Developmental Regulation of Eed Complex Composition Governs a Switch in Global Histone Modification in Brain*

Originally discovered as epigenetic regulators of developmental gene expression, the Polycomb (PcG) and trithorax (trxG) group of proteins form distinct nuclear complexes governing post-translational modification of histone tails. This study identified a novel, developmentally regulated interface between Eed and Mll, pivotal constituents of PcG and trxG pathways, respectively, in mouse brain. Although the PcG proteins Eed and EzH2 (Enhancer of Zeste protein-2) engaged in a common complex during neurodevelopment, Eed associated with the trxG protein Mll upon brain maturation. Comprehensive analysis of multiple histone modifications revealed differential substrate specificity of the novel Eed-Mll complex in adult brain compared with the developmental Eed-EzH2 complex. Newborn brain from eed heterozygotes and eed;Mll double heterozygotes exhibited decreased trimethylation at lysine 27 of histone H3, as well as hyperacetylation of histone H4. In contrast, adult hippocampus from Mll heterozygotes was remarkable for decreased acetylation of histone H4, which restored to wild-type levels in eed;Mll double heterozygotes. A physiological role for the Eed-Mll complex in adult brain was evident from complementary defects in synaptic plasticity in eed and Mll mutant hippocampi. These results support the notion that developmental regulation of complex composition bestows the predominant Eed complex with the chromatin remodeling activity conducive for gene regulation during neurodevelopment and adult brain function. Thus, this study suggests dynamic regulation of chromatin complex composition as a molecular mechanism to co-opt constituents of developmental pathways into the regulation of neuronal memory formation in adult brain.

Aside from differentiating mouse embryonic stem cells (14), the developmental regulation of Eed complex formation and activity remains largely unknown. This study detected dynamic changes in Eed isoform expression, complex composition, and histone modifying activity in the developing central nervous system (CNS) in mice. Specifically, Eed co-immunoprecipitated with EzH2 during CNS development but formed a complex with the trxG protein Mll in adult brain. Mll represents a 3,868-amino acid protein with multiple domains, including AT-hook domains, PhD zinc fingers, and a bromodomain (31). In addition, a C-terminal SET domain methylates H3-K4 in the regulatory regions of transcribed genes (12). Mll has been shown to interact directly with HDAC1 and -2 (32), providing an additional mechanism for Mll function in chromatin remodeling. In this study, the developmental switch from Eed-EzH2 to Eed-Mll complex formation resulted in stage-specific changes in global histone modification in eed and Mll mutant brains. Furthermore, a distinct physiological role of the Eed-Mll complex in adult brain was evident from the interdependent function of Eed and Mll in the regulation of hippocampal synaptic plasticity. Thus, the combined biochemical, genetic, and physiological results demonstrate a novel molecular interface between PcG and trxG function in mouse brain.

EXPERIMENTAL PROCEDURES
Mice-l7Rn5 3354SB represents N-ethyl-N-nitrosourea-induced point mutation allele, which encodes an L290P substitution in the second WD motif of Eed (15). Although homozygosity for this eed null allele causes lethality at gastrulation, heterozygous animals are viable (20,21). The mutant Mll allele was generated by gene targeting and harbors a lacZ insertion in the third exon, resulting in a truncated protein (33). Similar to other Mll alleles (34,35), homozygosity for the lacZ insertion causes embryonic lethality, whereas Mll heterozygotes are viable (33,36). eed and Mll mutant mice were genotyped as described (15,33). Single and double heterozygous brains were dissected at embryonic day 14.5 (E14.5) from timed mating, as well as at postnatal day 0 (P0) and P28. Adult brain and spleen were isolated from mice at 2-3 months of age.
Histone Preparation-Bulk histone preparation from P0 brain and adult hippocampus was performed as described (37). Electrophoresis of histone preparations and transfer onto polyvinylidene difluoride membranes (Bio-Rad) was followed by blocking for 2 h at room temperature with 5% bovine serum albumin in TBST. Primary incubation with ␣H4acK5,8,12,16, ␣H3m3K27, ␣H3m2K27, ␣H3m1K27, ␣H3m2K4, ␣H3acK14, ␣H3, or ␣H4 antibody was completed overnight at 4°C followed by anti-mouse horseradish peroxidase-conjugated secondary antibody incubation for 2 h at room temperature. Histone modifications were quantified as described previously (37). Levels of histone acetylation and methylation were normalized to the amount of total H3 or H4. Statistical analysis consisted of a one-sample t test with Bonferroni correction for multiple comparisons.
Immunohistochemistry-Dissected P0 and adult brain was fixed in Bouin's fixative and embedded in paraffin for sectioning at 5-7 m. Immunohistochemistry with ␣Eed, ␣Mll-N, or ␣Mll-C antibodies was performed on mutant and wild-type sections mounted side by side on the same slide to allow for comparison of expression levels between genotypes. Standard streptavidin-biotin immunoperoxidase detection was performed as described (38).
Quantitative Real Time RT-PCR-Total RNA was extracted from newborn brain or hippocampus from 3-month-old animals with the MELT total nucleic acid isolation system (Ambion). eed and Mll mRNA expression was evaluated by TaqMan One-step RT-PCR (Applied Biosystems) using the 7500 Fast System (Applied Biosystems). The following TaqMan MGB probes were employed: eed (exon junction 2-3, Mm469651_m1; Applied Biosystems), Mll (exon junction 28 -29, Mm1179231_m1; Applied Biosystems) and 18 S ribosomal RNA (4326313E; Applied Biosystems) as an endogenous control for each sample. The relative levels of eed and Mll mRNA expression were expressed as a ratio of the gene-specific probe to 18 S rRNA and normalized to a single wild-type ratio set to 1.
Electrophysiology was performed in an interface chamber (Fine Science Tools, Foster City, CA). Oxygenated ACSF (30°C) was perfused into the recording chamber at a rate of 1 ml/min. Electrophysiological traces were digitized and stored using a Digidata 1320A and Clampex software (Axon Instruments, Union City, CA). Extracellular stimuli were administered on the border of area CA3 and CA1 along the Schaffercollaterals using Teflon-coated bipolar platinum electrodes. fEPSPs were recorded in stratum radiatum with an ACSF-filled glass recording electrode (1-3 megohms). The relationship between fiber volley and fEPSP slopes over various stimulus intensities was used to assess base-line synaptic transmission. All subsequent experimental stimuli were set to an intensity that evoked an fEPSP with a slope of 50% of the maximum fEPSP slope. To assess NMDA receptor (NMDA-R)-dependent synaptic transmission, the ratio of fEPSP slope versus fiber volley slope over a range of intensities was measured in a modified ACSF (same as above except 4 mM CaCl 2 and no MgCl 2 ) in the presence of the AMPA-R antagonist 6-cyano-7-nitroquinoxaline-2,3-dione (20 M). High frequency stimulation LTP was induced by administering two 100-Hz tetani (1 s), with an intertetanus interval of 20 s. Synaptic efficacy was monitored 20 min prior to and 180 min following induction by recording fEPSPs every 20 s (traces were averaged for every 2 min interval). Slices that did not exhibit stable fEPSP slopes during the first 20 min of recording were excluded from the study.
Input/output relationships were analyzed using a single exponential function (Y ϭ slope max ϫ 1 Ϫ e ϪK ϫ X ) or linear regression where appropriate. Parameters used to fit I/O relationships were compared across genotypes using an F test. PPF, LTP, and PTP data were analyzed via 2-way ANOVA with repeated measures. For analysis of LTP and PTP, data acquired before and after induction were analyzed separately. Post hoc comparisons after 2-way ANOVA were made using the method of Bonferroni. Total depolarization induced by a 100-Hz tetanus was analyzed by comparing the area under the curve with a Kruskal-Wallis ANOVA, followed by a post hoc Dunn's multiple comparison test. Significance for all tests was set at p Ͻ 0.05.

Developmental Regulation of Eed Isoform Expression in
Brain-Immunoprecipitation of Eed at various stages of brain development revealed four Eed isoforms of varying abundance in the 50 -70-kDa molecular mass range, reminiscent of recent findings in 293 cells (13). As shown in Fig. 1, E14.5 brain and adult spleen demonstrated a similar pattern of Eed isoforms. At P0 and P28, the co-expression of developmental and adult brain isoforms at similar abundance suggested a gradual transition in Eed isoform expression. Interestingly, the isoforms identified by immunoprecipitation were undetectable in input lanes at both standard ( Fig. 1) and maximum loading capacity (data not shown). However, it should be emphasized that two Eed antibodies with different epitopes detected the four Eed isoforms, confirming the specificity of the immunoprecipitation (data not shown). In addition, the Western blot results correlate with the low level of Eed expression demonstrated by immunohistochemistry (Fig. 3A).
Developmental Transition from Eed-EzH2 to Eed-Mll Complex Formation in Brain-Differential Eed isoform expression correlated with developmental regulation of Eed complex composition. EzH2 co-immunoprecipitated with Eed from E14.5 FIGURE 1. Developmental transition from Eed-EzH2 to Eed-Mll complex formation in brain. Immunoprecipitation with an ␣Eed antibody detected several Eed isoforms in the 50 -70-kDa range at various stages of brain development, as well as in adult spleen. Asterisks denote isoforms of similar molecular weight in brain stages and spleen. Note the change in the relative abundance of Eed isoforms and the gradual transition from the Eed-EzH2 complex to the Eed-Mll complex during brain development. Antibodies against N-or C-terminal Mll epitopes detected an ϳ180-kDa band. Input consisted of 10 g of lysate. As a control for nonspecific binding, lysates were immunoprecipitated with free beads lacking the Eed antibody. and P0 brain, as well as adult spleen (Fig. 1). In contrast, the two proteins did not co-immunoprecipitate from adult brain. Instead, Eed formed a complex with Mll, as evident from Eed immunoprecipitation and Western blot analysis using both Nand C-terminal Mll antibodies ( Fig. 1). Similar to the Eed isoform pattern, the transition from the Eed-EzH2 complex in the developing brain to the Eed-Mll complex in adult brain appeared gradual, and both complexes co-existed in P28 brain (Fig. 1). These results implicate Eed isoforms in stage-and tissue-specific recruitment of Eed complex constituents. Unlike in transiently transfected 293T cells (32), the Mll and Bmi1 did not co-immunoprecipitate from adult brain (data not shown). Furthermore, attempts to confirm a possible association of HDACs with Eed-EzH2 and Eed-Mll complexes were inconclusive because of the high affinity of HDAC1 and -2 to agarose beads (data not shown).
A Developmental Switch in Global Histone Modification in Brain-The transition from Eed-EzH2 to Eed-Mll complex formation suggested a developmental switch in histone modifying activity from newborn to adult brain. This would be evident from stage-specific aberrations in global histone modification in eed and/or Mll mutant brains. Based on previous studies in other cellular contexts, defects in Eed-Ezh2 interaction would decrease levels of H3-K27 methylation (7-10, 13), whereas disruption of Mll SET domain activity would reduce H3-K4 methylation (12). Alternatively, given the known interaction of both Eed and Mll with HDAC1 and -2 (18,32), mutant brains may exhibit histone acetylation defects. Global histone modification was analyzed from whole newborn brain. To circumvent the significant regional heterogeneity in adult brain as a confounding variable, histones were isolated from dissected hippocampus, which presents a defined cellular composition. In addition, electrophysiology was employed as a well characterized assay for hippocampal function in wild-type and mutant brains.
Unlike P0 brain, adult hippocampus from eed heterozygotes did not exhibit significant changes in histone acetylation or methylation (all comparisons, p Ͼ 0.05) (Fig. 2). However, hippocampal histone preparations from Mll heterozygotes were remarkable for a significant decrease in H4 acetylation (p Ͻ 0.02). Defects in histone modification in adult Mll heterozy-FIGURE 2. A developmental switch in global histone modification in brain. Histone extracts from wild-type and mutant brain from newborns (A) and adult hippocampi (B) were probed with antibodies against H4acK5,8,12,16, H3tm3K27, H3m2K27, H3m1K27, H3meK4, and H3acK14. Representative Western blots of total and modified histones are shown below each graph. For each histone modification, the ratio of modified histone versus total histone H3 or H4 was normalized against the respective wild-type ratio expressed as 100%. Newborns brains were remarkable for hypomethylation of histone H3 in eed ϩ/Ϫ ;Mll ϩ/ϩ and eed ϩ/Ϫ ;Mll ϩ/Ϫ mutants, and hyperacetylation of histone H4 in eed ϩ/Ϫ ;Mll ϩ/ϩ animals. In contrast, adult eed ϩ/ϩ ;Mll ϩ/Ϫ hippocampi demonstrated hypoacetylation of histone H4 which returned to wild-type levels in eed ϩ/Ϫ ;Mll ϩ/Ϫ hippocampus. n represents the number of sets of wild-type and mutant newborn brains or adult hippocampi tested for each modification. Error bars indicate S.E. MARCH 2). Strikingly, H4 acetylation levels in eed;Mll double heterozygotes were indistinguishable from wild type (p Ͼ 0.05). This suggests that Mll requires Eed for histone modification and supports the biochemical findings wherein Mll and Eed form a common complex in adult brain.

Developmental Regulation of Chromatin Complexes in Brain
Increased Mll Protein Expression in Brain Is Consistent with a Gainof-Function Allele-Newborn and adult brain from eed and Mll heterozygotes revealed normal gross morphology (data not shown). Immunohistochemistry detected widespread nuclear expression of Eed and Mll in wild-type P0 and throughout adult brain (data not shown), including hippocampus (Fig. 3). For comparison of Eed and Mll expression across genotypes, immunohistochemistry was performed on sections from wild-type, eed, and Mll mutant brains mounted side by side on the same slide. Eed protein levels in eed heterozygous and eed;Mll double heterozygous hippocampi were indistinguishable from wild type (Fig.  3A). Importantly, the Eed antibody  employed in this study could not distinguish between wild-type Eed and the L290P mutant protein encoded by the eed null allele l7Rn5 3354SB (15). Nonetheless, because the L290P missense mutation renders the Eed protein nonfunctional (16,18), wild-type Eed activity decreases to ϳ50% in eed heterozygotes and eed;Mll double heterozygotes compared with wild type.
The mutant Mll allele was created by in-frame insertion of a lacZ cassette into the third exon of Mll (34). The N-terminal Mll/␤-Gal fusion fragment truncates the Mll protein and removes the PhD zinc fingers, the bromodomain, as well as the SET domain at the C terminus. Therefore, expression analysis using an antibody against C-terminal epitopes detected activity of the wild-type Mll protein in Mll heterozygotes. Strikingly, immunohistochemistry on sections from Mll heterozygotes and eed;Mll double heterozygotes revealed increased expression of the wild-type Mll protein in newborn and adult brain, including hippocampus (Fig. 3B). Similarly, an antibody against N-terminal epitopes of Mll manifested significantly higher Mll protein levels in Mll heterozygous and eed;Mll double heterozygous hippocampus compared with wild type (data not shown). This unexpected augmentation in wild-type Mll protein levels in Mll mutant brain should result in increased Mll activity, which is most consistent with a gain-of-function allele of Mll.
A Post-transcriptional Mechanism of Increased Mll Protein Expression-eed and Mll transcript levels were analyzed by quantitative real time RT-PCR (qRT-PCR) in newborn brain and adult hippocampus from mutant animals. At both stages, qRT-PCR did not detect significant changes in eed mRNA levels in mutant genotypes compared with wild-type littermates (Fig. 4), indicating stability of the l7Rn5 3354SB mRNA. In contrast, qRT-PCR with primers spanning an exon junction at the 3Ј end of the coding region revealed an ϳ50% decrease in Mll mRNA levels in Mll heterozygous and eed;Mll double heterozygous animals in both P0 brain and adult hippocampus (Fig. 4). This statistically significant decrease in Mll mRNA expression is consistent with the presence of truncated mRNA caused by insertion of a lacZ cassette with SV40 transcription termination sequences into exon 3b of Mll (33). In addition, mRNA levels of the paralogous gene Mll2, whose encoded protein shares many domains with Mll (39), were indistinguishable between wild type and Mll mutants (data not shown). Taken together, decreased wild-type Mll mRNA levels suggested a distinctly post-transcriptional mechanism for augmentation of Mll protein expression in Mll mutant brain.
Interdependence of Eed and Mll Function in the Regulation of Hippocampal Synaptic Plasticity-Synaptic plasticity represents the leading cellular candidate mechanism for memory storage in the adult CNS (40). The distinct phases of synaptic plasticity involve different mechanisms, including gene transcription during the late phase. Given PcG and trxG function in transcriptional regulation via histone modification, analysis of LTP at Schaffer-collateral synapses in area CA1 in eed and Mll mutant hippocampi investigated a specific physiological role of the Eed-Mll complex in the induction of synaptic plasticity.
Prior to measuring synaptic plasticity, control experiments in mutant hippocampi determined basal synaptic transmission. For example, input/output (I/O) relationships measured postsynaptic depolarization (fEPSP slope) as a function of presynaptic depolarization (fiber volley slope). As indicated by I/O relationships (Fig. 5A), hippocampal slices from Mll heterozygous (p Ͻ 0.2) and eed;Mll double heterozygous animals (p Ͻ 0.3) displayed normal levels of synaptic transmission. Analysis of I/O in hippocampal slices from eed heterozygotes revealed a significant reduction in synaptic transmission ( Fig. 5A; p Ͻ 0.005). To assess a potential mechanism for this defect, pairedpulse facilitation (PPF) was employed as an index of neurotransmitter release. Hippocampal slices from all genotypes exhibited normal PPF (Fig. 5B, eed, p Ͻ 0.2; Mll, p Ͻ 1.0; eed; Mll, p Ͻ 1.0). With regard to eed heterozygotes, normal PPF indicated that the defect in synaptic transmission was not because of a reduction in presynaptic neurotransmitter release. Mechanistically, heterozygosity for the eed null allele may reduce the expression and/or function of postsynaptic glutamate receptors.
Both I/O and PPF indicated that mutations in eed and Mll only minimally affected synaptic transmission. Subsequent experiments compensated for the reduced synaptic transmission in eed heterozygotes by stimulation at 50% maximal fEPSP for each particular animal. Therefore, every hippocampal slice presented the same potential for expression of synaptic plasticity, as assessed by long term potentiation with theta-burst stimulation and with high frequency tetanic stimulation. As shown in Fig. 5C, theta-burst LTP was normal in eed and Mll mutant hippocampus (all comparisons, p Ͻ 0.1), indicating that Eed and/or Mll do not affect the induction and/or expression of short term forms of synaptic plasticity.
Importantly, LTP induction with high frequency stimulation (2, 1-s, 100-Hz tetani) demonstrated a robust enhancement of early and late phase LTP in hippocampal slices from eed heterozygotes (p Ͻ 0.0001) (Fig. 6A). This LTP enhancement was apparent 40 min after induction and persisted for at least 180 min after induction. In contrast to eed, slices from Mll heterozygotes exhibited a significant deficit in LTP (p Ͻ 0.0001) wherein a gradual decrease in potentiation started at 20 min post-induction (Fig. 6B). Synaptic efficacy reached basal levels by 160 min after induction. Strikingly, LTP in hippocampal slices from eed;Mll double heterozygous animals was not significantly different from LTP in wild-type littermates (p ϭ 0.99) (Fig. 6C). The lack of a significant difference between eed;Mll double heterozygous and wild-type slices could derive from complementary effects on induction and/or expression of late phase synaptic plasticity. However, the increase in variance in eed:Mll double mutants raises the possibility that other, as yet unidentified, mechanisms may contribute to the apparent complementation in LTP.
A series of control experiments determined whether heterozygosity for Eed or Mll affected induction of LTP. The first parameter entailed depolarization during a 100-Hz tetanus, which represents an integrated metric of all synaptic mechanisms involved in the induction of LTP. As shown in Fig. 6, A-C, insets, wild-type and mutant hippocampi revealed no differences in the magnitude of depolarization during a 100-Hz . Schaffer-collateral synaptic transmission was significantly reduced in eed ϩ/Ϫ ;Mll ϩ/ϩ mice but appeared normal in eed ϩ/ϩ ;Mll ϩ/Ϫ and eed ϩ/Ϫ ;Mll ϩ/Ϫ animals relative to wild-type littermates. B, PPF, an index of presynaptic neurotransmitter release, was measured at Schaffer-collateral synapses of hippocampal slices in vitro. PPF was depicted as percentage of observed facilitation versus the interstimulus interval. PPF in eed ϩ/Ϫ ;Mll ϩ/ϩ animals was indistinguishable from wild type, suggesting that the reduction in synaptic connectivity in eed ϩ/Ϫ ;Mll ϩ/ϩ hippocampus (see Fig. 2A) was not because of a reduction in neurotransmitter release. Likewise, PPF was normal in eed ϩ/ϩ ;Mll ϩ/Ϫ and eed ϩ/Ϫ ;Mll ϩ/Ϫ animals. C, theta-burst LTP at Schaffer-collateral synapses of hippocampal slices in vitro. LTP was induced using a theta-burst stimulation paradigm (3 trains of 10 bursts, 4 stimuli/burst at 100 Hz, 200-ms interburst interval, 20-s intertrain interval). Synaptic efficacy was monitored 20 min before and 90 min after induction of LTP. Compared with wild type, theta-burst LTP appeared normal in eed ϩ/Ϫ ;Mll ϩ/ϩ , eed ϩ/ϩ ;Mll ϩ/Ϫ and eed ϩ/Ϫ ;Mll ϩ/Ϫ animals. tetanus (wild type, 977 Ϯ 162 versus eed heterozygotes, 1,008 Ϯ 169 mV ϫ ms; p ϭ 0.07; wild type, 977 Ϯ 162 versus Mll heterozygotes, 1,600 Ϯ 215 mV ϫ ms; p ϭ 0.07; wild type, 977 Ϯ 162 versus eed;Mll double heterozygotes, 1,418 Ϯ 321 mV ϫ ms; p ϭ 0.07). Such normal depolarization suggests that both eed and Mll mutants retained synaptic function with regard to the induction of LTP. Likewise, all genotypes demonstrated normal PTP, a short term form of synaptic plasticity, which occurs immediately after tetanus administration as follows: eed heterozygotes (p Ͻ 1), Mll heterozygotes (p Ͻ 0.6), and eed;Mll double heterozygotes (p Ͻ 0.6) (Fig. 5D). Activation of NMDA-R-mediated synaptic transmission constitutes the first critical step in induction of LTP. Although NMDA-Rdependent neurotransmission was normal in eed heterozygotes (p ϭ 0.11), hippocampal slices from Mll heterozygotes presented a significant deficit in NMDA-R-mediated synaptic transmission compared with wild type (p Ͻ 0.05) (Fig. 6, D and   E). NMDA-R-mediated transmission in eed;Mll double heterozygous hippocampi was indistinguishable from wild type (p ϭ 0.62) (Fig. 6F), indicating that a reduction in the dosage of wild-type Eed suppressed aberrant NMDA-R-mediated synaptic transmission in Mll heterozygotes.

DISCUSSION
An intricate interplay between biochemically distinct PcG and trxG chromatin complexes specifies segmental identity in embryos by means of transcriptional repression and activation, respectively, of Hox gene transcription (2,3,41). Best characterized in Drosophila, PcG and trxG complexes frequently co-localize to adjacent but physically separable regions in the vicinity of transcription units (42)(43)(44)(45)(46)(47). Although similar genomic elements await characterization in mammals, evolutionary conservation of antagonistic PcG and trxG function was evident from the rescue of reciprocal homeotic phenotypes in intercrosses between mutant alleles of Bmi1 and Mll (48) in mice. This study identified a novel, developmentally regulated interface between PcG and trxG function in the CNS (see model in Fig. 7). In the developing brain, Eed formed a complex with the histone methyltransferase EzH2, reminiscent of PRC2-4 (13,14), whereas Mll engaged in a physically separate complex of unknown function. In contrast, adult brain was remarkable for complex formation between Eed and Mll. The transition from the Eed-EzH2 complex to the Eed-Mll complex appeared gradual, as evident from their co-existence in P28 brain. Alternatively, the three proteins may be present in a single, transient complex at this stage. It should be emphasized that these findings do not exclude the existence of Eed-Mll and Eed-EzH2 complexes at lower abundance in the developing and adult CNS, respectively. However, the results underscore the importance of analyzing complex composition at various stages of pre-and postnatal development to detect dynamic regulation thereof.
Convergence from separate biochemical entities during CNS development to the presence of Eed and Mll in a common complex in adult brain correlated with a switch in Eed isoform expression. Previous studies implicated Eed isoforms in targeting EzH2 activity to H3-K27 or H1-K26 (13). Beyond regulation of histone methyltransferase activity, the present findings suggested a broader role for Eed isoforms in directing the recruitment of Eed complex constituents, such as Mll. Although complex formation between Eed and Mll represents a novel molecular interface between these two proteins, there is precedence for physical association of other PcG and trxG constituents. Trxl (trithorax-like), the Drosophila homolog of the GAGA factor, was detected in a complex with multiple PcG proteins in pre-blastoderm embryos (49). In addition, studies in transiently transfected 293T cells identified direct interaction of Mll with a number of transcriptional regulators, including the PcG proteins Bmi1 and HPC2 (32). In support of context-specific PcG and trxG interaction, co-immunoprecipitation did not detect Bmi1 in the Eed-Mll complex in adult brain. Thus, to the best of our knowledge, the present findings constitute the first in vivo evidence for spatio-temporal regulation of a molecular interrelationship between PcG and trxG proteins in mammals.
The developmental change in complex composition results in differential histone modifying activity. Although the Eed-EzH2 complex deacetylates histone H4 and trimethylates H3-K27 during neurodevelopment, the only known function of the Eed-Mll complex resides in deacetylation of histone H4 in adult hippocampus (see model in Fig. 7). Attempts to confirm association of HDACs with the Eed complexes were confounded by high affinity of HDAC1 and -2 to agarose beads, rendering co-immunoprecipitation inconclusive. The differential chromatin remodeling activity of the Eed complexes was evident from bulk histone preparations of eed and Mll mutant brain. At P0, eed heterozygous brain manifested decreased trimethylation of H3-K27 and hyperacetylation of histone H4, consistent with decreased EzH2 histone methyltransferase and HDAC activity, respectively. Similar defects in histone modification in eed;Mll double heterozygotes compared with eed single heterozygotes reflected the independence of Eed-EzH2 complex activity from Mll. These results are consistent with the presence of Eed and Mll in distinct biochemical entities in newborn brain. Strikingly, adult hippocampus from Mll heterozygotes was remarkable for decreased H4 acetylation, which restored to wild-type levels in eed;Mll double heterozygotes. These results strongly support interdependence of Eed and Mll function in adult hippocampus.
Given the direct interaction of Mll with an array of transcriptional repressors, such as the PcG proteins Bmi1 and HPC2, HDAC1 and -2, as well as the C-terminal binding protein (32), it is conceivable that in some chromatin complexes Mll contributes to gene repression. This might involve a decrease in Mll SET domain activity to prevent formation of a transcriptionally active chromatin environment by means of H3-K4 methylation (11,12). Indeed, despite increased Mll expression, levels of H3-K4 methylation were indistinguishable between Mll heterozygotes and wild-type littermates in both newborn and adult brain. Furthermore, H4 hypoacetylation in adult hippocampus strongly suggests increased HDAC function and, hence, elevated repressor activity of the Eed-Mll complex.
Opposite allelic effects govern differential abundance of Eed-Mll complex constituents in adult brain. l7Rn5 3354SB , a genetically and biochemically well characterized eed null allele (15,16,18), reduces the Eed wild-type dosage in eed heterozygotes by about 50%. In contrast, this study identified gain-of-function effects of the Mll allele. This was evident from an increase in wild-type Mll protein expression, despite decreased Mll mRNA levels in both newborn and adult brain from Mll mutants, implicating a post-transcriptional mechanism in augmentation of Mll protein levels. Recent studies showed that Mll undergoes proteolytic cleavage, and heterodimerization of the resulting N-and C-terminal fragments confers increased fragment stability (50,51). In addition, N-terminal fragments of human MLL exhibited significant stability following transfection into U937 cells (52). This suggests that the N-terminal Mll/␤-galactosidase fusion fragment (derived from the mutant Mll allele) may increase stability of wild-type C-terminal fragments (derived from the wild-type Mll allele), augmenting Mll protein levels in Mll heterozygous and eed;Mll double heterozygous brain.
These findings form the basis for a model that reconciles the opposite effects of the eed and Mll loss-and gain-of-function allele, respectively, with the presence of Eed and Mll in a common complex and levels of histone H4 acetylation in adult brain. In this model, Mll represents the rate-limiting component of complex assembly and HDACs transiently or permanently associated with the Eed-Mll complex. Based on this notion, a 50% decrease in Eed wild-type dosage in eed heterozy- FIGURE 7. Model of Eed complex composition and function in the developing and adult brain. Nucleosomes in the vicinity of a hypothetical target gene(s) of the two distinct Eed complexes are depicted as blue ovals. Transcription start sites are indicated by an arrow. Green pentagons and orange diamonds represent acetylation of histone H4 and trimethylation of histone H3, respectively. In the developing brain, the Eed-EzH2 complex deacetylates histone H4 and trimethylates H3-K27. In the adult brain, the histone methyltransferase EzH2 is absent from the Eed complex. Instead, in the presence of Mll, the only known function of the Eed complex involves deacetylation of histone H4. In this model, HDACs may transiently or permanently associate with the Eed complexes.
gotes would be less likely to limit complex formation and, consequently, H4 acetylation remained at wild-type levels. Conversely, up-regulation of wild-type Mll protein levels should facilitate Eed-Mll complex formation. In strong support of this hypothesis, overexpression of EzH2 in 293 cells promoted formation of PRC4 (14), indicating a reservoir of Eed isoforms available for increased complex formation upon overexpression of a rate-limiting constituent. Concomitant with increased Eed-Mll complex formation, HDAC activity would increase, as reflected by a decrease in hippocampal H4 acetylation levels in Mll heterozygotes. A significant reduction in wild-type Eed dosage in eed;Mll double heterozygotes would curtail the increase in Eed-Mll complex formation because of Mll up-regulation. As a result, H4 acetylation is restored to approximately wild-type levels in eed;Mll double heterozygotes. Thus, genetic analysis suggests a tightly regulated dosage requirement for Eed-Mll complex formation. Whereas Mll constitutes the ratelimiting factor under wild-type conditions, in turn, a rather small reservoir of Eed isoforms limits complex formation upon overexpression of Mll.
Recent studies identified potential downstream targets of PcG complexes in mammalian embryonic stem cells and embryonic fibroblasts (53)(54)(55). Although specific target loci in brain remain unknown, it is intriguing to speculate that the Eed-EzH2 complex regulates genes concerned with neurodevelopment, whereas the Eed-Mll complex governs gene regulation in support of specific functions of the adult CNS. Electrophysiological studies provided a well established experimental avenue toward elucidating the biological significance of Eed-Mll complex formation in mature brain. Indeed, interdependency of Eed and Mll function was evident from LTP induction at Schaffer-collateral synapses in hippocampal area CA1, a candidate mechanism for neuronal memory formation in vivo (40). Therein, eed and Mll heterozygotes exhibited opposite phenotypes in early and late phase LTP. Strikingly, eed;Mll double heterozygotes exhibited normal LTP, indicative of an intricate interplay between Eed and Mll in the regulation of synaptic plasticity.
Similar to other mutant strains, the question arises as to whether the aberrant synaptic plasticity in eed and Mll mutant mice derives from neurodevelopmental defects or reflects an acute requirement for Eed and Mll in neuronal function. In support of the latter, recent studies implicated changes in histone acetylation and phosphorylation in the induction of synaptic plasticity and long term memory in brain (37, 56 -59). Although not irreversible, methylation is considered a more permanent histone modification compared with acetylation (60). Hence, absence of the EzH2 histone methyltransferase from the predominant Eed complex could reflect a response to stage-specific functional requirements. Whereas the Eed-EzH2 complex would stably maintain gene repression during neurodevelopment, regulation of histone deacetylation by the Eed-Mll complex may provide the necessary molecular flexibility for the execution of higher order functions of the adult CNS, including synaptic plasticity. In conclusion, dynamic regulation of Eed complex composition may constitute a novel molecular mechanism to co-opt components of developmental chroma-tin-remodeling pathways into acute regulation of neuronal memory formation in the adult mammalian brain.