Macroscopic Differences in HMGA Oncoproteins Post-Translational

Mar 24, 2009 - Macroscopic Differences in HMGA Oncoproteins Post-Translational Modifications: C-Terminal Phosphorylation of HMGA2 Affects Its DNA ...
0 downloads 0 Views 2MB Size
Download PDF
Macroscopic Differences in HMGA Oncoproteins Post-Translational Modifications: C-Terminal Phosphorylation of HMGA2 Affects Its DNA Binding Properties Riccardo Sgarra,† Elisa Maurizio,† Salvina Zammitti, Alessandra Lo Sardo, Vincenzo Giancotti, and Guidalberto Manfioletti* Department of Life Sciences, University of Trieste, Trieste, Italy Received February 3, 2009

HMGA is a family of nuclear proteins involved in a huge number of functions at the chromatin level. It consists of three members, HMGA1a, HMGA1b, and HMGA2, having high sequence homology and sharing the same structural organization (three highly conserved DNA-binding domains, an acidic C-terminal tail, and a protein-protein interaction domain). They are considered important nodes in the chromatin context, establishing a complex network of interactions with both promoter/enhancer sequences and nuclear factors. They are involved in a plethora of biological processes and their activities are finely tuned by several different post-translational modifications. We have performed an LC/MS screening on several different cell lines to investigate HMGA proteins expression and their posttranslational modifications in order to detect distinctive modification patterns for each. Our analyses evidenced relevant macroscopic differences in the phosphorylation and methylation patterns of these proteins. These differences occur both within the HMGA family members and in the different cell types. Focusing on HMGA2, we have mapped its in vivo phosphorylation sites demonstrating that, similarly to the HMGA1 proteins, it is highly phosphorylated on the acidic C-terminal tail and that these modifications affect its DNA binding properties. Keywords: Cancer • Chromatin • Phosphorylation • Methylation • Mass spectrometry

Introduction Nuclear processes rely on the coordinated action of several factors which modulate chromatin dynamics. Besides the histones, which represent the molecules that for antonomasia are at the fundamental level of DNA structuring, another class of proteins which is involved in chromatin structure modulation is the High Mobility Group (HMG). This group is composed of several members that can be subdivided into three families on the basis of their DNA binding domains, that is, the HMGA, HMGB, and HMGN families (see ref 1 for a general review of HMG proteins). All these proteins share a common feature: the ability of binding to and inducing structural changes in DNA and chromatin.1 While HMGB and HMGN are ubiquitous, HMGA proteins are expressed at a high level during embryonic development, but are barely detectable in differentiated or nonproliferating cells, and it is noteworthy that they are highly re-expressed following neoplastic transformation.2 This family is composed of three members: HMGA1a and HMGA1b (together referred to as HMGA1), which are splicing variants of the same gene, and HMGA2, which is the product of another gene, showing a high degree of homology (about 50%) with * To whom correspondence should be addressed. Guidalberto Manfioletti, Department of Life Sciences, University of Trieste, Via L. Giorgieri, 1, 34127, Trieste (TS), Italy. Phone: +39 040 5583675. Fax: +39 040 5583694. E-mail: [email protected]. † These authors contributed equally to the work.

2978 Journal of Proteome Research 2009, 8, 2978–2989 Published on Web 03/24/2009

HMGA1. The causal involvement of HMGA proteins in cancer development has been well-established and recently experimental evidence highlights that HMGA actively participate in cell transformation by affecting different multiple proliferative and key regulative nodes.3 Indeed, it has been demonstrated that they affect the expression of several genes involved in neoplastic transformation and progression,3 alter E2F1 regulated gene expression,4 and deregulate p53 functions.5,6 In addition, proteomic data aimed at identifying their molecular partners evidenced that HMGA are highly connected in the chromatin protein network and suggested an involvement of these proteins also in RNA processing, DNA repair, and chromatin structure dynamics.7-9 The high levels of HMGA in both embryonic and cancer cells underline the concept that these proteins play multiple functions inside the nucleus and that they are involved in different activities. Their multifunctionality is probably due to their natively disordered state and their modular organization in five domains (three highly conserved DNA binding domains, a protein/protein interaction domain, and an acidic C-terminal tail; see Supporting Information Figure 1) that confer to them the ability to adapt to and interact with several different DNA sequences and protein partners. The three proteins differ from each other in the sequences that separate these five domains, both in terms of length and amino acid composition.2 10.1021/pr900087r CCC: $40.75

 2009 American Chemical Society

HMGA Macroscopic PTMs HMGA proteins, as well as histones, are heavily posttranslationally modified (for a recent review on HMG PTMs, see ref 10). It has been shown that HMGA1a and HMGA1b are phosphorylated by Casein kinase II (CK2) on three serines located in the C-terminal tail (S98, S101 and S102; numbers refer to the HMGA1a isoform, and since HMGA proteins always miss their initial Methionine, we consider the N-terminal Serine, which is constitutively acetylated, as aa number 1), but the function of these phosphorylations has never been uncovered. Several kinases modulate HMGA DNA-binding affinities through phosphorylation at specific sites. This is the case of p34cdc2, PKC, and HIPK2. It has been shown that, besides the N-terminal acetylation, HMGA1a can be acetylated by CBP and PCAF at lysine 64 and 70 resulting in the stabilization/ destabilization of macromolecular complexes (called enhanceosomes) in which HMGA proteins are key assembly factors. Recently, it has emerged that HMGA proteins, in particular HMGA1, can be also methylated at various residues; indeed, several Protein Arginine Methyl Transferases (PRMTs) display enzymatic activity versus HMGA. This is the case of PRMT6, which has been demonstrated to methylate HMGA1a protein at R57 and R59 both in vitro and in vivo and of PRMT1 and PRMT3 that can methylate HMGA1 proteins at R25 and R23 in vitro, respectively. Several other post-translational modifications (PTMs) have been mapped in HMGA proteins and they are all summarized in Supporting Information Table 1 and Supporting Information Figure 1. HMGA are thus heavily posttranslationally modified; in fact, HMGA1a, which is composed of 106 amino acid residues, shows up with 39 residues potentially affected by PTMs. Some of the above-mentioned modifications have been demonstrated to be transient and to regulate specific processes. This is, for instance, the case of HMGA1 acetylation by PCAF and CBP during the regulation of the expression of the interferon-β gene.11 Other PTMs are otherwise considered constitutive, such as C-terminal phosphorylations which are assumed to occur on the whole amount of HMGA proteins independently from the cellular context. In this work, we have systematically screened the posttranslational state of HMGA proteins, by LC/MS analyses, in several cell lines of different origin and phenotype. With our approach, we could evidence that, excluding the N-terminal acetylation, only C-terminal phosphorylations and the R25 methylation occur at high levels. The comparison of high quality mass spectra of the three HMGA proteins allows to evidence both intrafamily and intercellular type differences, both in the levels and in the nature of HMGA PTMs: (i) the level of C-terminal phosphorylation is cell-type dependent and strongly differs in the two HMGA1 isoforms (HMGA1a and HMGA1b) and HMGA2; (ii) HMGA1a is the only HMGA protein heavily methylated in a cell-type dependent manner. Moreover, we have demonstrated that C-terminal HMGA2 phosphorylation is involved in modulation of DNA binding properties.

Materials and Methods Cell Culture. PC-3, DU145, FRO, ARO, NIM 1, and TPC-1 were grown in RPMI 1640 (HyClone); HEK293, C-4I, ME-180, HeLa, Chang, Hep G2, Hep3B, PLC/PRF/5, HBL100, MDA-MB157, MDA-MB-468, MDA-MB-231, and MCF7 were grown in DMEM (Dulbecco’s Modified Eagle’s Medium). Growth media contained 10% fetal bovine serum, 2 mM L-glutamine, penicillin

research articles (100 U/mL), and streptomycin (100 µg/mL). PZ-HPV-7 cells were grown in Keratinocyte serum-free medium (Invitrogen). Cells were grown at 37 °C in humidified 5% CO2 incubator and collected under subconfluence conditions. Protein Extraction and LC/MS Analyses. Total HMG proteins and histone H1 from 20 different cell lines were selectively extracted with 5% perchloric acid (PCA) (w/v), precipitated by acetone-HCl (50 mM), resuspended in water and checked by SDS-PAGE (T ) 15%). For LC-MS analyses, PCA extracts were conditioned to a final concentration of 0.1% trifluoroacetic acid (TFA). Reverse-phase HPLC chromatography and LC-MS analyses were carried out as previously described12 with a PerkinElmer Life Sciences apparatus (Series 200 LC Pump and 785A UV-visible Detector) using a Vydac Protein C4 column (2.1 × 150 mm, 5 µm) and an interfaced single quadrupole mass spectrometer (PE SCIEX, API1). For the purification of the endogenous HMGA2, HPLC analyses were performed with a RP Supelco Discovery BIO Wide Pore C5 column (2.1 × 250 mm, 5 µm). In both cases, a water/acetonitrile gradient with 0.1% TFA as a modifier was used. Recombinant HMGA Proteins Production. Full-length recombinant human HMGA1a and HMGA2, and the human HMGA2 C-terminal deletion mutant (HMGA2 1-93) proteins were overexpressed in Escherichia coli as previously described.7 Recombinant proteins were extracted from bacteria by 5% PCA (w/v), purified by RP-HPLC on a Waters Symmetry Shield RP8 column (4.6 × 250 mm, 5 µm), and checked for their purity and molecular masses by both SDS-PAGE and MS. Purified proteins were quantified according to a modified Waddell method.7 In Vitro Phosphorylation of HMGA Proteins by CK2. CK2 phosphorylation was performed by incubating 5 µg of recombinant HMGA proteins with 100-250 Units of CK2 (New England BioLabs) in 50 µL of reaction volume (20 mM Tris/ HCl, pH 7.5, 50 mM KCl, 10 mM MgCl2, and 200 µM ATP) at 37 °C for 16 h. In order to obtain HMGA2 with different phosphorylation levels, we performed the same kind of phosphorylation assays, but using 2.5 Units of CK2/5 µg HMGA2 and stopping the reaction at 30 min, 45 min, 1 h, 2 h, 4 h, or 16 h. Enzymatic Digestions. In vitro phosphorylated proteins were purified by HPLC using a RP Supelco Discovery BIO Wide Pore C5 column (2.1 × 250 mm, 5 µm) as reported above. HMGA2 purified proteins (in vitro phosphorylated and the endogenous one) were dried in Speed-Vac concentrator and digested with trypsin (1/25 enzyme/substrate ratio) in 50 mM NH4HCO3 for 16 h at 37 °C. Phosphopeptides enrichment was carried out with the PhoshoProfile I Phosphopeptide Enrichment Kit (SIGMA) following the manufacturer’s instructions. LC-MS/MS Analyses. Either nonenriched and phosphopeptide enriched samples were analyzed by LC-MS/MS using a PerkinElmer Life Sciences apparatus (Series 200 LC Pump) interfaced to an HCT Ultra IT (Bruker Daltonics). HPLC separations were performed with a Waters Symmetry C18 column (1.0 × 150 mm, 3.5 µm) with a water/acetonitrile gradient (from 2% to 60% acetonitrile in 60 min) and 0.1% FA as a modifier. In order to obtain a flow of about 40 µL/min, a custom-made splitting system was used. MS/MS parameters were the following: full MS scan, m/z 100-1500; number of precursors selected, 3; fragmentation amplitude, 1.00 V with the Smart Frag active (from 50 to 200% of the fragmentation amplitude selected); MS/MS scan, m/z 100-2300; active excluJournal of Proteome Research • Vol. 8, No. 6, 2009 2979

research articles sion after accumulation of 2 MS/MS spectra for a period of 2 min; ion charge control (ICC) active allowing the storage of a maximum of 200 000 ions in a maximum accumulation time of 200 ms. LC-MS/MS were elaborated with Data Analysis software (v.3.4, Bruker Daltonics). Masses of tryptic peptides were searched by both exporting our analyses as mgf files as input for the MASCOT MS/MS Ion Search option (www.matrixscience.com) and by manual inspection of entire analyses. MS/MS spectra were interpreted by manual inspection with the auxilium of MASCOT MS/MS Ion Search results and Protein Prospector (http://prospector.ucsf.edu) MS-Product option. Electrophoretic Mobility Shift Assays (EMSAs). CK2 phosphorylated and unphosphorylated HMGA2 used for EMSA analyses were prepared starting from a common phosphorylation solution which was split in two equivalent aliquots. In one part, we added the active enzyme, whereas in the other one, we added the same amount of enzyme but heatinactivated (3 min at 95 °C). At the end of the phosphorylation reaction, an aliquot was used for checking the modification profile of HMGA2 by MS analyses. The remaining solutions were directly used for EMSAs. EMSAs were carried out by incubating proteins (1-16 pmol as indicated) with 100 fmol of labeled DNA in 20 µL reactions containing 180 mM NaCl, 1 mM MgCl2, 0.01% BSA, 8% glycerol, 10 mM Tris/HCl, pH 7.9, at room temperature for 10 min. After incubation, proteinbound DNA and free DNA were separated on a native 7% polyacrylamide gel run in 0.5× TBE at 15 V/cm at 4 °C. Visualization was achieved by autoradiography. The DNA probes used were a double-stranded B-type oligonucleotide (E3) and a four-way junction DNA (4WJ). The sequences of the probes are (for E3 only the upper strand sequence is shown): E3: 5′-AGAAAAACTCCATCTAAAAAAAAAAAAAAAAAAAAAAAAAAACA-3′ 4WJ Leg1: 5′-CCCTATAACCCCTGCATTGAATTCCAGTCTGATAA-3′ 4WJ Leg2: 5′-GTAGTCGTGATAGGTGCAGGGGTTATAGGGG-3′ 4WJ Leg3: 5′-AACAGTAGCTCTTATTCGAGCTCGCGCCCTATCACGACTA-3′ 4WJ Leg4: 5′-TTTATCAGACTGGAATTCAAGCGCGAGCTCGAATAAGAGCTACTGT-3′

Sgarra et al. Table 1. Cell Lines Used in LC/MS Screening expression human cell line

origin

characteristics

HMGA1 HMGA2

1

HEK293

Kidney

+

+

2 3

C-4 I ME-180

Cervix

+ +

n.d.a n.d.

+

n.d.

+ + +

n.d. + +

+ +

+ n.d.

+ +

n.d. n.d.

+

n.d.

Embryo kidney, tumorigenic Carcinoma, tumorigenic Epidermoid carcinoma, tumorigenic HeLa Adenocarcinoma, tumorigenic Chang Liver Normal hepatocyte Hep G2 Hepatocellular carcinoma Hep3B Hepatocellular carcinoma, tumorigenic PLC/PRF/5 Hepatoma, tumorigenic HBL 100 Breast Nontumor SV40 immortalized MCF7 Adenocarcinoma MDA-MB-157 Medulallary carcinoma, tumorigenic MDA-MB-468 Adenocarcinoma, tumorigenic MDA-MB-231 Adenocarcinoma, tumorigenic PZ-HPV-7 Prostate Immortalized normal epithelium PC-3 Adenocarcinoma, tumorigenic DU 145 Carcinoma, tumorigenic NIM 1 Thyroid Papillary carcinoma, low tumorigenic TPC-1 Papillary carcinoma FRO Anaplastic carcinoma, tumorigenic ARO Anaplastic carcinoma, tumorigenic

4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 a

+

n.d.

+

+

+

+

+ +

n.d. +

+ +

+ n.d.

+

n.d.

N.d.: not detected in our LC/MS analyses.

Results and Discussion Liquid Chromatography/Mass Spectrometry Analyses of HMGA and HMGN Proteins. Several cell lines (listed in Table 1), differing in their origin (kidney, cervix, liver, breast, prostate, and thyroid) and in their tumorigenic phenotype, have been selected for our LC/MS screening. HMGA, HMGB and HMGN proteins, together with Histone H1, have been selectively extracted from growing cells and analyzed by LC/MS. An example of an LC analysis from Hep G2 cells is shown in Figure 1, where it is possible to visualize the whole chromatographic separation. Mass spectra relative to chromatographic fractions F1-F5 (see Figure 1) have been deconvoluted to obtain reconstructed mass spectra (Figure 2, panels A-F). HMG proteins were identified by comparison between theoretical and experimental molecular mass values (see Supporting Information Table 2). The experimental conditions optimized for chromatographic separation allowed us to analyze HMGA proteins, which are our major interest, together with the coeluting HMGN proteins. Since with our LC/MS analyses we looked at full-length proteins, we could assume that the presence of phosphate groups or other PTMs does not substantially alter the ionization 2980

Journal of Proteome Research • Vol. 8, No. 6, 2009

Figure 1. Chromatographic separation of HMG proteins. Elution profile of PCA extracted HMG and histone H1 separated by RPHPLC. F1, HMGN2; F2, HMGN3 and HMGN4; F3, HMGA1b and HMGN1; F4, HMGA2; F5, HMGA1a. Detection was performed measuring absorbance at 220 nm.

efficiency of HMG proteins. The intensity of the various reconstructed peaks can therefore be considered as indicative of their relative abundance, at least within the same protein.12 Therefore, with this approach, we were able to investigate the three HMGA and the four HMGN (HMGN1, N2, N3, and N4) proteins. All the mass spectra relative to HMGA1a, HMGA1b, and HMGA2 proteins expressed by the different cell lines considered in this work are reported in Supporting Information Figure 2. Identification of HMGA and HMGN Post-Translational Modifications. From the reconstructed mass spectra reported in Figure 2 (and Supporting Information Figure 2), it is possible to note that the three HMGA and the four HMGN have substantially different macroscopic post-translational modifications patterns.

HMGA Macroscopic PTMs

research articles

Figure 2. Reconstructed mass spectra of HMGA and HMGN proteins from Hep G2 cells. PCA extracted HMGA and HMGN proteins were analyzed by LC-MS. For each peak of the chromatographic separation shown in Figure 1 (F1-F5), registered mass spectra were deconvoluted and reconstructed mass spectra obtained for each of the various HMG (panels A-F). The identity of each protein peak was obtained comparing experimental with theoretical molecular masses (Da). 1P, 2P, 3P, 4P, and 5P: mono-, di-, tri-, tetra-, and pentaphosphorylation.

HMGA are highly phosphorylated proteins showing relevant intrafamily differences. HMGA1a and HMGA1b share the same phosphorylation status being distributed among the two- and three-phosphorylated forms (2P and 3P), whereas HMGA2 shows a broader phosphorylation status (1P, 2P, 3P, 4P, and 5P) (Figure 2 and Supporting Information Figure 2). Our data are consistent with those previously reported10 that clearly demonstrate that HMGA1 proteins are constitutively phosphorylated by CK2 on their acidic C-terminal tail (S98, S101, and S102sHMGA1a isoforms...EEEEGIS98QES101S102EEEQ). On the other hand, HMGA2 C-terminal tail (...EET96EET99 S100S101QES104AEED) differs from that of HMGA1 because there are up to five potentially phosphorylatable residues (T96, T99, S100, S101, and S104) by CK2. Two of them, S100 and S104,

are within a canonical consensus sequence for CK2 (S/T-xxE/D), whereas the other three are embedded in an acidic context (phosphorylated S/T included), therefore becoming potential sites as well.13 It has already been demonstrated that HMGA2 is a phosphoprotein14 and that it can be a substrate for CK2,15 p34cdc2 15 and Nek216 kinases. However, no experimental data regarding HMGA2 in vivo phosphorylation degree and location were previously available. HMGN are prevalently unphosphorylated proteins; in fact, HMGN2 and HMGN4 were found to be completely unmodified, whereas only low-abundance of monophosphorylated forms of HMGN1 and HMGN3 and of diphosphorylated HMGN1 is detectable. Journal of Proteome Research • Vol. 8, No. 6, 2009 2981

research articles

Sgarra et al. and 10606.9 peaks, respectively) by +16 Da (10543.7 and 10623.0 peaks), which are consistent with hydroxylation at the level of a P, K, or D residue. Substantially, no other post-translational modifications are detectable in HMGA and HMGN proteins. However, our experiments are not in discordance with literature data describing several HMGA PTMs not detected in our screening. In fact, with our approach, we looked at the PTMs affecting the bulk of HMGA proteins, and therefore we refer to them as Macroscopic PTMs.

Figure 3. Comparison of HMGA1a, HMGA1b and HMGA2 sequences comprising the first two AT-hooks. HMGA sequences from the first to the second AT-hooks are shown. The arginine residue which is in vivo methylated in HMGA1a (R25) is evidenced in both HMGA1a and HMGA1b sequences. The corresponding arginine in HMGA2 sequence (R28) is also evidenced. The DNA-binding domains (AT-hooks) are underlined and the distance in terms of number of amino acid residues between R25 (28) and the first amino acid residue of the basic amino acid cluster of the second DNA-binding domain is indicated. Proline residues are evidenced in bold.

HMGA1a is the only HMGA protein that was heavily methylated (see Figure 2 panel F): mass peaks of 11760.4 and 11775.9 correspond to mono- and dimethylated forms of diphosphorylated HMGA1a, whereas mass peaks of 11840.3 and 11854.9 correspond to mono- and dimethylated forms of triphosphorylated HMGA1a. It is significant that monomethylation (as well as symmetric and asymmetric dimethylation) of HMGA1a was mapped to within the first DNA-binding domain at the site of R25.10,17 This domain is perfectly conserved both in the HMGA1b isoform and in HMGA2, but these proteins are substantially not methylated in vivo (Figure 2 and Supporting Information Figure 2). Such a difference could be due to a different docking of the PRMT responsible for R25 methylation (presumably PRMT118). Indeed, at least for PRMT1, it has been previously demonstrated that positively charged residues distal to the site of methylation are important for the high affinity interaction between this enzyme and its substrates.19 In the HMGA context, these positively charged residues could reasonably be constituted by the R and K residues of the second DNAbinding domain. It is noteworthy that R25 in HMGA1a is 29 amino acid residues from the second DNA-binding domain, whereas the corresponding R residues in HMGA1b and HMGA2 are only 18 and 17 positions, respectively, from the second DNA-binding domain. In addition, the amino acid sequences embedded between the modified R residue and the cationic amino acid cluster significantly differ in the three proteins, especially considering their content of the proline residues, which confer specific structural constraints (see Figure 3). The different size and flexibility of the stretch between the methylation sites and the basic clusters could explain the different methylation pattern. However, it remains to be clarified whether PRMT1 is the enzyme responsible for R25 methylation in vivo. Alternatively, the three HMGA proteins could have different chromatin localizations and consequently be differently accessible to PRMTs. In addition, the reconstructed spectrum reported in Figure 2 panel D indicates a peculiar characteristic of the HMGN1 protein in the presence of forms differing from the principal ones (unphosphorylated and monophosphorylated, 10527.0 2982

Journal of Proteome Research • Vol. 8, No. 6, 2009

Differences Among the Post-Translational Modifications of HMGA and HMGN Proteins in Cells of Different Origin and Phenotype. We have also considered the possibility of differences in HMG post-translational modifications related to the cell line type, origin and/or phenotype. No relevant differences were detected on HMGN proteins, and therefore, these will not be discussed further (data not shown). Even if it was not possible to individuate a clear correlation between cell line origin/phenotype and HMGA PTM profiles, HMGA proteins showed a high heterogeneity in their PTMs related to the cell type considered. Figure 4 shows selected comparisons of HMGA proteins from four different cell lines, which were chosen for their marked differences. Three considerations can be made: (1) The phosphorylation level for HMGA1 proteins is celltype dependent. For example, this is evident comparing the relative abundance of 2P and 3P peaks in panel A and B of Figure 4 (HMGA1a: cervix cancer cells C-4 I versus cervix cancer cells ME-180). This observation is even more evident taking into consideration all HMGA1a and HMGA1b reconstructed mass spectra (Supporting Information Figure 2). (2) HMGA1a (and to a slight extent also HMGA1b) can be monomethylated and its modification level is cell-type dependent. Compare the different relative abundance of M and 2M peaks in panel B and C of Figure 4 (HMGA1a: cervix cancer cells ME-180 versus prostate cancer cells DU 145). This is even more evident taking into consideration all the HMGA1 reconstructed spectra (Supporting Information Figure 2). (3) The HMGA2 protein is not always detectable, its expression being restricted to a relative small subset of cell lines (Table 1). This is in agreement with literature data confirming a different and restricted involvement of HMGA2 with respect to HMGA1 proteins. At variance with HMGA1, HMGA2 is always distributed with a wider phosphorylation range, from the monophosphorylated form up to the pentaphosphorylated one (Figure 4, panel D and Supporting Information Figure 2). In order to visualize the prevalent post-translational modification state of HMGA proteins, we generated “combined” reconstructed mass spectra for each of the three HMGA proteins which are the mean spectra resulting from all the acquired mass spectra in each different cell type. In addition, we calculated the standard deviation of the abundance of each HMGA post-translationally modified form, which can be taken as an indicator of their fluctuation. These combined spectra, including the standard deviation, are reported in Figure 5 where it is possible to note all the previously described differences among the three HMGA. For comparison, we present the combined spectra of HMGN1 as an example of a protein with no relevant variations in its PTMs.

HMGA Macroscopic PTMs

research articles

Figure 4. Exemplificative comparison of HMGA phosphorylation and methylation patterns. Reconstructed mass spectra of HMGA1a from cervix cancer cell line C-4I (A) and ME-180 (B), prostate cancer cell line DU 145 (C). In the upper-left part of each panel, the percentage of diphosphorylation (2P + 2P/M and 2M), triphosphorylation (3P + 3P/M and 2M) and methylation (Mtot) is indicated. Panel D shows reconstructed mass spectra of HMGA2 from hepatocarcinoma cell line PLC/PRF/5. 1P, 2P, 3P, 4P, 5P and 6P: mono-, di-, tri-, tetra-, penta-, and hexaphosphorylation. M and 2M: mono- and dimethylation.

HMGA2 Phosphate Group Mapping. Although the HMGA2 protein derives from a gene other than HMGA1a and HMGA1b, it shares a significant homology with the other two family members. However, despite a growing number of evidence for its fundamental role in many important biological processes, such as development, tumorigenesis and senescence, it is still largely uncharacterized from the point of view of its posttranslational modifications and their role in modulating HMGA2 function(s). Our in vivo screening evidenced that HMGA2 is widely phosphorylated, so we decided to further investigate this modification. Since its discovery, it has been demonstrated that human HMGA2 is a phosphoprotein14 and it was shown that CK2 could be responsible for phosphorylation of its acidic tail.15,20 However, neither the phosphorylation sites nor the stoichiometry of incorporated phosphates have been unambiguously assigned. We therefore decided to perform in vitro phosphorylation assays in combination with LC/MS analyses in order to analyze the phosphorylation pattern of CK2-treated HMGA2 protein. As can be noted from the reconstructed mass spectrum of Figure 6, panel A, CK2-phosphorylated HMGA2 bears up to five phosphate groups (P: phosphate group). Unmodified HMGA2 has a theoretical molecular mass of 11700.8 Da, so the 11940.7, 12020.8, and 12100.9 Da values are consistent with the addition of three, four, and five phosphates, respectively. In order to ascertain that all the phosphates are located only on the acidic

tail, we have also treated with CK2 a truncated form of HMGA2 lacking all the C-terminal domain (HMGA2 1-93). Panel B of Figure 6 clearly shows the complete absence of phosphates in this truncated protein with theoretical molecular mass 10019.5 Da. This confirms that all the CK2 phosphorylation sites reside on the acidic tail. HMGA1a (panel C) was used as a positive control since it is a well-known CK2 substrate.10 As we expected, CK2 added three phosphates to HMGA1a (mass of 11784.3 Da with respect to a theoretical molecular mass of unmodified HMGA1as11544.8 Dasis consistent with the triphosphorylated form). These data demonstrate that all the five S/T residues (...EET96EET99S100S101QES104AEED) within the acidic tail of HMGA2 are phosphorylatable by CK2 even if some do not fulfill the canonical CK2 consensus site. From our in vivo data (Figure 2, panel E and Supporting Information Figure 2, HMGA2) it is evident that HMGA2 distributes among several different phosphorylated forms. A time course phosphorylation assay, performed in vitro with recombinant HMGA2 and commercially available CK2, allowed us to obtain the same differentially modified forms (Supporting Information Figure 3). We purified the various phosphorylated proteins and subjected them to tryptic digestion and LC/MS analyses. The same was done also with the purified endogenous HMGA2. The peptides obtained allowed us to create two tryptic maps covering the entire HMGA2 sequence (data not shown). We were able to detect the 0P, 1P, 2P, 3P, 4P, and 5P forms of the Journal of Proteome Research • Vol. 8, No. 6, 2009 2983

research articles

Figure 5. Combined mass spectra of HMGA proteins and HMGN1. HMGA1a (A), HMGA1b (B), HMGA2 (C), and HMGN1 (D) mass spectra resulting from the LC-MS screening of all analyzed cell lines were combined together in order to obtain, for each single protein, the mean spectrum. Bars indicate standard deviations.

C-terminal peptide (aa 90-108) from the in vitro modified protein and the 0P, 1P, 2P, and 3P forms from the endogenous one. In addition, only in the endogenous protein, we found a further phosphorylated peptide not belonging to the C-terminal domain (peptide 33-45). No other phosphorylated peptides were found even if we performed phosphopeptide enrichment by Immobilized Metal Affinity Chromatography (IMAC). Phosphorylation of HMGA2 Acidic C-Terminal Tail. To establish which are the preferential CK2 modification sites on the HMGA2 acidic tail, we performed MS/MS sequencing analyses on every C-terminal form detected (from 0P to 5P). Data obtained both from the recombinant and the endogenous protein suggest the following phosphorylation order: S104 and S100 are the first residues to be phosphorylated followed by 2984

Journal of Proteome Research • Vol. 8, No. 6, 2009

Sgarra et al.

Figure 6. HMGA2 is in vitro phosphorylated by Casein Kinase II (CK2). Reconstructed mass spectra of HMGA2 (A), HMGA2 1-93 (B) and HMGA1a (C) in vitro treated with CK2. Molecular mass (Da) of each modified form is indicated together with the corresponding phosphorylation degree.

S101, T99, and finally T96. From the mass sequencing of m/z signals corresponding to the peptide of molecular mass 2123.8 Da, both in the recombinant and endogenous protein, we obtained clear fragmentation spectra (data not shown) from which it is possible to unambiguously assign these m/z values to the unphosphorylated C-terminal tail of HMGA2 (K90PAQEETEETSSQESAEED108). Peptides differing by 80 Da or its multiples from the molecular mass of K90-D108 likely represent differently phosphorylated forms. In addition, the loss of phosphoric acid (H3PO4) is a typical phenomenon observable during CID fragmentation of phosphopeptides, and indeed, in our analyses, we systematically observed such event during MS sequencing of di- and tricharged acidic C-terminal tail phosphopeptides (with the appearance of peaks differing by -49

research articles

HMGA Macroscopic PTMs

Figure 7. In vitro and in vivo HMGA2 acidic C-terminal tail phosphorylation. HMGA2 acidic C terminal tail (aa 90-108) phosphopeptides (from 1P to 5P) deriving from both in vitro CK2 phosphorylated and endogenous (Hep G2 cells) HMGA2 were analyzed by LC-MS/MS and phosphorylated residues mapped. The position of phosphate group(s) is indicated for each of the HMGA2 acidic C-terminal tail phosphorylated forms. P: phosphate group.

and -32.6 from the precursor ion, respectively; see MS/MS spectra reported in Supporting Information Figure 4), thus, confirming that the fragmented ions are phosphopeptides. These results are summarized in Figure 7 and all m/z fragmentation spectra, their interpretation schemes, and their descriptive interpretations are reported in Supporting Information (Supporting Information Figures 4 and 5). Phosphorylation of Endogenous HMGA2 Outside the C-Terminal Tail. Our LC-MS/MS analyses evidenced, as was previously suggested,20 that in the endogenous protein S43 can also be phosphorylated. The fragmentation spectrum (Figure 8, panel A) of the bicharged precursor ion at m/z 767.8 is dominated by a -49 neutral loss [peak number 2, (MH H3PO4)+2, m/z 718.8] clearly indicating that this is a phosphopeptide. Mass sequencing allows to attribute this tryptic peptide to HMGA2 peptide 33-45 (KQQQEPT39GEPS43PK) and clearly indicates that S43 is phosphorylated. Indeed the fragments of the b series until b9 (Figure 8, panel B) are consistent with unphosphorylated fragments and therefore exclude T39 as the phosphorylation site, whereas b11 and b12, which comprise S43, carry one phosphate group, indicating therefore that this serine is the phosphorylation site. Also the y series confirms the location of the phosphate since y4, y5, and y6 fragments, which only contain S43, carry one phosphate group. It was previously suggested that in the murine HMGA2 the same residue could be phosphorylated by the cyclin/cdk enzymes at the beginning of S-phase and in the G2/M phases of the cell cycle.20 It is worthwhile to evidence that S43 is the corresponding residue of T52 in human HMGA1a, which is well-known to be phosphorylated by p34cdc2 (also known as CDK1). The sequence in which S43 is embedded fits perfectly with the p34cdc2 consensus S/T-P-x-K/R22 and it was previously demonstrated in vitro that p34cdc2 phosphorylates murine HMGA2 at the level of S43 and S58 resulting in a strong decrease of its DNA binding affinity.15 Phosphorylation of the HMGA2 Acidic C-Terminal Tail Affects Its DNA Binding Properties. Since a peculiarity of HMGA2 is the high degree of phosphorylation of the acidic C-terminal tail, we asked whether this modification could play a role in modulating its DNA binding properties. To this end, we performed EMSA experiments with two typical DNA targets for HMGA proteins. On one hand, we took into consideration the ability of HMGA proteins to recognize noncanonical DNA structures using a Four Way Junction probe.23 On the other

hand, we considered the ability of HMGA to bind to the minor groove of AT-rich stretches of B-type DNA and we used as the DNA target a region of the Insulin receptor promoter (E3 probe24). We set up EMSA experiments in order to have exactly the same protein concentrations and buffer compositions for both unphosphorylated and CK2 phosphorylated HMGA2 (see Materials and Methods). As it is possible to observe from the mass spectra reported in Figure 9, panel A, after CK2 phosphorylation, we obtained a pentaphosphorylated HMGA2 (and a tetraphosphorylated minor form), whereas using a heatinactivated enzyme, we maintained the protein unmodified. The EMSA experiment performed with Four Way Junction probe clearly shows a change in the DNA binding properties of HMGA2 due to the C-terminal phosphorylation. Indeed, it is possible to observe a decrease in the DNA binding affinity of the phosphorylated protein compared with the unphosphorylated one (upper part of Figure 9, panel B, compare lanes 2-6 with 8-12 of the 4WJ experiment). Moreover, the complexes carrying the phosphorylated protein show less stability during electrophoretic migration as evident from the rather diffused bands of lanes 2 and 3 compared with the wellresolved bands of lane 8 and 9. In addition, comparing phosphorylated (lanes 2-5) with unphosphorylated (lanes 8-11) HMGA protein/DNA bands, it is also possible to observe a change in the mobility of complexes which could reflect alterations in the organization of protein-DNA complexes, a phenomenon which was already described for the HMGA1b isoform.25 EMSA experiments carried out using a B-type DNA probe (lower part of Figure 9, panel B) substantially confirm the change in the DNA binding properties of HMGA2 due to C-terminal phosphorylation.

Conclusions The HMGA architectural factor family is composed of three members which share a high degree of sequence and structural homology. However, several experimental evidence suggest that these proteins may have nonoverlapping functions. Differences in HMGA DNA binding properties,26 their nonoverlapping biological functions,27,28 and their different expression patterns3 have already been pointed out, and it was also suggested that their activity could be differentially modulated by posttranslational modifications.29 HMGA are highly abundant chromatin proteins, whose expression level, in particular during neoplastic transformation, can reach a 1/10 ratio with respect to histone H1.30 This means that, beside their involvement in regulating the expression of specific genes, they can play a widespread influence on chromatin structure.2,3 This has, for example, been recently underlined by their involvement in formation of senescence-associated heterocromatic foci (SAHFs).31 Beside their differences in the primary structure, what other difference affecting the bulk of HMGA could be responsible for their different actions at a general chromatin level? Our LC/MS screening, performed on a relevant number of cell lines differing both in phenotype and origin, evidenced macroscopic differences between the post-translational modifications affecting the three HMGA that are linked to the cellular context. This is relevant since changes in post-translational modifications are not observable in the highly related HMGN proteins. This kind of fluctuation suggests, even if does not directly demonstrate, that the activities of HMGA proteins could be mainly and generally modulated by Macroscopic PTMs. Journal of Proteome Research • Vol. 8, No. 6, 2009 2985

research articles

Sgarra et al.

Figure 8. HMGA2 is in vivo phosphorylated also at the level of serine 43. The MS/MS spectrum of precursor ion 767.82+ belonging to the monophosphorylated HMGA2 33-45 peptide is reported in panel A. The y and b ions together with their m/z values are indicated. 1, 2, 3, 4 correspond to precursor ion with losses of water or phosphoric acid. In B, a scheme of the found fragments is shown. -H3PO4, phosphoric acid loss (neutral loss); -H2O, water molecule loss; dotted lines, bicharged fragments. An asterisk indicates the presence of a phosphate group.

The acidic C-terminal tail phosphorylation has up to now been considered the only constitutive modification of HMGA protein, but our screening evidenced, at least for the HMGA1a isoform, that methylation (which has always been mapped at the level of Arginine 25) could also be considered an HMGA constitutive modification, the function of which, however, is completely unknown. Our data regarding C-terminal phosphorylation and R25 methylation are consistent with previous findings regarding HMGA1a post-translational modifications in human breast tumor specimens.32 Furthermore, our data regarding PTMs of HMGA2 protein support the above-mentioned nonoverlapping biological functions among HMGA proteins. Indeed, the phosphorylation pattern of HMGA2 is completely different from that of HMGA1 proteins. This is primarily linked to the different C-terminal aminoacidic composition of the two proteins; in fact, as we reported in the last section, HMGA2 possesses five phosphorylatable amino acids compared to HMGA1a which has only three. 2986

Journal of Proteome Research • Vol. 8, No. 6, 2009

HMGA phosphorylation at the C-terminal acidic tail suggests a modulatory role for this domain in regulating HMGA activity. In the past, several lines of evidence underlined the functional importance of the C-terminal domain both for HMGA1 and HMGA2 phosphoproteins.3 It is noteworthy that HMGA forms missing the acidic C-terminal tail, depending on the particular cellular system considered, either show an increased transformation potential33,34 or confer to cells an increased growth rate35 with respect to the full-length proteins. Therefore, the current idea is that the acidic C-terminal domain plays a negative modulatory role in the regulation of proliferative and transforming ability of HMGA proteins. A possible molecular mechanism at the basis of this negative modulation is suggested by the role of the C-terminal tail of another member of the HMG protein group, that is, HMGB1. In this protein, the acidic C-terminal tail contacts the Nterminal portion of the protein and modulates its DNA binding activities.36 It is reasonable to hypothesize that such a mechanism could be envisaged also for the HMGA proteins. In the

HMGA Macroscopic PTMs

research articles

Figure 9. Acidic tail phosphorylation affects HMGA2 DNA-binding properties. (A) HMGA2 was treated with CK2 or heat-inactivated CK2 in order to obtain either a fully phosphorylated or a nonmodified protein. Reconstructed mass spectra of the obtained proteins are shown. (B) EMSA experiments were performed with decreasing amounts (16, 8, 4, 2, and 1 pmol) of CK2- and CK2 inactivatedtreated HMGA2 with 32P labeled Four Way Junction (4WJ) and E3 DNA probes. Stoichiometry of the complexes detected is shown on the left.

light of the data from our screening, another question can be raised: is this modulation due solely to the acidic tail itself or is it somehow dependent on the phosphorylations of HMGA1 and HMGA2 C-terminal tails? It is possible to speculate that this modification could be involved in masking the inner portion of the molecules modulating in this way the DNA- or the protein-protein interaction domains of HMGA themselves. This is supported by the high content of arginine residues of HMGA proteins and by the unusual stability of the argininephosphate ionic interactions.37 EMSA experiments reported in Figure 9 substantiate these hypotheses. The influence of an acidic and phosphorylated C-terminal domain on the biological activity of a protein seems to be a rather frequent mechanism of protein function regulation. For example, the C-terminal domain of the tumor suppressor PTEN (Phosphatase and Tensin homologue) acts as an autoinhibitory domain of the catalytic activity of PTEN itself38,39 and the acidic C-terminal tail of the ssDNA-binding protein of bacteriophage T7 shields, intramolecularly, the positively charged DNA-

binding domain.40 Regarding the latter example, this mechanism has been proposed to be a general tool to prevent random bindings between positively and negatively charged protein/ nucleic acid surfaces.40 This work was aimed at systematically analyzing HMGA proteins in different cellular contexts. We were able to evidence relevant differences regarding the PTMs affecting HMGA proteins and have provided also, for the first time, data regarding in vivo HMGA2 PTMs. These modifications could be strongly correlated to the role played by HMGA in chromatin structure modulation. As happens for histones, HMGA posttranslational modifications could be part of a general epigenetic code regulating chromatin function. As often occurs, proteomic screenings represent relevant data on which functional hypothesis can be built, that necessarily will need the integration of dedicated biochemical approaches to support and further clarify the obtained results. Abbreviations: CK2, Casein kinase II; PTM, post-translational modification; CBP, CREB-binding protein; PCAF, P300/CBPJournal of Proteome Research • Vol. 8, No. 6, 2009 2987

research articles associated factor; HIPK2, Homeodomain interacting protein kinase 2; PKC, Protein kinase C; p34cdc2, p34 protein kinase/ cell division control protein 2 homologue; LC/MS, liquid chromatography/mass spectrometry.

Acknowledgment. We thank Dr. M. Pignataro for his technical assistance with mass spectrometry analyses and figures’ elaboration, Drs. S. Pegoraro and A. Tossi for careful reading of the manuscript. This work was supported by grants from: Universita` degli Studi di Trieste (Italy), Associazione Italiana per la Ricerca sul Cancro (AIRC), Milan (Italy), Progetto AITT 2007 Regione FVG, MIUR (PRIN 2007), and project ‘Rete Nazionale Proteomica’ FIRB 2008 RBRN07BMCT - to G. M.; MIUR (PRIN 2006) and MOMA project Agenzia Spaziale Italiana (ASI) to V.G. We performed mass spectrometry analyses at the ‘Mass Spectrometry and Proteomic Laboratory - Fondazione Kathleen Foreman Casali’ at the University of Trieste. Supporting Information Available: Supporting Information shows tables and figures regarding HMGA1 posttranslational modifications, reconstructed mass spectra of HMGA and HMGN1 proteins, reconstructed mass spectra of time course experiments, and fragmentation spectra. This material is available free of charge via the Internet at http:// pubs.acs.org. References (1) Hock, R.; Furusawa, T.; Ueda, T.; Bustin, M. HMG chromosomal proteins in development and disease. Trends Cell Biol. 2007, 17, 72–79. (2) Sgarra, R.; Rustighi, A.; Tessari, M. A.; Di Bernardo, J.; Altamura, S.; Fusco, A.; Manfioletti, G.; Giancotti, V. Nuclear phosphoproteins HMGA and their relationship with chromatin structure and cancer. FEBS Lett. 2004, 574, 1–8. (3) Fusco, A.; Fedele, M. Roles of HMGA proteins in cancer. Nat. Rev. Cancer 2007, 7, 899–910. (4) Fedele, M.; Visone, R.; De Martino, I.; Troncone, G.; Calmieri, D.; Battista, S.; Ciarmiello, A.; Pallante, P.; Arra, C.; Melillo, R. M.; Helin, K.; Croce, C. M.; Fusco, A. HMGA2 induces pituitary tumorigenesis by enhancing E2F1 activity. Cancer Cell 2006, 9, 459–471. (5) Pierantoni, G. M.; Rinaldo, C.; Esposito, F.; Mottolese, M.; Soddu, S.; Fusco, A. High Mobility Group A1 (HMGA1) proteins interact with p53 and inhibit its apoptotic activity. Cell Death Differ. 2006, 13, 1554–1563. (6) Frasca, F.; Rustighi, A.; Malaguarnera, R.; Altamura, S.; Vigneri, P.; Del Sal, G.; Giancotti, V.; Pezzino, V.; Vigneri, R.; Manfioletti, G. HMGA1 inhibits the function of p53 family members in thyroid cancer cells. Cancer Res. 2006, 66, 2980–2989. (7) Sgarra, R.; Tessari, M. A.; Di Bernardo, J.; Rustighi, A.; Zago, P.; Liberatori, S.; Armini, A.; Bini, L.; Giancotti, V.; Manfioletti, G. Discovering high mobility group A molecular partners in tumour cells. Proteomics 2005, 5, 1494–1506. (8) Pierantoni, G. M.; Esposito, F.; Giraud, S.; Bienvenut, W. V.; Diaz, J. J.; Fusco, A. Identification of new high mobility group A1 associated proteins. Proteomics 2007, 7, 3735–3742. (9) Sgarra, R.; Furlan, C.; Zammitti, S.; Lo Sardo, A.; Maurizio, E.; Di Bernardo, J.; Giancotti, V.; Manfioletti, G. Interaction proteomics of the HMGA chromatin architectural factors. Proteomics 2008, 8, 4721–4732. (10) Zhang, Q.; Wang, Y. High mobility group proteins and their posttranslational modifications. Biochim. Biophys. Acta 2008, 1784, 1159–1166. (11) Munshi, N.; Agalioti, T.; Lomvardas, S.; Merika, M.; Chen, G.; Thanos, D. Coordination of a transcriptional switch by HMGI(Y) acetylation. Science 2001, 293, 1133–1136. (12) Diana, F.; Sgarra, R.; Manfioletti, G.; Rustighi, A.; Poletto, D.; Sciortino, M. T.; Mastino, A.; Giancotti, V. A link between apoptosis and degree of phosphorylation of High Mobility Group A1a protein in leukemic cells. J. Biol. Chem. 2001, 276, 11354–11361. (13) Meggio, F.; Pinna, L. A. One-thousand-and-one substrates of protein kinase CK2? FASEB J. 2003, 17, 349–368.

2988

Journal of Proteome Research • Vol. 8, No. 6, 2009

Sgarra et al. (14) Giancotti, V.; Bandiera, A.; Buratti, E.; Fusco, A.; Marzari, R.; Coles, B.; Goodwin, G. H. Comparison of multiple forms of the high mobility group I proteins in rodent and human cells. Identification of the human high mobility group I-C protein. Eur. J. Biochem. 1991, 23, 211–216. (15) Schwanbeck, R.; Manfioletti, G.; Wisniewski, J. R. Architecture of high mobility group protein I-C.DNA complex and its perturbation upon phosphorylation by Cdc2 kinase. J. Biol. Chem. 2000, 275, 1793–1801. (16) Di Agostino, S.; Fedele, M.; Chieffi, P.; Fusco, A.; Rossi, P.; Geremia, R.; Sette, C. Phosphorylation of high-mobility group protein A2 by Nek2 kinase during the first meiotic division in mouse spermatocytes. Mol. Biol. Cell 2004, 15, 1224–32. (17) Sgarra, R.; Diana, F.; Bellarosa, C.; Dekleva, V.; Rustighi, A.; Toller, M.; Manfioletti, G.; Giancotti, V. During apoptosis of tumor cells HMGA1a protein undergoes methylation: identification of the modification site by mass spectrometry. Biochemistry 2003, 42, 3575–3585. (18) Zou, Y.; Webb, K.; Perna, A. D.; Zhang, Q.; Clarke, S.; Wang, Y. A mass spectrometric study on the in vitro methylation of HMGA1a and HMGA1b proteins by PRMTs: methylation specificity, the effect of binding to AT-rich duplex DNA, and the effect of C-terminal phosphorylation. Biochemistry 2007, 46, 7896–7906. (19) Osborne, T. C.; Obianyo, O.; Zhang, X.; Cheng, X.; Thompson, P. R. Protein arginine methyltransferase 1: positively charged residues in substrate peptides distal to the site of methylation are important for substrate binding and catalysis. Biochemistry 2007, 46, 13370– 13381. (20) Goodwin, G. The high mobility group protein, HMGI-C. Int. J. Biochem. Cell Biol. 1998, 30, 761–766. (21) Meggio, F.; Pinna, L. A. One-thousand-and-one substrates of protein kinase CK2? FASEB J. 2003, 17, 349–368. (22) Marin, O.; Meggio, F.; Draetta, G.; Pinna, L. A. The consensus sequences for cdc2 kinase and for casein kinase-2 are mutually incompatible. A study with peptides derived from the beta-subunit of casein kinase-2. FEBS Lett. 1992, 301, 111–114. (23) Bianchi, M. E. Interaction of a protein from rat liver nuclei with cruciform DNA. EMBO J. 1988, 7, 843–849. (24) Brunetti, A.; Manfioletti, G.; Chiefari, E.; Goldfine, I. D.; Foti, D. Transcriptional regulation of human insulin receptor gene by the high-mobility group protein HMGI(Y). FASEB J. 2001, 15, 492–500. (25) Piekielko, A.; Drung, A.; Rogalla, P.; Schwanbeck, R.; Heyduk, T.; Gerharz, M.; Bullerdiek, J.; Wis´niewski, J. R. Distinct organization of DNA complexes of various HMGI/Y family proteins and their modulation upon mitotic phosphorylation. J. Biol. Chem. 2001, 276, 1984–1992. (26) Maher, J. F.; Nathans, D. Multivalent DNA-binding properties of the HMG-1 proteins. Proc. Natl. Acad. Sci. U.S.A. 1996, 93, 6716– 6720. (27) Fedele, M.; Fidanza, V.; Battista, S.; Pentimalli, F.; Klein-Szanto, A. J.; Visone, R.; De Martino, I.; Curcio, A.; Morisco, C.; Del Vecchio, L.; Baldassarre, G.; Arra, C.; Viglietto, G.; Indolfi, C.; Croce, C. M.; Fusco, A. Haploinsufficiency of the Hmga1 gene causes cardiac hypertrophy and myelo-lymphoproliferative disorders in mice. Cancer Res. 2006, 66, 2536–2543. (28) Zhou, X.; Benson, K. F.; Ashar, H. R.; Chada, K. Mutation responsible for the mouse pygmy phenotype in the developmentally regulated factor HMGI-C. Nature (London) 1995, 376, 771– 774. (29) Edberg, D. D.; Adkins, J. N.; Springer, D. L.; Reeves, R. Dynamic and differential in vivo modifications of the isoform HMGA1a and HMGA1b chromatin proteins. J. Biol. Chem. 2005, 280, 8961–8973. (30) Giancotti, V.; Bandiera, A.; Ciani, L.; Santoro, D.; Crane-Robinson, C.; Goodwin, G. H.; Boiocchi, M.; Dolcetti, R.; Casetta, B. Highmobility-group (HMG) proteins and histone H1 subtypes expression in normal and tumor tissues of mouse. Eur. J. Biochem. 1993, 213, 825–832. (31) Narita, M.; Narita, M.; Krizhanovsky, V.; Nun ˜ ez, S.; Chicas, A.; Hearn, S. A.; Myers, M. P.; Lowe, S. W. A novel role for highmobility group a proteins in cellular senescence andheterochromatin formation. Cell 2006, 126, 503–514. (32) Zou, Y.; Wang, Y. Mass spectrometric analysis of high-mobility group proteins and their post-translational modifications in normal and cancerous human breast tissues. J. Proteome Res. 2007, 6, 2304–2314. (33) Fedele, M.; Berlingieri, M. T.; Scala, S.; Chiariotti, L.; Viglietto, G.; Rippel, V.; Bullerdiek, J.; Santoro, M.; Fusco, A. Truncated and chimeric HMGI-C genes induce neoplastic transformation of NIH3T3 murine fibroblasts. Oncogene 1998, 17, 413–418. (34) Li, Y.; Lu, J.; Prochownik, E. V. Dual role for SUMO E2 conjugase Ubc9 in modulating the transforming and growth-promoting

research articles

HMGA Macroscopic PTMs properties of the HMGA1b architectural transcription factor. J. Biol. Chem. 2007, 282, 13363–13371. (35) Pierantoni, G. M.; Battista, S.; Pentimalli, F.; Fedele, M.; Visone, R.; Federico, A.; Santoro, M.; Viglietto, G.; Fusco, A. A truncated HMGA1 gene induces proliferation of the 3T3-L1 pre-adipocytic cells: a model of human lipomas. Carcinogenesis 2003, 24, 1861– 1869. (36) Wang, Q.; Zeng, M.; Wang, W.; Tang, J. The HMGB1 acidic tail regulates HMGB1 DNA binding specificity by a unique mechanism. Biochem. Biophys. Res. Commun. 2007, 360, 14–19. (37) Woods, A. S.; Ferre´, S. Amazing stability of the arginine-phosphate electrostatic interaction. J. Proteome Res. 2005, 4, 1397–1402.

(38) Odriozola, L.; Singh, G.; Hoang, T.; Chan, A. M. Regulation of PTEN activity by carboxyl-terminal autoinhibitory domain. J. Biol. Chem. 2007, 282, 23306–23315. (39) Raftopoulou, M.; Etienne-Manneville, S.; Self, A.; Nicholls, S.; Hall, A. Regulation of cell migration by the C2 domain of the tumor suppressor PTEN. Science 2004, 303, 1179–1181. (40) Marintcheva, B.; Marintchev, A.; Wagner, G.; Richardson, C. C. Acidic C-terminal tail of the ssDNA-binding protein of bacteriophage T7 and ssDNA compete for the same binding surface. Proc. Natl. Acad. Sci. U.S.A. 2008, 105, 1855–1860.

PR900087R

Journal of Proteome Research • Vol. 8, No. 6, 2009 2989