Previously, it was shown by in vitro studies using purified STAT1 that tyrosine-phosphorylated STAT1 dimers bound to DNA are protected from the inhibitory activity of nuclear phosphatases and barred from nuclear exit. Although the physiological significance of this finding remains unclear, it has been suggested that a slow off-rate from genomic DNA critically compromises the STAT′s functions as potent transcription factors for a limited number of target genes. The corresponding DNA-binding mutants of STAT1 created so far display either a decreased affinity for DNA or a complete failure to discriminate between GAS and non-GAS elements. Thus, not surprisingly, both resulted in defective transcriptional activity[13, 41]. However, the behavior of a hypothetical DNA-binding mutant with preserved GAS recognition and enhanced DNA-binding affinity surpassing that of the wild-type protein has not been studied so far. It is surmised that such a DNA-binding mutant would be a useful tool to dissect STAT1 signal transduction in terms of nucleocytoplasmic shuttling, cytokine-induced nuclear accumulation, and activation of target genes. As presented here, we have created such a point mutant which has allowed us to systematically investigate the transcriptional consequences of enhanced DNA binding. Using this mutant, we have revealed a simple molecular mechanism that enables STAT dimers to dissociate from non-target DNA in order to continue their search for GAS sites. We have demonstrated that a high off-rate from genomic DNA is required as a key feature for target gene finding, which allows for the efficient transmission of extracellular signals into transcriptional responses.
In an attempt to characterize the biological effects of increased protein-DNA interactions at a transcriptional level, we performed a mutational study on STAT1 and assessed the resulting mutants for their capability to bind sequence-specifically to DNA and activate interferon-responsive target genes. Most of the STAT1 point mutants generated with substitutions in the DNA-binding domain showed a reduced affinity for DNA and were, therefore, inappropriate to test the functional consequences of high-affinity DNA binding for gene expression. However, we identified two single point mutants that fulfilled our expectations for an enhanced binding to GAS sites. Replacement of two glutamic acid residues in the DNA-binding domain, although not interfering with the recognition of GAS elements, independently stabilizes preformed STAT1-DNA complexes. The presence of negatively charged residues at position 411 and 421 is required for the release of STAT1 dimers from DNA, as their substitution with either alanine or a positively charged lysyl residue remarkably reduced the dissociation rate from both GAS and GAS-like elements. The striking finding that enhanced GAS binding is associated with a dramatically reduced gene expression in cytokine-stimulated cells clearly underlines the significance of intact nucleocytoplasmic shuttling for full transcriptional activation. Moreover, it suggests that a limited residence time in the nucleus is an inherent property of STAT1 signal transduction and, conversely, a reduced dissociation rate from GAS elements results in suppressed gene induction.
Available crystallographic data have revealed that the glutamyl residue 411 does not directly contact specific nucleotide bases or the sugar-phosphodiester backbone of DNA, but in the DNA-bound form it has nevertheless free access to the DNA molecule, suggesting that there may be some minor structural flexibility within the STAT1 DNA-binding domain (see Figure1A). It has been reported that residue 421 can accept hydrogen bonds from guanine in the minor groove, although the precise interface between the surface of the STAT1 DNA-binding domain and the DNA double helix in the proximity to E421 is not known due to the superimposition of non-equivalent base pairs at these positions. The functional relevance of the two glutamyl residues can best be regarded as an off-switch to release STAT1 dimers from DNA, so that they become a readily accessible substrate for the inactivating nuclear phosphatase. The presence of a glutamic acid residue with a terminal carboxyl group adjacent to phosphate groups in the DNA backbone facilitates the fast disassembly of STAT1-DNA complexes possibly via electrostatic repulsion. Interestingly, these residues are directly engaged in the discrimination between canonical and non-canonical binding sites, since its replacement by alanine results in a mutant with preserved GAS recognition and a broadened spectrum of potential binding sites (see Figure4C). This finding suggests that the repulsive effect on DNA binding exerted by these residues is independent of the underlying DNA sequences and occurs at classical GAS, GAS-like or even non-GAS sites. The native glutamyl residues seem to facilitate the release of STAT1 dimers from DNA via electrostatic interactions, thereby increasing the number of STAT1 molecules participating in productive nucleocytoplasmic shuttling.
In the wild-type molecule, the fast dissociation from DNA contributes to the coupling of DNA release and subsequent tyrosine dephosphorylation to transcriptional activation. Under conditions of cytokine stimulation the fast release from DNA ensures that the intracellular concentration of tyrosine-phosphorylated STAT1 is always limited due to the high tyrosine phosphatase activity in the nucleoplasma. In the DNA-binding mutants E411A/K and E421K, this coupling between the recruitment to genomic DNA and their fast dephosphorylation is critically disturbed, since these mutants are more than the wild-type protein stacked on genomic DNA in complexes, which may also contain co-expressed native STAT1 (see Figures1B and2A,B). Due to the reduced number of cycling STAT1 dimers (Figure3), their cytokine-induced transcriptional response is substantially limited (Figure6). The prolonged nuclear residence time of the glutamyl mutants following cytokine stimulation of cells (Figure5) appears to directly reflect their decreased tyrosine dephosphorylation (Figure4D), suggesting that they are retained in a DNA-bound state at transcriptionally inert genomic loci. Tyrosine-phosphorylated native STAT1 molecules form heterodimers with the co-expressed recombinant STAT1 mutants as detected by gelshift experiments (Figure2B), which are integrated into DNA-bound STAT complexes and protected from fast inactivation (Figure2A). Thus, paradoxically, despite their increased GAS binding and elevated concentration in the nuclear compartment, where transcription exclusively takes place, the mutants are nevertheless weaker transcriptional activators.
Interestingly, by introducing a neutral or a positively charged functional group at position 411, we generated a graduated series of STAT1 variants (glutamate > alanine > lysine) with stepwise diminished transcriptional activity at an artificial reporter gene construct. Thus, changing the electric charge of this residue permits interference with gene induction simply by shifting the amount of STAT1 dimers to a DNA-bound state in which they are prevented from freely shuttling between cytoplasm and nucleus. From our experiments, we cannot conclude whether the impaired transcriptional activity at native target genes detected for the mutants results from a diminished exchange rate at a single promoter or merely reflects decreased promoter occupancy due to predominant deposition at low-affinity DNA-binding sites. However, we observed that cytokine stimulation leads to high nuclear concentrations of mutant STAT1, which clearly exceed that of the wild-type protein (see Figure3A). This finding suggests that mutant STAT1 preferentially deposits outside transcriptionally active sites. In this scenario, a limited number of high-affinity GAS sites compete with the virtually unlimited amount of non-GAS sequences within the entire genome for binding to STAT1. In interferon-stimulated cells, phospho-STAT dimers retained in the nucleus may not be exclusively bound to GAS sites, but are additionally recruited to an overwhelming reservoir of unspecific, low-affinity DNA-binding sites, from which they are released with very high exchange rates. Interestingly, Lerner and colleagues had previously shown that STAT3 and glucocorticoid receptor assembled at the α2-macroglobulin promoter into an enhanceosome for which continued renewal of both transcription factors was required for full transcriptional activity.