Here, we have shown for the first time that phosphorylation of SF-1 is altered by its SUMOylation status. Removal of SUMO conjugation from SF-1 upregulates the level of S203 phosphorylation, and this effect depends on CDK7 and is likely due to alterations in CDK7 recruitment. Our findings highlight how posttranslational modifications within the hinge region of SF-1 play significant roles in regulating SF-1's transcriptional activity.
Previous reports have demonstrated that SF-1 is a target of SUMO1 modification in heterologous systems (11
). Whether other SUMO isoforms can modify SF-1 and whether this modification occurs in a more orthologous cellular system had not been examined. Our data indicate that SF-1 can be modified by SUMO3 and, importantly, that the endogenous machinery in adrenocortical cells can conjugate SUMO to SF-1. Whereas in heterologous systems species attributable to modification at both K119 and K194 can be detected (11
), SF-1 SUMOylation in Y1 cells is sensitive only to disruption of K194. This is consistent with the reported preferential modification of K194 (11
). Previous studies, as well as the present one, have indicated that replacement of K194 by an arginine residue prevents SUMOylation and leads to enhanced transcriptional activity. This is consistent with the reported inhibitory function of the underlying SC motif (40
). Lysine residues, however, can be the targets of multiple modifications. Our finding that fusion of SUMO to the N terminus of SF-1 mimics the effects of the SC motif argues that the normal inhibitory function of the motif is indeed mediated by SUMO. By examining the regulation of endogenous genes, our findings indicate that SUMOylation is an important regulatory mechanism to control the overall activity of SF-1.
Studies of multiple sequence-specific transcription factors indicate that phosphorylation of residues in the vicinity of SUMOylation sites can alter SUMOylation (17
). Thus, a subset of SUMOylation motifs in factors such as HSFs (29
) and ERR α and γ (63
) conform to the recently described phosphorylation-dependent SUMOylation motif, in which an adjacent proline-directed phosphorylation site lies downstream of the core SUMOylation motif (PsiKXEXXSP). In this regard, phosphorylation may provide an additional negative charge, which appears to favor SUMOylation in certain contexts (72
). Notably, whereas phosphorylation at S727 in STAT1 enhances SUMOylation at K703 (58
), tyrosine phosphorylation at Y701 is mutually exclusive with K703 SUMOylation. In this regard, our findings indicate that SF-1 K194 SUMOylation is not affected by S203 phosphorylation, and thus, that this motif is not a phosphorylation-dependent SUMOylation motif. The substantially larger separation between the sites may be the basis for this observation.
On the other hand, the relationship between phosphorylation and SUMOylation is, surprisingly, not reciprocal, since we found that preventing SUMOylation enhances the phosphorylation level at S203. This is accompanied by an increased ability of SF-1 to upregulate endogenous target genes. The enhanced phosphorylation at S203 depends on CDK7, and its interaction with SF-1 is enhanced when SUMOylation is prevented. This indicates that CDK7 plays a key role as a mediator of S203 phosphorylation in response to loss of SUMOylation. Given that SUMOylation of SF-1 is a negative influence on its phosphorylation, exploring the exact mechanism of this antagonism will be very informative. One possibility is that SUMOylation sterically hinders access to kinases such as CDK7. The low stoichiometry of steady-state SUMOylation, however, argues that such a mechanism would have to be highly compartmentalized or dynamic. Determining whether SUMOylation and phosphorylation are mutually exclusive, however, will require examining the phosphorylation state of the small SUMO-modified pool of SF-1. Since S203 phosphorylation is a strong positive signal for SF-1, our findings also provide insight into the mechanism of SUMO-dependent transcriptional inhibition. Thus, in addition to the direct inhibitory role of SUMO in transcription (14
), our data indicate that SUMOylation can play an inhibitory role by preventing positively acting posttranslational modifications. In addition, the enhanced promoter occupancy of the SUMOylation-deficient SF-1 indicates that SUMOylation may act by limiting promoter occupancy. The time scale of such an effect, however, is likely different from that of the cyclical pattern initiated by ACTH treatment, since SUMOylation does not appear to alter such dynamics (70
). To our knowledge, this is the first instance in which the SUMOylation status of a transcription factor governs its own phosphorylation. We anticipate, however, that this is likely to be a prevalent mechanism for other transcription factors or cell cycle regulators. Further studies in this area seem warranted.
Many studies support a model whereby ACTH activates ERK-dependent phosphorylation of SF-1 S203 and perhaps facilitates the generation of phospholipid ligands for SF-1. Interestingly, a recent study (44
), using proliferating cancer cell lines, indicated that SF-1 can be phosphorylated/activated on the identical S203 residue by CDK7. This enzyme functions both as a CDK-activating kinase (CAK) by phosphorylating cell cycle CDKs and also as a component of the general transcription factor TFIIH. In this context, CDK7 phosphorylates the C-terminal tail of the largest subunit of polymerase II (7
). While the C-terminal tail of polymerase II can be phosphorylated and hence activated by multiple kinases (including CDK8 and CKD9) in different promoter and cellular contexts, CDK7 is unique, as it participates in both cell cycle regulation and transcription. Thus, activation of CDK7 occurs only in the context of the CDK7-cyclin H-Mat1 complex in cells engaged in the cell cycle. Moreover, CDK7 is itself activated by its own targets, CDK1 and CDK2, supporting a feed-forward amplification. Such a mechanism is thus predicted to sustain proliferation induced by mitogenic stimuli (20
). In addition, CDK7 is a component of a variety of nuclear receptor complexes and has been shown to phosphorylate multiple nuclear receptors to facilitate active transcription and engagement with RNA polymerase II (10
). In this view, CDK7-mediated phosphorylation of nuclear receptors provides a mechanism to activate a unique subset of nuclear receptor genes that are coupled to proliferation. Our current finding that increased phosphorylation of SF-1 upon loss of SUMOylation is likely mediated by CDK7 reveals an additional mechanism by which CDK7 regulates the transcriptional activities of nuclear receptors. Future efforts will aim to characterize the molecular mechanisms by which SUMOylation regulates both CDK7 association and the phosphorylation of SF-1.
The transcriptional activities of steroidogenic enzymes regulated by SF-1 are thought to be dependent on the cell type, promoter context, and cell-signaling pathways (1
). The combined observations that loss of SUMOylation enhances the SF-1 occupancy and activity of target promoters and that ACTH does not appear to regulate SF-1 SUMOylation indicate that SUMOylation exerts a tonic inhibitory effect on SF-1 activity. The consequences of the interplay between ACTH-induced alterations in SF-1 function and SUMOylation, however, appear to be promoter dependent. Thus, for genes such as StAR and 3β-HSD, loss of SUMOylation is sufficient for nearly maximal activation, whereas for CYP21 and CYP17, ACTH can further enhance the elevated basal levels caused by loss of SUMOylation. These differences support a model in which loss of SUMOylation favors promoter occupancy by SF-1 as well as CDK7-mediated phosphorylation of SF-1. For some promoters, this is sufficient for nearly complete induction, whereas for others, ACTH engagement of additional signaling cascades further enhances their transcription. In this regard, the S203D mutation in SF-1 has been shown to mimic phosphorylation and to increase transcriptional activity in numerous studies (2
). However, our current data show that this substitution does not alter overall SF-1 promoter occupancy in the absence of ACTH stimulation. Thus, in addition to the tonic inhibitory effects of SUMOylation, other ACTH-initiated signals impinge on SF-1 through different pathways (ERK-mediated phosphorylation and/or generation of phospholipid ligands). The nature of such pathways and their significance will have to be evaluated in nontransformed cell lines and relevant in vivo systems. It is interesting that our recent studies have identified SF-1 response elements in the promoters of a number of upregulated kinases in adrenocortical cells, including the MAPK activator MAP4K2 (G. D. Hammer, unpublished observation). Moreover, additional kinases, including CDK10 and adenylate cyclase 4, are direct transcriptional targets of SF-1 in mouse Y1 adrenocortical cells (J. O. Scheys and G. D. Hammer, unpublished observation; 54
). Thus, additional signaling pathways may be engaged as a result of loss of SF-1 SUMOylation. Clearly, exploring the mutual relationships between posttranslational modifications of SF-1 and signaling cascades is an important area for future research efforts.
In summary, we have identified a novel nonreciprocal relationship between posttranslational modifications within the hinge region of SF-1 that contribute to transcriptional activity in adrenocortical cells. The findings that non-SUMOylated SF-1 leads to enhanced recruitment to chromatin, increased association with CDK7, and concomitant phosphorylation, together with amplified transcriptional activity, support a general model in which SUMOylation participates in transcriptional repression in part by preventing additional activating posttranslational modifications of nuclear receptors.