HMGA1a and HMGA1b are members of one subfamily of non-histone chromosomal highmobility group (HMG) proteins. They bind to various DNA-related substrates, including the minor groove of AT-rich duplex DNA sequences, and have been postulated to be architectural transcription factors functioning in a wide variety of cellular processes. Post-translational modifications of HMGA1 proteins, such as phosphorylation, acetylation, and methylation, are widely observed in tumor cells in vivo and correlated with the modulation of protein function. Here, we investigated the in vitro methylation of recombinant human HMGA1a and HMGA1b proteins by three members of the protein arginine methyltransferase (PRMT) family: PRMT1, PRMT3, and PRMT6. PRMT1 and PRMT3 showed a preference for methylating arginine residues in the first AT-hook of HMGA1 proteins, whereas PRMT6 methylated mainly residues in the second AT-hook. The initial sites of methylation catalyzed by PRMT1 and PRMT3 were mapped by tandem mass spectrometry to be Arg25 and Arg23, respectively, while we confirmed that the initial sites of methylation catalyzed by PRMT6 were at Arg57 and Arg59. Our results also revealed that binding of HMGA1 proteins to AT-rich duplex DNA, but not GC-rich duplex DNA, significantly inhibited the methylation efficiency of all of the PRMTs toward HMGA1 proteins. Moreover, C-terminal constitutive phosphorylation of HMGA1 proteins induced by protein kinase CK2 did not have any appreciable effect on the in vitro methylation of HMGA1. Our results suggest that PRMT1 might be involved in the previously reported methylation of Arg25 in HMGA1a in vivo.

High mobility group (HMG) A1 proteins are subject to a number of post-translational modifications, which may regulate their function in gene transcription and other cellular processes. We examined, by using mass spectrometry, the acetylation of HMGA1a and HMGA1b proteins induced by histone acetyltransferases p300 and PCAF in vitro and in PC-3 human prostate cancer cells in vivo. It turned out that five lysine residues in HMGA1a, i.e., Lys-14, Lys-64, Lys-66, Lys-70, and Lys-73, could be acetylated by both p300 and PCAF. We further quantified the level of acetylation by analyzing, with LC-MS/MS, the proteolytic peptides of the in vitro or in vivo acetylated HMGA1 proteins where the unmodified lysine residues were chemically derivatized with a perdeuterated acetyl group. Quantification results revealed that p300 and PCAF exhibited different site preferences for the acetylation; the preference of p300 acetylation followed the order of Lys-64ϳLys-70 Ͼ Lys-66 Ͼ Lys-14ϳLys73, whereas the selectivity of PCAF acetylation followed the sequence of Lys-70ϳLys-73 Ͼ Lys-64ϳLys-66 Ͼ Lys-14. HMGA1b was acetylated in a very similar fashion as HMGA1a. We also demonstrated that C-terminal phosphorylation of HMGA1 proteins did not affect the in vitro acetylation of the two proteins by either p300 or PCAF. Moreover, we examined the acetylation of lysine residues in HMGA1a and HMGA1b isolated from PC-3 human prostate cancer cells. Our results showed that all the above five lysine residues were also acetylated in vivo, with Lys-64, Lys-66 and Lys-70 in HMGA1a exhibiting higher levels of acetylation than Lys-14 and Lys-73. (J Am Soc Mass Spectrom 2007, 18, 1569 -1578