	Poly(rC)-binding protein 2	3′	2	9	0.0064	3
Dpysl2	Dihydropyrimidinase-related protein 2	3′	1	7	0.0065	2
Ppp1r8	Predicted: similar to protein phosphatase 1, regulatory (inhibitor) subunit 8	3′	2	11	0.0065	4
Cugbp1	CUG triplet repeat, RNA-binding protein 1	3′	1	4	0.0068	0
Cmpk	Cytidylate kinase	3′	2	11	0.0094	5
Ppp1cb	Protein phosphatase 1, catalytic subunit, β isoform	3′	2	16	0.0095	1
Ndrg1	N-myc downstream regulated 1	3′	1	4	0.0098	1

^aNumber of mismatches between 0 and 2.

^bThe p value is the probability of obtaining at least the same number of hits in a random sequence of the same composition.

^cThe secondary structure of RNA was predicted using RNAfold (53).

Validation of PTBP1 Binding Predictions—

To test whether PAI-RBP1, Staufen2, and PCBP2 are regulated by PTBP1, we analyzed their expression following Ptbp1 knockdown by RNAi or overexpression of PTBP1 tagged at its C terminus with a V5 epitope (PTBP1-V5). After Ptbp1 knockdown, only the levels of the 52-kDa isoform of Staufen2 were modestly increased, whereas they did not change upon PTBP1-V5 overexpression. The levels of the other Staufen2 isoforms, PAI-RBP1, and PCBP2 did not significantly differ in either condition (Fig. 6). In view of these findings, the possibility that PAI-RBP1, Staufen2, and PCBP2 are regulated by PTBP1 seems unlikely.

Fig. 6.

Validation of PTBP1 binding predictions by knockdown and overexpression of PTBP1. Western blots for PTBP1 (A, E, and F) and mRNA-binding proteins with predicted PTBP1 binding sites in their 5′- and 3′-UTRs are shown. In these cells, PTBP1 levels were either down-regulated by RNAi (A and B) or up-regulated by transfection of PTBP1-V5 (E–G). A and E show immunoblots obtained with the anti-PTBP1 antibody; F shows the immunoblot obtained with the anti-V5 antibody. For quantification, three lanes were probed for each protein under each condition. Equal loading was monitored by immunoblotting for γ-tubulin. C, quantification of PTBP1 knockdown as detected in A. The average level of PTBP1 in cells transfected with the control vector for RNAi was equal to 100%. D and I, quantification of immunoblots for the 52-, 59-, and 62-kDa isoforms of Staufen2 (Stau2) (70, 71), PAI-RBP1, and PCBP2 as detected in B and G. The level of these proteins in control cells was equaled to 100%. H, quantification of endogenous (endog.) and total PTBP1 levels as detected in E. The average PTBP1 level in cells transfected with the control vector was equaled to 100%. Each bar shows quantification from three independent experiments normalized to γ-tubulin (*, p ≤ 0.05; **, p ≤ 0.01; ***, p ≤ 0.001). Error bars represent standard deviation of three independent experiments.

DISCUSSION

Previous studies have shown that stimulation of β-cells and insulinoma cells rapidly increases the expression of many SG genes primarily by activating post-transcriptional mechanisms (11, 15–17). However, a comprehensive and unbiased proteomics analysis of these changes has not yet been reported (58). For this reason we have compared the proteomic profile of resting and stimulated INS-1 cells by 2-D DIGE followed by MS, which is a reliable platform for proteomics studies (37, 59). In total, we identified 165 spots whose levels significantly changed in response to stimulation for 2 h with 25 mm glucose and 1 mm IBMX, either alone or together, and in the presence or absence of AmD. We identified 117 (70.9%) of these spots, corresponding to 78 different proteins. This yield is comparable to that achieved in other proteomics studies in β-cells (40, 60, 61). The remaining 30% of the regulated spots could not be identified either because of their insufficient amount or because their spectrum could not be assigned to any entry in the queried databases. Thirteen of the identified spots (11.1%) included multiple proteins that co-migrated at the same position in the gel. In these cases it was not possible to determine the identity of the regulated protein(s).

Except insulin and other components of the secretory granules, none of the proteins identified in this study have been reported previously to exhibit rapid level changes in β-cells. Earlier studies investigated the proteome of mouse islets immediately after isolation (42, 62, 63) and following 24-h exposure to 11 mm glucose (40). More recent studies have focused on proteomic differences between glucose-responsive and non-responsive MIN-6 cells (64), INS-1 832/13 cells exposed to 16.7 mm glucose versus 2.8 mm glucose for 48 h (61), and INS-1E cells treated with different cytokines (60). Despite differences in experimental design, 27 (34.6%) of the 78 regulated proteins identified here were also found in other studies (see supplemental Table 8). According to their molecular functional classification in the PANTHER database, these 27 regulated proteins belong to the following groups: nucleic acid binding (five), cytoskeletal (four), chaperone (three), lyase/transferase (three), membrane trafficking (two), oxidoreductase (two), kinase (two), ion channel (two), protease (one), transfer/carrier (one), miscellaneous function (one), and molecular function unclassified (one). The inclusion in this list of enzymes involved in glucose metabolism and mitochondrial function, such as glyceraldehyde-3-phosphate dehydrogenase, pyruvate kinase L, aconitase, and glutamate dehydrogenase is not surprising given their abundance and relevant regulatory role in β-cells. Many of these 27 proteins were also shown to change levels after treatment of INS-1E cells with interleukin-1β and interferon-γ for 24 h (60), although these changes were often antithetical to those detected in our screen and in some cases below the threshold of 1.5-fold applied here. A remarkable example for this opposite regulation is hnRNP K. D'Hertog et al. (60) resolved hnRNP K in six spots, all of which were down-regulated in INS-1E cells exposed to cytokines. Here hnRNP K was separated in nine spots, five of which were up-regulated and four of which were down-regulated.

Notably our 2-D DIGE screen did not detect preproinsulin, pro-PC1/3, pro-PC2, prochromogranin A, or pro-ICA512 among the regulated proteins. The up-regulation of these SG components following stimulation was nevertheless confirmed using antibody-based approaches, such as radioimmunoassay/ELISA for insulin (not shown) or Western blotting for the other SG proteins. The proteomic composition of fractions enriched in insulin SGs of INS-1E cells was recently analyzed by one-dimensional gel electrophoresis and nano-LC-ESI-MS/MS (65). Of the 130 identified proteins, 110 had not been associated previously with SGs. Yet not all known SG components, such as PC1/3 and ICA512, were detected. On the other hand, five of the proteins identified, namely glyceraldehyde-3-phosphate dehydrogenase; actin; vitamin D-binding protein precursor; ATPase, H⁺-transporting, V1 subunit A, isoform 1; and heat shock protein 40 homolog (DnaJ), were in our list of rapidly regulated proteins.

The changes in the proteomic pattern that we observed in cells stimulated with glucose alone were less extensive than those produced by IBMX stimulation. Indeed only 13 spots were glucose-responsive, whereas 77 were IBMX-responsive. Thus, most of the proteins identified in our study are targets of cAMP regulation. Glucose-induced proteomic changes are most likely underestimated because of the reduced glucose responsiveness of INS-1 cells compared with primary β-cells, a limit shared with other insulinoma cell lines such as mouse MIN-6 cells (66). Co-stimulation of INS-1 cells with glucose and IBMX changed the levels of 75 spots of which 61% were still increased and 67% were decreased in the presence of the transcription inhibitor AmD, indicating the post-transcriptional nature of this change. These data add to the evidence that change in the expression profile of β-cells shortly after stimulation is largely driven by post-transcriptional mechanisms (5–8).

Phosphorylation is the most common post-translational modification, and many of the spots that changed in response to IBMX are either already known or likely targets of cAMP-dependent protein kinase A or other Ca²⁺-stimulated kinases, such as protein kinase C. Many proteins regulated in response to co-stimulation with glucose and IBMX, with or without AmD, migrated as multiple spots that characteristically shifted to a more acidic pI upon stimulation as expected in the case of phosphorylation. A prominent example of this behavior was again hnRNP K, whose overall pattern shifted toward a more acidic pH range. A similar shift towards lower pI was observed in the case of PAI-RBP1 and lamin A. Phosphorylation of hnRNP K and lamin A was further validated by treatment with alkaline phosphatase, which resulted in a downward shift of the protein doublet observed by SDS-PAGE (data not shown).

The most novel finding of our study is the identification among regulated proteins of seven heterogeneous nuclear ribonucleoproteins and seven additional mRNA-binding proteins, which account for 17.9% of the total rapidly regulated proteome. Some of them, namely hnRNP K, hnRNP H1, and splicing factor arginine/serine-rich 3 have already been shown to be affected following the long exposure of INS-1 832/13 cells to glucose (61) or INS-1E cells to cytokines (60). However, this is the first demonstration that mRNA binding factors represent a major class of rapidly regulated proteins in a β-cell model. Regulation of mRNA stability and translation by mRNA-binding proteins is emerging as a relevant post-transcriptional mechanism to rapidly increase insulin granule biosynthesis following β-cell stimulation (67, 68). In particular, stimulation of INS-1 and β-cells with either glucose or IBMX induces the nucleocytoplasmic translocation of PTBP1, an additional heterogeneous nuclear ribonucleoprotein (hnRNP I). Binding of cytosolic PTBP1 to the 3′- and 5′-UTR of mRNAs encoding SG components in turn promotes their stability and translation (11, 12).² The ability of IBMX to induce the phosphorylation of PTBP1 was here confirmed by 2-D immunoblotting. As in the case of PTBP1 (12, 69), the rapid phosphorylation of the other hnRNPs identified in this study could regulate their nucleocytoplasmic transport.

Using bioinformatics tools we found that the mRNAs of three regulated mRNA-binding proteins, i.e. PAI-RBP1, PCBP2, and Staufen2, include potential PTBP1 binding sites in their 5′- or 3′-UTRs. Neither up-regulation nor down-regulation of PTBP1, however, altered their levels except for a modest increase in Staufen2 following the Ptbp1 knockdown. Notably the glucose + IBMX-induced reduction of PCBP2 was AmD-sensitive, whereas the changed pI pattern of PAI-RBP1 by 2-DE was compatible with phosphorylation. Based on these data, the possibility that PTBP1 is involved in the post-transcriptional regulation of PAI-RBP1, PCBP2, and Staufen2 is unlikely.

In conclusion, our findings demonstrate that rapid modulation of mRNA-binding proteins is a major process following the stimulation of insulinoma cells with insulin secretagogues. Future studies will be necessary to analyze the significance of these changes on β-cell gene expression and function.