Deficient CDKN2B expression: A Double Hit for PAD

In 2007, multiple independent groups, utilizing genome wide association studies (GWASs), made a significant breakthrough in the field of cardiovascular genetics when they identified a coronary artery disease (CAD) risk locus on chromosome 9p21 (Chr9p21). Subsequently, additional studies examining a wide variety of patient cohorts have linked single nucleotide polymorphisms (SNPs) within Chr9p21 to several cardiovascular diseases including myocardial infarction (MI), stroke, aneurysms, and peripheral artery disease (PAD).¹ Importantly and perhaps unexpectedly, genetic variation at Chr9p21 has been shown to modify PAD risk independent of conventional atherosclerotic risk factors or preexistent MI.^2,3 While rates of CAD and first MI may be decreasing,⁴ PAD is showing no evidence of a decline, and with roughly 1/5 of the population carrying two copies of the 9p21 risk allele,⁵ a role for this locus in PAD pathogenesis is quite intriguing.

Enthusiasm over Chr9p21 was tempered by the reality that the roughly 58 kilobase (kb) haplotype block identified by the aforementioned GWASs lies within a region devoid of any protein-coding genes. The closest protein-coding genes lie a few kb proximal as part of the INK4/ARF locus.¹ This locus contains the cyclin-dependent kinase inhibitors CDKN2A and CDKN2B, as well as p14/ARF, a splice variant of CDKN2A. Could one of these genes be responsible for increased cardiovascular disease risk linked to the 9p21 locus? Early studies investigating CDKN2B have been promising, as its expression has been shown to be reduced in atherosclerotic plaques in human carriers of the 9p21 risk allele.⁶ In extending this knowledge, the Leeper group at Stanford utilized CDKN2B−/− mice to demonstrate increased SMC apoptosis and decreased SMC phagocytic clearance in mouse models of aneurysm and atherosclerosis, respectively.^5,6 In Nanda et al,⁸ published in this issue of Circulation Research, the Leeper group presents exciting evidence for a role of CDKN2B in PAD pathogenesis.

The authors first report the striking observation that human coronary atherosclerotic plaques from carriers of the 9p21 risk allele have increased intraplaque microvessel density but impaired α-SMA coverage of these microvessels. Increased microvessel density within atherosclerotic lesions has been linked to increased rates of intraplaque inflammation, hemorrhage and rupture.⁹ This human observation by Nanda et al⁸ suggests that it may not only be the presence of endothelial cells (ECs) within the plaque that is important but also whether or not the ECs are invested by perivascular cells, be it smooth muscle cells (SMCs) or pericytes. Throughout the study, the authors use α-SMA staining, which cannot distinguish between SMCs and pericytes, as pericytes can also express α-SMA, particularly in an inflammatory setting.¹⁰ In fact, distinguishing between SMCs and pericytes, at least in mouse, requires rigorous lineage tracing, use of multiple markers, and high-resolution confocal microscopy.¹⁰ Regardless, whether it is paracrine effects from ECs, the higher rates of EC growth, and/or defects within the perivascular cells themselves that leads to an increased number of immature blood vessels is an important area of investigation and potential therapeutic modulation. The authors’ initial observation linking 9p21 to increased microvessel density, coupled with their previous discovery that the 9p21 risk allele correlates with reduced CDKN2B expression in atherosclerotic plaques,⁶ together suggest a possible role for CDKN2B in atherosclerotic plaque progression and thus PAD.

The authors proceed to ask whether the perivascular cell coverage defect present in human atherosclerotic lesions can be recapitulated in a non-atherosclerotic setting and use hindlimb ischemia (HLI), a well-established mouse model of PAD, to answer this question. Upon induction of HLI in global CDKN2B−/− mice, they see decreased -SMA coverage of new blood vessels, which is accompanied by impaired blood flow recovery and increased hindlimb tissue loss. They also note decreased numbers of total microvessels in CDKN2B−/− mice two weeks after HLI. Since the authors later show that ECs deficient in CDKN2B have a greater angiogenic capacity, this suggests that the relative loss of ECs may be due to the lack of a perivascular cell layer to stabilize them.

In well over 90% of patients, PAD is caused by atherosclerosis leading to occlusions of the large inflow vessels, and in a sizeable fraction of patients with symptomatic PAD there is a total occlusion along the single path that blood takes from the aorta to the distal leg.¹¹ As such, perfusion becomes entirely dependent on a patient’s ability to effectively remodel vasculature to permit distal tissue perfusion. In many ways, much like the response to HLI in mice is under genetic control, the response to the total occlusion in inflow (iliac or femoral) arteries is highly variable from patient to patient. The greater the neovascular response, the less the patient is afflicted. It is interesting to note that roughly half of PAD patients, despite having a reduced ankle-brachial index (ABI), report no symptoms, suggesting their ability to mount an effective neovascular response.¹¹ In light of the findings in Nanda et al,⁸ could reduced CDKN2B expression lead to impaired functional neovascularization and therefore worse outcomes for PAD patients? Conversely, if an iliac or proximal femoral artery occlusion occurs in a patient, can robust CDKN2B expression allow sufficient revascularization so that the PAD becomes undiagnosed? If so, reduced CDKN2B expression, by both accelerating atherosclerotic disease progression and impairing an effective neovascular response, may be a ‘double hit’ for PAD.

The authors next turn to in vitro studies to determine how reduced CDKN2B expression leads to impaired neovessel maturation. Using siRNA-mediated knockdown of CDKN2B in individually cultured ECs and SMCs exposed to hypoxic (2% oxygen) conditions, they find that ECs have increased angiogenic properties including increased migration and proliferation. Of note, these effects are severely reduced in the absence of hypoxia. This adds to other data suggesting that hypoxic endothelium in hind limb muscle may well respond very differently than ECs that are not under hypoxic stress.¹² In vascular SMCs, CDKN2B knockdown has minimal effects on migration and proliferation when grown under hypoxia. When they co-inject CDKN2B deficient ECs and SMCs using a matrigel plug assay, they are able to recapitulate the phenotype observed in vivo, i.e. poor SMC coverage of endothelium.

This study is now the third by the Leeper group to use CDKN2B−/− mice and demonstrate a unique SMC phenotype.^6–8 Although aneurysm and atherosclerotic plaque development are often associated with defective SMC function, the role of the SMC in neovessel formation/angiogenesis is clearly understudied. As a point of clarity, the authors occasionally use the term ‘angiogenesis’ as a more general term for neovascularization, or new vessel formation. Their in vivo histologic methods of analysis do not distinguish between location and sizes of vessels and thus the neovascularization, or lack thereof, they see in vivo following HLI cannot be defined as strict angiogenesis, per se, typically defined as the sprouting of new capillaries from existing vessels.¹¹ To return to the original point, the fact that CDKN2B KO leads to impaired SMC investment of new vessels, which correlates with impaired perfusion recovery and increased tissue loss, highlights the importance of SMCs to functional blood flow. SMCs, after all, must detach from the vessel wall, proliferate, migrate, and ultimately reattach to form new functional vessels capable of blood flow. The finding that global CDKN2B−/− seems to positively impact ECs (pro-angiogenic effects) but negatively impact SMCs (poor SMC coverage of vessels) and yet still lead to a net negative overall phenotype is further support for the essential role of perivascular cells to functional blood flow. In the future, it may be interesting to use cell-specific KO of CDKN2B to tease out the relative contributions of ECs versus SMCs to the observed phenotype.

The fact that Nanda et al⁸ has uncovered a novel mechanism whereby reduced CDKN2B expression impacts both atherosclerotic and non-atherosclerotic disease, via inhibition of neovessel maturation, is exciting, particularly in light of recent GWASs. Several studies have linked SNPs within the 9p21 locus with PAD independent of atherosclerotic risk factors.^2,3 Could this be the mechanism behind it? While the data is encouraging, the story is likely much more complicated than a single SNP in a single gene leading to various cardiovascular diseases. Recall that another cell cycle inhibitor, CDKN2A, and its splice variant p14/ARF, are also located just proximal to the 9p21 haplotype block. Mouse models utilizing knockout of this complex have been linked to accelerated atherosclerosis.^13,14 Another gene with exons overlapping the INK4/ARF locus, methylthioadenosine phosphorylase (MTAP), has also been linked with SMC proliferation and apoptosis.¹ Additionally, the 9p21 locus itself contains a long intergenic noncoding RNA, termed antisense non-coding RNA in the INK4 locus (ANRIL). ANRIL is capable of recruiting transcriptional repressive complexes to epigenetically repress various loci, including the CDKN2B promoter.¹⁵ Taken together, it is possible that each of these genes may contribute to disease risk independently or through combinatorial mechanisms.

Finally, the authors use a series of cDNA microarrays to identify nearly 250 genes significantly dysregulated in CDKN2B deficient ECs and SMCs. Pathway analysis shows that the majority of dysregulated processes involve angiogenesis or transforming growth factor beta (TGFβ) signaling. To confirm involvement of the TGFβ signaling pathway, they analyze human carotid endarterectomy samples and show an inverse correlation between CDKN2B and TGFβ expression. They then use a series of ELISAs, PCR and Western blots to further explore how the TGFβ signaling pathway is altered under hypoxic conditions in the setting of reduced CDKN2B expression. Hypoxic ECs and SMCs have decreased expression of the inhibitory factor SMAD7, upregulation of TGFβ1, increased SMAD3 activation, and ultimately upregulation of the focal adhesion molecule TGFβ1i1. They then return to the EC-SMC matrigel plug assay and demonstrate that simultaneous siRNA-mediated inhibition of both TGFβ1i1 and CDKN2B leads to a normalization of vessel maturation, i.e. that the phenotype can be rescued in an in vitro setting through modulation of the TGFβ pathway.

PAD is a growing public health problem for which no medical therapies exist that are effective in improving perfusion to the lower extremities.¹¹ Could modulation of TGFβ signaling be a potential therapeutic target to effectively improve perfusion by promoting neovascularization? It is probably not so simple, as TGFβ signaling is responsible for a wide range of cell and context-dependent effects. For example, TGFβ signaling has been shown to have pro- or anti-angiogenic effects on ECs depending on whether signaling occurs through ALK1 and SMAD1/5 or through ALK5 and SMAD2/3, respectively. The story is similar in vascular SMCs, where TGF1 can either promote the contractile state through myocardin/SRF interactions at CArG boxes or the synthetic state via effects on proliferation and extracellular matrix synthesis.¹⁶ In the present study, the authors also note cell-dependent effects, which may complicate future therapeutic strategies. Additionally, although only ECs and SMCs were studied, TGFβ signaling can also affect function and viability of skeletal muscle, an often over-looked component of an effective neovascular response following HLI.¹⁷ As a downstream effector molecule, TGFβ1i1 may prove to be a more promising therapeutic target. In the future, it will be informative to see whether increased TGFβ1 and/or TGFβ1i1 expression correlates with the 9p21 risk allele and, in turn, reduced CDKN2B expression in other human tissue samples.

In summary, this study provides a novel mechanism linking the 9p21 risk allele with reduced CDKN2B expression, increased TGFβ signaling, and impaired neovessel maturation. Importantly, in line with previous GWAS reports, these correlations appear to be present under atherosclerotic and non-atherosclerotic conditions, suggesting that CDKN2B may both promote atherosclerosis progression and impair functional neovascularization, effectively a ‘double hit’ for PAD pathogenesis.