Expansion and remodelling of the cerebral vasculature via the induction of pro‐angiogenic genes is a crucial issue in many pathologies. Urrutia et al¹ demonstrate how NG2‐derived cells in the brain sense oxygen to promote neurovascular angiogenesis.

Master regulator of cellular adaptation to hypoxia, which includes angiogenesis, is the Hypoxia inducible factor (HIF). Three isoforms exist of the cellular oxygen sensor: HIF prolyl hydroxylases (PHD)‐1,‐2 and ‐3 (EGLN‐2, ‐1, ‐3, respectively), enzymes that hydroxylate prolyl residues of the regulatory subunit HIF‐α,^2-4 which, thus tagged, is easily captured by the von Hippel Lindau protein (VHL),⁵ and subsequently degraded by the proteasome (review in Ref.6). PHDs are ubiquitously expressed, as is the HIF system. Highly specific, synthetic, orally available, small molecule PHD‐inhibitors are being tested in large phase III clinical trials for treatment of renal anaemia (review in Ref.7). Historically, erythropoietin (EPO) led to the discovery of HIF⁸ and PHD,^{2, 3} while in the future, PHD‐inhibitors may complement or even replace EPO. PHD inhibitors are potent erythropoiesis stimulating agents, comparable to EPO and its related compounds.⁹ Moreover, preclinical data suggest that PHD inhibition can improve tissue responses to hypoxia, and this may ameliorate highly relevant clinical conditions like ischemic stroke (review in Ref.10), myocardial infarction (review in Ref.11), acute kidney injury (review in Ref.12) etc Major outcome data from the above mentioned clinical trials are pending. In the meantime, it's worth taking a closer look at PHD inhibition in basic science, as it likely will help us understand upcoming clinical data. In this regard, Urrutia et al¹ present a meticulous assessment of brain angiogenesis, a highly relevant topic in ischemic and degenerative central nervous disorders.¹³ With help of knockout technology ranging from single to quadruple, they dissect the contributions of PHD‐1 to ‐3 and EPO in this process. To the best of our knowledge, this is the most thoroughly conducted study addressing this topic. But, before we review their results, we need a little closer look at the HIF‐PHD system: The transcription factor HIF is a heterodimer of an oxygen‐dependent, variable alpha subunit (HIF‐1α, ‐2α or ‐3α) and a constitutive beta subunit (HIF‐β, ARNT). HIF‐1α and ‐2α, the most important and best studied alpha subunits, each have two oxygen dependent degradation domains (ODD), one at the C‐terminus (CODD), and another at the N‐terminus (NODD) (review in Ref.6). PHDs are dioxygenases using molecular O₂ as substrate, and 2‐oxoglutarate as co‐substrate. Their KmO₂ is above normal arterial O₂ concentration.¹⁴ PHD‐2 is considered the most important isoform, responsible for HIF‐α degradation in normoxia. PHD‐3 and ‐2 are HIF target genes, and hence, as a negative feedback loop, help terminate the HIF response engendered by hypoxia. The four clinically tested PHD inhibitors roxadustat, vadadustat, daprodustat and molidustat are highly specific, but differ in kinetics and their ability to target HIF‐1α vs ‐2α, and the respective CODD vs NODD.¹⁵ Hence, at least to a certain extent, the four compounds may exhibit different clinical effects. As a transcription factor, HIF acts through its target genes, the precise number of which remains unclear. Confusingly, the consensus DNA binding motif for HIF, the so‐called hypoxia response element (HRE; RCGTG, with R = A or R = G) appears more than one million times in the genome.¹⁶ Clearly, most of these potential HREs are not accessible to HIF, but covered by chromatin. To make things more complex, some HREs of proven relevance locate quite far from the transcription initiation site. Furthermore, for full gene transactivation, binding of HIF to a single HRE seems insufficient, but rather requires binding to multiple HREs, or contribution from additional co‐factors. Chromatin binding assays suggest that both HIF‐1α and HIF‐2α regulate some hundreds of genes, with partial overlap.¹⁷ Hypoxia response, thus, varies between underlying conditions, tissues and even individual cells. Altogether, interventions into the HIF‐PHD system, whether pharmacological or genetical, are nearly unpredictable, which brings us back to the in vivo studies of Urrutia et al¹: They chose NG2‐Cre mice to specifically knockout genes of interest in central nervous cells with particular relevance to angiogenesis, namely pericytes and NG2 glia cells. VHL knockout, a robust model for combined HIF‐1α and HIF‐2α up‐regulation, led to enhanced angiogenesis and EPO. In the past, the latter has been attributed angiogenic effects in the brain, but a plethora of clinical trials (review in Ref.18) proved that exogenous EPO was unable to ameliorate brain damage. By double knockout (VHL/EPO) Urrutia et al¹ show that the angiogenic phenotype is independent of EPO. Interestingly, knockout of PHD‐2, which carries the workload of HIF degradation in normoxia, did not lead to angiogenesis while double knockout (PHD‐2/PHD‐3) did. The authors speculate that PHD‐3, as a HIF target gene, had compensated for the single PHD‐2 knockout. Knockout of PHD‐1 alone had no angiogenic effect, most likely since preserved PHD‐2 activity was sufficient to degrade HIF. However, combined knockout of all three PHD isoforms further enhanced angiogenesis with respect to the PHD‐2/PHD‐3 knockout. Hence, as the authors put it, each PHD isoform has its contribution to neurovascular homeostasis. And, finally, quadruple knockouts (PHD‐1/PHD‐2/PHD‐3/HIF‐2α) proved that the angiogenic phenotype was fully dependent on HIF‐2α.

Aside from sophisticated and refined methodology, the study by Urrutia et al¹ broadens our vision of the HIF‐PHD components working together to shape a potentially relevant clinical effect. It seems that, indeed, cells make use of their triple oxygen sensor.

在许多病理学中，通过促血管生成基因的诱导来扩大和重建脑血管系统是至关重要的问题。Urrutia等人¹证明了大脑中NG2衍生的细胞如何感觉到氧气以促进神经血管血管生成。

缺氧诱导因子（HIF）是细胞适应缺氧（包括血管生成）的主要调节剂。细胞氧传感器存在三种同工型：HIF脯氨酰羟化酶（PHD）-1，-2和-3（分别为EGLN-2，-1，-3），使调节亚基HIF-α的脯氨酰残基羟化的酶，^2-4，其因此被标记，很容易被von Hippel Lindau蛋白（VHL）⁵捕获，随后被蛋白酶体降解（参考文献6）。PHD和HIF系统一样被普遍表达。高特异性，合成的，口服的小分子PHD抑制剂正在大型III期临床试验中用于治疗肾性贫血（参考文献7的综述））。从历史上看，促红细胞生成素（EPO）导致发现了HIF ⁸和PHD，^2、3，而在将来，PHD抑制剂可能会补充甚至替代EPO。PHD抑制剂是有效的促红细胞生成剂，与EPO及其相关化合物相当。⁹此外，临床前数据表明，PHD抑制可改善组织对缺氧的反应，这可能会改善高度相关的临床状况，例如缺血性中风（参见参考文献10），心肌梗塞（参见参考文献11），急性肾损伤（参考文献11）。在参考文献12中）等上述临床试验的主要结果数据尚待确认。同时，值得深入研究基础科学中的PHD抑制作用，因为它可能有助于我们了解即将到来的临床数据。在这方面，Urrutia等人¹对脑血管生成进行了仔细的评估，这是缺血性和退行性中枢神经疾病中高度相关的主题。¹³借助从单一到四倍的敲除技术，他们剖析了PHD-1至-3和EPO在此过程中的贡献。据我们所知，这是针对该主题进行的最彻底的研究。但是，在我们审查其结果之前，我们需要更仔细地研究HIF-PHD系统：转录因子HIF是氧依赖性可变α亚基（HIF-1α，-2α或-3α）和一个组成性β亚基（HIF-β，ARNT）。HIF-1α和-2α是最重要和研究得最好的α亚基，每个都有两个氧依赖性降解域（ODD），一个在C末端（CODD），另一个在N末端（NODD）（综述参考文献6）。PHD是使用分子O _2的双加氧酶作为底物，2-氧戊二酸酯作为底物。它们的KmO ₂高于正常动脉O ₂浓度。¹⁴ PHD-2被认为是最重要的同工型，在正常氧中导致HIF-α降解。PHD-3和-2是HIF的靶基因，因此，作为负反馈环，有助于终止缺氧引起的HIF反应。四种经过临床测试的PHD抑制剂roxadustat，vadadustat，daprodustat和molidustat具有高度特异性，但在动力学及其针对HIF-1α和-2α以及分别针对CODD和NODD的能力方面有所不同。¹⁵因此，至少在某种程度上，这四种化合物可能表现出不同的临床效果。作为转录因子，HIF通过其靶基因起作用，其确切数目尚不清楚。令人困惑的是，HIF的共有DNA结合基序，即所谓的缺氧反应元件（HRE； RCTGG，R = A或R = G）在基因组中出现了超过一百万次。¹⁶显然，这些潜在的HRE中的大多数都无法被HIF使用，而是被染色质覆盖。为了使事情变得更复杂，一些已证明具有相关性的HRE距离转录起始位点很远。此外，对于全基因反式激活，HIF与单个HRE的结合似乎不足，但需要与多个HRE结合或来自其他辅因子的贡献。染色质结合试验表明，HIF-1α和HIF-2α都可调节数百个基因，部分重叠。¹⁷缺氧反应因此在基础条件，组织甚至单个细胞之间变化。总之，无论是药理学还是遗传学，对HIF-PHD系统的干预几乎都是不可预测的，这使我们回到了Urrutia等人^1的体内研究中。他们选择了NG2-Cre小鼠来特异性敲除与血管生成特别相关的中枢神经细胞中感兴趣的基因，即周细胞和NG2胶质细胞。VHL基因敲除是HIF-1α和HIF-2α联合上调的稳健模型，可增强血管生成和EPO。过去，后者被归因于大脑中的血管生成作用，但是大量的临床试验（参考文献18的综述）证明，外源性EPO不能减轻脑损伤。通过双敲除（VHL / EPO）Urrutia等人¹表明血管生成表型独立于EPO。有趣的是，PHD-2的敲除携带了正常氧中HIF降解的工作量，但并未导致血管生成，而双重敲除（PHD-2 / PHD-3）却导致了血管生成。作者推测，作为HIF靶基因的PHD-3已补偿了单个PHD-2敲除。单独敲除PHD-1没有血管生成作用，这很可能是因为保留的PHD-2活性足以降解HIF。然而，相对于PHD-2 / PHD-3敲除，所有三种PHD亚型的组合敲除进一步增强了血管生成。因此，正如作者所说，每种PHD亚型都对神经血管稳态具有贡献。最后，四重敲除（PHD-1 / PHD-2 / PHD-3 /HIF-2α）证明血管生成表型完全依赖于HIF-2α。

除了先进和完善的方法外，Urrutia等人¹的研究拓宽了我们对HIF-PHD成分协同作用以形成潜在相关临床效果的视野。实际上，细胞似乎利用了其三重氧传感器。