Histone Variants and Their Chaperones in Hematological Malignancies

Epigenetic changes in hematological malignancies

Epigenetic perturbations in cancer include dysregulated DNA methylation, alterations in chromatin composition, and abnormal patterns of PTMs on histone tails.^7. Indeed, in hematological malignancies, numerous frequently mutated genes belong to gene families encoding proteins involved in the epigenetic machinery. A prominent example of this is the polycomb-associated gene ASXL1, which is found to be recurrently mutated in various hematological malignancies including clonal hematopoiesis, myelodysplastic syndromes (MDS), acute myeloid leukemia (AML), and chronic myelomonocytic leukemia (CMML).^8.-10. ASXL1 mutations are associated with inferior overall survival.^11.,12. Furthermore, in some cases, mutations in chromatin regulators can co-occur. For instance, co-occurrence of DNMT3A and TET2 mutations is observed at a high frequency in T-cell lymphomas.^13. Moreover, certain oncogenic fusion proteins, such as PML::RARα, have DNA-binding capacity and are able to aberrantly recruit chromatin regulators and alter gene expression driving leukemogenesis.^14.,15. Alterations to the epigenetic machinery are considered in latest risk-scoring systems for MDS and myeloproliferative neoplasms.^16.,17.

Given the high degree of epigenetic dysregulation in hematological malignancies, pharmacological targeting of chromatin regulators, frequently referred to as epigenetic therapies, has attracted significant interest in recent years. For example, the nucleoside cytidine analogue Azacytidine acts as a hypomethylating agent by inhibiting DNA methyltransferases, and is used in the clinic to treat patients with high-risk MDS and AML.^18.,19. A recent report showed that the Menin inhibitor Revumenib efficiently induced remissions in pediatric and adult AML patients with KMT2A rearrangements or NPM1 mutations.^20. These results stand out as they represent the first efficient epigenetic therapy that is based on the eviction of protein complexes from chromatin. However, not all patients respond to epigenetic therapies, and responses to such therapies are frequently transient. For example, clinical trials involving the preclinically promising DOT1L inhibitor Pinometostat have shown very limited clinical benefit in both children^21. and adult^22. patients, and clinical testing was discontinued. Thus, efforts toward the development of more efficient intervention strategies targeting the epigenetic machinery, as well as the combination of these agents with other targeting approaches are needed and currently ongoing.

Histone variants

An important part of epigenetic regulation in cells occurs on the level of the nucleosome, involving PTMs of histone tails^23. and the replacement of replication-coupled histones by histone variants.^24. The replacement of replication-coupled histones by histone variants is still one of the least understood epigenetic mechanisms. Replication-coupled histones H2A, H2B, H3, and H4 are highly conserved among vertebrates, and are encoded by multicopy gene clusters that lack introns.^25. Synchronized expression during S-phase provides messenger RNAs (mRNAs) of replication-coupled histones lacking poly-A tails for the synthesis of large amounts of histones that are required to efficiently package the newly replicated genome.^26. In contrast, genes encoding histone variants resemble regular protein-coding genes, and are transcribed by RNA polymerase II (RNA Pol II), expressed throughout the cell cycle, contain introns, and in some cases, feature several splice variants. Histone variants share a histone-fold domain with replication-coupled histones, but otherwise differ in sequence and structure with high variability in their N- and C-termini.^23. This bestows on histone variants the ability to interact with distinct chaperones and chromatin remodeling complexes, contributing to their locus-specific deposition, which will be discussed in more detail below.

Nucleosomes generally provide a barrier for transcription, limiting transcription factor binding and RNA Pol II elongation. Histone variants affect transcription by altering the dynamics and biophysical properties of nucleosomes. Nucleosome-destabilizing histone variants facilitate gene expression, while nucleosome-stabilizing histone variants exert the opposite effect. The same is true for all other biological processes requiring access to the DNA template, such as DNA repair and replication. Additional complexity and regulatory capacity is introduced through the fact that histone variants can have specific binding partners and specific PTMs.^27.,28.

While many and diverse variants exist for histones H2A and H3, very few variants for H2B and H4 have been identified, with minimal variation and restricted expression patterns (Figure 1B). This suggests that a higher evolutionary pressure induced a greater molecular diversification of histones H2A and H3.^29. Several special features of H2A and H3 exist that could provide an explanation for this phenomenon. First, H2As and H3s engage in more contacts within the nucleosome, sharing interfaces with both their own dimers and other histone proteins. Second, both the N-terminus of H3 and the C-terminus of H2A are close to the DNA entry and exit site of the nucleosome, respectively, and this surface needs to be accessed by most factors involved in epigenetic regulation.^30.

Here, we discuss the current knowledge about the role of H1, H2A, and H3 histone variants in hematopoiesis and hematopoietic malignancies (Tables 1 and 2; Figure 2). For the role of histone variants in other contexts, we refer readers to reviews by the groups of Bernstein on solid cancers,^75. Banaszynski on molecular aspects and developmental syndromes,^76. Buschbeck and Hake on cancer, differentiation, and reprogramming,^24. and Skoultchi and Bai on linker histones.^77. More detailed information on specific histone variants and their functions can be found in a special review collection edited by Frederic Berger.^78.

H2A.X guards genome integrity in lymphocytes

H2A.X can be regarded as an intermediate between a replication-coupled histone and a histone variant.^80. It can be incorporated into chromatin in a replication-coupled manner, which explains its broad genome-wide distribution. The phosphorylation of H2A.X at Ser139 by the kinase ataxia-telangiectasia mutated is an early signal at sites of DNA damage, and frequently referred to as γ-H2A.X.^83. Additional modifications of H2A.X fine-tune the intensity and length of the DNA repair signal.^28.,82. Further replication-independent deposition of H2A.X at DNA damage sites by the facilitates chromatin transcription (FACT) complex potentiates the DNA damage signal and repair process.^84. Apart from DNA damage, several studies have reported that H2A.X can affect the transcription of particular sets of genes, both in a positive and negative manner that is still poorly understood.^80.

H2A.X knock out (KO) mice have compromised immune functions due to reduced number of lymphocytes.^35. These animals are prone to the development of T-cell lymphomas, which is accelerated in the absence of p53 and attributed to increased genomic instability.^36. Furthermore, γ-H2A.X has been extensively used to assess the induction of DNA damage by various therapeutic agents.^85. Additionally, it has been recently reported that the phosphorylation state of H2A.X is capable of controlling human stem and progenitor cell fate decisions, and that its pharmacological modulation could help to overcome differentiation blockade in human leukemia.^86.

H2A.Z is a highly regulated and polyvalent histone variant involved in gene transcription

H2A.Z was reported to have diverse and, to some extent, opposing functions in gene expression. Two independent genes and 1 event of alternative splicing give rise to 3 H2A.Z isoforms.^87.-90. The chromatin remodeling ATPase complexes p400 and SRCAP are responsible for the deposition of H2A.Z into chromatin,^91.,92. while ANP32E and INO80 mediate its removal.^93.,94. In vivo KO studies showed that H2A.Z is essential for early embryonic development in mammals.^38. Whether H2A.Z is required for hematopoiesis has not been tested in lineage-specific KO studies, but loss of p400 induces severe defects in hematopoietic development.^95.

H2A.Z destabilizes nucleosomes at promoters and enhancers to facilitate transcription. Yet, it was also shown to have an antagonistic role, as it can form repressive and stable nucleosomes (discussed in ref.^79.). The explanation for this paradox lies in the context: the consequence of H2A.Z deposition into chromatin is dependent on its PTMs, the presence of binding partners, and the composition of the nucleosome, and in particular, the presence of other histone variants. For instance, the combination H2A.Z with H3.3 at active promoters and enhancers positively regulates gene expression.^96.,97. Similarly, acetylation of H2A.Z by p300 or Tip60 promotes gene expression.^98.,99.

Until now, little is known about a possible role of H2A.Z in blood cancers, and the available evidence is indirect. The acetyltransferase Tip60 is required for embryonic and adult hematopoietic stem cell (HSC) maintenance and mixed lineage leukemia (MLL) fusion-driven leukemia.^55.,56. In both cases, Tip60 functions are associated with H2A.Z acetylation on key effector genes such as HOXA9 in leukemia. Conversely, in other fusion-protein driven leukemias, binding sites of AML::ETO and PML::RARα are characterized by low H2A.Z acetylation.^58. One study suggested that H2A.Z could have a higher affinity for atypical retinoids, thus linking DNA damage with histone variants in PML::RARα-positive leukemia cells.^57. In the context of a lymphoid cancer, a reduction of H2A.Z levels and its extensive genomic redistribution accompanied Myc-driven B-cell transformation.^59.

H2A.B is a nucleosome destabilizer involved in Hodgkin lymphoma

H2A.B is a short H2A isoform lacking a C-terminal tail. It appeared late in the evolution of eutherian mammals, is fast evolving, and its expression is restricted to the testis.^100. Incorporation of H2A.B leads to very unstable nucleosomes, thus promoting open chromatin structures and gene expression.^80. Testis-specific genes, including several of those encoding chromatin regulators, are frequently re-expressed in cancer.^101. In line with this, H2A.B was found to be upregulated in Hodgkin lymphoma cell lines and primary samples.^52.,53. H2A.B binds proliferating cell nuclear antigen (PCNA) and accelerates DNA replication, thereby leading to increased susceptibility for DNA damage.^54. In addition, H2A.B promotes protein synthesis by increasing the transcription of ribosomal RNAs and protein-coding mRNAs by RNA Pol I and II, respectively.^52. As H2A.B is most abundantly expressed in the testis, KO mice display male subfertility associated with abnormal sperm function.^37. A hematopoietic phenotype in these mice has not been reported.

Multidomain histone variants with a role in hematopoiesis

MacroH2A is unique among H2A histone variants, as it possesses a 60 amino acid long unstructured C-terminal linker followed by a globular macrodomain. Like H2A.Z, macroH2A exists in 3 forms that are encoded by 2 independent genes, and 1 event of alternative splicing. A splicing switch from macroH2A1.2 to macroH2A1.1 expression is frequently associated with cell differentiation.^102. In contrast to H2A.B, macroH2A is frequently linked to gene repression^103. and contributes to the 3-dimensional compaction of repressed heterochromatin.^104. The mechanisms regulating the genome-wide distribution of macroH2A are not well understood. However, it has been shown that the FACT complex removes macroH2A from transcribed regions, and ATRX is a negative regulator of macroH2A deposition on subtelomeric regions.^105.,106. In terms of biological function, a large number of studies on reprogramming, differentiation, and development have suggested that a major function of macroH2A is to promote and stabilize the differentiated state of cells.^24.

Mice lacking both macroH2A genes are viable, but display reduced pre and postnatal growth, which was attributed to altered expression of metabolic genes in the liver.^41. While the same study did not observe any phenotype in the hematopoietic system, deletion of the MACROH2A1 gene alone or only the alternatively-spliced exon encoding macroH2A1.1 caused reduced capacity of HSCs to differentiate into B and T cells in competitive transplantation assays.^39.,42. Interestingly, mutations in the splicing factor U2AF1 occur frequently in MDS, and cause a splice isoform switch from macroH2A1.1 to macroH2A1.2. This is functionally relevant, as reintroduction of macroH2A1.1 was sufficient to partially rescue defects in erythroid and B-cell differentiation in cells and mice bearing mutant U2AF1.^39.,60.,61. MacroH2A2 was shown to have a role in erythroid development, as downregulation of macroH2A2 expression caused reduced erythroid differentiation in a cell culture model.^62.

In patients, anecdotal evidence has linked macroH2A to disease etiology and risk. In a single AML case, macroH2A1 has been found as part of a fusion with the MECOM gene encoding the transcription factor EVI1.^65. MacroH2A1 is one of the many genes on chromosome 5q that is frequently lost in MDS.^42. In T-cell lymphoma and AML, macroH2A1 mRNA and macroH2A2 protein, respectively, are part of signatures that are associated with elevated risk and poor prognosis.^63.,64. Taken together, the knowledge about the involvement of H2A variants in hematopoietic malignancies until now is limited, but point to tumor-promoting roles of H2A.B, while macroH2A1.1 might act as a possible differentiation-promoting tumor suppressor.

Replication-coupled H3 variants and their chaperones in hematopoiesis and leukemogenesis

Because the canonical H3.1 variant plays an essential role in replication-coupled nucleosome assembly, its correct expression and localization is essential for a healthy gene expression profile. Indeed, profiling of histone mutations in primary human AML samples by next-generation sequencing (NGS) has identified K27M, K27I, and Q69H amino acid substitutions in patients with de novo AML and AML secondary to myelofibrosis, occurring due to mutations in genes encoding H3.1 (HIST1H3H, HIST1H3F, and HIST1H3A, respectively).^66. Moreover, the same study has found that H3.1 Q69H and K27M/I mutations alter HSC frequency and differentiation potential in vitro and in vivo. When examined for leukemic potential, it was established that H3.1 K27M/I mutations increase AML cell proliferation and disease aggressiveness.^66.

The chaperone CAF-1 for replication-coupled histone deposition is composed of 3 separate functional subunits, one of which is the p60 subunit named chromatin assembly factor 1B (CHAF1B).^119. Investigations in CHAF1B KO mice revealed that this chaperone subunit is required for normal hematopoiesis, as its loss resulted in the depletion of hematopoietic stem and progenitor cells (HSPCs) in the bone marrow (BM). The same study found that CHAF1B overexpression, however, resulted in much higher proliferation rates of HSCs and a leukemic phenotype. Furthermore, overexpression of CHAF1B in MLL::AF9 fusion oncoprotein-expressing cells drastically enhanced leukemia development in mice, while the deletion of this chaperone subunit induced differentiation of MLL::AF9 AML cells.^67. These results highlight the CAF-1 canonical histone chaperone acting as an important molecular switch, regulating normal hematopoiesis and leukemogenesis. Additionally, ASF1A, the other replication-coupled histone chaperone, was identified as an enhancer of differentiation arrest in chronic myeloid leukemia (CML) cells. Mechanistically, ASF1A was proposed to play an active role in initiating Notch signaling and upregulating H3K56 acetylation, which in turn resulted in CML acceleration and blast crisis.^68. Roles of H3.2 and CENP-A in regulating hematopoiesis or driving hematological malignancy have not yet been described in literature. Taken together, the above-mentioned studies suggest that H3.1 and its chaperones play a role in the maintenance of healthy hematopoiesis, and perturbations to their healthy state can drive or accelerate leukemias.

H3.3 and its chaperones regulate hematopoietic homeostasis

Several studies have identified the H3.3 variant and its chaperones to play a role in the fine-tuned regulation of hematopoiesis. Recently, Guo et al^43. reported that the double KO of the mouse orthologs of H3F3A and H3F3B resulted in reduced numbers of long-term HSCs, together with the expression of a granulocyte/monocyte progenitor-like transcriptional signature, increased bias toward the myeloid lineage over lymphopoiesis, splenomegaly, and a loss of terminally differentiated erythroid cells. The authors, thus, concluded that H3.3 maintains HSC stemness and restrains their myeloid bias. A complementary study accounted for the requirement for the H3.3-specific chaperone Hira in adult mouse HSCs self-renewal, and for the incorporation of H3.3 at genomic regions that are enriched for binding sites of hematopoietic transcription factors to restrain erythroid differentiation.^71. Additionally, in mouse embryonic stem cells (mESCs), Hira was identified to interact with Runx1, a transcription factor (TF) that is essential for the early hematopoietic specification from hemogenic endothelium.^120. Through this interaction, Hira was shown to modulate hematopoiesis by regulating Runx1 expression and localization through enhancing the enrichment of H3.3 within the intronic enhancer of Runx1.^70. Moreover, mice with Cre-mediated deletion of Hira in HSCs present with reduced numbers of BM HSCs, anemia, thrombocytopenia, and lymphocytopenia. Further analyses in this mouse model demonstrated that Hira is required for the expression of TFs and signaling molecules that are essential for HSC development and maintenance, and that Hira establishes a HSC-specific DNA accessibility signature, including Spib/PU.1 sites.^121. Indeed, Cre-mediated KO of the other H3.3-specific chaperone Daxx in mice further revealed that loss of Daxx changes the global genomic distribution of H3.3, as well as histone modifications in hematopoietic progenitors. These conditional KO mice showed an activation of PU.1-dependent transcriptional programs with biased myelopoiesis, causing neutrophilia and inflammation,^72. both of which are crucial drivers of oncogenesis.^122.

Perturbations in H3.3 and its chaperones as a driver of hematological malignancies

In addition to their roles in the regulation of normal hematopoiesis, the H3.3 variant and its chaperones have been found to be implicated in hematological malignancies. For instance, Boileau et al^66. have identified K27M and A26P mutations in the H3.3 variant-encoding H3F3A gene in patients with secondary AML and AML secondary to CMML. A study investigating the role of H3 mutations in leukemic transformation showed that the H3.3 K27M/I mutations drastically accelerated AML1::ETO-driven AML and decreased disease latency. However, this was not observed in mice with MLL::AF9-driven AML, hinting at context-dependent effects of H3.3 mutants in leukemic transformation.^73. Additionally, expression of HIRA was found to be upregulated in the CML cell line K562 and in the BM of patients with CML; and its downregulation in K562 cells induced cell cycle arrest, reduced proliferation, as well as differentiation into megakaryocytes.^69. Finally, deep NGS of cancer-associated genes in samples from pediatric acute lymphoblastic leukemia (ALL) patients identified ATRX as a novel cancer driver gene in B-cell precursor ALL patients.^74. Taken together, the H3.3 variant and its chaperones stand as an interesting candidate for further exploration of their roles in leukemogenesis and the regulation of normal hematopoiesis.