Myb repeats were first discovered in the DNA-binding domains of proteins typified by the products of the v-Myb retroviral oncogene and its normal cellular progenitor, c-Myb. Such Myb domains adopt a helix-turn-helix fold found in many DNA-binding proteins in which basic residues interact with the negatively charged phosphate backbone of the DNA double helix. However, Myb-related domains are present in a wide variety of nuclear proteins, some of which do not themselves bind directly to DNA. We used phylogenetic analysis to identify a conserved acidic patch within the first of the three helices of most Myb domains. Remarkably, these acid patches lie on surfaces of DNA-binding Myb proteins that face away from DNA.
We have used alanine scanning mutagenesis to determine the function of the conserved acidic patches in the c-Myb protein. We chose to focus particularly on the acidic patch within the third Myb repeat of c-Myb because a mutant of this patch that did not affect protein stability via salt bridge formation, nuclear localization, or DNA binding, nevertheless was defective in transcriptional activation both of transfected reporter genes and of an endogenous target gene. We used biochemical analyses to demonstrate that the specific defect in transcriptional activation by this conserved acidic patch mutant correlated with a decrease in chromatin association in vivo. This mutant protein also displayed a reduced ability to interact with the histone H4 tail in vitro.
Although we observed a complete defect in activation of the endogenous target by the mR3-3 mutant, we detected only a 50% decrease in the ability of the third repeat mutant to bind chromatin in vivo. One possible explanation for our results is that c-Myb may utilize different methods of binding at distinct chromatin domains. The in vitro histone binding experiments demonstrated a specific defect of mR3-3 for binding the histone H4 tail but not the histone H2B tail. If the interaction of c-Myb with chromatin requires the H4 tail at a subset of sites and conversely the H2B tail at a different subset of sites, then the reduced ability to bind nucleosomes by mR3-3 in the chromatin immunoprecipitation assays would represent the loss of binding at the subset of sites requiring the c-Myb interaction with the H4 tail.
Another possibility is that all sites require both H4 and H2B tail association for stable c-Myb chromatin binding. In this case, the observed decrease in association with nucleosomes would represent a weakened interaction of the mR3-3 mutant with chromatin at all sites. Weakened association with nucleosomes would lead to a higher off-rate for the mutant protein bound to chromatin. However, the chemical crosslinking used in chromatin immunoprecipitation experiments might allow us to capture mutant protein bound to chromatin even if the mutant cannot occupy its chromatin site long enough to effectively activate transcription, thereby explaining the observed complete loss of activity in the context of a partial chromatin binding defect. The reduced association of the mR3-3 mutant with nucleosomes containing a variety of specific histone modifications was similar to that see for total core histones, supporting a global chromatin binding defect.
Although we observed a decreased binding of the mutant Myb protein to histone H4 tails in vitro, it is also possible that the defects in transcriptional activation and in chromatin association in vivo are due to interactions with non-histone proteins. For better or worse, the major histone proteins in metazoans are encoded by large clusters of tandemly repeated histone genes, making the generation and study of substitution mutants impossible with current technology. This is a general problem with all studies of covalent histone modifications and histone binding proteins. At present, the simplest model that explains all of our data is that the conserved acidic patch in Myb-related proteins is required for interaction with the basic region of the H4 tail, and that a loss of this interaction in the acidic patch mutant results in decreased chromatin occupancy which in turn results in a failure of transcriptional activation in vivo.
We note that genome-wide studies of chromatin occupancy by the sole Myb protein of Drosophila have shown that in a single cell line, greater than 30% of all promoters are occupied by a Myb protein complex . One can hypothesize two classes of promoters: those which require the Myb acidic patch for initial access in their native chromatin environment; and those in which another protein creates an apparently nucleosome-free promoter region and thus do not require the Myb acidic patch for DNA occupancy by Myb protein. Such a model might explain why the VP16 activation domain can bypass the absence of the Myb domain acidic patch in poorly chromatinized templates for transcription typicial of transient transfection assays, whereas no such bypass occurs on the endogenous mim-1 gene in its native chromatin context.
Others have presented data arguing that the c-Myb but not the v-Myb DBD can specifically bind the histone H3 tail in vitro and have proposed that disruption of H3 tail binding by v-Myb is due to three mutations of hydrophobic residues in the second Myb repeat (N91I, H106L, D117V) . Specific H3 tail binding by the c-Myb DBD was not observed in our experiments under several different conditions. However, by including competing proteins (1% nonfat dried milk), glycerol to disrupt hydrophobic interactions (10%), and a reducing agent (5 mM DTT), we did observe specific binding to the H4 and H2B tails, arguing that histone tail binding is an important function of the c-Myb DBD. Alanine mutagenesis of the acidic patch (mR3-3) led to a selective disruption of H4 tail binding. We did not see binding to the H3 tail by c-Myb in the presence of non-specific competitor proteins unless we used a concentration of GST-H3 tail ten-times higher than our standard conditions which did permit binding to H4 and H2B (data not shown).
The acidic patch of c-Myb was identified due to high conservation of these amino acids among the diverse group of Myb-related proteins, suggesting a common conserved function. Our results demonstrating chromatin and histone binding properties of the c-Myb DNA-binding domain are consistent with activities reported for other proteins containing Myb-related repeats. SANT domains represent a subset of Myb-related repeats in proteins that lack specific DNA-binding capacity . Mutagenesis and deletion analysis of the Myb-related repeat, or SANT domain, in the ADA2 component of yeast GCN5 histone acetylase complex led to impaired histone acetylase activity and disrupted H3 tail binding of the complex [8, 9]. Myb-related repeats in components of the ATP-dependent chromatin remodeling SWI/SNF and RSC complexes were also essential for in vivo activity .
The ISWI protein provides the enzymatic component of several ATP-dependent chromatin remodeling complexes and contains two Myb-related repeats. Nucleosome and histone H4 tail stimulated activity in vitro required the Myb-related repeats. One of the ISWI Myb-related repeats, termed the SLIDE domain, was also essential for stimulation of ISWI ATPase activity by free DNA, indicating this Myb-related repeat contacts DNA as well as histones in chromatin . Interestingly, deletion of the Myb-related repeat in the TFIIIB-B" subunit of RNA polymerase III disrupted transcription on native chromatin in vivo but not on a naked DNA template in vitro . Myb-related domains are also present in protein complexes that associate with and regulate inactive heterochromatin, suggesting that Myb-related motifs can function in the context of different chromatin domains [44, 45].
Our data may reflect a predilection for histone H4 binding by complexes containing the three-repeat Myb proteins. Consistent with this idea is the finding that the Drosophila MMB complex (Myb-MuvB/dREAM) specifically interacts with unmodified histone H4 tails . Histone H4 constitutes a prime target for complexes involved in global regulation of chromatin because it is the only histone without known variants . In contrast, histones H3 and H2A have relatively abundant nonallelic variants, some of which define particular chromatin domains. Association of the Myb domain with histone H4 may reflect the ability of Myb-related proteins to bind chromatin despite the structural context since nucleosomes in all chromatin domains would carry the same invariant H4.
An intriguing group of Myb-related proteins in yeast may provide further insight into the role of Myb family protein in vertebrates. Yeast general regulatory factors (GRF) consist of a group of four proteins (ScRAP1, ScREB1, ScTBF1, ScABF1) identified due to their involvement in multiple nuclear processes including replication, transcription, silencing, and telomere maintenance [48–50]. The presence of many binding sites throughout the genome and the existence of Myb-related DNA-binding domains in three of these proteins (ScRAP1, ScREB1, ScTBF1) are common to this group [51, 52]. Data suggest that GRFs have a common mechanism since the binding site of one GRF can substitute for another and swapping of protein domains leads to functional GRF chimeric proteins [51, 53]. GRFs were first identified because in vivo binding led to a nucleosome free region at promoters, giving rise to the hypothesis that GRFs function by local opening of chromatin structure . Recent studies have implicated the REB1 GRF in establishing "poised" chromatin domains at about 60% of budding yeast promoters via localization of the variant histone H2A.Z at the boundaries of a nucleosome free region [52, 54–56]. Association with an invariant H4 tail might be required to facilitate this replication-independent exchange of H2A variants. GRFs also function by insulating chromatin domains, such as between silenced heterochromatin and genes primed for activation [48, 49]. The ability to properly establish and regulate such chromatin domains is essential for genomic stability and the loss of any single yeast GRF leads to cell death, possibly due to genomic instability.
Besides the obvious presence of a Myb DNA-binding domain, the Myb family of proteins share many common properties with yeast GRFs. Myb proteins have a short consensus binding site (PyAACG/TG) present throughout the genome . Visualization of GFP-tagged Dm-Myb revealed many cytologically visible binding sites on polytene chromosomes . In addition, genome-wide studies have recently shown that this Myb complex is present at greater than 30% of all promoters in a single cell type .
Our finding of a conserved acidic motif essential for the biological function and required for efficient chromatin and histone association support a role for Myb proteins in binding to and possibly regulating chromatin structure. Consistent with this are the observations that other Myb-related proteins utilize their Myb domains for chromatin regulation. The Myb-like domain in yeast GRFs has been implicated in genomic partitioning and regulation of chromatin domains . The Myb-related domains in SANT proteins have been shown to be required for in vivo function and are known to serve a role in histone binding. The Myb family is required for the regulation of multiple genomic processes, is essential for genomic stability, and we have demonstrated that it contains a conserved acidic motif necessary for transcriptional activity and efficient chromatin association. Because this acidic patch is among the most conserved features common to all Myb-related domains, our findings support the idea that the primary conserved function of Myb-related domains may be histone-binding rather than DNA-binding. We postulate that the acidic patch of the Myb-related proteins may interact with the same basic region of the N-terminal histone H4 tail that was bound to an acidic patch created by the H2A/H2B core in the crystal structure of the nucleosome .