A potential histone-chaperone activity for the MIER1 histone deacetylase complex

Abstract Histone deacetylases 1 and 2 (HDAC1/2) serve as the catalytic subunit of six distinct families of nuclear complexes. These complexes repress gene transcription through removing acetyl groups from lysine residues in histone tails. In addition to the deacetylase subunit, these complexes typically contain transcription factor and/or chromatin binding activities. The MIER:HDAC complex has hitherto been poorly characterized. Here, we show that MIER1 unexpectedly co-purifies with an H2A:H2B histone dimer. We show that MIER1 is also able to bind a complete histone octamer. Intriguingly, we found that a larger MIER1:HDAC1:BAHD1:C1QBP complex additionally co-purifies with an intact nucleosome on which H3K27 is either di- or tri-methylated. Together this suggests that the MIER1 complex acts downstream of PRC2 to expand regions of repressed chromatin and could potentially deposit histone octamer onto nucleosome-depleted regions of DNA.


INTRODUCTION
The histone deacetylase enzymes, HDACs 1 and 2 act as epigenetic 'erasers' removing acetyl gr oups fr om histone tails ( 1 ). This results in the loss of binding sites for regulatory 'reader' proteins and causes chromatin compaction by restoring the positi v e charge on histone tails. HDACs 1 and 2 are highly homologous enzymes that are assembled into six distinct multiprotein complexes: NuRD, CoREST, Sin3A / B , MiDAC , RERE and MIER (2)(3)(4)(5)(6)(7). There appears to be little or no redundancy between these complexes since mice lacking unique components typically die during embryo genesis ( 5 , 8-10 ). A part from the HDAC enzyme, the complexes differ in subunit composition, but often contain chromatin and / or transcription factor interaction activities. The MIER complex has to date recei v ed relati v ely little at-tention, but is likely to have an important regulatory function since it is present from nematodes to man.
The MIER complex is named after the MIER cor epr essor scaffold proteins that directly recruit HDAC1 / 2 through a canonical ELM2-SANT domain ( 11 ). There are three paralogous genes mier1, 2 and 3. MIER1, also known as Mesoderm Induction Early Response protein 1, was first identified as a fibroblast growth factor regulated gene in Xenopus laevis ( 4 , 12 ). The central ELM2-SANT domain of MIER1, 2 and 3 is highly conserved; the amino-terminal region is moderately conserved (Supplementary Figure S1a) and the carboxy-terminal region is poorly conserved. The MIER proteins ar e pr edominantly located in the nucleus ( 12 ).
Se v eral studies hav e shown that the BAH domaincontaining protein, B AHD1, inter acts with the MIER1:HDAC1 complex ( 13 , 14 ). BAH domains are found in many chroma tin-associa ted proteins including se v eral proteins involv ed in transcriptional r epr ession such as SIR3, MTA1, RERE, BAHCC1 and BAHD1 ( 15 , 16 ). BAH domains have been shown to mediate interaction both with intact nucleosomes and with specific histone marks such as H4K20me2 and H3K27me3 (17)(18)(19)(20). Both B AHCC1 and B AHD1 recognise H3K27me3 suggesting that they are 'reader' domains acting downstream of the PRC2 complex ( 10 , 21 ). It is unclear, howe v er, whether these BAH domains simpl y reco gnise the H3K27me3 histone mark, or whether they can bind to intact nucleosomes in an analogous fashion to the BAH-containing proteins ORC1 and SIR3 ( 17 , 20 ).
Her e, we r eport a biochemical and functional study of the MIER1:HDAC1 and MIER1:HDAC1:BAHD1 complexes. We find that the MIER1 amino-terminal domain mediates interaction with histones H2A and H2B and is able to bind to an intact histone octamer. The MIER1:HDAC1:BAHD1 complex co-purifies with the multi-functional protein C1QBP and with an intact nucleosome specifically bearing H3K27me2 / 3 marks. The finding that a histone deacetylase complex binds histone octamer and may ther efor e potentially act to recruit or deposit histones is unexpected, but makes sense in terms of increasing nucleosome density as part of a transcriptional r epr ession activity.

Mammalian protein expression and purification
MIER1 constructs were cloned into pcDNA3.0 expression vectors containing an amino-terminal 10xHis-3xFLAG tag followed by a TEV protease cleavage site. Full length HDAC1(aa:1-482) and BAHD1(aa:525-780) were cloned without affinity tags into the same vectors. Constructs were co-transfected in HEK293F suspension grown cells (Invitro gen) using pol yethylenimine (PEI; Sigma) as a transfection reagent and harvested after 48 h. Cells were lysed in buffer containing 50 mM Tris / HCl pH 7.5, 50 mM potassium acetate, 5% v / v glycerol, 0.4% v / v Triton X-100 and Complete EDTA-free protease inhibitor (Roche). The lysate was clarified by centrifugation and applied to Anti-FLAG M2 affinity resin (Sigma Aldrich) for 30 min and washed three times with 50 mM Tris / HCl pH 7.5, 50 mM potassium acetate and 5% v / v glycerol and three times more with 50 mM Tris / HCl pH 7.5, 50 mM potassium acetate and 0.5 mM TCEP before being cleaved by TEV protease overnight. The protein complex es wer e purified further by gel filtration using a Super de x S200 column (Cytiva) with buffer containing 25 mM Tris / HCl pH 7.5, 50 mM potassium acetate and 0.5 mM TCEP.

E. coli protein expression and purification
Truncated MIER1 was cloned into pGEX e xpression v ectors containing an amino-terminal GST-tag followed by a TEV protease cleavage site and the recombinant protein was ov er e xpressed in Esc heric hia coli strain Rosetta (DE3). Bacterial cultur es wer e grown at 37 • C to an optical density of ∼0.6, the temperature of the culture was reduced to 20 • C and protein expression was induced by addition of isoprop yl-␤-D -thiogalactop yranoside (IPTG) to a final concentration of 40 M. The cells were harvested after the cultur es wer e grown for a further 16 h. The cells were lysed in lysis buffer containing phosphate buffered saline (PBS), 1% v / v Triton X-100, 1 mM Dithiothreitol (DTT) and Complete EDTA-free protease inhibitor (Roche). The soluble fraction was then incubated with Glutathionine Sepharose (Cytiva) resin for 1 hour at 4 • C with gentle agitation. The protein-bound resin was then washed three times with wash buffer (1 × PBS, 1% v / v Triton X-100, 1 mM DTT) and then three times more with cleavage buffer (1 × PBS, 1 mM DTT). The GST-tag was removed by incubation with TEV protease overnight at 4 • C. The TEV-cleaved protein was concentrated to around 500 l using a 4 ml Amicon ® Ultra centrifugal filter concentrator (Merck Millipore) with a membrane molecular weight cut-off of 10 kDa. The concentrated protein was filtered through a 0.22 m centrifugal filter (Merck Millipore) prior to loading onto the gel filtration column. The protein sample was purified on a Super de x S200 (10 / 300) GL column (Cytiva) with buffer containing 25 mM Tris / Cl pH 7.5, 50 mM potassium acetate and 0.5 mM TCEP. 500 l fractions were collected and 10 l of the fractions was taken for analysis on NuPAGE 4-12% Bis-Tris gels (Invitrogen). Fractions containing the purified protein were concentrated using a 0.5 ml Amicon Ultra centrifugal filter (Merck Millipore) with a membrane molecular weight cut-off of 10 kDa.

Histone purification
Human histones cloned into pET3a (plasmids were a gift from Martin Browne and Andrew Flaus NUI Galway) were expressed separately in Rosetta2 (DE3) pLysS (Novagen). Cell pellets wer e r esuspended in 30 ml histone wash buffer which contained 50 mM Tris / Cl pH 7.5, 100 mM NaCl, 5 mM DTT and Complete EDTA-free protease inhibitor (Roche). After sonication, the insoluble fractions were pelleted by spinning a t 4 • C a t 30 000 g for 15 min and washed twice with wash buffer containing 50 mM Tris / Cl pH 7.5, 100 mM NaCl, 5 mM DTT, 1% Triton X-100 and Complete EDTA-free protease inhibitor (Roche) and three times with wash buffer without Triton X-100. To resuspend the inclusion bodies 0.5 ml of DMSO was then added and incubated at room temperature for 30 min on a roller then 5 ml unfolding buffer containing 7 M guanidine HCl, 20 mM Tris / Cl pH 7.5 and 10 mM DTT was added, and the incubation continued for an additional 1 h.
After the pellet was fully dissolved, 70 ml 7 M urea, 50 mM Tris / Cl pH 7.5 and 1 mM EDTA was added and the sample was centrifuged at 35 000 × g for 20 min. The histone solution was then filtered through a 0.22 m filter prior to loading onto a HiTrap Q HP. The flow through was then loaded onto a HiTrap SP HP. The histones were eluted with a gradient to 2 M NaCl. Fractions containing histone were then dialysed with three changes into buffer containing 0.1% acetic acid and 5 mM DTT. They were then freeze dried in aliquots.

Histone octamer r ef olding and nucleosome reconstitution
Equimolar amounts of H2A, H2B, H3 and H4 were dissolved in unfolding buffer which contained 7 M guanidine, 20 mM Tris pH 7.5 and 10 mM DTT and mixed with a small excess of H2A and H2B. The histone mix was dialysed against high salt buffer (20 mM Tris pH 7.5, 2.0 M NaCl, 1 mM EDTA and 5 mM mercaptoethanol) and then purified by size exclusion chromato gra phy using a Super de x S200 column.

Size e x clusion chromatography with multi-angle light scattering (SEC-MALS)
Gel filtrated pure MIER1(aa:171-350):HDAC1 complex was reapplied to a Super de x S200 column and monitored with an Optilab T-rEX differential Refracti v e inde x detector connected to a DAWN HELEOS MALS detector (Wyatt Technology). The mass of the complex was calculated using ASTRA software version 6.1.

Boc-lys(ac)-AMC HDAC assay
A fluorescent based HDAC assay was used to determine HDAC activity of the MIER1 complex with BOC-lys(Ac)-AMC (BaChem) as the substrate. For these experiments, 20 nM MIER1(aa:171-350):HDAC1 complex (alone or with 100 M Ins(1,2,3,4,5,6)P 6 or 30 M SAHA or 30 M MS275) in 25 mM Tris / Cl, pH 7.5 and 50 mM NaCl were incubated in a shaking incubator at 37 • C for 30 min before adding 100 M substrate for a further 30 min. The assay was de v eloped by the addition of de v eloper solution (2 mM TSA, 10 mg / ml Trypsin, 50 mM Tris pH 7.5, 100 mM NaCl). Fluorescence was measured at 335 / 460 nm using a Victor X5 plate reader (Perkin Elmer). Data was analysed using GraphPad Prism 9.0.

Isothermal titration calorimetry (ITC)
The ITC experiment was performed using a VP-ITC instrument (MicroCal). H2A / H2B dimer (4.6 M) was loaded in the sample cell and 170 M of MIER1(aa:1-177) in the syringe. The titration experiment was performed at 25 • C and consisted of 30 injections of 5 l each with a 5 min interval between additions. The raw data were integrated, corrected for nonspecific heat, normalized for the molar concentration and analysed according to a 1:1 binding model.

NMR measurement
For NMR sample preparation, MIER1(aa:1-177) was uniformly labelled by ov ere xpression in M9 minimum medium, containing 15 NH 4 Cl as the sole nitrogen source. The expression and purification of labelled MIER1 was performed as described above. MIER1(aa:1-177) (200 M) with 10% D 2 O in a final volume of 330 l of PBS buffer was placed in 3 mm NMR tubes (Norell). ( 15 N-1 H)-HSQC spectra wer e measur ed on a Bruker AVIII-600 MHz spectrometer equipped with a cryoprobe. Data were processed using Topspin (Bruker) and analysed with Sparky. H2A:H2B dimer was added to the labelled MIER1(aa:1-177) protein at the ratio of 1:1 and incubated for 30 min at 4 • C. The MIER1(aa:1-177):H2A:H2B complex was then purified on a Super de x S200 column before collecting ( 15 N-1 H) HSQC NMR data.

Western blot
Western blotting was carried out using NuPAGE 4-12% Bis-Tris gels (Invitrogen) and a semi-dry transfer system with nitrocellulose membr anes. Membr anes were blocked for 1 h at room temperature using 5% skimmed milk TBS / T and incubated with the primary mouse antibody C1QBP (1:1000 H-9:sc-271200, Santa Cruz) diluted in blocking buffer overnight at 4 • C. Secondary antibody goat antimouse 680RD (1:10 000, LI-COR), was incubated with membranes for 1 h at room temperature and detected using an Odyssey CLx digital imaging system (LI-COR).

GST-pull down experiment
GST-MIER1(aa:1-177) and GST tag alone were purified separately without eluting the proteins from the glutathione sepharose resin (Cytiva) in PBS, 1 mM TCEP and Complete EDTA-free protease inhibitor (Roche). Excess octamer and nucleosome wer e mix ed either with the GST-tagged MIER1 or with the GST-tag alone and incubated for 30 min at 4 • C. After washing 5 times with PBS, 0.5 mM TCEP, the samples were analysed on an 18% SDS-PAGE gel. complex. This was incubated for 30 min at 37 • C and EDTA to 10 mM was added to stop the reaction. 10 l of proteinase K and 1% SDS were added and incubated for 30 min at 37 • C. This was then adjusted to 1 M NaCl and phenol:chloroform:isoamyl alcohol extracted. The DNA was ethanol precipitated, analysed on a 1 × TBE, 1% agarose gel and visualised with ethidium bromide.
(ADMA: asymmetric dimethyl arginine; Ab u: aminob utyric acid; ed: ethylene diamine linker) The assays were performed in black 384-well assay plates with a non-binding surface (Corning Life Sciences). Multiple titrations were performed using a fixed concentration of 5 nM histone peptide with increasing concentrations of E. coli expressed BAHD1(aa:588-780) (0-142 M) in a final volume of 25 l in 25 mM Tris / Cl, pH 7.5 and 50 mM NaCl. The plate was incubated in the shaking incubator for 15 min a t 37 • C , then centrifuged at 61 g for 10 s and the data wer e acquir ed in a Victor X 5 plate r eader (Perkin Elmer) with an excitation wavelength of 480 nm and an emission wavelength of 535 nm. The subsequent data were analysed with GraphPad Prism 9.0. K d values were calculated by nonlinear curve fitting with a one-site binding model.

Mass spectrometry analysis
Propionylation of histones and in-gel digestion. Gel bands were derivatised according to ( 23 , 24 ) with some modifica-Nucleic Acids Research, 2023, Vol. 51, No. 12 6009 tions. Briefly, 50 l of 50 mM ammonium bicarbonate was added to the gel pieces and incubated with 16.6 l of propionylation reagent (1:3 v / v propionic anhydride in acetonitrile) for 15 min at 37 • C with agitation in a thermomixer at 900 rpm (Eppendorf, UK). The supernatant was then removed and the deriva tisa tion process was repeated. After discarding the supernatant, the gel was de-stained, and proteins wer e r educed with 10 mM dithiothr eitol for 1 h at 55 • C and then alkylated with 55 mM iodoacetamide for 20 min at room temperature in the dark. Protein in the gel was digested with 100 l of trypsin (0.02 g / l) at 37 • C overnight. Peptides were then extracted twice with 100 l of 3.5% formic acid in 30% ACN then followed by 5% formic acid in 50% ACN. Eluted peptides were dried in a vacuum concentrator and stored at −20 • C until MS analysis.
MS and data analyses. Peptides were resuspended in 0.5% formic acid and analysed using an Orbitrap Elite (Thermo Fisher) hybrid mass spectrometer equipped with a nanospray source, coupled to an Ultimate RSLCnano LC System (Dionex). Peptides were desalted online using a nano trap column, 75 m I.D. × 20 mm (Thermo Fisher) and then separated using a 120-min gradient from 5 to 35% buffer B (0.5% formic acid in 80% acetonitrile) on an EASY-Spray column, 50 cm × 50 m ID, PepMap C18, 2 m particles, 100 Å pore size (Thermo Fisher). The Orbitrap Elite was operated with a cycle of one MS (in the Orbitrap) acquired at a resolution of 120 000 at m / z 400, with the top 20 most abundant m ultipl y charged (2 + and higher) ions in a gi v en chromato gra phic window subjected to MS / MS fragmentation in the linear ion trap. An FTMS target value of 1e6 and an ion trap MSn target value of 1e4 were used with the lock mass (445.120025) enabled. Maximum FTMS scan accumulation time of 200 ms and maximum ion trap MSn scan accumulation time of 50 ms were used. Dynamic e xclusion was enab led with a repea t dura tion of 45 s with an exclusion list of 500 and an exclusion duration of 30 s. Raw mass spectrometry data were analysed with MaxQuant version 1.6.10.43 ( 25 ). Data wer e sear ched against a human UniProt r efer ence proteome (downloaded June 2022) using the following search parameters for standard protein identification: enzyme set to Trypsin / P (2 mis-cleavages), methionine oxidation and N-terminal protein acetylation as variable modifications, cysteine carbamidomethylation as a fixed modification. For histone PTM analysis the following settings were used: enzyme set to Trypsin / P with up to 5 missed cleavages, methionine oxidation, lysine propionyla tion, lysine acetyla tion, lysine mono, di and trimethylation were set as variable modifications, cysteine carbamidomethylation as a fixed modification. A protein FDR of 0.01 and a peptide FDR of 0.01 were used for identification le v el cut-offs based on a decoy database searching stra tegy. MaxQuant calcula ted intensities of peptides containing PTMs (lysine acetylation, lysine mono, di and trimethylation) were normalised (divided) to unmodified versions of corresponding peptides. Biological triplicates were perf ormed f or each group, mean and standar d de viation (SD) were then determined using GraphPad Prism 9.0. Raw MS data and searched results from MaxQuant are available from PRIDE ( https://www.ebi.ac.uk/pride/ ) with identifier PXD040685.

MIER1 binds to HDAC1 / 2 through the ELM2-SANT domain
We hav e pre viously shown that the ELM2-SANT domains from MTA1, MIDEAS and RCoR1 mediate interaction of the NuRD, MiDAC and CoREST complexes with HDACs 1 and 2 ( 5-7 ). To confirm that the homologous ELM2-SANT domain from MIER1 is also capable of binding HDAC1 / 2 we co-expressed FLAG-tagged MIER1(aa:171-350) with full-length HDAC1 in HEK293F cells (Figure 1 ). Af finity purifica tion on anti-FLAG resin, followed by size exclusion chromato gra phy confirmed the expected interaction (Figure 1 B). The elution profile is consistent with a monomeric 1:1 complex. This was confirmed using size exclusion chromato gra phy coupled with multi-angle light scattering (SEC-MALS). The complex molecular weight was determined to be 73.6 ± 1.5 kDa (Figure 1 C), which is consistent with a monomeric complex (HDAC1 55 kDa + MIER1(aa:171-350) 21 kDa) and contrasts with the dimeric MTA1 and tetrameric MIDEAS complexes ( 5 , 7 ). HDAC assays confirmed that the complex is active; can be further activated by inositolhexakis-phosphate and is inhibited by hydroxamic and benzamide-based inhibitors (Figure 1 D). We used Al-phafold2 multimer ( 26 ) to predict the structure of the MIER1-ELM2-SANT domain bound to HDAC1 (Figure 1 E). The region of the ELM2 domain that media tes dimeriza tion in MTA1 is clearly unable to support dimerization in MIER1. Helix 2 is absent in MIER1 and the non-polar residues loca ted a t the MTA1 dimer interface are mostly replaced with basic residues in MIER1 (Figure 1 F).

The amino terminal region of MIER1 co-purifies with endogenous H2A:H2B histone dimers
To characterise the nati v e comple x, we co-e xpressed FLAGtagged full-length MIER1 in HEK293F mammalian cells, together with full-length HDAC1. Following purification, we observed the expected 1:1 interaction with HDAC1 (Figure 2 A). Unexpectedly however, in addition to the bands for MIER1 and HDAC1, we observed two low molecular-weight proteins that co-purify with the complex through both the affinity and size exclusion purifications. Mass-spectrometry (MS) analysis of these bands identified these proteins to be endogenous histones H2A and H2B (Figure 2 A).
This unexpected interaction with histones H2A and H2B was not observed with the isolated ELM2-SANT domain (Figure 1 B) suggesting that either the aminoterminal or carboxy-terminal regions mediate this interaction. To identify the interaction region, we expressed FLAG-tagged MIER1(aa:1-177) and FLAGtagged MIER1(aa:346-512). The H2A:H2B dimer copurified with the amino-terminal construct and not with the carboxy-ter minal construct. Further more, the coomassie staining is consistent with a stoichiometric 1:1:1 complex (Figure 2 B).
MIER1 has two paralogues MIER2 and MIER3. Whilst the ELM2-SANT domain is highly conserved (63% identity), the amino-terminal region is less well conserved ( Figure 2 E -a cross-species alignment is shown in Supplementary Figure S1a). To determine whether histone binding is a conserved feature of the MIER proteins, we expressed both full-length and the amino-terminal domain of MIER3 in HEK293F cells. Both constructs of MIER3 co-purified with histones H2A:H2B (Figure 2 C) suggesting that histone binding is a common and important property of this family of HDAC co-r epr essor proteins. Inter estingly, in se v eral e xperiments, we found that histones H3 and H4 also co-purify with the MIER proteins (Figure 2 D). Since this interaction does not survi v e gel filtration, we belie v e that this is a relati v ely weak interaction ( in vitro ) mediated by interaction between H2A:H2B heterodimer with H3:H4. In contrast, we infer from the fact that the H2A:H2B heterodimer co-purifies with MIER1 over both affinity and size-exclusion columns that the interaction must be relati v ely strong, or at least the off-rate is slow. Using isothermal calorimetry we measured a dissociation constant of between 227 and 383 nM (Figure 3 A). Further confirmation of interaction was obtained using NMR spectroscopy. 15 N-labelled MIER1(aa:1-177) was expressed in E. coli and was mixed in a 1:1 ratio with unlabelled H2A:H2B dimer (expressed separately). The limited 1 H chemical shift distribution of MIER1 in an HSQC spectrum is typical of an intrinsically disordered protein (Figure 3 B). Following addition of the H2A:H2B dimer, the HSQC spectrum clearly shows both chemical shift perturbations and a general broadening consistent with MIER1 binding the folded histone dimer (Figure 3 C). Approximately half the amino acids in MIER1(aa:1-177) appear to be involved in the interaction.

Histone binding activity maps to two regions of conservation in the amino-termini of MIER proteins
To investigate which regions of the amino-terminus of MIER1 mediated the interaction with H2A:H2B, we hypothesised that regions conserved between MIER1, MIER2 and MIER3 would be most likely to mediate the interaction. Sequence alignment of the three isoforms re v eals three regions with modera te conserva tion (Figure 2 E). Deletion of either region 1 or 2, but not 3, resulted in failure of MIER1 to co-purify with the H2A:H2B heterodimer (Figure 3 D-G, Supplementary Figure S1b). Since regions 1 and Nucleic Acids Research, 2023, Vol. 51, No. 12 6013 2 contain many negati v ely charged residues, it is important to rule out a non-specific interaction with the positi v ely charged histone dimer. Ther efor e, to confirm that this interaction is specific we made MIER1 amino-terminal constructs in which the order of amino acids in regions 1 or 2 was scrambled. In both cases, MIER1 no longer co-purifies with the H2A:H2B dimer confirming a stereo-specific interaction (Figure 3 H, I).
To obtain a structural model of the interaction of MIER1(aa:17-75) regions 1 and 2 with H2A:H2B, we used Alphafold2 multimer to predict the mode of binding (Figure 4 A-C, Supplementary Figure S2). The resulting model is structurally convincing and re v eals how the MIER1 sequence has complementary stereochemistry to the H2A:H2B dimer surface. Overall the interface area between MIER1 and the histone dimer is 2005 Å 2 ( 27 ). Region 1 is predicted with high confidence and adopts an extended configuration interacting with H2B through a conserved phenylalanine interacting in a non-polar pocket with residues Y43, I55 and M60 in H2B. Region 2 is also predicted with high confidence, to form a helix making numerous non-polar interactions with the other end of H2B. Between regions 1 and 2 there are three additional helices containing many negati v ely charged residues that interact with the positi v ely charged surface of the H2A:H2B dimer that interacts with DNA in the assembled nucleosome. Importantl y, Alphafold2 m ultimer predictions in w hich region 3 was included, suggested that region 3 plays no role in the interaction with the H2A:H2B histone dimer. To support the Alphafold model we designed mutations in regions 1 and 2 ( Figure 4 c and Supplementary Figure S3). As expected, mutations S71A / L72A did not disrupt the complex since they are oriented away from the histone dimer. Mutations I66A / L69A / L70A in region 2 essentially abolished interaction as expected since they make multiple interactions with the histone dimer. Mutations F22A / M28A / L29A in region 1 were not sufficient to disrupt the complex. This is surprising gi v en that scramb ling or deletion of region 1 did abolish interaction. Potentially the non-polar alanine mutations allow some residual interaction of this region and the other interactions are sufficient to maintain interaction.

The MIER proteins can bind to H2A:H2B in the context of a histone octamer, but not when assembled into a nucleosome
As mentioned above, we noticed that the MIER proteins sometimes co-purify not only with H2A and H2B, but also with histones H3 and H4 (Figure 2 D). This suggests that the MIER proteins can bind to H2A:H2B in the context of a full histone octamer. Gi v en that H3:H4 are always lost during gel filtration, we suggest that this is a relati v ely weak, indirect interaction mediated by H2A:H2B binding. Interestingly, the Alphafold2 model of MIER1 regions 1 and 2 (aa:17-75) bound to the H2A:H2B dimer appears to partially occlude the surface of H2B that interacts with histone H4 in the H3:H4 tetramer (Figure 4 B and F).
To explore this further, we expressed MIER1(aa:1-177) with a GST-tag in bacterial cells. This protein is significantly proteolyzed during purification from bacteria, consistent with being an intrinsically disordered protein. However, the degraded mixture is clearly able to pull-down a histone oc-tamer, albeit with some dissociation of H3:H4, likely due to reduced stability of the octamer in low-salt buffer ( Figure  4 D). In contrast, no histone octamer binding was observed for the GST-only control (Figure 4 D).
Gi v en that we e xperimentally observ ed that the MIER1 amino-terminal region is able to bind to a histone octamer, we generated a further Alphafold2 multimer model including MIER1(aa:17-75) together with eight histone proteins. In this model MIER1 binds to the H2A:H2B dimer in the octamer conte xt, e xactly as the isolated H2A:H2B dimer, with the exception that the helix corresponding to the conserved region 2 is displaced and now interacts with histone H4. Figure 4 F shows the Alphafold2 model in the context of a full histone octamer compared with the DN A-wra pped nucleosome (Figure 4 G). Importantly, neither GST-MIER1(aa:1-177), nor the GST-only control were able to interact with an intact DN A-wra pped nucleosome (Figure 4 E).

The BAH domain from BAHD1 interacts with the MIER1:HDAC1 complex and recruits endogenous C1QBP
The Bromo-Adjacent-Homology-Domain containing protein 1 (BAHD1) has been identified as being a core component of the MIER1:HDAC1 complex ( 13 , 14 ). BAHD1 consists of a carboxy-terminal BAH domain with an intrinsically disordered amino-terminal region containing a proline-rich region (Figure 5 A). We co-expressed the BAH domain from BAHD1(aa:525-780) together with the MIER1:HDAC1 complex confirming that the BAH domain is sufficient for interaction with the MIER1:HDAC1 complex (Figure 5 B).
Interestingly in these purifications we consistently observed an additional band with a molecular weight of around 28 kDa that co-purifies with a fraction of the complex. MS analysis of this gel band identified it as the protein C1QBP (aka: ASF / SF2-associated protein p32) (Supplementary Table S1). This 282 aa protein is known to undergo a ma tura tion process tha t removes a mitochondrial targeting sequence (aa:1-73) from the amino terminus leaving a mature protein of around 23 kDa. To confirm the identity of this protein we performed a western blot with an antibody against the full-length protein. Figure 5 C clearly shows that C1QBP is present in the lysate (lane 1) and strongly enriched in the purified complex (lane 2). To confirm the specificity of the antibody, we expressed a full-length C1QBP protein with an amino-terminal HA-tag that blocks ma tura tion of the protein. This has an expected molecular weight of 33 kDa (Figure 5 C, lanes 3 and 4). From this, it seems that the co-purified version of C1QBP is the mature form, consistent with the MS data (Supplementary Table S1) which did not identify any peptides from the amino-terminal 73 amino acids. fractions contain only H2A:H2B. We initially hypothesised that a full histone octamer was binding to the amino-terminal region of MIER1 and that the presence of BAHD1:C1QBP might partly stabilise histone octamer binding so that it survi v es size exclusion chromato gra phy.
Howe v er, the histones appear to be supra-stoichiometric suggesti v e that the complex may have pulled down polynucleosomes, which would not be the expected behaviour for the amino-terminal region of MIER1.
To explore this further we co-expressed MIER1(aa:171-512) lacking the amino-terminal region with HDAC1 and BAHD1. In this experiment the MIER1 still co-purified with all four histones (Figure 5 D). This supports the concept that the BAH domain from BAHD1 (or perhaps the co-purified C1QBP) is able to recruit nucleosomes. To confirm this, we asked whether or not the histones that we copurified in this complex are assembled with nucleic acid. We treated the peak fraction with micrococcal nuclease and analysed the released DNA on an agarose gel. We observed a smear between 100 and 250 bp which is consistent with nucleosomal DNA (Figure 5 E).
It has previously been shown that the BAH domain from BAHD1 recognises H3 histone tails bearing a K27me3 modification ( 10 ). We confirmed this interaction using a fluor escence anisotrop y assay (Figure 5 F). We also used MS analysis to quantify the post-transla tional modifica tions on the histones tha t copurified with the MIER1(aa:171-512):HDAC1:BAHD1-BAH:C1QBP complex (Supplementary Table S2). Importantly we found that the vast majority of histone H3 that co-purifies with the complex carries either di-and trimethylation on H3K27, consistent with specific recognition by the BAH domain ( Figure 5 G). Interestingly no acetylated lysine residues were detectable, consistent with efficient deacetylation by HDAC1 in the complex.
The finding that the MIER1 amino-terminal region has H2A:H2B / histone octamer binding activity suggests that the complex may potentially play a role as a histone chaperone. This, in turn, suggests that the complex may be involved in either depositing or removing histones from chromatin. The Alphafold2 model of the MIER1 aminoterminal region bound to an H2A:H2B dimer makes excellent stereochemical sense, giving confidence that it is likely to be accurate. Importantly, the model also has numerous features in common with other peptide chaperones that bind H2A:H2B. Like ANP32E, Swr1, YL1, Chz1 and APLF, MIER1 has numerous acidic residues that interact with the DNA-binding surface of H2A:H2B (28)(29)(30)(31)(32). The MIER1 region (aa:17-75) overlaps closely with the other chaperones (except Chz1) which all have an aromatic residue (Y / F / W) that interacts with a non-polar pocket in the histone dimer (Supplementary Figure S4). MIER1(aa:17-75) follows a trajectory round the histone dimer similar to YL1 and Chz1 although the details of the interactions ar e differ ent ( 33 ). Inter estingly the model of the interaction of MIER1 with H2A:H2B has the largest solv ent-e xcluded surface of all these chaperone complexes, consistent with the finding that the complex co-purifies over se v eral columns.
The experimental finding that MIER1 can bind to a complete histone octamer fits with the Alphafold2 model, but r equir es a r eorientation of the r egion 2 alpha helix to avoid occluding the surface of the H2B that interacts with histone H4. This reorientation was seen in the Alphafold2 model of MIER1 bound to a complete octamer. Experimental structural biology would be r equir ed to confirm these models. The ability of a peptide to act as a chaperone for an intact histone octamer has only been observed once before, for the APLF peptide ( 32 , 34 ). The MIER1:octamer complex would appear to allow initial binding of DNA to the H3:H4 dyad position. We presume the DNA could then wrap around the octamer in both directions displacing the MIER1 complex.
The histone-binding activity of MIER1 is unexpected and it appears that, of the six known HDAC1 / 2 comple xes, only the MIER comple x has this acti vity. Importantly, this activity is common to all three MIER proteins. Indeed, MIER homologues in D. melanogaster (Uniprot A1Z6Z7) and C. elegans (Uniprot P91437) show 47% and 37% identity compared with MIER1, respecti v ely, in the amino-terminal region that we have shown interacts with the H2A:H2B histone dimer (Supplementary Figure S5). This suggests that histone-binding activity is conserved. Interestingly, MIER1 knockout mice are viable, albeit with metabolic dysfunction ( 13 ). It may of course be that MIER2 and MIER3 are able to compensate for the loss of MIER1 in these animals and that mice lacking all three genes would not be viable.
Genome-targeted HDAC complexes typically contain domains or sequence motifs, or accessory proteins that mediate interaction with chromatin ( 35 ). The MIER complex has been reported to associate with the chromodomain protein CDYL and the BAH domain protein BAHD1 ( 10 , 13 , 36 ). When we co-expressed MIER1 with HDAC1 and the BAH domain from BAHD1 we found that the endogenous protein C1QBP co-purified with the complex in a pparentl y stoichiometric amounts. This association of C1QBP with the MIER1 complex is unexpected and raises questions as to its role. C1QBP is an enigmatic, multi-functional protein present in multiple cellular compartments ( 37 ). It has an amino-terminal 73 residue mitochondrial-targeting sequence that is cleaved during ma tura tion ( 38 ). The carboxy-terminal 208 residues form a trimeric ring, doughnut-like structure ( 39 ). It has been reported to have roles in the regulation of apoptosis, pre-mRNA-splicing, mitochondrial protein synthesis and to act as a receptor for C1q at the cell membrane ( 40 , 41 ). Perhaps more relevant for the MIER complex, C1QBP has also been reported to have a role in transcriptional regulation and homologous recombination ( 42 ). A recent report suggests that, intriguingly, C1QBP can act as a chaperone of histones H3 and H4 ( 43 ). This is supported by proteomic studies which identify that C1QBP interacts with all four Figure 6. Proposed collaboration between the MIER1 and PRC2 complexes. PRC2 adds di-and tri-methyl marks to H3K27 to induce transcriptional r epr ession. The MIER1 complex is r ecruited to chr omatin thr ough interaction with the H3K27me2 / 3 marks recognized by the BAH domain in BAHD1. This results in histone deacetylation and deposition of histone octamer feeding forward to enable the PRC2 complex to add further H3K27me2 / 3 marks. cor e histones and, inter estingly, four proteins in the PRC2 r epr ession complex ( 44 , 45 ).
Both this study and previous studies by others, suggest that the MIER1 complex is recruited to chromatin via in-teraction of the BAHD1-BAH domain with H3K27me3 marks ( 10 ). These marks are deposited by the PRC2 repressor complex leading to heter ochr oma tin forma tion and gene r epr ession. Inter estingly, MIER1 has also been r eported to form a complex with CDYL ( 13 , 36 ), another reader of H3K27me2 / 3 marks. The finding that a deacetylase complex is recruited to this mark, suggests a feedforward mechanism through which HDAC1 / 2 removes proximal H3K27ac marks, and perhaps other Kac marks such as H3K9ac ( 46 ), shutting down transcription and facilita ting further methyla tion by PRC2. Since acetyl-lysine marks are associated with promoters and enhancers that typically have reduced nucleosome occupancy, it is tempting to speculate that the histone-binding activity of the MIER complex allows it to act as a chaperone to deposit histone octamer at these sites thereby increasing nucleosome density and contributing to the formation of heter ochr omatin ( Figure 6 ).
The intriguing observation of the juxtaposition of a deacetylase enzyme together with a potential histone octamer chaperone and H3K27me2 / 3 nucleosome-binding activity is both unexpected and unprecedented, but would fit well with a role for the MIER complex in shutting down acti v e transcription.

DA T A A V AILABILITY
Raw MS data and sear ched r esults from MaxQuant are available from PRIDE ( https://www.ebi.ac.uk/pride/ ) with identifier PXD040685.