Modifying Histones and Initiating Meiotic Recombination

Cell 404

transition an all-or-nothing event or are there transition states? Will the protein components whose folds remain to be determined yield more novel structures? Does the tail transition involve local refoldings as well as wholesale rigid-body motions? Above all—what is the reason for its extreme complexity? Much remains to be gleaned from closer comparison of the pre- and postinfection tail complexes of T4. Alasdair C. Steven Laboratory of Structural Biology, NIAMS Building 50, Room 1517 50 South Drive MSC 8025 National Institutes of Health Bethesda, Maryland 20892 Selected Reading Conway, J.F., Wikoff, W.R., Cheng, N., Duda, R.L., Hendrix, R.W., Johnson, J.E., and Steven, A.C. (2001). Science 292, 744–748. Coombs, D.H., and Arisaka, F. (1994). In Molecular Biology of Bacteriophage T4, J.D. Karam, ed. (Washington, D.C.: American Society for Microbiology), pp. 259–281. Jacob, F. (1988). The Statue Within (New York: Basic Books Inc.), p. 224. King, J., and Chiu, W. (1997). In Structural Biology of Viruses, W. Chiu, R.M. Burnett, and R. Garcea, eds. (New York: Oxford University Press), pp. 288–311. Kostyuchenko, V.A., Leiman, P.G., Chipman, P.R., Kanamaru, S., van Raaij, M.J., Ariska, F., Mesyahnzhinov, V.V., and Rossmann, M.G. (2003). Nat. Struct. Biol. 10, 688–693. Leiman, P.G., Kanamaru, S., Mesyahnzhinov, V.V., Arisaka, F., and Rossmann, M.G. (2003). Cell. Mol. Life Sci. 60, 2356–2370. Leiman, P.G., Chipman, P.R., Kostyuchenko, V.A., Mesyanzhinov, V.V., and Rossmann, M.G. (2004). Cell 118, this issue, 419–429. Monod, J. (1971). Chance and Necessity (Glasgow: Collins Sons & Co.), pp. 85–86. Van Raaij, M.J., Schiehn, G., Burda, M.R., and Miller, S. (2001). J. Mol. Biol. 314, 1137–1146.

Modifying Histones and Initiating Meiotic Recombination: New Answers to an Old Question It is well documented that the formation of the DNA double-strand breaks (DSBs) that initiate meiotic recombination is influenced by chromatin and larger scale chromosome organization, but the molecular nature of this influence has remained elusive. Several recent studies, including Reddy and Villeneuve (2004, this issue of Cell), shed light on this issue by revealing roles for posttranslational histone modifications in promoting DSB formation. Meiosis enables sexual organisms to halve their chromosome number. Usually, accurate chromosome segregation at the first meiotic division requires the formation of physical connections between homologous parental chromosomes that result from reciprocal exchange of genetic material. This exchange occurs via homologous re-

combination, and is initiated by DSBs formed by a conserved topoisomerase-like protein, Spo11 (Keeney, 2001). Several observations tie DSB formation to the structural organization of chromosomes. In budding yeast, DSBs form preferentially within localized regions of ⵑ100-250 bp, called hotspots. Although the rules dictating hotspot location and activity are not fully understood, local chromatin structure is one important determinant: hotspots correspond to nuclease-hypersensitive chromatin regions and in several cases DSB formation requires an open chromatin structure (Petes, 2001; Keeney, 2001). Moreover, the chromatin at hotspots displays a meiosis-specific increase in nuclease sensitivity before DSB formation, indicating that alterations in chromatin structure may contribute to hotspot activity. Recombination events in mammals are also highly localized within small (1–2 kb) hotspots (Kauppi et al., 2004). DSB formation is also influenced by higher order chromosome architecture. DSBs are nonrandomly distributed along chromosomes such that there are alternating large domains of high and low recombination frequencies. Hot domains usually correlate with a G⫹C-rich DNA base composition (Petes 2001; Kauppi et al., 2004). In addition, sister chromatids are organized into a series of loops; the bases of these loops and their associated proteins form the structural axes of meiotic chromosomes. Recent studies indicate that DSB formation occurs within the chromatin loops, but homologous recombination proceeds in intimate contact with the chromosome axes. Thus, higher order chromosome structure contributes to the spatial organization of meiotic recombination (see, for example, Kleckner et al., 2004). Although the links of DSB formation to chromatin and chromosome organization are well established, the molecular basis for these links is not understood. Several recent studies provide new insight by pointing to certain histone modifications as being important for DSB formation. In this issue of Cell, Reddy and Villeneuve (2004) provide an intriguing characterization of the him-17 gene in C. elegans (Reddy and Villeneuve, 2004). Null him-17 mutants are defective for meiotic recombination and chromosome segregation. Similar to spo-11 mutants, him-17 mutant oocytes fail to accumulate chromosomeassociated complexes of the strand exchange protein RAD-51, and DNA breaks caused by ionizing radiation can rescue the meiotic defects of him-17 mutants. These and other phenotypes provide a compelling case that HIM-17 is required for DSB formation. Two lines of evidence also connect HIM-17 to posttranslational histone modifications. First, him-17 null mutants display reduced and delayed accumulation of histone H3 dimethyl lysine 9 (H3MeK9). This defect is not simply a consequence of failure to make DSBs because spo-11 mutants do not show this pattern. Second, HIM17 is a modular protein with multiple copies of a conserved sequence motif related to a DNA binding domain in the Drosophila P element transposase. This domain is found in several other C. elegans proteins, many of which interact genetically with the worm homolog of the human retinoblastoma (Rb) protein, LIN-35. Based on this similarity, lin-35 mutations were tested and shown to exacerbate the recombination defect caused by a

Previews 405

non-null allele of him-17. Because Rb proteins can modulate chromatin structure through interactions with histone modifying enzymes, these findings provide a tantalizing, albeit indirect, connection between HIM-17 and chromatin structure. As pointed out by the authors, the precise molecular connection remains to be elucidated between HIM-17 functions in histone modification and DSB formation. H3MeK9 might be required for DSB formation per se, or DSB and methylation defects might be separate consequences of loss of HIM-17. Notably, HIM-17-dependent H3 methylation is quite late relative to DSB formation. This apparent timing discrepancy may simply reflect limitations of the antibody staining used to detect histone modifications. Nevertheless, some caution in interpreting the cause-effect relationship between these events is warranted. Two recent studies in S. cerevisiae have also implicated histone modifications in DSB formation. Set1 is a methyltransferase that methylates histone H3 on lysine 4. Deletion of the SET1 gene significantly reduces DSB formation at several hotspots (Sollier et al., 2004). In another study, ubiquitylation of histone H2B was shown to be important for DSB formation: elimination of Rad6 (an E2 ubiquitin-conjugating enzyme), Bre1 (an E3 ubiquitin ligase), or the ubiquitin-acceptor lysine on histone H2B each causes a delay in the timing and a reduction in the frequency of DSBs at some but not all hotspots (Yamashita et al., 2004). These two studies may be tied together by the fact that Rad6-dependent ubiquitylation of histone H2B is required for H3MeK4 formation (Fischle et al., 2003). Finally, a correlation between histone acetylation, DSB formation, and local chromatin structure has been reported at a meiotic recombination hotspot in S. pombe (Yamada et al., 2004). DSB formation at the well-characterized ade6-M26 hotspot depends on binding of a heterodimeric transcription factor, Atf1-Pcr1. Previous work showed that the nuclease sensitivity of chromatin at ade6-M26 is altered in meiosis. Ohta and colleagues have now demonstrated that histones H3 and H4 in the vicinity of ade6-M26 site are hyperacetylated during meiosis, and that much of this H3 acetylation requires the S. pombe HAT (histone acetyltransferase) Gcn5 (Yamada et al., 2004). Changes in nuclease hypersensitivity of hotspot chromatin are delayed and recombination is reduced in gcn5-deleted strains. Mutation of snf22, which encodes a putative ATP-dependent chromatinremodeling factor (ADCR), causes even more profound defects. Putting these observations together, the authors propose that the transcription factor Atf1-Pcr1 recruits HATs and ADCRs to assist chromatin structural changes and DSB formation. Because snf22 single and snf22 gcn5 double mutants are similar to one another and are more severe than gcn5 single mutants, it is likely that another HAT(s) in addition to Gcn5 also acts at this hotspot. These findings provide further evidence for a connection between DSB formation and histone modifications. However, although gcn5 deletion reduces genome-wide H3 acetylation, recombination is not affected globally, so this regulatory mechanism may be specific for ade6-M26. These new studies in worms and fungi clearly establish correlations between meiotic DSB formation and certain histone modifications. Formally, it is difficult to

rule out indirect effects, such as through defects in expression of meiotic genes important for DSB formation. However, direct measures of gene expression argue against this interpretation in several of the studies. Another caveat is that posttranslational modifications of proteins other than (or in addition to) histones may be important. For example, HATs can also acetylate nonhistone proteins such as p53 (Brown et al., 2000). However, if we assume that histone modifications promote meiotic DSB formation more directly, what are the possible mechanisms? As discussed earlier, DSB formation throughout the genome is governed at several levels, both locally within and near DSB hotspots, and more globally along chromosomes. It is possible then that histone modifications could work at one or the other of these levels. At the local level, histone modifications might regulate accessibility of the chromatin substrate to Spo11 and other DSB factors through alterations in chromatin structure at hotspots (e.g., through recruitment of chromatin-remodeling activities or by direct effects of histone modifications on the positioning, stability, or occupancy of nucleosomes). Alternatively, histone modifications might provide local “marks” that recruit DSB proteins by direct protein-protein interactions. A review by Petes (2001) provides a detailed discussion of these and other possibilities. At a more global level, histone modifications might promote DSB formation through effects on higher order chromatin compaction (Reddy and Villeneuve, 2004). For example, Kleckner and colleagues (2004) have proposed that many aspects of chromosome function can be explained on the basis of mechanical forces generated by the expansion of chromatin against structural features of the chromosome that constrain that expansion. Based on the relationship of Spo11 to archaeal topoisomerases, such mechanical forces could promote DSB formation by favoring disruption of a Spo11 dimer interface and converting a reversible Spo11-DNA complex to an irreversible Spo11 bound DSB (see Keeney 2001 for more details). In this light, it is interesting to note an intriguing paradox between the worm and fungal studies. The HIM17-dependent modification in C. elegans (H3MeK9) is implicated in processes more often associated with a “closed” chromatin state, such as in heterochromatin or other transcriptionally repressed regions (reviewed in Fischle et al., 2003). In contrast, the modifications implicated in yeast recombination (H3MeK4 and hyperacetylated H3 and H4) are more often associated with more “open,” transcriptionally active chromatin. These distinctions might reflect fundamental differences in the manner in which Spo11-dependent breaks are regulated in nematodes versus yeasts. But there are alternative ways to think about the situation as well. Reddy and Villeneuve (2004) proposed an interesting possibility, suggesting that chromatin compaction promoted by histone H3 methylation in one region of the chromosome might result in a compensatory expansion that promotes DSB formation elsewhere. Such an action-at-a-distance model for HIM-17 might help to explain the organization of hot and cold chromosomal domains and loop-axis relationships. On the other hand, the paradox might only be an apparent one, resulting from exploration of only

Cell 406

the tip of the modification iceberg: HIM-17 might also influence other modifications; likewise, roles of other modifications remain to be tested in yeast. Finally, it is almost certainly too simplistic to divide particular histone modifications into classifications as either open or closed chromatin, as evidenced by the fact that substantial functional crosstalk connects transcriptional silencing with histone modifications associated with transcriptionally active chromatin (Fischle et al., 2003). None of these molecular models is mutually exclusive. Whether the connection is local or global, direct or indirect, it is now clear that the complex array of posttranslational histone modifications must be taken into account in order to fully understand the mechanism of meiotic recombination initiation.

Shohreh Maleki and Scott Keeney* Molecular Biology Program Memorial Sloan-Kettering Cancer Center 1275 York Avenue Box 97 New York, New York 10021

Selected Reading Brown, C.E., Lechner, T., Howe, L.A., and Workman, J. (2000). Trends Biochem. Sci. 25, 15–19. Fischle, W., Wang, Y., and Allis, C.D. (2003). Curr. Opin. Cell Biol. 15, 172–183. Kauppi, L., Jeffreys, A.J., and Keeney, S. (2004). Nature Rev. Genet. 5, 413–424. Keeney, S. (2001). Curr. Top. Dev. Biol. 52, 1–53. Kleckner, N., Zickler, D., Jones, G.H., Henle, J., Dekker, J., and Hutchinson, J. (2004). Proc. Natl. Acad. Sci. USA, in press. Petes, T.D. (2001). Nature Rev.Genet. 2, 360–369. Reddy, K.C., and Villeneuve, A. (2004). Cell 118, this issue, 439–452. Sollier, J., Lin, W., Soustelle, C., Suhre, K., Nicolas, A., Ge´li, V., and de la Roche Saint-Andre´, C. (2004). EMBO J. 23, 1957–1967. Yamada, T., Mizuno, K.I., Hirota, K., Kon, N., Wahls, W.P., Hartsuiker, E., Morofushi, H., Shibata, T., and Ohta, K. (2004). EMBO J. 23, 1792– 1803. Yamashita, K., Shinohara, M., and Shinohara, A. (2004). Proc. Natl. Acad. Sci. USA 101, 11380–11385.

Little Things that Count in Transcriptional Regulation

Although the NF-␬B transcription factor tolerates significant sequence variations in its DNA binding site, NF-␬B sites show remarkable evolutionary stability. In this issue of Cell, Leung et al. (2004) demonstrate that a single nucleotide change in an NF-␬B binding site changes the dependence of the promoter on a specific coactivator without affecting NF-␬B binding. These data suggest that binding sites may impart a specific configuration to bound transcription factors.

Interactions between a transcription factor and its cognate binding site are often described quantitatively, usually in terms of affinity of the transcription factor for a certain DNA sequence. In this regard, most transcription factors are tolerant to sequence variations, as they are able to bind several degenerate binding sites with comparable affinity. A fundamental question is if this sequence variability among high-affinity binding sites simply reflects the functional neutrality of small nucleotide changes or if these variations are biologically relevant. If the latter is the case, binding sites should bear information that transcription factors can decipher and broadcast to other components of the transcriptional machinery. The new findings reported by Leung, Hoffmann, and Baltimore in this issue of Cell (Leung et al., 2004) indicate that the specific sequence of a binding site indeed imparts characteristic functional properties to the bound transcription factor. A single nucleotide change in an NF-␬B binding site did not modify the ability of the promoter to recruit NF-␬B dimers in vivo, consistent with the well-known tolerance of NF-␬B to sequence variations; remarkably, however, this small variation was sufficient to change the dependency of an NF-␬B-regulated promoter on a specific coactivator. Baltimore and coworkers show that two NF-␬B-dependent genes encoding the chemokines MCP-1 and IP-10 and bearing ␬B sites differing in a single nucleotide have different responsiveness to NF-␬B dimers and different coactivator requirements. MCP-1 can be activated by both p65/p50 heterodimers and p65 homodimers and does not require the bcl-3 coactivator for transcription in response to TNF. Conversely, the IP-10 promoter strictly requires p50/p65, and, although in cells lacking p50 it can recruit p65 homodimers in response to TNF stimulation, it is not activated. This is because p65 homodimers cannot support a functional interaction with bcl-3, which is required for IP-10 induction by TNF. If the MCP-1 ␬B sites are swapped for the IP-10 ␬B sites, MCP-1 acquires NF␬B dimer and cofactor requirements identical to those of IP-10: it is not activated by p65 homodimers and strictly requires bcl-3 for transcriptional induction by TNF. Finally, although p65 homodimers bound to IP-10 promoter cannot productively interact with bcl-3, they can promote recruitment of a different transcription factor, namely IRF3, which in this context works as a coactivator (i.e., it is recruited to promoter via protein-protein interactions and not via binding to DNA). Since IRF-3 is selectively induced by LPS and not by TNF, the IP10 promoter loaded with p65 homodimers retains full responsiveness to LPS stimulation. The final message is that a single nucleotide difference in the ␬B sites of the two genes is the key to their specific NF-␬B dimer and cofactor requirements. How does such a molecular switch work? Side chains of critical amino acids in DNA binding domains of transcription factors make highly specific contacts with the accessible chemical groups of the base pairs exposed in the minor and major grooves of DNA (Garvie and Wolberger, 2001). Even a single nucleotide change causes detectable differences in the chemical groups exposed for interaction with the transcription factor. Therefore, the transcription factor has to adopt different conformations to optimize the contacts between its amino acids and