Distinct DNA Elements Contribute to Rap1p Affinity for its Binding Sites

Abstract

The essential Saccharomyces cerevisiae regulatory protein Rap1 contains two tandem Myb-like DNA binding sub-domains that interact with two defined DNA “hemisites”, separated by a trinucleotide linker sequence. We have mapped the thermodynamically defined DNA-binding site of Rap1 by a primer extension method coupled with electrophoretic separation of bound and unbound DNAs. Relative to published consensus sequences, we detect binding interactions that extend 3 bp beyond the 5′-end of the putative DNA-binding site. This new site of interaction is located where the DNA minor groove faces the protein, and may account for the major DNA bending induced by Rap1p that previous studies have mapped to a site immediately upstream of the consensus binding site. In addition, we show that a minimal DNA-binding site made of one single consensus hemisite, preceded or followed by a spacer trinucleotide that interacts with the unstructured protein linker between the two Rap1p DNA binding domains, is able to bind the protein, although at lower affinity. These findings may explain the observed in vivo binding properties of Rap1p at many promoters that lack canonical binding sites.

Introduction

The majority of ribosomal protein (RP) genes and a number of the translation factors genes contain binding sites for the essential Rap1 protein in their upstream regions.1., 2. Rap1p binding at these and many other promoter sites in vivo was recently demonstrated by global (genome-wide) chromatin immunoprecipitation analyses.3., 4. The set of genes whose upstream sequences bind Rap1p is heavily skewed towards highly expressed, growth-related genes, and, in the few cases examined in detail, deletion analyses indicate that the Rap1p binding sites are required for full activation.⁵ Nevertheless, direct proof for an activation role of Rap1p at specific genes has been difficult to obtain.⁶ Recent global transcription analyses have revealed a remarkably robust coordinate regulation of RP gene transcription in response to a wide range of nutrient or stress conditions,7., 8., 9. emphasizing the importance to the yeast cell of controlling RP gene transcription, and ultimately ribosome production, with perceived growth potential.¹⁰ Interestingly, the down-regulation of RP gene transcription in response to one such stress condition, a block to secretion,¹¹ has been directly linked to Rap1¹² and this protein is also important for this response at RP gene promoters devoid of its binding sites.¹² Moreover, in recent papers, a role of Rap1p in the transcriptional repression of Pol I genes due to secretory defects has been shown.13., 14. An understanding of the molecular mechanisms underlying Rap1p action at promoters, both direct and indirect, will thus have important implications for growth regulation in general.

The mechanism(s) by which Rap1p contributes to transcriptional activation at RP genes are still unclear, although several models have been proposed.15., 16., 17., 18., 19. Rap1p exerts a strong effect on the chromatin structure of the promoters to which it binds, and has been proposed to act, at least in some cases, as a chromatin “opener”.20., 21. At promoters for some glycolytic enzymes genes Rap1p appears to promote activation by a direct interaction with the co-activator Gcr1p, which requires Rap1p to bind in the context of chromatin.19., 22. It should also be pointed out that in different chromatin contexts, namely at telomeres and silent mating-type loci, Rap1p binding contributes directly to the formation of Sir-dependent heterochromatin.2., 23.

Rap1 is a large protein (827 aa) that contains two distinct but structurally similar Myb-like DNA-binding domains (DBD).²⁴ These two DBD interact with two (often imperfect) direct repeats present in known Rap1 DNA-binding sites,24., 25. which display considerable sequence variability. Detailed analysis of three different Rap1-DNA co-crystal structures indicates that Rap1 recognizes different sequences through subtle re-positioning of side-chains at the protein–DNA interface rather than global movements of the two DNA-binding sub-domains.²⁵ Two similar consensus Rap1 binding-site sequences have recently been published: ACAYCCRTACATY²⁶ and ACACCCRYACAYM.³ The latter was shown to be predictive for Rap1p binding within promoter regions.³ However, Rap1p binding to promoters devoid of canonical binding sites has also been observed in vivo.3., 4. Both consensus sequences are characterized by the two tandem hemisites ACACC or ACAYC and ACATY or ACAYM, connected by a trinucleotide linker sequence (CRT or CRY). Previous analyses using gel electrophoresis or scanning tunneling microscopy suggest that Rap1p causes a large DNA bend (90–110° according to gel estimates), whose vertex is located 5′ of its consensus DNA-binding site.27., 28., 29. Interestingly, this bending is neither observed in the crystal structures of protein–DNA complexes with the isolated DBD,24., 25. nor in solution studies of the DBD alone, but appears instead to require sequences amino-terminal to the DBD.²⁹

Rap1p binding induces the formation of KMnO₄ hypersensitive sites so specific as to be potentially diagnostic for the binding itself.²⁷ We have adapted this method to the in vivo analysis in Saccharomyces cerevisiae cells and have used it to determine the promoter occupancy by Rap1p during transcription regulation.²¹ Our data showed that three copies of Rap1p are simultaneously bound to the TEF2 promoter in exponentially growing cells, and that aminoacid starvation does not cause loss of Rap1p from the complex with DNA. However, in vivo UV-footprinting reveals the occurrence of structural modifications of the complex. Moreover, low-resolution micrococcal nuclease digestion shows that the chromatin of the entire region is devoid of positioned nucleosomes but is susceptible to changes in accessibility to the nuclease upon aminoacid starvation. Thus, the modification of the function of the Rap1p-controlled promoters studied reveals a defined set of in vivo modifications caused not by loss of protein occupancy but, possibly, by structural modifications of its complex with DNA, as postulated by Pina.⁵ In order to better understand the mechanism of action of this widespread and relevant protein, whose regulatory roles often appear to be contradictory, we have performed a detailed analysis of the DNA elements involved in its function. The results provide an explanation of the biological role paradoxically exerted by Rap1p at several promoters devoid of canonical binding sites.