Histone H3 variants at the root of metastasis

ABSTRACT

Chromatin remodeling is at the root of any cell fate decision, laying the foundation for the necessary reprogramming to occur. Our work shows histone H3 variants as a new addition to the ever-growing body of epigenetic regulators, one that is essential for cell fate transitions in carcinoma cells to promote tumor progression and metastasis.

Over the last decade it has become clear that histone variants confer a higher degree of chromatin complexity with their unique functions within the chromatin. Histone variants are more prominent in the H2A and H3 families and contain sequence and structural variations from their canonical counterparts that allow for distinct post-translational modifications and the recruitment of different chromatin-modifiers. Incorporation of canonical histone and histone variants are regulated through distinct spatial and temporal mechanisms further highlighting the additional layer of complexity that they provide to the chromatin structure.¹ In the recent years several reports have suggested that histone variants might be at play in complex diseases such as cancer.^2,3 Histone H3.3 (a variant of the H3 family) in particular has been shown to have oncogenic mutations that drive tumorigenesis in several types of cancer^4,5 thus leading to the establishment of the term “oncohistones”.

While at the protein level histone H3.3 and canonical histone H3 (also referred to as H3.1/H3.2) have only four amino acid substitutions, the mechanisms that govern their synthesis and incorporation onto chromatin are completely distinct. H3.1/H3.2 are encoded by multiple genes that are organized in an operon-like fashion and their transcription is coupled to DNA replication during S phase of cell cycle or during DNA repair. On the other hand, H3.3 is encoded by two genes whose transcription is independent of DNA replication and can occur throughout the cell cycle.⁶ In addition to the differences in temporal regulation, their incorporation into and eviction from nucleosomes is also regulated by different histone chaperones, likely contributing to the differential localization patterns of the H3 variants within the genome. Chromatin assembly factor 1 (CAF-1) promotes canonical histone H3 and histone H4 deposition onto the newly synthesized DNA, while other histone chaperones, histone cell cycle regulator (HIRA) and ATRX chromatin remodeler (ATRX)/death domain associated protein (DAXX), are involved in the deposition of the H3.3 at genic or telomeric regions, respectively.^7,8

Considering the fundamental differences between histone H3 variants and the profound implications that their differential incorporation onto chromatin can elicit for gene expression, we sought to understand their dynamics during tumor progression and their role in the transcriptional reprogramming necessary for acquisition of metastatic properties. We found that in several carcinoma cells in response to metastatic signals, such as transforming growth factor beta 1 (TGFβ) and tumor necrosis factor (TNF, or previously known as TNFα) treatment, the levels of chromatin-associated canonical histones (H3, H4 and H2A) were severely reduced thus rendering a more accessible chromatin throughout the genome.⁹ Moreover, and contrary to the general decline in canonical histones, chromatin-bound levels of H3.3 were significantly upregulated suggesting that H3.3 was “gap-filling” naked DNA to maintain chromatin integrity as it has been previously shown in other models of canonical histone decline.¹⁰ Surprisingly, we discovered that the enrichment of H3.3 was not random as it has been previously suggested, but instead we found it to be targeted to specific locations in the genome. Specifically, we observed that upon metastatic signaling or in metastasis-derived cells, H3.3 localizes primarily to the promoters of many pro-aggressive genes including many pro-aggressive transcription factors that regulate key genetic programs necessary for tumor progression and metastasis formation such as snail family transcriptional repressor 1 (SNAI1), zinc finger E-box binding homeobox 1 (ZEB1), SRY-box transcription factor 4 and 9 (SOX4 and SOX9), and Kruppel like factor 9 (KLF9) (Figure 1). Consistently with the idea that H3.3 acts as the effector of metastatic signals to endow cancer cells with metastatic potential, knockdown of H3.3 or its histone chaperone HIRA completely prevented the acquisition of metastatic properties.⁹

Figure 1.

Histone H3.3 is essential for tumor progression and metastasis formation. During tumor progression, metastatic signals induce a mitogen activated kinase 1 (MAPK1, commonly known as ERK2)-mediated suppression of chromatin assembly factor 1 (CAF-1) complex. Suppression of CAF-1 causes a reduction in canonical histone H3.1/H3.2 incorporation into the chromatin while triggering specific deposition of H3.3 in a histone cell cycle regulator (HIRA)-dependent manner. Consequently, H3.3 enrichment in the promoters of pro-aggressive transcription factors leads to a transcriptional reprogramming that lays the foundation for the acquisition of aggressive and pro-metastatic traits necessary for a primary tumor to evolve into a metastatic tumor. (TGFβ: transforming growth factor beta 1, TNFα: tumor necrosis factor (current nomenclature is TNF), P: phosphorylation indicating activation of ERK2, TF: transcription factor).

Compelled by these observations, we went on to understand how this histone H3 switch was regulated. Our results showed that a reduction of the canonical H3 chaperone complex CAF-1 and an increase in the H3.3 chaperone HIRA were observed in response to metastatic stimuli.⁹ Highlighting the complexity of this regulatory circuit, we went on to determine that upregulation of H3.3 or HIRA alone, was not able to induce metastatic properties in carcinoma cells. Rather, we found that reduction of canonical histone incorporation due to suppression of the CAF-1 complex was necessary to trigger the specific enrichment of H3.3 at the pro-metastatic transcription factors and consequently promote metastasis. As a consequence of this mechanism, we showed that the ability of triple negative cancer cells to form metastatic colonies in xenograft models is a function of CAF-1 levels and validated these observations in matched patient samples from primary and metastatic sites. Lastly, we provided a mechanism of CAF-1 regulation by metastatic signaling that impinges on CAF-1 through the regulation of the p60 subunit (CHAF1B) promoter by a mitogen activated kinase 1 (MAPK1, commonly known as ERK2)-mediated early growth response 1 (EGR1)/SP1 transcription factor (SP1) mechanism of facilitated inhibition.⁹

Together our work identifies the CAF-1 complex as an important node of signaling integration during tumorigenesis that functions as a barrier to determine whether a primary carcinoma remains localized or progresses into a metastatic state, and highlights the possibility of targeting H3.3 deposition as an efficacious and much needed new therapeutic avenue for metastatic carcinomas (Figure 1). Taking into consideration the crucial role of the H3 chaperones and the histone H3 switch for cancer progression, an important line of questioning emerges; what defines the specificity of H3.3 incorporation at the promoters of metastasis-inducing transcription factors upon CAF-1 suppression? What are the differences brought onto the chromatin by H3.3 that mediate the transcriptional activation of said transcription factors? Furthermore, is there a role for H3.3 in other components of the tumor microenvironment? The answers to these questions remain unknown but are essential for our understanding of the epigenetic bottlenecks that condition the various tumor stages, which may indeed prove vital for advances in precision medicine and the treatment of cancer.

Funding Statement

A.P.G is supported by a Susan G. Komen Postdoctoral fellowship [PDF17481555] and a Pathway to Independence Award [K99CA218686]. The National Institutes of Health [RO1: GM051405], National Institutes of Health [RO3: CA212562] and National Institutes of Health (US) [RO1: CA046595] grants provide reseach support for the Blenis laboratory.

Acknowledgments

We apologize to those whose work was not discussed and cited in this article due to limitations in space and scope.

Disclosure of Potential Conflicts of Interest

No potential conflicts of interest were disclosed.