Histone modifiers are oxygen sensors

See allHide authors and affiliations

Science  15 Mar 2019:
Vol. 363, Issue 6432, pp. 1148-1149
DOI: 10.1126/science.aaw8373

Approximately 2.6 billion years ago, during the Proterozoic period, the evolution of photosynthesis in cyanobacteria led to the introduction of the by-product of this reaction, oxygen, into Earth's atmosphere (1). This great oxidative event heralded the rise of multicellular organisms, which are almost totally dependent on oxygen as an efficient fuel for metabolism and as a cofactor in many critical physiological enzymatic reactions. Central to this adaptation, and to allow cellular physiology across a wide range of oxygen concentrations (tensions), metazoans have evolved the highly conserved hypoxia-inducible factor (HIF) pathway (2). This is important for both physiological and pathological processes that occur in a hypoxic microenvironment, including embryogenesis, stem cell homeostasis, cancer, and cardiovascular disease. It has long been observed that hypoxia induces histone lysine hypermethylation, a form of epigenetic chromatin modification. However, whether this represents a direct sensing of oxygen tension or an indirect effect, perhaps through the HIF pathway, has not been established (3). On pages 1222 and 1217 of this issue, Batie et al. (4) and Chakraborty et al. (5), respectively, resolve this question, demonstrating in different cellular systems that the activity of the lysine-specific demethylases (KDMs) KDM5A and KDM6A is oxygen sensitive, and thereby identifying them as oxygen sensors.

In ambient normoxic conditions, HIF-1α, the DNA-binding component of the HIF heterodimeric transcription factor complex, is targeted for ubiquitylation and destruction. This occurs through hydroxylation on proline residues in HIF-1α by the EglN family of prolyl hydroxylases (PHDs), which are 2-oxoglutarate– and oxygen-dependent dioxygenase enzymes that sense physiological changes in oxygen tension and are activated in normoxia. However, in hypoxic conditions, PHD activity is lost and HIF-1α is stabilized so that it can bind to its partner ARNT (aryl hydrocarbon nuclear translocator; also called HIF-1β). The HIF complex translocates to the nucleus and induces hypoxia-specific gene expression programs that mediate altered cellular metabolism and survival through binding to specialized hypoxia response elements (HREs) in target gene promoters. The family of 2-oxoglutarate– and oxygen-dependent dioxygenases is large, with more than 60 members (6), and also includes the TET and JmjC (Jumonji-C) KDM families of epigenetic regulators.

Using biochemical analysis of recombinant proteins, Batie et al. and Chakraborty et al. have added KDM5A and KDM6A to the list of dioxygenases that have low oxygen affinities (KM values) comparable to those of the EglN PHD family. Batie et al. also used time course experiments to show that histone methylation changes after the induction of hypoxia were rapid and preceded subsequent transcriptional events. Using cellular systems expressing loss-of-function and gain-of-function mutant proteins in the HIF pathway and through documenting the speed of HIF-1α stabilization following induction of hypoxia, histone methylation changes were found to be independent of HIF as well as independent of other known hypoxia-inducible inhibitors of KDM activity, such as reactive oxygen species and 2-hydroxyglutarate.

Linking direct histone hypermethylation to cellular function, Chakraborty et al. demonstrated that hypermethylation of lysine 27 of histone H3 (H3K27), a histone change that is associated with gene repression, prevented differentiation in different cell line model systems. Conversely, these effects could be antagonized by inhibition of the reciprocal histone methyltransferase EZH2 (enhancer of zeste homolog 2). Batie et al. focused on histone methylation modifications associated with gene activation: trimethylation of lysine 36 of histone H3 (H3K36me3) and, in particular, trimethylation of lysine 4 of histone H3 (H3K4me3). They identified KDM5A as responsible for the hypermethylation of H3K4 in their HeLa human cervical cancer cell line system and linked H3K4me3 to the induction of enhancer (long-range promoter of transcription) activity and both HIF-dependent and HIF-independent promoter function. Both studies report preliminary data regarding the structural basis of differences in oxygen affinity between KDMs.

Hypoxia-mediated alterations in transcription

Under normoxic conditions, the hypoxia-inducible factor 1α (HIF-1α) subunit is targeted for destruction by prolyl hydroxylases (PHDs) and ubiquitylation (Ub) by the von Hippel–Lindau tumor suppressor protein (VHL). In hypoxic conditions, PHDs are inactivated, allowing HIF-1α to dimerize with ARNT, translocate to the nucleus, and activate HIF target genes. The lysine demethylases KDM6A and KDM5A are also direct oxygen sensors that are inactivated during hypoxia. This allows hypermethylation of H3K27 (KDM6A target) and gene repression, as well as hypermethylation of H3K4 (KDM5A target) and gene activation. P, phosphorylation; OH, hydroxy


These two complementary studies further clarify the regulatory mechanisms of histone demethylases, and specifically how they directly—rather than through the HIF pathway or via the influence of metabolic intermediates—coordinate a range of epigenetic alterations, transcriptional outputs, and cell fate decisions in response to external environmental changes (see the figure). However, these studies raise as many important questions as they answer. The alteration of a large number of histone modifications evident in the multiplexed mass spectrometric assay upon hypoxia and described in the literature (4, 7, 8) suggests that the full identity of oxygen-sensitive JmjC KDMs is not yet known. For example, the increased H3K36me3 in response to hypoxia is not linked to loss of KDM5A or KDM6A activity, and so the responsible KDM needs to be identified. Further studies are therefore warranted, with this knowledge not only informing biology but also facilitating elucidation of the structural basis of oxygen affinity in KDMs.

Additionally, the HIF pathway has been implicated in an array of physiological and pathological cellular processes, and it is likely that the direct oxygen-sensing KDM pathways are similarly implicated in a number of these processes. However, the exact nature of the pathways involved, and whether they function independently and/or cooperatively with HIF-mediated transcriptional programs, remains to be determined. Malignancies often develop in and/or metastasize to hypoxic environments, and activation of the HIF pathway is frequently observed in cancer (9, 10). Moreover, multiple mutations of epigenetic regulators are described in malignancies, and loss-of-function mutations in many histone methyltransferases and KDMs (11, 12), potentially mimicking hypoxia, have been observed. Moreover, therapeutics that target chromatin modifiers are being evaluated for treating certain cancers in clinical trials (13). We speculate that the direct oxygen-sensing KDM pathways are also aberrant in malignancy and other pathological states, such as cardiovascular disease. Targeting the HIF pathway with small-molecule inhibitors is currently being explored in cancer therapy and in the treatment of nonmalignant conditions such as renal disease. These studies suggest that oxygen sensing by KDMs might also be therapeutically targeted, if the mechanistic basis for this sensing can be determined.

Interestingly, Batie et al. and Chakraborty et al. speculate, on the basis of phylogenetic sequence conservation, that the direct oxygen-sensing KDM pathways may evolutionarily predate the HIF pathway. It is likely that both pathways have more recently coevolved and function in a coordinated and temporally defined manner to modulate the cellular response to low oxygen tensions. This is suggested by the immediate hypermethylation of H3K4 at the promoters of HIF target genes. However, further delineating the interaction between these pathways and how this might be modulated will be important to understand. Together, these observations have profound implications for our understanding of how microenvironmental changes, and specifically oxygen concentrations, might affect both physiological and pathological cell fate decisions and phenotypes through direct effects on chromatin structure.

References and Notes

Acknowledgments: Research in the Huntly lab is funded by an ERC Consolidator award (COMAL–647685), a CRUK program grant (C18680/A25508), MRC, Bloodwise, and KKLF. P.G. is funded by a CRUK Advanced Clinician Scientist Fellowship (C57799/A27964) and an ASH Global Research Award.
View Abstract

Stay Connected to Science

Navigate This Article