A possible role of transglutaminase 2 in the nucleus of INS-1E and of cells of human pancreatic islets.

Transglutaminase 2 (TG2) is a multifunctional protein with Ca(2+)-dependent transamidating and G protein activity. Previously we reported that the role of TG2 in insulin secretion may involve cytoplasmic actin remodeling and a regulative action on other proteins during granule movement. The aim of this study was to gain a better insight into the role of TG2 transamidating activity in mitochondria and in the nucleus of INS-1E rat insulinoma cell line (INS-1E) during insulin secretion. To this end we labeled INS-1E with an artificial donor (biotinylated peptide), in basal condition and after stimulus with glucose for 2, 5, and 8min. Biotinylated proteins of the nuclear/mitochondrial-enriched fraction were analyzed using two-dimensional electrophoresis and mass spectrometry. Many mitochondrial proteins involved in Ca(2+) homeostasis (e.g. voltage-dependent anion-selective channel protein, prohibitin and different ATP synthase subunits) and many nuclear proteins involved in gene regulation (e.g. histone H3, barrier to autointegration factor and various heterogeneous nuclear ribonucleoprotein) were identified among a number of transamidating substrates of TG2 in INS-1E. The combined results provide evidence that a temporal link exists between glucose-stimulation, first phase insulin secretion and the action of TG on histone H3 both in INS-1E and human pancreatic islets. BIOLOGICAL SIGNIFICANCE: Research into the role of transglutaminase 2 during insulin secretion in INS-1E rat insulinoma cellular model is depicting a complex role for this enzyme. Transglutaminase 2 acts in the different INS-1E compartments in the same way: catalyzing a post-translational modification event of its substrates. In this work we identify some mitochondrial and nuclear substrates of INS-1E during first phase insulin secretion. The finding that TG2 interacts with nuclear proteins that include BAF and histone H3 immediately after (2-5min) glucose stimulus of INS-1E suggests that TG2 may be involved not only in insulin secretion, as suggested by our previous studies in cytoplasmic INS-1E fraction, but also in the regulation of glucose-induced gene transcription.

Introduction

Translutaminase 2 (TG2; EC 2.3.2.13) is a multifunctional and ubiquitous enzyme belonging to the transglutaminase family [1]. Its primary enzymatic activity resides in Ca2 + dependent transamidation reaction in which the γ carboxyamide group of peptide bound glutamine residues serves as acyl donor (Q donor), while primary amino groups of several compounds function as acceptor substrates [2]. The two best assessed functions of TG2, namely transamidation and G protein activities, are finely regulated in cells by Ca2 + and GTP concentrations, which respectively promote and inhibit transamidation activity [3]. A regulatory role for transamidation activity of TG2 has also been proposed for ATP, but this is still a matter of debate [3].

In various cells TG2 is predominantly a cytoplasmic protein, but different, less clear cellular localization has also been reported as for example in the nucleus, mitochondria, on the plasma membrane and in the extracellular matrix [4-7]. Regarding this issue it has been shown that TG2 dynamically translocates to these various subcellular compartments as a function of cell proliferation state or in response to the elevation of intracellular calcium concentrations [8,9]. However, the underlying mechanism by which the enzyme is translocated remains unclear, especially because TG2 is devoid of the signal peptide and of other export sequences. For example, importin-α 3 has been suggested as a mediator of active TG2 nuclear transport [9], but other transport mechanisms have also been described, such as an association of TG2 with p65 subunit of NFkB [10], the formation of a complex with VEGFR-2 [11] and an association between TG2 and the eukaryotic translation initiation factor 5 [12]. The way by which TG2 could translocate into the mitochondrion remains a very vague issue too; a possible mechanism could be the interaction of the functional BH3 domain present in the TG2 with cytoplasmic Bax which might address the protein complex in the mithocondria. [13]. Many mitochondrial proteins are post-translationally modified by TG2 [14-16], and several nuclear proteins have been identified as its potential substrates, such as retinoblastoma protein [17,18], p53 [19] and histones [20-22]. Moreover, nuclear TG2 mainly seems to alter the regulation of gene expression via post-translational modification of and/or interaction with transcriptional factors and related proteins, including hypoxia inducible factor (HIF) 1 [23], Sp1 [24] and histones [20-22], thereby controlling cell growth or survival, differentiation and apoptosis. One of the best studied TG2 activities in the nucleus is histone cross-linking, an event which has also been proposed to mediate chromatin condensation [22].

In our previous study [25] researching the role of TG2 in cytoplasmic and granular compartments of INS-1E, a rat insulinoma cell line, we learned that the basal TG2 transamidation activity, present in resting cells, was increased during the glucose-induced first phase insulin secretion (FPIS) and was closely related to the increase in Ca2 + [25].

In the present study we investigate the kinetics of transamidation reaction of TG2 in the nuclear/mitochondrial enriched fractions of INS-1E during FPIS. We show that TG2 is active as transamidase in these compartments and we identify some of its natural substrates. The emerging picture indicates that in the mitochondrion TG2 is probably involved in the sensing of Ca2 + levels, and in the nucleus TG2 might contribute to the regulation of the condensed state of chromatin and in the stability of pre-mRNA.

Materials and methods

Reagents and media

All chemicals were from Sigma Aldrich, unless differently specified. All media and cell supplements were from Invitrogen.

Immunocytochemistry

INS-1E grown on coverslips were fixed in ice-cold 4% paraformaldehyde in PBS, washed with PBS, permeabilized with 0.1% Triton X-100, blocked in 5% BSA in PBS for 30 min, and then incubated 12 h with monoclonal mouse anti-TG2 CUB 7401 (Thermo Scientific, 1/100) or polyclonal rabbit anti-TG2 (CovAlb, 1/50) antibodies. After PBS washes, samples were incubated with Alexa 488-conjugated goat anti-mouse-Ig or goat anti-rabbit-Ig polyclonal antibodies (Molecular Probes) for 1 h, at RT. Double staining was performed using the following antibodies: mouse anti-cytochrome c oxidase subunit II (COXII; Santa Cruz Biotechnologies, 1/500), goat anti-receptor-type tyrosine–protein phosphatase 1A2β granule regulator (PTP1A2β; Santa Cruz Biotechnologies, 1/30) and anti-synaptosomal-associated protein 25 (SNAP25; Santa Cruz Biotechnologies, 1/50). In this case the second antibody could be either Alexa 555-conjugated goat anti-mouse or donkey anti-goat IgG (Molecular Probes) for 1 h, at RT.

To examine the TG2 localization within the extracellular matrix, INS-1E and dermal fibroblasts were fixed in ice-cold 4% paraformaldehyde, blocked in 5% BSA in PBS, then double labeled using anti-TG2 and anti-fibronectin (Santa Cruz Biotechnologies) antibodies.

Nuclear staining was carried out with 1 μg/ml Hoechst 33342 (Molecular Probes). All the experiments were repeated at least three times. Negative controls were performed using 1% PBS in BSA without the primary antibody.

Images were acquired using an Olympus Fluoview FV1000 confocal microscope, equipped with a 60 × (numerical aperture: 1.42) oil objective. Optical single sections were acquired with a scanning mode format of 1024 × 1024 pixels, sampling speed of 40 μs/pixel, and 12 bits/pixel images. Fluorochrome unmixing was performed by acquisition of automated-sequential collection of multi-channel images, in order to reduce spectral crosstalk between channels. Z-reconstructions of serial single optical sections were performed with a scanning mode of 1024 × 1024 pixels with a 0.207 μm/pixel size, sampling speed of 40 μs/pixel, Z stack of 0.4 μs/slice and 12 bits/pixel images. Each group of images was processed and analyzed using the same settings (i.e. laser power and detector amplification). Images were processed using Adobe Photoshop software (Adobe Systems Inc.).

Human pancreatic islets

Pancreata were obtained from heart-beating cadaveric multiorgan donors with approval from the Ca' Granda Niguarda Hospital Ethical Committee (Milan). Islets were isolated according to the automated method [26], purified by continuous gradient with refrigerated COBE processor as previously described [27] and maintained in M199 medium, supplemented with serum and antibiotics, at 37 °C, 5% CO2 until use. M199 was chosen because its glucose concentration is 5.5 mM.

INS-1E and human pancreatic islets labeling with A25 peptide

INS-1E cells (passage 55–90) cultured as previously described [25] were labeled with 1 mM biotin TVQQL-OH (A25 peptide, Zeidra or PRIMM) as reported in [25]. Briefly, when the rat insulinoma cells INS-1E reached 70% confluence, they were grown in RPMI at 5.5 mM glucose and 10% serum for one night at 37 °C and 5% CO2. The day after, cells were quickly washed with RPMI without glucose and serum, and medium was replaced with RPMI at 2.5 mM glucose and no serum for 1 h at 37 °C and 5% CO2. This medium was replaced again with new RPMI at 2.5 mM glucose, 1 mM A25 peptide and no serum for 1 h. Cells were then collected by scraping after 2, 5, 8 and 15 min of stimulation with 15.5 mM glucose [25] or after incubation for the same time in basal condition (2.5 mM glucose). Cells were washed in PBS and homogenated by Dounce in 0.250 M sucrose, 60 strokes on ice and passed through needles of different sizes (19G; 22G and 27G, 20 strokes/each on ice). A nuclear/mitochondrial-enriched fraction was separated by 60 minute centrifugation at 1500 rpm at 4 °C. Cytosolic calcium level after glucose stimulus was not directly measured in our experiments and was assumed to be equal to the level assessed previously in Dr. Maechler's laboratory [28]. In separate experiments, INS-1E cells pre-incubated with A25 peptide were stimulated or not stimulated with either 15.5 mM glucose in the presence of 100 μM diazoxide (an ATP-sensitive potassium channel opener), or 30 mM KCl or 10 μM tolbutamide (a potassium channel blocker which acts as an oral hypoglycemic drug) as previously described [25].

Human islets were maintained for 24 h at 37 °C in M199 medium supplemented with 10% serum. On the day of the experiment the medium was replaced with M199 without serum for 1 h, and then substituted for 1 h with serum free M199 supplemented with 1 mM A25 peptide. After this labeling the control group of human islets was maintained in M199 plus A25 peptide, while glucose up to 11 mM was added to the islets of the stimulated group. Stimulation was performed for 3 min. Human islets were then collected, centrifugated, washed with PBS and homogenized in 0.250 M sucrose as previously described.

2DE, Western blot and image analysis

For 2DE analysis 200 μg proteins from nuclear/mitochondrial enriched fraction were acetone-precipitated and resuspended in a RB-Thio buffer (6 M Urea, 2 M Thio-Urea, 4% w/v CHAPS), 65 mM DTT, 0.2% v/v IPG buffer 3–10 NL (GE Healthcare) and 0.05% bromophenol blue. Samples were applied on 7 cm IPG strips 3–10 NL by in-gel rehydration at 18 °C; focusing and second dimension (12% acrylamide SDS-PAGE) were performed as described [25,29]. Proteins were electrotransferred to nitrocellulose at constant 25 V overnight at 4 °C. Biotinylated proteins were then revealed by incubation with streptavin-HPR (1/2000 in 1% BSA in 100 mM Tris, 150 mM NaCl pH 7.5, 0.1% Tween 20, TTBS). Peroxide reaction was performed using ECL Prime (GE Healthcare). Blot images were acquired using ChemiDoc XRS + (BioRad) or using film exposure. Apparent molecular weights (MW) were estimated by comparison with MW referent markers (Precision, BioRad). 2DE protein patterns were analyzed using PDQuest 7.4.0 (BioRad). Histograms were generated using Prism V4.03 software (GraphPad Inc., San Diego, CA).

Western blots were performed either on 2DE or one dimensional SDS-PAGE resolved proteins using: a rabbit polyclonal serum antihistone H3 (Santa Cruz Biotechnology) 1/1000 in 1% BSA in TTBS; a mouse monoclonal serum anti ATP synthase subunit alpha (MitoScience) 1/2000 in 1% BSA in TTBS; a rabbit polyclonal serum anti Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) 1/1000 in 1% BSA in TTBS; a rabbit polyclonal serum anti TG2 (CovAlb) 1/1000 in 1% BSA in TTBS and a goat polyclonal serum anti LaminA/C (Santa Cruz Biotechnology) 1/2000 in 1% BSA in TTBS. All specific secondary polyclonal antibodies HPR conjugated were from Dako and diluted 1/2000 in 1% BSA in TTBS.

Biotinylated protein purification

In order to purify TG2 substrates in the nucleus and mitochondria of INS-1E during FPIS, 3 mg of nuclear/mitochondrial-enriched fraction proteins from cells stimulated at 15.5 mM glucose for 2 min were dialyzed for 12 h at 4 °C to remove free A25 peptide and then purified using affinity chromatography on streptavidin linked to magnetic beads (Invitrogen). Biotinylated proteins were detached from the beads adding 310 μl of RB-Thio buffer and incubating for 5 min at 95 °C. The volume of the obtained supernatant was adjusted again to 310 μl (if necessary) and supplemented with 0.2% v/v IPG buffer 3–10 NL, 65 mM DTT and 0.05% bromophenol blue. This sample was applied on 18 cm IPG strips 3–10 NL by in-gel rehydration at 18 °C for 2DE analysis; focusing and second dimension were performed as described [25,29]. Second dimension was performed on 20 × 20 cm 12% acrylamide gels and stopped until BFB reached the bottom of the gels. Proteins were then visualized by staining with Coomassie Blue Silver [30].

Protein identification

Protein spots were excised from 2DE gels with EXQuest™ spot cutter (Bio-Rad). All visible biotinylated protein spots were picked from Coomassie Blue Silver stained gels. Spot picking was performed from two different gels, derived from two independent affinity chromatography purifications. Spots were processed and gel-digested with modified trypsin from bovine pancreas (Roche) as described [31]. All procedures, after desalting and concentration with ZipTip® pipette tips with C18 resin (Millipore), were carried out on an automated protein digestor DigestPro MS (Intavis AG). Peptide mass fingerprinting and MS/MS analysis were done on a 4800 MALDI TOF/TOF™ mass spectrometer (Applied Biosystems). The mass spectra were internally calibrated with trypsin autolysis fragments. The five most abundant precursor ions were selected for MS/MS analysis from the exclusion mass list (ions from human keratin and trypsin). The combined MS and MS/MS data were submitted by the GPS Explorer v.3.6 software (Applied Biosystems) to the MASCOT database search engine (Version 2.1, Matrix Science) and searched with the following parameters: Swissprot 55.2 × database over all Rattus norvegicus protein sequences deposited, no fixed modifications, but allowing as possible modifications: a) carboamidomethylation of cysteine and oxidation of methionine, b) one missed trypsin cleavage, and c) a mass tolerance of ± 0.1D for the peptide mass values and of ± 0.3D for the MS/MS fragment ion mass values. A protein was regarded as identified if the MASCOT protein score, based on combined MS and MS/MS data, was above the 5% significance threshold for the database (score > 51) [32].

Results

TG2 distribution in INS-1E cells

First of all we performed a set of experiments aimed at establishing the subcellular distribution of TG2 in INS-1E. As inferred by Western blot analysis, TG2 localized prevalently in the cytoplasmic fraction but, less abundantly, also in the nuclear/mitochondrial enriched fraction (Fig. 1A and B), as indicated by the correct segregation of specific subcellular markers (i.e. ATP synthase subunit alpha for mitochondria; Lamin A/C for nuclei and GAPDH for cytoplasm) (Fig. 1C–H).

Fig. 1

TG2 localizes in nuclear/mitochondrial enriched fraction. INS-1E proteins from the nuclear/mitochondrial (A, C, E and G) and cytoplasmic (B, D, F and H) enriched fractions were resolved by 2D-electrophoresis and transferred to nitrocellulose for Western blot analysis. TG2 segregation in the subcellular fractions was inferred by Western blot using a rabbit polyclonal serum anti TG2 (panels A and B). The quality of the fraction enrichment was assessed by using: a mouse monoclonal anti ATP synthase subunit alpha as marker for mitochondria (panels C and D); a rabbit polyclonal serum anti GAPDH as cytoplasmic marker (panels E and F); and a goat polyclonal serum anti LaminA/C as nuclear marker (panels G and H). Reactivities were revealed using specific HRP-conjugated secondary antibodies and chemiluminescence reaction. Blot images were collected using ChemiDoc XRS + exposure time 30 s.

Immunocytochemistry analysis supported the observation that TG2 segregates in mitochondria showing that TG2 signals co-localized with the mitochondrial marker COX II (Fig. 2A–C). As previously shown by electron microscopy in human β pancreatic cells [25], this experiment also confirms that TG2 localizes around the insulin granules in the cytoplasm of INS-1E, as indicated by the co-localization with PTP1A2β, a protein which belongs to the membrane of insulin granule, and SNAP25, a protein which localizes at the plasma membrane and is involved in vesicles docking and fusion between plasma membrane and insulin granule membrane (Fig. 2 D–F). In addition, TG2 clearly localizes inside the nucleus (Fig. 3) as inferred by both confocal analysis in single sections (Fig. 3 A and C) and in Z reconstruction (Fig. 3 B and D). It is worth noting that no differences in TG2 distribution were observed regardless of the antibody used for TG2 detection, (namely the polyclonal serum, panels A and B, or the monoclonal CUB7402, panels C and D), although the nuclear localization of TG2 is more evident using the rabbit polyclonal serum.

Fig. 2

TG2 distribution in INS-1E cells. Confocal single sections of INS-1E cells immuno-staining with polyclonal (A) and monoclonal CUB7402 (D and G) antibodies directed against TG2 protein. A–C: TG2 co-distributed in several points (merge yellow signal in C, high magnification of arrowed cell in the inset) with COXII (B), a marker of mitochondria. D–F: TG2 resulted concentrated in cytoplasm with a granular distribution (D), which co-localized (F, inset) with PTP1A2beta, a protein of the membrane of the insulin granule (E). G–I: Double staining of TG2 (G) and SNAP25 (H), showed the cytoplasmic peripheral distribution of TG2 under the plasma membrane (I, inset). The intra-nuclear TG2 distribution was observed with both polyclonal (A) and monoclonal (G) antibody. Nuclear staining performed with Hoechst (blue). Bar: 10 μm.

Fig. 3

Nuclear and cytoplasmic TG2 distribution in INS-1E cells. Confocal single sections (A and C) and Z-reconstructions (B and D) of INS-1E cells using polyclonal (A–B) and monoclonal CUB7402 (C–D) anti-TG2 antibodies. Orthogonal XZ- and YZ-reconstructions (A–B and C–D) showed a higher concentration of the granular TG2 distribution in the cytoplasm and a lower distribution inside nuclei (A and C). Arrows indicate the sites of major density nuclear labeling with Hoechst (blue). Bars: 10 μm.

Lastly, as shown in other cell types, immunocytochemistry shows that TG2 localizes at INS-1E extracellular matrix (Fig. 4A–C), where it is organized in fibril-like filaments (Fig. 4 A, high magnification of arrowed area in the inset) following the pattern of fibronectin (Fig. 4 C, merge orange signal in the inset). The emerging picture indicates that the extracellular matrix of INS-1E cells is less extensive and organized than that observed in fibroblasts (Fig. 4 D–E). Specifically, dermal fibroblasts secrete a more conspicuous quantity of TG2 (D) that shows a strong co-localization (F, yellow signal) with FN fibrillar network (E).

Fig. 4

TG2 deposition in the extracellular matrix. Confocal Z-reconstructions of INS-1E cells (A–C) and human dermal fibroblasts (D–F) double stained with rabbit anti-TG2 (A and D, green) and goat anti-fibronectin (FN, B and E in red) polyclonal antibodies. Fibril-like filaments were observed in TG2-labeled INS-1E cells (A, high magnification of arrowed area in the inset) co-localizing with FN fibrils (C, merge orange signal in the inset). Dermal fibroblasts secrete a more conspicuous quantity of TG2 (D) which showed a strong co-localization (F, yellow signal) with FN fibrillar network (E). Bar: 10 μm.

Transamidation activity of TG2 during FPIS in the nucleus and mitochondria of INS-1E

The kinetics of TG2 transamidation activity in the nuclear/mitochondrial-enriched fractions of INS-1E during FIPS (one out of three experiments) is shown in Fig. 5 and summarized in the graph (Fig. 5G). The peak of TG2 activity is reached at 2 min after stimulation with 15.5 mM glucose (Fig. 5B), with a slight decrease at 5 and 8 min (Fig. 5C and D) and a sharp decrease at 15 min of stimulation (Fig. 5E), when the number of spots observed in the Western blot is similar to that seen at 2.5 mM glucose (Fig. 5G). The changes in the number of spots for each time point parallel the Ca2 + intracellular concentration changes expected to occur in glucose stimulated INS-1E cells [25] (Fig. 5F).

Fig. 5

Kinetic of transamidation reaction of TG2 during FPIS. 200 μg of INS-1E proteins from the nuclear/mitochondrial enriched fraction were first resolved on 7 cm strips 3–10 NL and then electroblotted. Cells labeled with A25 peptide. Western blot was revealed using streptavidin–HPR 1/1000 in 2% BSA in TTBS followed by ECL. Blot images were collected using ChemiDoc XRS +. Exposure time: 60 s. A: INS-1E maintained at 2.5 mM glucose; B: INS-1E stimulated for 2 min; C: INS-1E stimulated for 5 min; D: INS-1E stimulated for 8 min at 15.5 mM glucose; E: INS-1E stimulated for 15 min. Stimulation performed at 15.5 mM glucose F: The graph shows the changes of Ca2 + intracellular concentration during FIPS. Ca2 + levels were not directly measured and were assumed to be equal to that assessed previously in this cell line [28]. G: Histogram indicates the changes in the number of spots for each condition. The number is reported as the mean and SE of three independent experiments.

Software assisted image analysis indicated that the number of biotinylated spots during FPIS was about 10% of the total number of spots visible in the Coomassie stained gels of nuclear/mitochondrial enriched fractions (data not shown). In addition, software-assisted image analysis confirmed that transamidated proteins across different experiments were consistently the same.

To better investigate the dependence of transamidation reaction of TG2 on intracellular Ca2 + concentration, the stimulation experiment of INS-1E with 15.5 mM glucose for 2 min was repeated in the presence of either diazoxide, KCl or tolbutamide. The same experiment was also performed maintaining cells in resting (2.5 mM glucose) condition. Fig. 6 shows that when the glucose stimulus is given in the presence of diazoxide, a compound that blocks the Ca2 + entrance in the cell, the TG2 transamidation activity in the nuclear/mitochondrial compartment is almost abolished/suppressed, both in basal condition as well as after 15.5 mM glucose stimulus. In contrast, KCl, a compound that promotes the massive influx of Ca2 + in the cell, elicits a maximal transamidation activity at basal (2.5 mM), non-stimulatory glucose concentration (Fig. 6D and I). In addition, tolbutamide stimulates TG2 activity in INS-1E in resting condition (Fig. 6B) with no difference observed at 15.5 mM glucose (Fig. 6F) as proved by the number of the spots (Fig. 6I). It has previously been shown that in response to KCl and tolbutamide treatments insulin is secreted in the medium of INS-1E, while in the presence of diazoxide the secretion is inhibited [25].

Fig. 6

Effects of Ca 2 + concentration on transamidation reaction of TG2 in nucleus and mitochondria. 200 μg of INS-1E proteins from the nuclear/mitochondrial enriched fraction were first resolved on 7 cm strips 3–10 NL and then electroblotted. Western blot was revealed using streptavidin–HPR 1/1000 in 2% BSA in TTBS followed by ECL. Cells labeled with A25peptide. All blots exposed on film for 3 min except for D exposed for 15 s. A–D: INS-1E maintained at 2.5 mM glucose; E–H: INS-1E stimulated at 15.5 mM glucose for 2 min. B and F: 10 μM tolbutamide on TG2 transamidation results in an activation of TG2 already at 2.5 mM glucose; C and G: 100 mM diazoxide on TG2 transamidation results in a completely blunted activation; D: 30 mM KCl on TG2 transamidation results in a maximal activation. I: The graph summarizes the kinetic trends of the TG2 transamidation reaction in function of changes in number of spots per condition. The number is reported as the mean and SE of two independent experiments.

Identification of TG2 substrates using MALDI TOF/TOF

The TG2 substrates biotinylated by the transamidation reaction were purified by affinity chromatography from nuclear/mitochondrial enriched fractions of glucose-stimulated INS-1E. Since we loaded 3 mg of proteins from the enriched fraction onto the streptavidin affinity chromatography column, and we recovered about 300 μg of biotinylated proteins, we estimated that roughly 10% of total proteins were targeted by transamidation activity. These purified proteins were resolved using 2DE and spots analyzed using MS. Of 120 picked spots from two gels stained with Coomassie Blue Silver, 74 (61.6%) had identification (score > 51) using MALDI TOF/TOF. As shown in Supplemental Table 1 some proteins were identified more than once, while some were identified once only. As a result, 32 different proteins were identified (Fig. 7A). Among these 32, 10 belong to the mitochondrion, 3 are residents in the reticulum, 9 are nuclear and 9 are constituents of cytoskeleton (Fig. 7B). GAPDH could have both cytoplasmic and nuclear localization.

Fig. 7

Spot identification by MS and MS/MS: A: 300 μg of biotinylated proteins from two different affinity chromatography purification on magnetic beads coupled with streptavidin was resolved using 2DE starting from nuclear/mitochondrial enriched fraction (2 min at 15.5 mM glucose). Gel stained with Coomassie Blue Silver. All clear defined spots were picked. Identified proteins were indicated by SwissProt number. TUB = Tubulin: P69897/Q3KRE8/P85108/Q6P9T8/Q4QRB4. B: Schematic representation of the distribution of the identified proteins.

The TG2 substrates biotinylated by the transamidation reaction were purified by affinity chromatography from nuclear/mitochondrial enriched fractions of glucose-stimulated INS-1E. Since we loaded 3 mg of proteins from the enriched fraction onto the streptavidin affinity chromatography column, and we recovered about 300 μg of biotinylated proteins, we estimated that roughly 10% of total proteins were targeted by transamidation activity. These purified proteins were resolved using 2DE and spots analyzed using MS. Of 120 picked spots from two gels stained with Coomassie Blue Silver, 74 (61.6%) had identification (score > 51) using MALDI TOF/TOF. As shown in Supplemental Table 1 some proteins were identified more than once, while some were identified once only. As a result, 32 different proteins were identified (Fig. 7A). Among these 32, 10 belong to the mitochondrion, 3 are residents in the reticulum, 9 are nuclear and 9 are constituents of cytoskeleton (Fig. 7B). GAPDH could have both cytoplasmic and nuclear localization.

Identification of histone H3 as TG2 substrate during FPIS

Previously it has been shown that histone H3 is a TG2 substrate in different cells types [20-22] and that it plays a role in regulating the expression of insulin gene [33]. We therefore focused our attention on the possible presence of histone H3 among the low molecular weight TG2-substrates. Because it is difficult to resolve proteins with a strong basic pI using 2DE, we addressed this issue using one dimensional electrophoresis.

As already visible in 2DE analyses (Figs. 5–7), we confirmed that proteins with low molecular weight are TG2 substrates in transamidation reaction both in resting (Fig. 8A1) and glucose-stimulated cells (Fig. 8A2). Western blot analysis with a specific antibody performed on purified biotinylated substrates showed that histone H3 is one of the low molecular weight substrates of TG2 both in resting and glucose-stimulated INS-1E (Fig. 8B1 and B2). The same result was obtained in human pancreatic islets where histone H3 is present among TG2 substrates of low molecular weight in resting (5.5 mM glucose) islets (Fig. 8C1) and in stimulated (11 mM glucose) islets as well (Fig. 8C2).

Fig. 8

Histone H3 is a TG2 substrate during FPIS in INS-1E and in human pancreatic islets. 70 μg of INS-1E proteins taken from the nuclear/mitochondrial enriched fraction were first resolved using classical SDS-PAGE. 15% acrylamide gel 7 × 7cm and then electrotransferred. Cells labeled with A25 peptide. Images collected using ChemiDoc XRS + exposing for 60 s. A: Total biotinylated INS-1E nuclear/mitochondrial proteins. Western blot performed on the total lysate of nuclear/mitochondrial fraction revealed using streptavidin–HPR; 1: Resting cells. 2: Cells stimulated for 2 min with 15.5 mM glucose. B: Purified biotinylated proteins from INS-1E cells. 1: Resting cells. 2: Cells stimulated for 2 min with 15.5 mM glucose. C: Purified biotinylated proteins from human pancreatic islets. 1: Human pancreatic islets grown in 5.5 mM glucose. 2: Human pancreatic islets stimulated 3 min with 11 mM glucose. In B and C the Western blot was revealed using rabbit polyclonal antiserum against histone H3. Arrows indicate bands corresponding to histone H3 at the expected MW according to the manufacturer. In C lower bands corresponding to histone H3 are over exposed and for this reason appear white.

Discussions

Transglutaminase 2 is the most abundantly expressed member of the transglutaminase family and in general it is considered capable of exerting effects on cell growth, differentiation, apoptosis, cell adhesion [34-36] and scaffold activities [5] via multiple activities, including primarily transamidase [1-4] and GTPase function [37,38], but also more controversial ones such as protein disulfide isomerase [39] and kinase [40]. The presence of TG2 has been described in various parts of a cell such as plasma membrane, cytosol, mitochondria and nucleus, as well as the extracellular matrix [4-7]. In normal cells with intracellular Ca2 + concentrations in the 10–20 nM range, TG2 acts mainly as a GTPase and participates in the transmembrane signaling of phospholipase Cδ, thereby supporting cell growth [37,38]. In β pancreatic and rat insulinoma INS-1E cells, where Ca2 + concentration is around 100 nM already at resting condition (2.5 mM glucose) [28], TG2 mainly functions as a transglutaminase, increasing its transamidating activity during first phase insulin secretion [25] when Ca2 + concentration reaches the value of 450 nM in INS-1E [28] and 300 nM in isolated islets [41]. In the model we previously proposed [27], TG2 during first phase insulin secretion acts as a modifying enzyme by inserting post translational modification(s) in various proteins both in the cytoplasm and in the granular/microsomal fraction, [25,42].

According to previous reports [4,5], in this study we showed that in INS-1E insulinoma cell line the TG2 localizes mainly in the cytoplasm in association with insulin granules but also in mitochondria and nucleus where it acts as transamidase.

At variance with what is seen in fibroblasts [7] that have an extended and organized extracellular matrix with an important presence of TG2, in INS-1E this enzyme is scarcely distributed in the extracellular matrix, where it is organized in fibril-like filaments that follow the pattern of fibronectin. This observation, which confirms what has been shown previously in human pancreas using electron microscopy [25], seems to suggest a marginal role of TG2 in the organization of extracellular matrix of INS-1E.

In the present work we have shown that TG2 is capable of transamidating activity also in the mitochondria and in the nucleus of INS-1E. In addition activity was coordinated with the changes of concentration of cytoplasmic Ca2 + expected during first phase insulin secretion. Of note, the kinetics of the transamidation reaction was comparable to what we observed in the cytoplasm and in the microsomal/granular compartments [25]. This coordination of TG2 action in different compartments seems attributable to changes in Ca2 + concentration, because diazoxide is able to suppress transamidase activity in the nuclear/mitochondrial fraction during first phase insulin secretion, as it does in the cytoplasm [25], while the opposite is observed in the presence of KCl. To the best of our knowledge, this is the first reported observation of the same enzymatic activity taking place in a coordinated manner in the cytoplasm, in mitochondria and in the nucleus during first phase insulin secretion. We have hypothesized that the increased TG2 transamidating activity – i.e. Ca2 + dependent – observed in the nucleus and in mitochondria is associated with the active transport of Ca2 + by specific channels similar to those active on plasma membrane [43-46]. Interestingly, Ca2 + homeostasis in the mitochondrion is maintained, among other proteins, by the action of VDACs, and we identify two members of VDAC protein family as TG2 substrates during first phase insulin secretion in mitochondrion (Suppl. Table 1 and Fig. 7). We have also identified VDAC as TG2 substrates in the granular/microsomal cellular fraction [25], and this is due to the fact that some of these proteins are constituents of the insulin granule membrane [47]. Furthermore, VDACs are part of a network [48] which include the inositol 1,4,5 triphosphate receptor, Stress-70 protein (also known as GRP75), a mitochondrial chaperone that facilitates mitochondrial Ca2 + uptake, and calreticulin. Our group previously identified these last two proteins as substrates of TG2 during first phase insulin secretion in INS-1E [25]. These observations together with the fact that this network works in conjunction with tricarboxylic acid cycle and respiratory complexes suggest that TG2 could also play a regulatory role in the mitochondrion of INS-1E cells as reported previously in other cellular systems [14,15]. This role could contribute to maintaining the integrity of respiratory chains and in the functioning of this network.

In the present work we have shown that TG2 is capable of transamidating activity also in the mitochondria and in the nucleus of INS-1E. In addition activity was coordinated with the changes of concentration of cytoplasmic Ca2 + expected during first phase insulin secretion. Of note, the kinetics of the transamidation reaction was comparable to what we observed in the cytoplasm and in the microsomal/granular compartments [25]. This coordination of TG2 action in different compartments seems attributable to changes in Ca2 + concentration, because diazoxide is able to suppress transamidase activity in the nuclear/mitochondrial fraction during first phase insulin secretion, as it does in the cytoplasm [25], while the opposite is observed in the presence of KCl. To the best of our knowledge, this is the first reported observation of the same enzymatic activity taking place in a coordinated manner in the cytoplasm, in mitochondria and in the nucleus during first phase insulin secretion. We have hypothesized that the increased TG2 transamidating activity – i.e. Ca2 + dependent – observed in the nucleus and in mitochondria is associated with the active transport of Ca2 + by specific channels similar to those active on plasma membrane [43-46]. Interestingly, Ca2 + homeostasis in the mitochondrion is maintained, among other proteins, by the action of VDACs, and we identify two members of VDAC protein family as TG2 substrates during first phase insulin secretion in mitochondrion (Suppl. Table 1 and Fig. 7). We have also identified VDAC as TG2 substrates in the granular/microsomal cellular fraction [25], and this is due to the fact that some of these proteins are constituents of the insulin granule membrane [47]. Furthermore, VDACs are part of a network [48] which include the inositol 1,4,5 triphosphate receptor, Stress-70 protein (also known as GRP75), a mitochondrial chaperone that facilitates mitochondrial Ca2 + uptake, and calreticulin. Our group previously identified these last two proteins as substrates of TG2 during first phase insulin secretion in INS-1E [25]. These observations together with the fact that this network works in conjunction with tricarboxylic acid cycle and respiratory complexes suggest that TG2 could also play a regulatory role in the mitochondrion of INS-1E cells as reported previously in other cellular systems [14,15]. This role could contribute to maintaining the integrity of respiratory chains and in the functioning of this network.

To identify nuclear substrates we used both 2DE and monodimensional polyacrylamide gel. Among nuclear proteins identified as putative TG2 substrates we found four members of the heterogeneous nuclear ribonucleoprotein family (hnRNP), barrier to autointegration factor (BAF), the heat shock cognate 71 kDa protein, the coiled-coil domain-containing protein 105, the CCR4-NOT transcription complex subunit 10 and the bifunctional protein NCOAT. All these proteins are newly identified substrates of TG2 and are involved at different levels in the regulation of gene expression. The hnRNPs are RNA binding proteins and they complex with heterogeneous nuclear RNA. These proteins associate with pre-mRNAs in the nucleus and appear to influence pre-mRNA processing and other aspects of mRNA metabolism and transport [49,50]. Thus, if nuclear TG2 can modulate INS-1E gene expression during first phase insulin secretion through the post-translational modification of hnRNPs (and histone H3, see below), it is conceivable that the specific reduction of this enzymatic activity associated with mutations which are located in close proximity to the TG2 transamidation catalytic site [51] could affect the stability of pre-mRNAs and/or their maturation.

Another protein which acts on RNA and RNA processing is CCR4-NOT transcription complex subunit 10. This complex is one of the major cellular mRNA deadenylases and is linked to various cellular processes including bulk mRNA degradation, miRNA-mediated repression, translational repression during translational initiation and general transcription regulation [52].

BAF plays fundamental roles in nuclear assembly, chromatin organization, gene expression and gonad development, may strongly compress chromatin structure and may be involved in membrane recruitment and chromatin decondensation during nuclear assembly [53].

Bifunctional protein NCOAT is an interesting protein which catalyzes the hydrolysis of terminal non-reducing N-acetyl-d-hexosamine residues in N-acetyl-beta-d-hexosaminides and binds both acetylated and non-acetylated histone H4 tail, even if acetylation on ‘Lys-8’ of histone H4 abolishes binding [54]. Moreover, NCOAT seems to play a role in insulin secretion and may be involved in secretion of other glucose-regulated hormones [55,56].

The post-translational modifications on these nuclear proteins performed by TG2 could be another level of control gene expression in INS-1E.

Because of highly basic pI of histone and all nuclear proteins, which directly interact with DNA, we were unable to separate this group of proteins using 2DE. Consequently, we used a different approach: a combination of classic SDS-PAGE and Western blot. We looked specifically for histone H3 because on the one hand it has been described as a TG2 substrate [20,57] and on the other it has been considered important in transcription of insulin gene(s) [33,58].

In this work we showed that histone H3 is a TG2 substrate in living cells, both INS-1E cells and human pancreatic islets, in resting and glucose-stimulated conditions. Previously it has been shown that TG2 acts on the N-terminal tails of histone H3 when native nucleosomes are used in an in vitro reaction [20]. It is known that modification of histone N-terminal tails, such as lysine acetylation, is strongly connected to active transcription of genes [59]. Moreover, H3 methylation by methyltransferase Set7/9 modulates transcription of Ins1 and Ins2 genes in βTC3 cells by maintaining the euchromatin structure at these genes [58].

Interestingly, it has been proposed that TG2 might phosphorylate histone H3 on Ser 10 [57] a site that has previously been shown to be important in the expression of immediately early genes [60] and crucial for chromosome condensation and cell cycle progression [61,62]. It is worth noting that the monosaccharide beta-N-acetylglucosamine can be alternatively added on Ser10 of histone H3, resulting in an O-glycosylation of this residue [63]. This post translational modification can take place on serine and threonine residues competing with phosphorylation [64]. Therefore interplay exists between phosphorylation and O-glycosylation. More generally, transamidase activity of TG2 might regulate gene expression by affecting the different histone H3 post-translational modifications (acetylation, methylation, O-glycosylation and phosphorylation) which can result in a change of DNA architecture.

The evidence that TG2 might phosphorylate histone H3, as the possible action of TG2 on the N-terminal tails of histone H3, has been obtained in in vitro reactions. Our results indicate that histone H3 could be a substrate of TG2 in living cells (both INS-1E and human pancreatic islet) in a transamidation reaction. In fact, due to the high value of Ca2 + concentration in both INS-1E and human pancreatic islets, TG2 can only catalyze transamidation reactions.

In addition, since we never observed the appearance of high molecular weight bands on Western blot, including histone H3, we hypothesized that during first phase insulin secretion TG2's transamidation reaction does not lead to the formation of covalently linked homo/hetero-dimers or -polymers including histone H3, as on the contrary previously reported [22]. Furthermore, as previously shown in the cytoplasm [25], we also hypothesized that a large part of TG2's action on histone H3 and on other nuclear substrates is not a covalent modification but it can be removed at the end of first phase insulin secretion when Ca2 + concentration reaches lower values.

We suppose that during first phase insulin secretion, TG2 in physiological condition catalyzes post-translational modifications, which might represent a mechanism by which INS-1E and human islet cells could coordinate different events in the nucleus, possibly through the action and the role performed by histone H3 and other nuclear proteins which can belong to an extended network and which are also TG2 substrates during first phase insulin secretion itself. For instance, recent studies have reported that BAF could participate in an articulated network which consists, among other proteins, in BAF itself, histone H3, hnRNPs, nucleophosmin and actin [53,65]. The two latter proteins have already been identified as TG2 substrates [25]. All together these results suggest that TG2 might participate in the nucleus in the modulation of the expression of different genes, such as insulin gene itself, and genes involved in insulin secretion and/or glucose metabolism, and that this role could be performed not only in a cellular model such as INS-1E, but also in human pancreatic islets as shown previously [25]. How TG2 may play a specific role in this interaction remains to be established, but the temporal link between glucose-stimulated, first-phase insulin secretion and TG2's action on H3 warrants further studies.

We are aware that the data presented in this work along with those previously published by us [25,66] are in conflict with those of Iismaa and collaborators who recently reported normal glucose metabolism and insulin secretion in two TG2 knock-out mice with different genetic background (B6 and 129) [67]. To reconcile our findings with those of Iismaa, one could hypothesize that TG2 in the pancreatic beta cell acts as an accessory enzyme that facilitates/coordinates a number of events occurring during insulin secretion, but it is – nevertheless – dispensable. In addition taking into account that TG2 could play a different role in insulin secretion among different species, it is conceivable to think that, if TG2 transamidating activity is reduced or absent, human pancreatic beta cell can function properly for some time, but when the lack of all these “accessory” activities piles up, this may result in a defect in insulin secretion. In support of this view, two probands carrying TG2 mutations we previously described developed diabetes as adults [51].

Conclusions

In conclusion, our results support the idea that TG2 acts as a transglutaminase in the nucleus and in the mitochondrion of INS-1E cells and human pancreatic islets during first phase insulin secretion, catalyzing a reaction which results in post-translational modification(s) of its substrates.

In the mitochondrion TG2 may be involved in Ca2 + homeostasis. In the nucleus there are two groups of TG2 effectors: the proteins which act on the condensation status of chromatin and on epigenetic modifications such as BAF and histone H3, and the proteins which contribute to the assembly of nucleosome and have a role in the maturation of pre-mRNA. Both these functions may be involved in the transcriptional regulation of beta-cell genes, including the insulin gene.

The following is the supplementary data related to this article.

Supplementary Table 1

Complete list of all identified proteins from INS-1E starting from nuclear/mitochondrial enriched fraction. Cells stimulated at 15.5 mM glucose for 2′.

Supplementary data to this article can be found online at http://dx.doi.org/10.1016/j.jprot.2013.11.011.