Metazoans organisms utilize molecular oxygen during cellular respiration to produce sufficient amounts of ATP, as continuous and proper oxygen supply is an absolute requirement for survival. In response to hypoxia, defined as the reduction of oxygen levels below normal values, complex mechanisms are triggered to allow adaptation to this environmental condition. These adaptive responses comprise changes at the systemic, tissue and cellular levels that tend to reduce oxygen consumption, and improve oxygen supply. Cardinal adaptations to hypoxia include angiogenesis, erythropoiesis, a switch from oxidative to glycolytic metabolism, and a partial shut-down of major energy consuming cellular processes [1–3].

Remarkably, hypoxic conditions take place during normal physiological processes, being a necessary stimulus for embryonic development and stem cell maintenance [4–7]. During different human pathological conditions, hypoxia also plays a major role. The pathologies include cancer, stroke, myocardium infarction and chronic kidney disease being the most relevant ones [3,8–11].

The cellular response to hypoxia is largely dependent on changes in gene expression, which are mainly commanded by a unique family of transcription factors named HIFs, for Hypoxia Inducible Factors [12–13]. HIFs are heterodimeric transcription factors composed of a constitutively expressed HIF-1β subunit and one of three oxygen-sensitive alpha subunits (HIF-1α, HIF-2α or HIF-3α) [14–15]. Under normal oxygen levels (normoxia), HIF-α subunits are hydroxylated in two key prolyl residues by specific prolyl-hydroxylases named PHD1, PHD2 and PHD3 [16–18]. These enzymes utilize molecular oxygen and 2-oxoglutarate as co-substrates for the reaction [18]. Hydroxylated HIF-αs are recognized by the von Hippel-Lindau protein (VHL), the substrate recognition subunit of an E3 ubiquitin ligase complex; HIF-αs are then ubiquitinated and degraded at the 26S proteasome [19–21]. Under hypoxia, when oxygen is limiting, PHD activity is reduced, HIF-α subunits escape hydroxylation and proteolysis, leading to heterodimerization with the β-subunit and induction of hundreds of target genes through the binding of specific DNA sequences termed HREs (for Hypoxia Response Elements) [14]. Whereas HIF target genes are upregulated under oxygen deprivation, general transcription in the cell is largely inhibited [22].

Eukaryotic DNA is complexed with different histone and non-histone proteins to form the chromatin [23–24]. Histones play an essential role in DNA compaction as well as in the regulation of all DNA-related processes, including transcription, replication and DNA repair. The structural and functional unit of the chromatin is the nucleosome, which consists of an octamer of two molecules of each of the histones H2A, H2B, H3 and H4, around which 147 base pairs of DNA are wrapped [23–24]. Nucleosomes are connected by linker DNA which is associated with linker histones, usually H1 and H5 [24]. Nucleosomes together with the linker DNA are progressively folded and compacted into structures of higher-order [25–27]. Highly compacted chromatin has an inhibitory effect on transcription, since it limits the accessibility of the transcriptional machinery to the DNA [26]. All nucleosomal histones have a globular domain that forms the nuclesome core, and an N-terminal tail that protrudes away from the DNA [23]. Histone tails are subjected to multiple posttranslational modifications, including acetylation, methylation, phosphorylation and ubiquitination [28–29]. These histone marks have effects on both chromatin compaction and recruitment of different proteins that regulate transcription [30–31].

Besides the posttranslational modifications of the histone tails, chromatin can also be modified by various ATP-dependent chromatin remodeling complexes, which change the accessibility of the transcription machinery to particular loci, by promoting nucleosome sliding, nucleosome exchange or DNA exposure [32–34].

Alteration of the gene expression profile is a common cellular response to different types of stress. Changes in gene expression under stress are intimately associated with alterations in chromatin structure, mediated by histone modifying and chromatin remodeling complexes [35]. In this review we will discuss the chromatin alterations that take place under hypoxia and the mechanisms by which they regulate gene expression to restore homeostasis under this condition.

In order for transcription to take place, an initiation complex conformed by the RNA Polymerase II and general transcription factors must overcome the physical barrier posed by the chromatin. It has been demonstrated that HIF recruits co-activators that modify the chromatin structure thereby facilitating access of the transcriptional machinery to the DNA. These so called HIF co-activators include the histone acetyltransferases (HAT) p300 and CREB-binding protein (CBP) [36–40], some histone deacetylases (HDACs) [41–42] and the chromatin remodeling complex SWI/SNF [43].

Histones can be methylated in specific lysine (K) or arginine (R) residues, which can appear as mono-, di-, or tri-methylated forms. Methylated histone residues induce alterations in compaction of the chromatin, and also provide binding sites for different proteins that regulate gene expression, resulting in the inhibition or enhancement of transcription [84–87]. Although some exceptions have been reported [30,88], in general, methylation at lysines 4, 36 or 79 of H3 are hallmarks of chromatin active regions, whereas methylation of lysines 9 and 27 of H3, as well as of H4 lysine 20 are associated with transcriptional repression and heterochromatin formation [29,86]. Work carried out over the last ten years has shown that histone methylation is a tightly regulated process, dependent on the activity of both histone methyltransferases and histone demethylases (HDMs) [29,86].

The Jumonji C (JmjC)-domain containing histone demethylases (JHDM) constitute the largest family of lysine demethylases. JHDM family members utilize molecular oxygen (O₂), Fe(II) and 2-oxoglutarate to remove methyl groups from specific histone residues through an hydroxylation reaction. Several lines of evidence have linked JHMD function to hypoxia [64,89–90].

As JHDMs utilize O₂ as a co-substrate in the demethylation reaction, one could anticipate that their enzymatic activity is compromised under hypoxia. In agreement with this, it was demonstrated that the activity of over-expressed JMJD1A and JMJD2B was diminished, although not completely abrogated, when cells were exposed to 0.2% O₂ [91]. Interestingly, most members of the JDHM family have been reported to be transcriptionally induced under hypoxic conditions [64,89,91–93]. It was therefore proposed that JHDM induction under hypoxia tends to restore histone methylation homeostasis, in such a way that when oxygen availability is limiting, by increasing the expression of these enzymes, reduction of enzymatic activity due to oxygen scarcity is totally or partially compensated [89]. In agreement with this model, 17 out of the 22 JHDM family members are induced under hypoxic conditions [89], and at least 4 of them, JMJD1A, JMJD2B, JMJD2C and JARID1B, are HIF direct targets, as evidenced by chromatin immunoprecipitation (ChIP) assays [64,89,91–93].

Results reported by Xia et al. [89] support a role of a JHDM named JARID1B in histone methylation homeostasis: They have shown that under hypoxia, genome-wide levels of tri-methyl lysine 4 of histone H3 (H3K4me3) are increased, and that this increase is enhanced if cells are additionally depleted from HIF-β. As the effect on histone methylation elicited by the depletion of HIF can be over-came by JARID1B overexpression, it was proposed that JARID1B HIF-dependent induction is necessary to prevent H3K4 hypermethylation under hypoxia. Considering that some JHDMs, including JMJD1A and JMJD2B, are induced by hypoxia to a higher extent than other members of the family [89,91–92], and taking into account that JHDMs display residue specificity, hypermethylation of some histone lysines under hypoxia could be specially prevented.

It has been proposed that in addition to a possible role of the JHDMs in global histone methylation homeostasis under hypoxia [89], some family members might play a direct role in specific gene expression [93]. In fact, Krieg and colleagues [64] have recently demonstrated that JMJD1A is necessary for full hypoxic induction of 53 genes, including several HIF targets such as ADM, EDN1, SERPINE1, PLAUR and HMOX1. Moreover, they have shown that JMJD1A contributes to hypoxia-dependent gene expression by demethylating H3K9me2 in certain hypoxia responsive promoters, such as ADM and GDF15 [64] (Figure 1A, lower panel). Notably, JMJD1A recruitment reduces, although it does not completely prevent, hypoxic induction of H3K9me2 on selected promoters [64].

It was reported that hypoxia induces global histone methylation possibly through partial inhibition of JHDMs [22,65,67], while it reduces histone methylation of certain hypoxia-responsive promoters [22]. Tausendschön et al. have shown that in macrophages, hypoxia provokes an increase in H3K9 di- and tri-methylation in promoter regions of several hypoxia-repressed genes, while hypoxia has no effect on these modifications in the promoter region of the hypoxia-inducible gene ADM [68].Gene-specific recruitment of JMJD1A (and probably other JHDMs) can explain this dual effect. It remains to be determined the molecular mechanisms that target JMJD1A to specific promoters.

Interestingly, JMJD1A is a HIF target gene itself. Krieg and colleagues [64] suggested that regulation of JMJD1A by HIF may represent a feed-forward mechanism for favoring HIF-dependent gene expression. They proposed that JMJD1A maintains an active epigenetic pattern in target promoters, thereby minimizing the energy required to support expression.

The specific regulation of just a subset but not all hypoxia-inducible genes by JMJD1A suggests that additional promoter-specific mechanisms account for differential gene induction. It is tempting to speculate that other JHDMs are also involved in the activation of specific hypoxia responsive genes. Further research is required to determine if this is indeed the case.

Although progress has been made in characterizing JHDM induction under hypoxia, detailed studies on the biological roles of this family of enzymes in the response to hypoxia needs to be carried out. RNAi silencing experiments targeting individual or multiple JHDMs, and posterior analysis of global as well as promoter-specific histone methylation will help to clarify the role of these enzymes in hypoxia-dependent regulation of transcription.

Johnson et al. [22] analyzed hypoxia-induced epigenetic changes in the promoter regions of genes activated or repressed under oxygen deprivation. Hypoxic induction of the HIF target genes VEGF and EGR1 correlates with a marked increase in H3K9ac and H3K4me3 levels and a decrease in H3K27me3 levels in their promoters, three events usually associated with transcriptional active loci (Figure 1A, lower panel). The same or similar epigenetic changes have also been observed at the promoters of the hypoxia-inducible genes EPO, HMOX1 and DAF [43,62,65] (Table 1). Hypoxic induction of H3K4me3 seems to rely on the inhibition of the JARID1A demethylase [65].

In contrast, promoters of several hypoxia-repressed genes show decreased levels of H3K9ac [22,66] and/or increased levels of H3K9me2 [22,66–68] under hypoxia, events typically associated with transcriptional repression (Table 1). Available evidence suggests that H3K9me2 increase is supported in part by the hypoxic-dependent induction of the G9a methyltransferase [67] (Figure 1C).

Of interest, hypoxia also increased H3K9me2 levels at the promoters of some genes induced in hypoxia [64] (Table 1), as well as it provokes modifications typically associated with transcriptional activation (increased H3K4me3 and decreased H3K27me3 levels) at the promoters of some hypoxiarepressed genes [22] (Figure 1C and Table 1). These observations are not paradoxical, since patterns of histone modifications can be differentially interpreted by transcriptional factors and co-factors, depending on the epigenetic and cellular contexts [88]. Then, a so called “positive” histone mark can have repressive effects in particular physiologic situations, or vice versa, a typical “negative” modification can favor transcription in a certain physiologic context. Further research is required to assess the biological outcome of hypoxia-induced epigenetic changes.

In addition to promoter-specific chromatin alterations, cells exposed to hypoxia show global epigenetic changes [22,65–68,89,94–95]. Several groups have shown that the global acetylation and methylation profiles of H3 and H4 [22,67,94] and also methylation of the DNA (recently reviewed in [96]) are oxygen sensitive. Remarkably, among the hypoxia-induced global epigenetic alterations, histone posttranslational modifications usually associated to either transcriptional repression or activation have been reported [22,65–68,89,94–95] (Table 2).

It has been shown that hypoxia provokes general reduction of gene transcription [22]. However, it remains to be determined to what extent global epigenetic alterations account for this global gene downregulation in situations of oxygen shortage. Johnson and colleagues [22] have determined that HePa 1–6 cells incubated in strong hypoxia (0.2% O₂ for 48 h) reduce their overall rate of transcription to about 50% of their usual normoxic levels. This sharp reduction in mRNA synthesis correlates with several histone modifications, that are not only associated with transcriptional repression, but also with transcriptional induction. Again, as a particular histone mark can promote different, even opposite biological effects, it cannot be ruled out that these “positive” histone marks globally induced under oxygen deprivation might have negative effects in the particular physiological and epigenetic contexts that take place under this condition.

An increase in di- and tri-methylation of H3K9, two typically repressive marks, are among the best characterized global modifications induced under hypoxia [22,66–68,94,97] (Table 2). Chen et al. [67] have demonstrated that the hypoxia-dependent increase in H3K9me2 stems from both increased methylation and decreased demethylation of H3K9. These authors have shown that at 0.2% O₂ the activity of the methyltransferase G9a increases, supporting in part the accumulation of methylated species. In addition, they have shown that demethylation of H3K9me2 is reduced under hypoxia, although the identity of the particular demethylases whose activity diminishes is unknown. As has been mentioned above, JDHM activity is partially inhibited at 0.2% O₂, so one possibility is that reduction of H3K9me2 demethylation is a consequence of partial loss of activity of one or more JHDM.

Remarkably, deacetylation of H3K9, another chromatin modification associated with transcriptional repression, globally increases under oxygen deprivation conditions [22,66–67,89,94]. Thus, taken together, these observations suggest that in hypoxia H3K9 acetyl groups are globally replaced by di- or tri- methyl groups establishing a mechanism that might account for the overall transcriptional repression seen in response to oxygen deprivation (Table 2).

In hypoxia, global induction of H3K4 di- or tri- methylation, both of them modifications usually associated with transcriptional activation, have also been well documented [22,66–68,89,94] (Table 2). Experiments carried out by Zhou et al. [65] suggest that hypoxia provokes an increase in H3K4me3 by inhibiting JARID1A demethylase activity, rather than by inducing methylation.

In summary, oxygen deprivation induces a widespread combination of histone modifications that are usually associated with either transcriptional repression or transcriptional activation. The field is now beginning to define more precisely these modifications, although more research is required to fully understand their biological roles, as well as to identify the enzymes and signaling pathways involved. It will be highly relevant to assess the impact of preventing the global epigenetic changes that occur under hypoxia, over general transcriptional repression and specific gene induction.

In order to activate gene transcription, HIF recruits a range of gene-specific co-factors that acetylate/ormethylate histones, or change chromatin structure. It would be very interesting to define the mechanisms that recruit co-factors to specific hypoxia-inducible genes. On the other hand, the hypoxic stimulus induces a general repression of transcription, presumably supported by global chromatin changes. Therefore, it seems that epigenetic mechanisms play a dual role in gene regulation under hypoxia, controlling HIF target gene induction and the downregulation of general transcription.