Sumoylation is a common post-translational modification of transcription factors, but in several proteins the physiological functions and molecular mechanisms of this modification have remained enigmatic. Several lines of evidence support the concept that SUMO serves as a negative regulator of STAT1 [10, 11, 13, 14]. Furthermore, the results demonstrating that sumoylation also negatively regulates STAT5-mediated signaling and the only STAT transcription factor in Drosophila melanogaster, Stat92E, indicates that sumoylation is an evolutionary conserved post-translational modification for some STAT transcription factors [18, 19].
Sumoylation is a highly reversible covalent modification that is regulated through conjugating and deconjugating enzymes. Several studies support the importance of PIAS1-mediated sumoylation of the proteins. Recently, it was shown that PIAS1 regulates oncogenic signaling by sumoylating promyelocytic tumor suppressor (PML) that leads to its ubiquitination and proteosomal degradation . PIAS1 has also been shown to associate with protein-tyrosine phosphatase 1B (PTP1B) and to catalyze its sumoylation, which resulted in down-regulation of the phosphatase activity and inhibited dephosphorylation of the insulin receptor. PIAS1-mediated negative regulation of PTP1B was reversed by SENP1, an isopeptidase that was also shown to regulate sumoylation of STAT5 [18, 28]. Senp1 knock out mice were found to have severe defects in early T and B cell development. The defect in lymphoid development was likely caused by enhanced level of STAT5 sumoylation that subsequently led to decreased STAT5 transcriptional activity . In our experiments removal of conjugated SUMO-1 by SENP1 increased STAT1-mediated reporter gene expression, thus confirming the negative regulatory role of sumoylation for STAT1 (Figure 1B).
STAT1 homodimerization is required for the optimal IFN-γ-mediated gene activation and STAT1 homodimers form a nutcracker-like structure that binds to DNA. The monomers are held together by interface between Tyr701 phosphorylated C-terminal tail segment of one monomer and the SH2 domain of the other . The Lys703 is located adjacent to the dimerization interface and this prompted us to investigate if sumoylation of the Lys703 could affect dimerization or DNA-binding of STAT1. Analysis of the SUMO conjugation consensus site in STAT1 dimer revealed that side chain of Lys703 formed a projection towards DNA. Both Lys703 and Glu705 residues have hydrophilic side chains, which are converted away from the hydrophobic core of SH2 interface. Additionally, the β-sheet structure between two C-tail segments of STAT1 dimer is not likely to be affected by Lys703 mutation to Arg, while this mutation will interrupt the formation of covalent bond with SUMO. The structural analysis revealed that side chain of Lys703 has an interaction phase with Glu632 residue in the SH2 domain of the adjacent monomer. Most probably this interface prevents rotation of this flexible side chain and keeps orientation favorable for the SUMO conjugation. The finding of controlled position of Lys703 also supports the importance of Lys703 as an SUMO acceptor site (Additional file 2: Figure S2). Sumoylation has been shown to impede Tyr701 phosphorylation of STAT1 and subsequent SH2 domain-phospho-Tyr701-mediated homodimerization, leading to formation of semi-phosphorylated dimers that interact through their N-terminal domains [15–17]. Our experimental data indicated that sumoylated STAT1 can form dimers, but it remains to be determined if the interaction is mediated through their N-terminal domains or through the SH2 domains (Figure 4).
To predict how SUMO-1 would structurally orientate in SH2 domain-phospho-Tyr701 interaction-mediated STAT1 dimers, we reconstructed the structure of the disordered loop 684–699 of STAT1, and made a molecular model of sumoylated STAT1 dimer using x-ray structure of TDG-SUMO-1 as a template . This model suggests that the position of SUMO under the loop structure is directed towards DNA and can inhibit interaction with nucleic acids (Figure 2C). Furthermore, results from the oligonucleotide pull down experiments indicated that sumoylation interferes STAT1 binding to STAT1-responsive promoters, as sumoylation deficient STAT1 E705Q showed increased DNA-binding to both Gbp-1- and Irf-1-oligos (Figure 3A and B), and as SUMO-1 overexpression hindered STAT1 binding to Irf-1-oligo (Additional file 3: Figure 3F). The difference in the DNA-binding properties between STAT1 WT and E705Q mutant was not caused by altered Tyr701 phosphorylation (Figure 3A and B) [15–17]. In addition, another sumoylation deficient STAT1 mutant K703R also showed increased binding to Gbp-1 oligo (data not shown). Taken together, the oligoprecipation experiments are supporting the molecular model where SUMO moiety interferes with DNA-binding of STAT1 (Figure 3A and B, Additional file 3: Figure S3F).
Sumoylated STAT1 was not detected in our oligoprecipitation experiments (Additional file 3: Figure S3G) and this result is consistent with results by Song et al. (2006), showing that sumoylated STAT1 does not bind to DNA, or that the bound fraction is very small . In their EMSA studies Song et al. also found that sumoylation deficient STAT1 K703R mutant shows prolonged binding to GAS-probe, but unexpectedly sumoylation deficient E705A mutant had similar DNA-binding profile than STAT1 WT . We chose to use STAT1 E705Q mutant in the DNA-binding experiments because the mutant has been reported to have minimal SUMO-independent effects on STAT1 when compared to K703R and E705A mutations . Our results with STAT1 E705Q suggest that sumoylation inhibits DNA-binding properties of STAT1. Supporting this and previously published results of Song et al. (2006), we observed that STAT1 K703R has enhanced binding to Gbp-1-oligo when compared to STAT1 WT as well (data not shown). Furthermore, sumoylation deficient STAT1 showed enhanced histone H4 acetylation on Gbp-1 promoter (Figure 3C), thus functionally confirming the enhanced STAT1 promoter binding. Whether sumoylation also alters the interaction with histone acetyl transferases, such as CBP, remains to be determined.
It has become evident that sumoylation participates in regulation of STATs and the precise molecular mechanisms and physiological functions are gradually being revealed. Several studies have analysed sumoylation in STAT1, and sumoylation has been shown to inhibit STAT1 activity by different mechanisms. SUMO conjugation to Lys703 inhibits phosphorylation of Tyr701 [15, 16] and prevents paracrystal formation, thereby increasing solubility of STAT1 which subjects STAT1 for dephosphorylation [14, 17]. Our results suggest an additional regulatory mechanism for sumoylation and indicate that SUMO moiety is directed towards DNA and can inhibit DNA-binding of STAT1.