What (H)IF isoform matters? A deubiquitinase can tune the hypoxic response

Context‐specific control mechanisms of hypoxia‐inducible transcription factors HIF‐1alpha and HIF‐2alpha in tumors exposed to oxygen shortage remain incompletely understood. In this issue, Zhang et al (2022) identify a deubiquitinase that differentially stabilizes HIF‐2alpha in stem‐like glioblastoma cells, suggesting potential implications for regulation of the hypoxic response in a wide array of tissues and cancers.

Recent work uncovers deubiquitinase USP33 as oxygen level‐dependent control factor for HIF‐2alpha in brain tumor stem‐like cells.

HIF‐1alpha and HIF‐2alpha are key transcription factors behind the lion’s share of gene expression changes induced at the cellular level in response to oxygen shortage (Lee et al, 2020). The HIF‐1beta subunit is consistently expressed, whereas the alpha subunits are rapidly degraded in an oxygen‐dependent process that involves hydroxylation by the prolyl hydroxylases PHD1‐3 and subsequent ubiquitination by the von Hippel Lindau (VHL) complex (Fig 1A) (Lee et al, 2020). The transcriptional activity of HIF is further regulated by factor inhibiting HIF‐1 (FIH‐1)‐mediated hydroxylation (Lee et al, 2020). While the HIF‐alpha isoforms share these basic mechanisms of oxygen‐dependent regulation, differences in HIF‐1alpha and HIF‐2alpha expression levels exist based on tissue and cell type, oxygen levels, and time under hypoxic conditions (Hu et al, 2003; Holmquist‐Mengelbier et al, 2006; Keith et al, 2011). Furthermore, dysregulation of the pathway can lead to HIF‐1alpha and/or HIF‐2alpha expression despite sufficient oxygenation in many cancers (Kim & Simon, 2022). In light of advances over the past decades deciphering distinct roles and target genes for HIF‐1alpha and HIF‐2alpha (Keith et al, 2011), it is clear that such differences can influence the nature and potency of the cellular response to hypoxia, leading to differential hypoxic phenotypes of, e.g., stem cells and their differentiated progeny in health and disease. The mechanisms underlying differential HIF expression and activity between tissues and cell types are incompletely understood, but can include cell‐type‐specific transcriptional co‐factors, differential HIF gene expression, or variations in expression and functionality of HIF regulatory genes in the PHD/VHL pathway.

Figure 1. Oxygen‐dependent regulation of HIF‐1alpha and HIF‐2alpha.

(A) When oxygen is abundant, PHD1‐3‐mediated hydroxylation targets HIF‐1alpha and HIF‐2alpha for recognition by the VHL complex. VHL‐mediated ubiquitination in turn targets both HIF‐alpha isoforms for proteasomal degradation. FIH‐1‐mediated hydroxylation can further inhibit any HIF transcriptional activity in the presence of oxygen. (B) Zhang et al (2022) found that under hypoxic conditions, USP33 can deubiquitinate HIF‐2alpha specifically, thereby saving it from oxygen‐dependent degradation. In intermediately oxygenated tissues, HIF2alpha, but not HIF1alpha, may thus bind HIF‐1beta and activate transcription of target genes. (C) Under severely hypoxic conditions, PHD/FIH‐1‐mediated hydroxylations cannot occur, and both HIFs are rapidly stabilized, and can interact with HIF‐1beta to activate downstream target genes.

Differential expression of HIF‐1alpha and HIF‐2alpha is known to contribute to tumor heterogeneity in gliomas. In particular, stem‐like and treatment‐resistant glioma cells were shown to preferentially upregulate HIF‐2alpha in response to hypoxia, whereas non‐stem‐like glioma cells display an overall less potent hypoxic response (Li et al, 2009). This HIF‐2alpha‐driven hypoxic response appears to at least partially control the aggressive and stem‐like state of these cells, and stem‐like cells were shown to be able to stabilize HIF‐2alpha even under relatively well‐oxygenated conditions (around 5% O₂, which likely represents the oxygen tension near end capillaries in the brain). In this issue, Zhang et al (2022) found that the ubiquitinase USP33 is preferentially expressed in stem‐like glioma cells, and that its expression is regulated by hypoxia. Intriguingly, like HIF‐2alpha but unlike other studied deubiquitinases, they found that induction of USP33 was evident even under moderate hypoxic conditions like 5% O₂. Inhibition of USP33 in stem‐like glioma cells abolished HIF‐2alpha stabilization in hypoxia, whereas overexpression of USP33 elevated HIF‐2alpha levels even under normoxic conditions. Together, these experiments support a role for USP33 in regulating HIF‐2alpha protein levels specifically, in a manner that is only partially dependent on oxygen levels. Indeed, the authors found that USP33 interacted directly with HIF‐2alpha under hypoxic conditions, and that USP33 inhibition dramatically increased polyubiquitination of HIF‐2alpha. From a mechanistic point of view, USP33 induction under hypoxia preceded that of HIF‐2alpha, and the interaction between the two was facilitated by polyubiquitination and ERK1/2‐mediated phosphorylation of S484 in the oxygen‐dependent degradation domain of HIF‐2alpha. ERK1/2 activation preceded that of HIFs under hypoxic conditions in these glioma cells. The findings by Zhang et al (2022) suggest that HIF‐2alpha stabilization can be controlled or fine‐tuned by activation of the ERK pathway, and may link aberrant signaling from receptor tyrosine kinases to enhanced HIF‐2alpha expression. Finally, in keeping with a role for HIF‐2alpha in controlling the aggressive stem‐like state of glioma stem cells, USP33 knockdown affected performance in a range of functional assays: Stem‐like glioma cells with USP33 knockdown displayed diminished self‐renewal in the sphere forming assay even under normoxic conditions, and importantly inhibited tumor growth in a murine xenograft model.

While the data from Zhang et al (2022) show increased USP33 protein levels in an oxygen‐dependent manner even under severely hypoxic conditions, the mechanism proposed would seem the most relevant at intermediate oxygen tensions such as the 5% O₂ used throughout their study, considering that severe hypoxia likely prevents most PHD‐dependent HIF hydroxylation and, thus, VHL‐mediated targeting for proteasomal degradation (Fig 1B and C). Stabilization of HIF‐2alpha at higher oxygen levels than HIF‐1alpha has been frequently described in cancer (Holmquist‐Mengelbier et al, 2006) and specifically in glioma (Li et al, 2009). Mechanistically, HIF‐2alpha activity at higher oxygen levels can be influenced by the greater affinity of FIH‐1 for HIF‐1alpha than HIF‐2alpha (Bracken et al, 2006; Lee et al, 2020); FIH‐1 remains active at lower oxygen tensions than the PHDs, suggesting that FIH‐1 can suppress HIF‐1alpha transcriptional activity at oxygen tensions that allow HIF‐2alpha to be active. Hydroxylation by FIH‐1 affects HIF binding to cofactors like CBP and p300, and additional proteins including proposed glioma stem cell markers CD44 and Notch1 have been shown to interact with HIFs specifically in a manner dependent on FIH‐1 (Johansson et al, 2017; Landor & Lendahl, 2017). Together with USP33‐dependent deubiquitination, these mechanisms provide a potential framework for explaining the unexpectedly high HIF‐2alpha activity in glioma stem cells as compared to more differentiated glioma cells. In the context of tumors and brain tumors in particular, it is intriguing to consider how HIFs can be specifically regulated in proposed stem cell niches such as the perivascular and perinecrotic niches that are thought to maintain stemness of resident tumor cells (Hambardzumyan & Bergers, 2015). HIF‐1alpha and HIF‐2alpha protein stainings of gliomas do not support a strictly oxygen‐dependent expression pattern, as HIF‐2alpha (but not HIF‐1alpha) has been found in both the well‐oxygenated perivascular niche and in the hypoxic perinecrotic regions, but not in the presumably intermediately oxygenated areas in between (Li et al, 2009; Johansson et al, 2017). While the present study by Zhang et al (2022) did not address HIF‐2alpha regulation in well‐oxygenated tumor areas, ERK1/2‐facilitated interactions between HIF‐2alpha and USP33 provide a potential mechanism for how HIF‐alpha levels can be controlled in specific microenvironmental niches such as the perivascular or perinecrotic niche, provided that niche cells can induce ERK1/2 activity in adjacent tumor cells by, e.g., activating receptor tyrosine kinase signaling in the tumor cell. Similarly, niche‐dependent activation or expression of HIF‐2alpha‐specific or FIH‐1 hydroxylation‐sensitive cofactors could potentially contribute to HIF‐2alpha stability and activity under oxygenated conditions.

While a number of HIF‐1alpha‐specific deubiquitinases have been previously described (Mennerich et al, 2019), the findings by Zhang et al (2022) are among the first to describe deubiquitination of HIF‐2alpha under hypoxia by a HIF‐2alpha‐specific deubiquitinase. Complementary and sometimes contradictory roles of HIF‐1alpha and HIF‐2alpha are likely to play important roles in fine‐tuning the hypoxic response of cells in many tissues, and dysregulation of the pathways controlling HIF stability may influence tumor biology and therapeutic responses of cancers. A better understanding of oxygen‐dependent as well as ‐independent regulation of individual HIF‐alpha isoforms will be crucial to achieve context‐relevant HIF‐targeting strategies, which has been a longstanding goal in cancer therapeutics as well as other therapeutic areas.

Disclosure and competing interests statement

The author declares that he has no conflict of interest.

The EMBO Journal (2022) 41: e110819.