polycomb complexes prc1 and their function in hematopoiesis

PcG products assemble into diverse multiprotein complexes that fit into either of two major types of biochemical entities, characterised by non-overlapping sets of subunits: type I and t

Polycomb complexes PRC1 and their function in hematopoiesis

Miguel Vidal, Katharzina Starowicz

PII: S0301-472X(16)30774-3

DOI: 10.1016/j.exphem.2016.12.006

Reference: EXPHEM 3498

To appear in: Experimental Hematology

Received Date: 10 October 2016

Revised Date: 19 December 2016

Accepted Date: 20 December 2016

Please cite this article as: Vidal M, Starowicz K, Polycomb complexes PRC1 and their function in

hematopoiesis, Experimental Hematology (2017), doi: 10.1016/j.exphem.2016.12.006.

This is a PDF file of an unedited manuscript that has been accepted for publication As a service to our customers we are providing this early version of the manuscript The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Miguel Vidal*, Katharzina Starowicz

Department of Cellular and Molecular Biology

Centro de Investigaciones Biológicas

Hematopoiesis, the process by which blood cells are continuously produced,

is one of the best studied differentiation pathways Hematological diseases are associated to reiterated mutations in genes encoding important gene expression regulators, including chromatin regulators Among them, the Polycomb group (PcG) of proteins is an essential system of gene silencing involved in the maintenance of cell identities during differentiation PcG

proteins assemble into two major types of Polycomb repressive complexes (PRC) endowed with distinct histone tail modifying activities PRC1 complexes are histone H2A E3 ubiquitin ligases and PRC2 trimethylate histone H3 Established conceptions about their activities, mostly derived from work in embryonic stem cells, are being modified by new findings in differentiated cells Here we focus on PRC1 complexes, reviewing recent evidence on their intricate architecture, the diverse mechanisms of their recruitment to targets and on the different ways in which they engage in transcriptional control We also discuss hematopoietic PRC1 gain-of- and loss-of-function mouse strains, including those that model leukemic and lymphoma diseases, in the belief that these genetic analysis provide the ultimate test bench for molecular

mechanisms driving normal hematopoiesis and hematological malignancies

Blood cells are produced during hematopoiesis, a process where integration

of developmental and homeostatic signals guides the generation of

appropriate numbers of fully differentiated cells [1, 2] A network of defined DNA sites, enhancers and promoters, and the transcription factors (TF) that bind to them, results in cell type-specific gene expression patterns [3] All hematopoietic cell types originate in the bone marrow, through hierarchical differentiation from a minute pool of immature hematopoietic stem cells [4, 5]

TF orchestrate differentiation by establishing and activating a vast collection

of enhancers Establishment of new enhancers is initiated in early stages of cell lineage commitment During cell specification, some enhancers are of transient nature whereas others become cell type-specific after their

maintenance in one but not other lineages [6, 7] Accessibility of TF to their cognate sites and interactions with positive and negative transcriptional

cofactors is affected by nucleosome density, histone modifications and other molecular alterations corresponding to defined chromatin structures A large collection of chromatin regulators dynamically delineate such structures, from local to high-order, representing the core of epigenetic regulation of gene expression [8-10] Genetic mouse models and the study of mutations in

patients underline the crucial impact of chromatin regulation in normal and malignant hematopoiesis [11, 12]

The Polycomb system of chromatin regulators has attracted a great deal of interest and it has been recounted in recent excellent reviews [13-16] Here

we summarise advances on Polycomb system activities carried out by

Polycomb Repressive Complex(es) of type I (PRC1) and their impact in adult hematopoiesis Much of what is known about the Polycomb system originates from work in other models, mainly mammalian embryonic stem cells (ESCs)

and the fly Drosophila melanogaster It is worth noting that while core

concepts may be conserved throughout models, Polycomb regulation may hold a great deal of context-specific diversity With this in mind, we first outline PRC1 assemblies, their components and structural motifs in them of potential functionally Then, we consider PRC1 recruiting to targets and ways in which

The Polycomb system

Polycomb products conform an evolutionary conserved system of chromatin regulators, mainly known for their involvement in developmental processes, from plants to metazoans They are encoded by genes initially identified in mutations leading to morphological transformations in embryos and adults of

D melanogaster [17] Phenotypically, the alterations are consistent with gains

of function of homeotic products, and consequently the notion of their function

as repressors Genes mutated in new fly strains sharing phenotypic features

of the original mutant, Polycomb, were collectively branded as the Polycomb

group (PcG) of genes [18, 19] The name makes a reference to the additional number of sex combs, a structure present in legs of adult male flies, observed

in mutant flies Hence, the names for many of such genes: Posterior sex combs, Sex combs extra, Sex comb on midleg, etc The genetic analysis of fly

development revealed another group of mutations, in the so-called Trithorax

group (TrxG) of genes, as counteractors of Polycomb phenotypes [20] TrxG

genes and their products usually act as activators of gene expression and are not considered here (see reviews by [21, 22])

PcG products assemble into diverse multiprotein complexes that fit into either

of two major types of biochemical entities, characterised by non-overlapping sets of subunits: type I and type II Polycomb Repressive Complexes (PRC1 and PRC2, respectively) Associated to chromatin through mechanisms not fully understood, they are endowed with histone modifying activities: a protein ligase that monoubiquitylates histone H2A (PRC1), and a lysine

methyltransferase specific of histone H3 (PRC2) How PRC1-dependent H2A modification relates to transcriptional control is under active investigation [23-25], whereas PcG-modification of histone H3 is known to be central to gene repression [26, 27] The working paradigm, accepted for a long time,

sustained that PRC1 activity followed PRC2 Recent research, however, has

domains in heterochromatin [30] Characteristically, the Polycomb system silences targets important for the maintenance of cell identity during

transitions between cell states [31] The extent of Polycomb-modified

chromatin varies with cell state, being much larger in differentiated cells than

in cells with developmental potential [32], possibly in relation to restriction of developmental potential through transcriptional repression Of note, while the Polycomb system has been associated to transcriptional silencing, increasing evidence, for PRC1 at least, links it also to transcriptional activity (see below) Polycomb proteins participate also in genome stability, through DNA

replication and DNA damage repair functions [33-35] but here we will focus on transcriptional control

PRC1 assemblies: canonical and non-canonical complexes

The first Polycomb complex, isolated from Drosophila embryos, was termed

PRC1 [36, 37] Mammalian cells contain similar complexes [38], except for the presence of additional subunits corresponding to the multiplicity of homologs Subsequent purifications, however, showed that the term PRC1 encompasses

a larger collection of biochemical entities, containing subunits of unexpected relation to PcG products [39-41] All of these complexes share RING finger proteins, while diverging in their content of other subunits In an attempt to organise this heterogeneity, complexes that contain subunits found in the original purifications were termed canonical PRC1 complexes The rest, were grouped into a non-canonical category The division also reflects that, at the time, the little that was known about PRC1 pertained to subunits found in complexes to be named canonical Subsequent attempts to get a more

As with all classification attempts, these PRC1 categories use some

simplification They are as snapshots determined by the cell type from which the complexes are isolated, and although its systematic core may hold, it is already known that PRC1 complexes are very dynamic structures that evolve with progression between cell states [43, 44] An idea of the difficulties in getting these categories right is the presence of (non-canonical) RYBP in preparations of proteins associated to (canonical) PCGF2/MEL18 [45] The simplest picture is that of a core PRC1 complex, an ubiquitin ligase module, to which a reduced numbers of subunits associate: proteins with oligomerizing and histone reading abilities (PHC, CBX), for canonical forms, and DNA

binding proteins, for non-canonical forms Additionally, each of these minimal complexes may contain a heterogeneous collection of subunits, sometimes in

a substoichiometrically manner, conforming a pool of PRC1 complexes that varies with cell type

The core of PRC1 complexes: RING1-PCGF heterodimers that

monoubiquitylate histone H2A

The RING finger proteins at the centre of each PRC1 complex are structurally related, sharing a N-terminal RING finger, a specialised type of Zn2+ binding motif of the Cys3HisCys4 type [46] and a C-terminal RAWUL (from Ring-finger And WD40 associated Ubiquitin-Like, [47]) domain The paired RING finger motifs make E3 ubiquitin ligases and the RAWUL motifs act as binding platforms for other PRC1 components

The PRC1 ubiquitin ligase module (Figure 2) is made of either RING1A or its paralog RING1B, and one of the six PCGF proteins: PCGF1/NSPC1;

PCGF2/MEL18; PCGF3; PCGF4/BMI1; PCGF5 and PCGF6/MBLR Each

exclusively through contacts in RING1A or RING1B, away from the PCGF moiety [51] Mutations in either of the RING1-PCGF or RING1-E2 surfaces can impair ubiquitin ligase activity Characteristically, canonical PRC1

complexes contain RING1-PCGF2 and RING1-PCGF4 E3 ligases, whereas the remaining E3 ligases are constituents of non-canonical PRC1 complexes [42]

PRC1-dependent modification of histone H2A involves the monoubiquitylation

of (predominantly) lysine 119 (K119Ub) or of lysine 120 [52] Histone H2 variant H2A.Z is also monoubiquitylated by PRC1 at lysines 120, 121 and 125 [53-55].The activity of PRC1 E3 ligases is regulated by complex assembly, i.e., their activity follows the establishment of a multiplicity of contacts

involving the dimeric RING module, the nucleosome, DNA and the E2 ligase [27] PRC1 recognition of the nucleosome occurs through binding of the E3 ligase module that, as other nucleosome-interacting proteins, bind the acidic patch defined by histone pairs H2A and H2B, mostly through contacts

involving the RING1 protein These interactions, and the organization of the C-terminal tail of H2A in the nucleosome, position the Ub carrier E2 in the vicinity of the appropriate lysine residue, thus determining the specificity to the modification of the histone substrate [27] (histone H2A can also be

ubiquitylated at lysines 127 and 129 by another E3 ligase, [56]) In fact, it appears that E3 ligases in canonical PRC1 complexes may be autoinhibited until engaged with nucleosomal substrates [57] It has been suggested that E3 ligases in canonical, but not in non canonical PRC1, use some

stereoselective step previous to their activation, thus explaining their poor performance modifying H2A following forced recruitment to chromatin [58] Most monoubiquitylation of histone H2AUb depends on PRC1 E3 ligases, as

The shared C-terminal motif of PRC1 RING finger proteins is an Ub-like

folded domain (RAWUL) that participates in interactions with other PRC1 subunits The Ub-like motif in RING1A/RING1B binds through a common surface either chromobox (CBX) or RYBP/YAF2 proteins [63] Hence the presence exclusive of one or the other in isolated PRC1 complexes Ub-like motifs of PCGF proteins, instead, associate with specific PRC1 components For example, PCGF2/MEL18 and PCGF4/BMI1 interact with PHC proteins, whereas PCGF1/NSPC1 binds BCOR [64] It is likely that, as for other

multiprotein complexes, PRC1 aggregates form following a driven, hierarchical association of preassembled subunits [65, 66] For

cooperativity-example, PCGF1 binds a heterodimer of PRC1 subunits SKP1 only if previously associated to BCOR [67] Subsequently, the complex grows larger after contacts with heterodimers made of RING1A (or RING1B)-RYBP/YAF2 or CBX subunits The coexistence of modules may be reflected

KDM2B/FBXL10-in the wide range of sizes estimated for PRC1 complexes [42] and KDM2B/FBXL10-in the chromatin binding patterns of individual subunits [43]

PRC1 subunits preferentially present in canonical complexes

Mammalian homologs of Drosophila Polyhomeotic, Polyhomeotic-like 1, 2 and

3 (PHC1, 2 and 3) can participate in polymerizing protein-protein interactions thanks to the their C-terminal SAM (sterile alpha motif) domain [68] This protein motif, present in many other proteins too, has two surfaces for

interaction with the same or different SAM-containing proteins and therefore permit the formation of associations of high structural complexity [69] In addition to the SAM domain, PHC paralogs share a so-called homology

domain, a FCS-type zinc finger for binding to RING1/PCGF [70] The ability to make large molecular aggregates is evidenced in vivo, both in mammalian

and Drosophila cells [70-72] The tendency of PHC proteins to aggregate is

regulated by addition of O-linked β-N-acetylglucosamine (O-GlcNAc) to

serine/threonine strings [73], an activity dependent of the product of the GlcNAc transferase (Ogt) gene, the homolog of Drosophila super sex combs

O-[74]

A number of proteins of the malignant brain tumor (MBT) family, homologs of

the products of Drosophila PcG genes Sex comb on midleg (SCM) and SCM with four MBT (SfMBT), also have a C-terminal SAM domain and appear,

substoichiometrically, in preparations of canonical PRC1 [38, 42] It is though that the presence of SCML1, SCML2, SCMH1, SFMBT1 and SFMBT2L, in PRC1.4 complexes [75] is due contacts between their SAM domain and the SAM domain of PHC subunits [76]

RYBP/YAF2 and DNA-binding proteins, subunits of non canonical PRC1 complexes

RING1 and YY1 binding protein (RYBP) [77] and its paralog, YY1-associated factor 2 (YAF2) [78], are two small, basic proteins with a N-terminal RANBP2-type zinc finger motif They interact with the RAWUL domain of RING1A or RING1B through their C-terminal regions [77] Both RYBP/YAF2 and CBX proteins, despite their dissimilar conformation, contact the same region of RING1B [63] RYBP binds non specifically DNA [79], although the

physiological relevance of this ability is unknown Instead, RYBP/YAF2 were related initially to a possible recruiting activity, through their partner YY1, the

homolog of Drosophila Pleiohomeotic, known to participate in PRC1 recruiting

functions in flies [80] In mammalian cells, however, YY1 does not copurify with PRC1 components [42] nor colocalizes with PRC1 or PRC2-bound

chromatin [81]

A subset of non-canonical PRC1 complexes contain subunits with dedicated DNA-binding domains The best studied is lysine-specific demethylase 2B (KDM2B, or F-box and leucine-rich repeat protein 10, FBXL10), a PRC1.1 component It binds CpG islands (CGI) [82], a singular collection of DNA segments enriched in non-methylated CpG dinucleotides, permissive to

transcription that contain embedded over half of all promoters [see 83]

to those recruiting substrates to a subset of E3 ubiquitin ligases [86] KDM2B becomes an integral component of PRC1.1 through the association of its LRR regions with surfaces formed in the specific association of PCGF1 and BCOR [67] S-phase kinase-associated protein 1 (SKP1) bound to the F-box of

KDM2B leaves open, spatially, the possibility of a hypothetical interaction with cullin ubiquitin ligases [67] KDM2B histone H3 demethylase activity has been described as specific for lysine 36, or lysine 4, or both [84, 87, 88]

Other DNA-binding proteins in non-canonical complexes are heterodimeric transcription factors well known in other contexts, present in PRC1.6 One of the pairs is made of Max gene-associated protein (MGA) and interacting partner MAX, two helix-loop-helix leucine zipper proteins, that can bind E-box sequences recognized by MYC [89] In addition, MGA displays a second DNA binding domain, a T-box motif [89] The second pair, made of transcription factors and E2F6 and TFDP1, have potential to bind sites recognized by proteins of the E2F family In vitro, PRC1.6 associates to DNA fragments bearing MYC-, BRACHYURY- and E2F-binding sites [39] It is not clear that such potential to recruit PRC1 complexes to their in vivo targets is used

effectively, even in cell types such as ESC where MGA, MAX, E2F6 and TFCP1 are prevalent RING1B interactors [44] Recent work, however, reports

transcriptional derepression of meiosis-related genes in Max-null ESCs,

accompanied by defective recruitment of PRC1.6 subunits to targets [90]

Polycomb chromobox (CBX) proteins and other chromatin readers in PRC1 The mammalian homologs of Drosophila Polycomb, CBX2/M33, CBX4/PC4,

CBX6, CBX7 and CBX8, share a N-terminal chromo (CHRromatin

favored coimmunoprecipitation of CBX7 with PHC1 or PCGF2 from extracts which also contain CBX2 or CBX4 [94] CBX proteins recognition of the

trimethylated form of lysine 27 in the tail of histone H3 (H3K27me3), the

product of PRC2 activity, has been the basis of the long preferred mechanism for PRC1 recruiting to chromatin [94, 95] Despite high structural relatedness, CBX chromodomains bind methylated histones with different affinities and cannot discriminate between H3K27me3 and H3K9me3 [96, 97] Surprisingly,

in vivo confirmation of H3K27me3-dependent association of CBX proteins to chromatin is rather scarce (an exception, CBX7 in ES cells [94]) Indeed, in vivo imaging studies in ESCs show that H3K27me3 is effectively used for chromatin recruitment of CBX7 and CBX8, but not so much of CBX2, CBX4 and CBX6 [99] Moreover, it appears that in vivo accessibility of the aromatic cage that accommodate K27me3 varies with each CBX subunit [98] This is consistent with individual CBX proteins showing little functional overlap [94,

100, 101] Although CBX proteins appear preferentially in canonical PRC1 isolates [41, 42, 102], sometimes they have been found in non-canonical complexes [103] It is hard to tell whether these preparations are more

heterogeneous than anticipated and CBX8-non canonical PRC1 are entities of their own, or whether the presence of CBX8 is of substoichiometric nature

Chromodomains of Polycomb CBX proteins can recognize not only

methylated lysines on histones but also RNA, although in a non-specifically fashion RNA [96] In vivo, the best documented example is the binding human non-coding RNA ANRIL to CBX7[104] An additional feature of Polycomb CBX proteins is a conserved sequence, in the proximity of the chromodomain, that contacts the histone acetyltransferase (HAT) domain of CRE-binding protein (CBP) Such an interaction interferes with activating CBP

autoacetylation and contributes to transcriptional repression [105]

recognise histones through its MBT repeats [107, 108]

Recruiting PRC1 to targets

Polycomb-defined chromatin states are cell type-specific, in response to differentiation programs [43, 44, 109, 110] Steady state levels for Polycomb-dependent modifications result from a balance between the activities of PRC1 and PRC2 and that of specific histone deubiquitinases and demethylases (chromatin erasers) Here we will refer only to localization of Polycomb

modifications influenced by recruiting processes

Polycomb-modified chromatin encompasses large CGIs with H2AUb and H3K27me3-enrichment peaking around promoters These promoters usually are transcriptionally silent or with very little activity, and are also enriched in H3K4me3 Being these histone H3 modifications usually associated to

opposed, repressed and active transcriptional states the genomic domains thus marked were termed bivalent [111, 112] During differentiation some of these bivalent domains are resolved into silent, H3K27me3 only, or active, H3K4me3 only, marked promoters [109] Although H2AUb1 localizes too to bivalent domains, many H3K27me3-enriched regions lack H2AUb1 [113] First described in ESCs, bivalent domains are also present in somatic

oligopotent adult cells, were the ESCs-view that bivalency poises genes for expression at a subsequent differentiation stage does not hold [114] In

contrast, a large number of PRC1 targets, in differentiated cells, are occupied CGIs devoid of H3K27me3 (Figure 1B) [44, 115], most of which are transcriptionally active [43, 44, 116-119] To this date, the question of

RING1B-Polycomb recruiting to targets remains a big puzzle under active investigation

been identified in Drosophila as Polycomb response elements (PREs), or

regions enriched in sites for DNA-binding proteins that confer Polycomb

recruitment in transgenic studies (see [123] for a review) In mammalian cells, PRC1 recruiting to specific sites can take place through association to pre-bound transcription factors such as RUNX1, in megakaryoblastic and tumoral

T cells [124], or by YY1 in ESCs [125] and REST [126, 127] (disputed in [128]) A well-studied case is that of non-canonical PRC1.1 recruitment to BCL6-binding sites in lymphoid cells, mediated by BCOR [103] These

examples, however, account for a minority of Polycomb recruiting In fact, while not exactly PRE-like sites, unmethylated GC rich sequences without identifiable DNA sequence features are able to confer ectopically, in a

mammalian transgenic setting, PRC1 and PRC2 engaging activity [81, 129] Surely, recruiting to CGI sites can take place through CGI-binding KDM2B, the caveat being that whereas KDM2B binds to all CGIs, Polycomb products are enriched only in a fraction them [115, 119, 130, 131] Thus, additional, as yet unknown players act in Polycomb recruiting Among candidates, long non-coding RNAs (lncRNAs) have received attention, in particular after a great deal of work linking them to PRC2 recruiting Nonetheless, the actual impact

of promiscuous PRC2 binding to RNA on gene regulation is still being

debated (see [132, 133]) Alternatively, the perception that the transcriptional status of a given genomic region, through uncharacterized mechanisms, influences Polycomb localization is at the basis of a responsive model for recruitment According to this notion, Polycomb complexes would sample chromatin continuously and associate to transcriptionally repressed, not

active, genomic sites [134] Whether the presence of Polycomb complexes on transcriptionally active, KDM2B occupied promoters [119] is an intermediate step in the recruiting cycle or the consequence of a dedicated mechanism remains to be clarified

The study of PRC1 recruiting through KDM2B has modified the widely

accepted relationship between PRC1 and PRC2 interactions Forced

experimental setting, KDM2B also mediates PRC2-dependent H3K27me3 enrichment This was unexpected considering the prevalent notion that PRC1 recruiting to H3K27me3 sites was mediated by CBX subunits The

explanation for recruitment of PRC2 lies in the finding that, at least in vitro, reconstituted PRC2 complexes can bind H2AUb-decorated nucleosomes Moreover, this association stimulates the catalytic activity of EZH2, the PRC2 histone methyl transferase [135] Interestingly, since H3K27me2 is not the EZH2 preferred substrate (compared to H3K27me1, [136]), such an activation step may be pertinent to the existence of H3K27me3 peaks within vast

genomic regions modified with H3K27me2 [137] Thus, together with the known ability of PRC1 CBX subunits to associate to H3K27me3 nucleosomes, H2AUb facilitation of PRC2 activity conforms a feed forward mechanism to stabilise chromatin states at Polycomb targets

There are, however, observations that break the linearity of KDM2B-mediated Polycomb recruiting In fact, although the highest RING1B density has been found to coincide with H3K27me3-enriched CGIs [115], supporting self-

reinforced interactions, not all ESC sites bound by KDM2B and RING1B show H3K27me3 enrichment, perhaps an evidence of self reinforcing interactions mentioned above Also, the impact of KDM2B downregulation on RING1B binding and H3K27me3 deposition is a milder than predicted by the model [115, 119, 130] At the same time, severe down regulation of H2AUb levels correlates with decreased H3K27me3 enrichment at targets [25] and,

conversely, the absence of H3K27me3 marks is accompanied by a drastic but incomplete loss of RING1B and CBX7 from targeted promoters [45, 94, 138]

It is worth noting that the above scenario is that of ESCs A quite different situation is that of differentiated cells, where most RING1B-bound sites

correspond to CGIs (and other regions, such as enhancers) without

H3K27me3 [42, 44, 139] This is the clearest evidence that PRC1 is recruited

to targets in a PRC2-independent manner, possibly by mechanisms

compatible with the presence of KDM2B on active CGIs In summary,

Polycomb recruiting uses interrelated and context-dependent DNA binding, histone recognition and protein-protein interactions It’s likely that this diffuse picture will get more complicated when recruitment at targets in somatic adult cells gets better known

Transcriptional functions of PRC1 at silent and active loci

Work on transcriptional control by PRC1 has addressed mostly their role as repressors on silent loci Several mechanisms have been involved, from the transcription process itself, accessibility of DNA binding proteins and

transcription machinery to the modulation of chromatin topology determined

by compaction and contacts between distant genomic sites In contrast, the study of PRC1 activity on transcriptionally active sites is currently at the early, descriptive stages

oligonucleosomes [141] PRC1 complexes can affect the assembly of the initiation complex (PIC) For instance, recruitment of Mediator, a multi protein complex that communicate transcription factors and RNA pol II on the

pre-template [142] is blocked by PRC1 ln contrast, association to the pre-template of Transcription factor IID (TFIID), the key regulator of de novo transcription initiation (through promoter recognition, organization of the formation of the preinitiation complex (PIC), and interaction with transactivators [143]) or of its subunit TATA-box binding protein (TBP) is not altered by PRC1 [144] In vivo tests of predictions derived from these biochemical assays are difficult

Comparison of chromatin binding patters of PIC components and of PRC1 subunits in ESCs shows extensive overlap of promoters bound by TBP and RING1B Interestingly, assessing Mediator enrichment at these sites, as

binding of its Med1 subunit, identifies a correlation between the lowest

enrichment in Med1 and low transcriptional output [144] These results would

be consistent with PRC1 repression occurring through interference with the formation of a functional RNA poll II PIC assembly

Pausing, another critical step of the RNA pol II transcription cycle, has also been linked to PRC1 repression RNA pol II pausing is a widespread

mechanism by which, shortly after initiation, the polymerase is halted at

promoter proximal sites by the activity of pause-inducing complexes NELF and DSIF and other factors Productive mRNA elongation takes place after recruitment, prompted by activators, of P-TEFB kinase which phosphorylates NELF and RNA pol II (see [145]) Engaged, paused RNA pol II is

phosphorylated at serine 5 (S5p) in the repeats in its C-terminal domain, and upon P-TEFB mediated release, acquires phosphorylation at serine 2 (S2P) The study of a collection of Polycomb-regulated, bivalent promoters in ESCs finds them enriched in RNA pol II S5p As derepression of these promoters,

after inactivation of Ring1A and Ring1B, correlates with an increase on

promoter-associated RNA pol II, without changes in content of S5P or S2P forms [146], an implication of RING1 proteins in RNA pol II pausing was

suggested However, engaged RNA pol II location, assessed by genome wide run-on experiments in ES cells, indicates a poor association with sites bound

by PRC1 subunits [147], making little likely a role for PRC1 in gene silencing through RNA pol II pausing In fact, promoter proximal RNA pol II pausing in naive ES cells is not enriched in developmental genes repressed by PRC1 [148, 149] Moreover, in contrast with ESCs, bivalent promoters at

differentiated cells lack RNA pol II [114]

PRC1 repression through regulation of chromatin topology

Interactions between chromatin-bound proteins can affect chromatin

configurations RING1B shows homodimerizing ability in vitro [150, 151] but its actual impact on the assembly of multiprotein assemblies is expected to be

of lesser importance than that of SAM-domain containing PRC1 subunits Indeed, microscopical clustering of PRC1 complexes, an indication of

chromatin compaction, decreases if the cells express a PHC subunit bearing

protein-structures is disrupted, along with partial gene derepression, in Drosophila

cells that express a Polyhomeotic variant whose SAM domain of cannot polymerize [72]

High order PRC1-multiprotein structures not only regulate gene expression through chromatin compaction but also participate in the stability of contacts between distant genomic sites In ESCs, RING1B enriched genomic sites coalesce three-dimensionally through a network of promoter-promoter and promoter-enhancer contacts [152] The involvement of PRC1 activity in

maintaining this structure is shown by the loss of promoter-promoter

associations in the most prominent network of contacts (Hox genes) that

follows depletion of RING1A and RING1B [152] Within the disrupted

structure, enhancers still touch promoters, but their histone modifications reveal a transition from poised (H3K27me3) to active (H3K27Ac) states that is

in agreement with the observed gene derepression Transcriptional control through PRC1-regulated contacts between distant genomic regions is also

appreciated in the murine Meis1 locus Here, RING1B participates in the

formation of exclusive contacts between enhancer and site at the 3' end of the locus, a developmentally regulated process that

promoter-determine the timing of Meis1 expression in embryonic midbrain [153]

PRC1-H2AUb function in a repression to activation switch?

The long standing acceptance of H2AUb as an important element in

Polycomb repression [52] is being re-assessed in the light of new evidence from mouse and fly cells The first discrepancy with the expected link between H2AUb and gene repression was seen in RING1B-deficient ESCs, where de-

repression of Hox genes is rescued by complementation with an E3-inert RING1B variant.[154] The observation is not restricted to Hox clusters,

dispensability, studying imaginal disc cells that contain only

non-ubiquitylatable forms (lysine to arginine mutations) of histone H2A or H2Av (homolog of mammalian H2AZ), shows that repression of typical Polycomb

targets Ubx and Abd-B is maintained in the absence of H2AUb [23] The

unexpected observation that these loci normally lack H2AUbK118 [155], however, makes the observation inconclusive

Despite the apparent disconnection between PRC1-dependent H2A

modification and gene repression, there are abundant examples correlating gene activation with decreased H2AUb levels For instance, androgen-

receptor activation of gene expression through a complex that couples histone acetylation and deubiquitylation by a complex containing p300/CBP HAT and MYSM1 histone H2AK119 deubiquitylase, respectively [156] In quiescent splenocytes, RING1B-binds transcriptionally active genes in a context of concerted down regulation and upregulation of RING1B E3 ligase and USP16 deubiquitylase activities that lead to minimization of H2AUb enrichment [118]

An additional example has been shown for the PRC1.5 complex that can activate transcription through p300/CBP HAT recruiting, bound to subunit AUTS2, and downregulation of RING1B E3 ligase by casein kinase II-

mediated phosphorylation of RING1B [157]

In a function alternative to gene repression, histone H2AUb could act as a signal for chromatin on-site remodeling of protein complexes A recent report proposes that repressed, H2AUb marked genes in ESCs are activated during differentiation in a process that involves H2AUb-binding Zuotin related factor 1 (ZRF1) In this case of gene activation, ZRF1 displaces RING1B from a pre-assembled complex with an inactive form of Mediator while recruiting

activating Mediator module CDK8 [158] The scope of this on-site remodeling

of PRC1 and associated complexes is not known but it may be related to the

Loss-of-associated to deregulated expression in gain-of-function models Bone

marrow transplantation and cellular models, however, make wider use of of-function approaches, possibly due to experimental ease With the advent of genomic edition technologies (CRISPR-Cas9 and related) the collection in Table 2 is expected to grow rapidly, not only with single but also compound modifications

Cell type-specific differences in PRC1 protein levels are already suggestive of distinctive roles for the various PRC1 complexes Table 3 shows quantitative estimations for primitive, lineage negative (Lin-) ckit+ Sca1+ (LKS) progenitors,

a pool enriched in hematopoietic stem cells (HSCs) and their immediate differentiated derivatives, LKS- cells Protein levels in cells from bone marrow and secondary hematopoietic organs are also shown The data show that the highest levels of PRC1 subunits correspond to the Lin- populations, the

compartment with higher developmental potential, and then decrease to a variable extent with differentiation The cells with lowest content in PRC1 subunits are neutrophils, probably related to their short lifespan Note that levels vary in ways that do not fit the simplified categories of canonical and non-canonical complexes described above

As it could be anticipated from protein levels, function(s) of PRC1 complexes

in primitive cells are critical In many cases, loss of PRC1 subunits leads to

through regulated quiescence [163] Therefore, defective proliferation control,

as in Kdm2b or Bmi1 mutations, results in an abnormally high proliferative

rate of LKS cells [162, 164] Excessive proliferation is known to exhaust the HSC reservoir, thus explaining not only its reduced size but also the poor performance of PRC1 mutant cells in reconstitution experiments Double

inactivation of Ring1A and Ring1B leads to defective growth in colony assays

with Lin- cells [33], although this seems related to functions in DNA replication [33, 34] Such defect in consistent with the aplastic anemia observed in mice

with Ring1A and Ring1B-deficient hematopoiesis (Figure 3) Another

antiproliferative effect can result from excessive oxidative stress and the DNA damage response associated that result in hematopoietic cells lacking BMI1

[165] The increased repopulating ability of Mel18 -/- progenitors, despite their increased content in cycling cells, has been associated to upregulation of the

Hoxb4 gene in HSCs [166], the product of which was previously known to

promote expansion of HSCs [167]

Deficient control of proliferation in mutant PRC1 models affects not only HSCs but other immature progenitors and more differentiated cells Reduced

expansion of lymphoid cells associated to inactivation of M33, Bmi1 or Ring1B

is related to the derepression of the locus encoding tumor suppressors

p16INK4a and p19ARF [168-170] KDM2B is also a repressor of

Cdkn2b/Ink4b, encoding the negative regulator of the cell cycle p15INK4B

[171] Hematopoietic neoplasia caused by over expression of certain PRC1 subunits (see below, Table 2) concur with downregulation of products of the

Cdkn2a,b/Ink4a,b cluster [169, 171, 172] In other models that develop

proliferative defects, however, no involvement of Ink4 genes has been

appreciated [100, 173] PRC1 actions on cell cycle regulators can be

differentiation stage-dependent These have been seen in the depletion of MEL18 and RING1B, where hypocellularity of distinct extent in bone marrow

lacking BMI1 or KDM2B [162, 174] Mechanistically, Bmi1deletionactivates

prematurely lymphoid specification genes (Ebf1, Pax5) in HSPC, accelerating

their differentiation leading to augmented B lymphopoiesis [174] Likewise, ectopic KDM2B, enhances lymphoid commitment, in a function that relies on its JmJC domain [162] A clear example of PRC1-dependent maintenance of cell fate is the conversion of T-cell progenitors to B cells following compound

deletion of Ring1A and Ring1B in immature T-cells [175] This result suggests

that PRC1-mediated repression of the B-cell lineage program, commanded by

Pax5, is required for T-cell lineage commitment Another developmental

alteration in the lymphoid lineage results from the early disruption of non

canonical Rybp As a consequence, progenitors of the B1 lineage, a

population that decreases in the fetal to adult transition, becomes enlarged and at the same time the pool of progenitors of the B2 lineage, responsible for

acquired immunity, is reduced [176] Surprisingly, the Rybp mutation shows

no genetic interaction with Ring1A or Ring1B On the other hand, inactivation

of non-canonical PRC1 BCOR in bone marrow cells upregulates myeloid

differentiation genes [177] Finally, PCGF1 downregulation in Runx1

-/-progenitors enhances their self-renewal ability and, in parallel, the HoxA cluster gets derepressed [179] Runx1 is a master regulator of adult (and

embryonic) hematopoiesis frequently disrupted in leukemia [178] It is thought that its inactivation predisposes progenitors to immortalization in cooperation with downregulation of PCGF1, a scenario in which silencing of a

transcriptional program in primitive cells prior to scheduled differentiation is defective

Despite the wealth of information derived from these models, there are

constrains due limitations inherent to genetic analysis Besides lethality,

redundancy, when one of the paralogs compensates for the inactivation of the

other, is a common situation This is well illustrated by the combined

inactivation of paralogs Ring1A and Ring1B, where the hematopoietic

phenotypes of single mutations are rather mild [170], but the combined

inactivation results in severe aplasia and prompt lethality (Figure 3) The limited availability of Cre-driver lines, for better controlled conditional

inactivation precludes cell stage/differentiation studies worth exploring For example, cell-type specific functions for PRC1 complexes are evidenced, in a gain-of-function approach, that shows distinct repression of progenitor (LKS-) and stem cell programs (LKS+) by CBX7 and CBX8, respectively [100]

Another example of differentiation stage-specific function has been noted after

Cbx8 conditional inactivation in germinal centre cells, showing cooperation

with PRC2 in the transient silencing of plasma cells transcriptional programs [103]

The study of mouse models bearing inactivating mutations in genes encoding histone H2AK119Ub deubiquitylases (DUB), unfortunately, adds little to a general understanding of the role of this histone modification in

hematopoiesis DUB mutations lead to phenotypes that are very difficult to

interpret in relation to PRC1 due to pleiotropic functions (Bap1, [180]) or to the poor overlap of targets Defective lymphoiesis and HSC exhaustion in Mysm1 - /-

mice correlate with failed activation of commitment regulators (Ebf1, [181])

or of negative regulators of the cell cycle (Gfi1, [182]) The association of repression with H2AUb, however, is less clear in Usp16 and Mysm1

mutations where Cdkn1a or Cdkn2a/Ink4a (p19ARF) are upregulated,

respectively [183, 184] It is possible that these are the result of effects

secondary to cellular stress

alterations in the absence of a PRC1 product, using a mouse model for

conditional inactivation of Ring1B [170] The results showed ample changes,

affecting many PRC1 components, that increase or decrease their levels after RING1B depletion (Fig 4) As possible causes are the deregulation of PRC1-encoding genes (as seen ES cells [94]) or that of factors acting on protein stabilization/destabilization It is worth noting that although PRC1 products are considered gene repressors, transcriptome analysis of mutant hematopoietic cells shows a large number of down-regulated targets, sometimes a larger set than that of upregulated, genes [162] Altogether, the data support a

preponderant role of PRC1 in hematopoiesis through cell proliferation control, and also through the execution of transcriptional programs

PRC1 in models of malignant hematopoiesis

Here we are not describing PRC1 mutations in patients (see [185]) as a way

to understand PRC1 roles in hematopoietic disorders Instead, we summarise results from experiment with models of aberrant hematopoiesis in which

PRC1 functions are perturbed The overexpression of some PRC1 products can alter hematopoietic homeostasis, as mentioned above (Table 2)

However, the manipulation of the Polycomb system in a context of malignant hematopoiesis not only can provide insight into Polycomb function but also widen the range of opportunities in a quest for therapeutic intervention The list of modeled hematopoietic malignancies together with gain-of or loss-of-PRC1 function we have compiled is shown in Table 4 Most of these models correspond to acute myeloid leukemia (AML), in which fusion proteins,

resulting from translocations involving the Mll1 (mixed lineage leukemia 1)

locus, commonly found in AML patients, are overexpressed The resulting chimeric proteins fuse the N-terminal region of MLL1/KMT2A to a variety of subunits of the superelongation complex, such as AF9 or ENL and are

powerful transcriptional activators (reviewed in [186]) Leukemic models can also be induced by overexpression of targets upregulated by MLL-fusion

proteins, such as the Hoxa9 and Meis1 genes [210], or of or mutant RAS

proteins [211] Lymphomas are modeled by lymphoid-targeted expression of

the c-Myc oncogene [187]

defective transformation associated to PRC1 inactivation are labeled by ON,

as if the involved subunits were acting as oncogenes) These contrasting functions, however, are not gene-specific because the same gene product may act as an oncogene or a tumor suppressor, in a cell context-dependent manner

Examples of PRC1 subunits cooperating with hematopoietic transformation are seen for MLL fusion proteins and CBX8 [188, 189] Also in the aberrant repopulating ability and ex-vivo expansion of progenitors, induced by HOXA9-MEIS1 or Promyelocytic leukemia zinc finger-retinoic acid receptor a (PLZF-RARA), respectively, in the absence of BMI1 [190, 191] Even just a reduced dosage of BMI1 (heterozygosity of a null allele) abrogates the enlargement of the HSC pool induced by expression of DNA methyl transferase DNMT3a-

R882H [192] or lymphoma development in Eu-Myc transgenic mice [169] In agreement with this, ectopic BMI1 promotes lymphoma in E µ -Myc transgenic

progenitors [172] and shortens the latency of appearance of myelodisplastic syndromes induced by expression of RUNX1 D171N [194] Similarly, KDM2B appears necessary to maintain expansion of cell lines established from

lymphoid malignancies [162] and the downregulation core PRC1 subunits RING1A or RING1B interferes with MLL-AF9 initiation of leukemic

transformation in human progenitors [173] (note a discrepancy with a similar

experiment with mouse progenitors [188] The double inactivation of Ring1A and Ring1B in murine Lin- progenitors makes them refractory to

transformation by MLL-Af9 or PML-RARα [193] However, given the dramatic effect that the compound inactivation of RING1 paralogs has on in vitro [33] and in vivo (Figure 3A) hematopoietic growth is difficult to tell whether these are dominant on those derived from genuine chromatin regulation As a

whole, the impairment of hematopoietic malignization in loss-of-PRC1 function

moieties of MLL fusion proteins [196] Thus, impaired transformation in the absence of CBX8 may be a direct consequence of an altered transcriptional output of the transforming proteins Thus, murine CBX8-deficient progenitors

that express MLL-AF9 fail to upregulate its target Hoxa9 [188], whereas

ectopic CBX8 represses it in MLL-ENL progenitors [189]

PRC1 loss of function has also been found to promote hematopoietic

transformation, highlighting the complexity of PRC1 function Among

examples, is the accelerated onset of lymphoma in conditionally deficient

Ring1B mice, associated to the inactivation of the Cdkn2a/Ink4a (encoding

P16INK4a + P19ARF) locus [170] It is possible that the apparent

contradiction with facilitation of MLL-AF9 transformation of progenitors [173] relates to the different cell types involved An analogous disagreement, the

expedite growth of KRASG12D-induced leukemia seen in Kdm2b mutant mice

[162] versus the impairment of MLL-AF9 leukemogenesis upon KDM2B

downregulation [171, 173] may be due to differences in transforming

pathways or in the experimental set ups Yet, facilitated malignancy is seen

with another PRC1 subunit, BMI1, which under Cdkn2a/Ink4a inactivation can

induce extramedular hematopoiesis and myelofibrosis In this case, the

upregulation of chromatin high mobility group protein HMG2A, was linked to pathological development in the absence of BMI1 [197] Accelerated

leukemogenesis in the absence of KDM2B takes place with upregulation of positive regulators of the cell cycle and of PRC2-bound genes, together with decreased expression of lymphoid-specific transcription factors [162] Thus, PRC1 complexes participate in pathways that promote or restrain

hematopoietic transformation at their initiation and maintenance stages, and further work is needed to understand the paradox posed by malignant

hematopoiesis combining self-renewal and (partial) differentiation

Conclusion and perspectives

The progress achieved throughout the last years in the study of PRC1 and PRC2 complexes system is incontrovertible However, the importance of questions without satisfactory answers is but an indication of the rudimentary state of the current understanding of how the Polycomb system plays out In particular, it is becoming urgent to widen the field of study to attain an

accurate scenario in the hematopoietic compartment, instead of that intensely influenced by the large body of knowledge derived from pluripotent ESCs Increasingly, evidence from differentiating embryonic and adult cell types often shows inconsistencies with ESC-derived models Obviously, this is of relevance to the deciphering of Polycomb workings in normal and aberrant hematopoiesis Functionally, the irreplaceable genetic analysis to approach these questions, will have to become more sophisticated The generation of allelic series should help uncoupling events intertwined in null mutations, for example proliferation and control of differentiation programs The impact of functions with ample effect on cell physiology, such as Polycomb roles in metabolism and DNA replication/DNA damage responses will also have to get better attention Luckily, these challenges will benefit enormously from

genomic edition methodologies developed in the last few years

Mechanistically, the participation of individual Polycomb subunits, as

mentioned above for PRC1, is very much in need of clarification While cell heterogeneity and scarcity make hematopoietic progenitors, the compartment more sensitive to PRC1 activity, a demanding model for study, the increased sensitivity of mass spectrometry, massive DNA sequencing and microscopy methodologies are making more accessible single cell approaches Knowing about Polycomb functions in the hematopoietic compartment will help

understanding normal differentiation programs and assist in intervention when gone awry in the hematological disease

Acknowledgements

We are grateful to Claudia Pérez for histology work We apologize to

colleagues whose work has not been quoted due to limited space Work in the lab is supported by grants SAF2013-47997-P and SAF2016-80486-P

from the Ministry of Economy and Competitiveness

3 Pimanda JE, Göttgens B Gene regulatory networks governing

haematopoietic stem cell development and identity Int J Dev Biol 2010;54:1201–1211

4 Sawai CM, Babovic S, Upadhaya S, et al Hematopoietic Stem Cells

Are the Major Source of Multilineage Hematopoiesis in Adult Animals Immunity 2016;45:597–609

5 Busch K, Klapproth K, Barile M, et al Fundamental properties of

unperturbed haematopoiesis from stem cells in vivo Nature

2015;518:542–546

6 Lara-Astiaso D, Weiner A, Lorenzo-Vivas E, et al Chromatin state

dynamics during blood formation Science 2014;345:943–949

7 Huang J, Liu X, Li D, et al Dynamic Control of Enhancer Repertoires

Drives Lineage and Stage-Specific Transcription during

Hematopoiesis Dev Cell 2016;36:9–23

8 Chen T, Dent SYR Chromatin modifiers and remodellers: regulators of

cellular differentiation Nat Rev Genet.2014;15:93–106

9 Allis CD, Jenuwein T The molecular hallmarks of epigenetic control

Nat Rev Genet 2016;17:487–500

10 Hu D, Shilatifard A Epigenetics of hematopoiesis and hematological

malignancies Genes & Dev 2016;30:2021–2041

11 Shih AH, Abdel-Wahab O, Patel JP, Levine RL The role of mutations

in epigenetic regulators in myeloid malignancies Nat Rev Cancer 2012;12:599–612

12 Butler JS, Dent SYR The role of chromatin modifiers in normal and

malignant hematopoiesis Blood 2013;121:3076–3084

13 Aranda S, Mas G, Di Croce L Regulation of gene transcription by

Polycomb proteins Sci Adv 2015;1:e1500737

14 Entrevan M, Schuettengruber B, Cavalli G Regulation of Genome

Architecture and Function by Polycomb Proteins Trends Cell

Biol.2016;26:511–525

15 Comet I, Riising EM, Leblanc B, Helin K Maintaining cell identity:

PRC2-mediated regulation of transcription and cancer Nat Rev

Cancer 2016;16:803–810

16 Koppens M, van Lohuizen M Context-dependent actions of Polycomb

repressors in cancer Oncogene 2015;35:1341–1352

17 Lewis EB A gene complex controlling segmentation in Drosophila

Nature 1978;276:565–570

18 Jürgens G A group of genes controlling the spatial expression of the

bithorax complex in Drosophila Nature 1985;316:153–155

19 Gaytán de Ayala Alonso A, Gutiérrez L, Fritsch C, Papp B, Beuchle D,

Müller J A genetic screen identifies novel polycomb group genes in Drosophila Genetics 2007;176:2099–2108

20 Ingham PW trithorax and the regulation of homeotic gene expression

in Drosophila: a historical perspective Int J Dev Biol 1998;42:423–429

21 Schuettengruber B, Chourrout D, Vervoort M, Leblanc B, Cavalli G

Genome regulation by polycomb and trithorax proteins Cell

2007;128:735–745

22 Piunti A, Shilatifard A Epigenetic balance of gene expression by

Polycomb and COMPASS families Science 2016;352:aad9780

23 Pengelly AR, Kalb R, Finkl K, Müller J Transcriptional repression by

PRC1 in the absence of H2A monoubiquitylation Genes & Dev

2015;29:1487–1492

24 Kahn TG, Dorafshan E, Schultheis D, et al Interdependence of PRC1

and PRC2 for recruitment to Polycomb Response Elements Nucleic Acids Res 2016;gkw701–18

25 Illingworth RS, Moffat M, Mann AR, et al The E3 ubiquitin ligase

activity of RING1B is not essential for early mouse development

Genes & Dev 2015;29:1897–1902

26 Pengelly AR, Copur Ö, Jäckle H, Herzig A, Müller J A histone mutant

reproduces the phenotype caused by loss of histone-modifying factor Polycomb Science 2013;339:698–699

27 McGinty RK, Henrici RC, Tan S Crystal structure of the PRC1

ubiquitylation module bound to the nucleosome Nature

2014;514:591–596

28 Kadoch C, Williams RT, Calarco JP, et al Dynamics of BAF–Polycomb

complex opposition on heterochromatin in normal and oncogenic

states Nat Genet 2016;doi 10.1038/ng.3734

29 Ram O, Goren A, Amit I, et al Combinatorial Patterning of Chromatin

Regulators Uncovered by Genome-wide Location Analysis in Human Cells Cell 2011;147:1628–1639

30 Ho JWK, Jung YL, Liu T, et al Comparative analysis of metazoan

chromatin organization Nature 2014;512:449–452

31 Reynolds N, O'Shaughnessy A, Hendrich B Transcriptional repressors:

multifaceted regulators of gene expression Development

2013;140:505–512

32 Zhu J, Adli M, Zou JY, et al Genome-wide chromatin state transitions

associated with developmental and environmental cues Cell

2013;152:642–654

33 Piunti A, Rossi A, Cerutti A, et al Polycomb proteins control

proliferation and transformation independently of cell cycle checkpoints

by regulating DNA replication Nat Commun 2014;5:3649

34 Bravo M, Nicolini F, Starowicz K, et al Polycomb RING1A- and

RING1B-dependent histone H2A monoubiquitylation at pericentromeric regions promotes S-phase progression J Cell Sci 2015;128:3660–

3671

35 Vissers JHA, van Lohuizen M, Citterio E The emerging role of

Polycomb repressors in the response to DNA damage J Cell Sci 2012;125:3939–3948

36 Shao Z, Raible F, Mollaaghababa R, et al Stabilization of chromatin

structure by PRC1, a Polycomb complex Cell 1999;98:37–46

37 Saurin AJ, Shao Z, Erdjument-Bromage H, Tempst P, Kingston RE A

Drosophila Polycomb group complex includes Zeste and dTAFII

proteins Nature 2001;412:655–660

38 Levine SS, Weiss A, Erdjument-Bromage H, Shao Z, Tempst P,

Kingston RE The core of the polycomb repressive complex is