TET2 protects against VSMC apoptosis and intimal thickening in transplant vasculopathy

Abstract

Background:

Coronary allograft vasculopathy (CAV) is a devastating sequelae of heart transplant in which arterial intimal thickening limits coronary blood flow. There are currently no targeted therapies to prevent or reduce this pathology that leads to transplant failure. Vascular smooth muscle cell (VSMC) phenotypic plasticity is critical in CAV neointima formation. TET methylcytosine dioxygenase 2 (TET2) is an important epigenetic regulator of VSMC phenotype, but the role of TET2 in the progression of CAV is unknown.

Methods:

We assessed TET2 expression and activity in human CAV and renal transplant samples. We also employed the sex-mismatched murine aortic graft model of graft arteriopathy (GA) in wild type and inducible smooth muscle-specific Tet2 knockout mice; and in vitro studies in murine and human VSMCs using knockdown, overexpression, and transcriptomic approaches to assess the role of TET2 in VSMC responses to IFNу, a cytokine elaborated by T cells that drives CAV progression.

Results:

In the present study, we found that TET2 expression and activity is negatively regulated in human CAV and renal transplant samples and in the murine aortic graft model of GA. IFNу was sufficient to repress TET2 and induce an activated VSMC phenotype in vitro. TET2 depletion mimicked the effects of IFNу, and TET2 overexpression rescued IFNу-induced dedifferentiation. VSMC-specific TET2 depletion in aortic grafts, and in the femoral wire restenosis model, resulted in increased VSMC apoptosis and medial thinning. In GA, this apoptosis was tightly correlated with proliferation. In vitro, TET2 deficient VSMCs undergo apoptosis more readily in response to IFNγ and expressed a signature of increased susceptibility to extrinsic apoptotic signaling. Notably, enhancing TET2 enzymatic activity with high-dose ascorbic acid rescued the effect of GA-induced VSMC apoptosis and intimal thickening in a TET2-dependent manner.

Conclusions:

TET2 is repressed in CAV and GA, likely mediated by IFNу. TET2 serves to protect VSMCs from apoptosis in the context of transplant vasculopathy or IFNу stimulation. Promoting TET2 activity in vivo with systemic ascorbic acid reduces VSMC apoptosis and intimal thickening. These data suggest that promoting TET2 activity in CAV may be an effective strategy for limiting CAV progression.

Introduction:

Organ transplants provide life-saving therapy for end stage heart, renal, and lung disease. Graft arteriopathy (GA) – or coronary allograft vasculopathy (CAV) in the context of heart transplant - is the major challenge and leading cause of mortality that limits the long-term success of heart transplantation. CAV is characterized by circumferential neointimal hyperplasia in the vasculature due to chronic rejection. CAV occurs in 50% of transplanted hearts by 10 years after surgery¹, and ischemic injury to the myocardium is responsible for 30% of all deaths in heart transplant recipients². Neointimal lesions in CAV are largely composed of smooth muscle alpha-actin (ACTA2) positive cells³. These cells are heterogeneous in origin (reviewed in⁴), but are thought to arise primarily from dedifferentiated vascular smooth muscle cells (VSMCs) originating from the donor arterial tunica media. ^{3, 5–7}.

VSMCs are remarkable in that these differentiated myocytes retain the ability to dedifferentiate, proliferate, and migrate in response to pathologic stimuli. While this characteristic is necessary for growth and wound healing, it also contributes to pathologic vascular remodeling. The stimuli that initiate VSMC dedifferentiation in transplant vasculopathy originate from the host’s immune system. ‘Non-self’ antigens expressed by transplanted donor tissue initiate both innate and adaptive immune alloreactivity. T cells recruited to the graft are activated directly or indirectly by alloantigen presentation (reviewed in⁸) to secrete interferon-γ (IFNу), a well-established promoter of neointimal hyperplasia⁹. Interestingly, while IFNγ is sufficient to induce GA in animal models¹⁰, it does not directly induce VSMC proliferation¹⁰. Several studies in animal models have shown that VSMC apoptosis in GA occurs along with, and exacerbates, intimal hyperplasia^11–17. Indeed, there is a positive association between apoptosis, or medial thinning, and CAV severity in human specimens^18–22. The mechanism whereby IFNγ induces VSMCs to dedifferentiate and contribute to the formation of neointimal lesions in CAV – and whether this is mediated through apoptosis of a population of medial VSMCs - is not well understood.

We reported previously that VSMC phenotype is regulated by the methylcytosine dioxygenase TET2, an enzyme that promotes modification of methylated DNA and subsequent chromatin remodeling to alter gene expression²³. We showed that TET2 expression and activity, as measured by the 5-hydroxymethylcytosine (5-hmC) epigenetic mark, is repressed in the murine femoral artery denudation model of restenosis. TET2 promoted 5hmC modification of the promoters of key VSMC differentiation genes, including myocardin, SRF, and contractile genes. This study also found an inverse correlation between TET2 expression and severity of atherosclerotic disease in human tissue samples²³. The role of TET2 in the pathogenesis of GA - where IFNγ plays a significant role in promoting VSMC phenotypic modulation, unlike in restenosis or atherosclerosis – has not been investigated. Treatment of CAV is currently limited to modulating immunosuppressive dosing or surgical revascularization, although these are of limited efficacy compared to retransplantation, which is limited by availability of donor organs²⁴. New insights into the pathophysiology of GA are urgently needed to develop new therapeutic interventions. In studying the role of TET2 and the relationship between apoptosis and neointima formation in CAV, we identify novel underlying mechanisms of GA that suggest potential treatment and prevention strategies.

Methods:

The data, analytic methods, and study materials will be made available to other researchers for purposes of reproducing the results or replicating the procedure by contacting the corresponding author. Additional experimental details are available in the Supplemental Methods section. All antibodies and primers are listed in Tables I and II in the Supplement, respectively.

Animal experiments:

All procedures involving animal subjects were approved by the Institutional Animal Care and Use Committee of Yale University.

All founder mice were obtained from Jackson Laboratory: TET2 floxed (LoxP sites at the 5’ and 3’ sides of exon 3; B6;129S-Tet2^tm1.1Iaai/J, 017573)²⁵, MYH11-Cre^ERT2 (B6.FVB-Tg(Myh11-cre/ERT2)1Soff/J), Rosa26-mTmG (B6.129(Cg)-Gt(ROSA)26Sor^{tm4(ACTB-tdTomato,-EGFP)Luo}/J, 007676), and B6 (C57BL/6J, 000664). The TET2 floxed strain was crossed with the MYH11-Cre^ERT2 strain to generate tamoxifen-inducible smooth muscle-specific TET2-deficient mice (“iKO”). TET2-deficient hematopoietic cells made from the TET2 floxed strain have previously been shown to exhibit catalytic loss of function²⁶. Male donor animals were derived from homozygous floxed breeding; all donor mice expressed Cre and were homozygous floxed, therefore control animals received corn oil vehicle while TET2-deficient mice received tamoxifen (Sigma, T5648). The TET2-deficient strain was further crossed to the mTmG reporter strain to allow for smooth muscle lineage tracing. Male donor animals were derived from heterozygous floxed breeding; all donor mice expressed Cre and a single copy of the mTmG transgene. Control mice were homozygous wild type and TET2-deficient mice were homozygous floxed; all donor animals received tamoxifen. Tamoxifen was given at 1mg/mouse/day intraperitoneally (IP) for 10 days. When vehicle was administered it was given at the same volume as tamoxifen. Tamoxifen was dissolved in 10% ethanol/90% corn oil vehicle.

Aortic interposition grafts were performed as described elsewhere²⁷. Briefly, approximately 4mm-long segments of male donor thoracic aorta were grafted to the aorta of female recipient mice with end-to end anastomoses. Intercostal vessels exiting the donor segment were ligated with un-braided 5–0 nylon suture material. Donor and recipient animals were of the same background, except in saline/ascorbic acid experiments where all donors were on a mixed 129-B6 background and recipients were on a pure B6 background. 0.5g/ml ascorbic acid (AA) for in vivo administration was prepared in 1x PBS buffered with equimolar sodium bicarbonate and was pH-adjusted to 7.5 with 30mM NaOH. Animals received 4g/kg AA IP, or equivalent volume saline, daily beginning three days after surgery.

Statistical analysis

Values are presented as mean ± standard error of the mean (SEM). Two-sample datasets, where the data is expressed relative to the control, were analyzed by one-sample t tests. For two-sample datasets, equality of variance was determined by F-test; Welch’s t-tests were used when variances were significantly different, otherwise Student’s t-tests were used. Datasets with two independent variables were analyzed by two-way ANOVA followed by Sidak’s multiple comparisons testing as indicated. Datasets with two independent variables where variances were significantly different between conditions (as determined by Brown-Forsythe test) were analyzed by serial Welch’s t tests, followed by Bonferroni correction to account for multiple comparisons. Statistical analysis was performed using Graph-Pad Prism7. Probability values of P<0.05 were considered significant.

Results:

TET2 protein expression and global 5-hmC is repressed in human coronary allograft vasculopathy, and in a murine model of GA:

We assessed TET2 and 5-hmC by immunostaining in autopsy specimens of coronary vessels from heart transplant recipients with CAV and age-matched controls, as well as from kidney transplant recipients with renal artery graft arteriopathy (GA) and control arteries. There was robust TET2 staining in the nuclei of medial SMC in the controls, but notably reduced TET2 expression in the CAV/GA coronaries (Fig. 1A, Figure I-A in the Supplement). There was also less intense 5-hmC and smooth muscle alpha-actin (ACTA2) staining in the transplant samples (Fig. 1B, Figure I-B in the Supplement). To allow for mechanistic studies, we employed the murine sex-mismatched aortic interposition model of GA, where a male graft is transplanted into a female receipient²⁷. We found, as in human samples (Figure I-A in the Supplement), there was a dramatic reduction in nuclear TET2 protein in medial and neointimal cells compared to the medial VSMCs in pre-transplant donor vessels (Fig. 1C). Time course studies showed a repression in medial TET2 beginning at 3 days, and a more pronounced repression at 14, 21, and 28 days after transplant. 5-hmC immunostaining revealed similar kinetics to TET2 expression (Figure II-A-B in the Supplement). The repression of TET2 and 5-hmC correlates temporally with the recruitment of T cells to the graft and subsequent IFNу signaling, as assessed by immunostaining for CD3 and activation of STAT1 (phospho-STAT1; pSTAT1), respectively (Figure II-C-D in the Supplement). The majority of cells expressing high levels of TET2 in the neointima of human CAV samples expressed CD3 (T cells) (Fig. 1F, Figure III in the Supplement). TET1 and TET3 were not repressed in grafts compared to donor control tissue (Figure IV in the Supplement). Quantification of 5-hmC signal intensity from human samples is shown in the left panel of Figure 1E. Co-immunostaining for 5-hmC and pSTAT1 in mouse aorta grafts showed that low TET activity (low 5-hmC) was inversely correlated with nuclear pSTAT1 (Fig. 1D–E). These data reveal that T cell recruitment and canonical IFNγ signaling are associated with decreased TET2 expression and activity in smooth muscle cells in GA.

Figure 1: TET2 expression and 5-hmC are repressed in allograft vasculopathy; coincident with canonical IFNγ signaling.

Formalin-fixed artery tissue was obtained from similarly aged, deceased patients who had received heart transplants (CAV, N=7), kidney transplants (N=3), or who died from non-cardiovascular diseases and had mild atherosclerotic intimal thickening (Control coronaries N=10, control kidney N=1)). Sections of coronary arteries were immunostained for A) TET2 or B) 5-hmC and smooth muscle alpha actin (ACTA2), with DAPI nuclear counterstain. Representative images shown. Cryosections of 28day sex-mismatched murine aortic interposition grafts were similarly immunostained for C) TET2 (L denotes lumen), or D) 5-hmC and phospho-STAT1 (pSTAT1), with DAPI. Representative images shown. N=3 E) Quantitation of Left) relative 5-hmC signal intensity in human samples (Control N=11, Transplant (Xplant) N=10), Right) relative 5-hmC and pSTAT1 signal intensity at 28days after aortic transplant in mouse. (Donor N=5, Graft N=6). Scale bars: A) 50μM, B) 50μm, C) 25μm, D) 10μm. F) Normal, and transplant, human coronary artery sections were immunostained for TET2 and CD3. Representative images shown (Control N=11, CAV (and kidney transplant GA) N= 10). Yellow dotted line indicates internal elastic lamina (IEL), green dotted line indicates external elastic lamina (EEL). Scale bar: 20μm. E) Left: Welch’s t test, Right: one sample t test. **P<0.01, ***P<0.001.

IFNγ treatment is sufficient to cause repression of TET2 and an activated phenotype in cultured VSMCs:

Treatment of primary human aortic vascular smooth muscle cells (hVSMCs) in vitro with 100ng/ml IFNγ resulted in a significant repression of TET2 mRNA and protein compared to control conditions (Fig. 2A, Figure V-A in the Supplement). IFNγ treatment also induced a de-differentiated hVSMC phenotype including repression of MYOCD, SRF, and MYH11 mRNAs, ACTA2 and CNN1 proteins, and concurrent induction of KLF4, PDGFRA, and PDGFRB mRNAs, and PDGF receptor proteins (Fig. 2A, Figure V-A in the Supplement). We observed a similar effect of IFNγ stimulation on these genes in primary murine aortic VSMCs (mVSMCs) (Figure VI-A in the Supplement). Additionally, TET2 knockdown is sufficient to recapitulate much of the IFNγ-induced changes in gene expression associated with VSMC de-differentiation (Figure VI-B in the Supplement, and shown previously²³). We assessed the effect of IFNγ treatment on total genomic 5-hmC content in hVSMCs and mVSMCs by ELISA. IFNγ stimulation significantly decreased 5-hmC content in VSMCs from both human and mouse (Fig. 2B). Furthermore, we found that adenovirus-mediated overexpression of TET2 in hVSMCs prevented the IFNγ-induced changes in gene and/or protein expression of MYH11, ACTA2, CNN1, KLF4, and PDGFRs (Fig. 2C, Figure V-B in the Supplement). In these experiments, upregulation of STAT1 serves as a positive control readout for IFNу stimulation. These data suggest that repression of TET2 may mediate some of the IFNγ-induced effects on VSMC de-differentiation in GA.

Figure 2: IFNγ treatment causes TET2 and 5-hmC repression, and an activated phenotype in VSMCs.

Primary human or mouse vascular smooth muscle cells (VSMCs) were cultured under control conditions (Ctrl or “-“) or treated with 100ng/ml recombinant human or mouse IFNγ (IFNγ or “+”) for 48hrs, respectively. A) Left panel: mRNA levels for indicated genes from hVSMCs determined by quantitative PCR (qPCR) normalized to RNA18SN5, expressed as fold change relative to control (dotted line) (N=4), Right panel: representative western blot (N=4). B) Quantitation of genomic 5-hmC content, by ELISA, 48hr after treatment of VSMCs with 100ng/ml IFNγ. Left panel: human VSMCs (N=3), Right panel: mouse VSMCs (N=3). C) hVSMCs were infected with adenoviruses expressing GFP control or TET2b protein prior to IFNγ treatment as described in 2A-B. Left panel: mRNA levels for indicated genes determined by qPCR normalized to RNA18SN5 (N=4), Right panel: representative western blot (N=4). A-B) one sample t test, C) Two-way ANOVA: *P<0.05, **P< 0.01, ***P<0.001.

TET2 deficiency in donor VSMCs exacerbates Intima:Media ratio in GA:

Because TET2 plays an important role in regulating VSMC phenotype²³, and because TET2 is repressed in GA, and in vitro by IFNу, we hypothesized that genetic depletion of TET2 in murine aorta grafts would exacerbate GA. To test this hypothesis, we again employed the sex-mismatched aorta graft model. We generated smooth muscle-specific, inducible, TET2-deficient donor mice (Figure VII-A in the Supplement). TET2-deficient cells generated from this floxed mouse exhibit TET2 loss-of-function with respect to DNA hydroxymethylation (Figure VII-B in the Supplement, and shown elsewhere²⁶). TET2-deficient or control vessels from male donors were transplanted into female recipients on the same background. 28 days after transplant, the intima:media ratio was increased in TET2 deficient grafts relative to controls (Fig. 3A). Unexpectedly, this was due to a decrease in area of the tunica media in the TET2-deficient grafts rather than a significant increase in neointimal area (Fig. 3A). This suggested that TET2 deficient VSMCs may undergo cell death more readily compared to TET2-competent VSMCs in GA. To confirm that the TET2-deficient VSMCs were being lost from the graft medial layer, we crossed the TET2-deficient mice to the Rosa26-mTmG reporter strain to label medial VSMC and repeated the 28day graft experiment. We found there were markedly fewer medial donor derived (GFP⁺) VSMCs remaining in the TET2-deficient grafts compared to controls (Fig. 3B). Donor vessels from control and TET2-deficient mice were equivalent with respect to ACTA2 expression, proliferation, and apoptosis at baseline (Figure VII-C in the Supplement). As the donor derived VSMCs do not appear to have migrated to a different vascular layer, these data suggest that TET2 is required for VSMC survival in GA. Further, since the neointima is at least as thick in the TET2-deficient grafts as in controls, these data suggest enhanced pro-neointima signaling from the fewer remaining donor TET2-deficient VSMCS.

Figure 3: Depletion of Tet2 in donor VSMCs exacerbates pathologic vascular remodeling in GA due to medial cell loss.

Control and TET2-deficient thoracic aorta segments from male donor mice were transplanted into sex-mismatched recipients of the same strain. A) Elastin Von-Geisson (EVG) staining of control (Ctrl=Tet2^FFMyh11-Cre, vehicle-treated; N=6) or TET2-deficient (Tet2^FFMyh11-Cre, tamoxifen-treated; N=7) grafts 28 days after surgery; graphs show associated quantitation of intima:media ratio, intima area, and media area. B) The Tet2 floxed-Cre strain was bred to the Rosa26-mTmG reporter line. Micrographs of GFP fluorescence and ACTA2 immunostaining in littermate Tet2⁺⁺mTmG^mut+Myh11-Cre (Ctrl; N=7) and Tet2^FFmTmG^mut+Myh11-Cre (iKO; N=5) grafts 28 days after surgery (red bars indicate medial layer); graphs show associated quantitation of intima:media ratio, intima area, media area, and percent GFP-positive (%GFP+) area in the intima-media combined region. Student’s t test: *P<0.05, **P<0.01.

Baseline sensitization of TET2 deficient cells to apoptosis results in medial cell loss in the femoral guidewire denudation model of restenosis:

The above results are surprising because previous work from our group found that local virus-mediated shRNA depletion of TET2 in a murine model of restenosis exacerbated intimal thickening²³. This discrepancy may arise for several reasons. In the prior study, TET2 deletion was not confined to VSMC. In the present study, it is possible that TET2-deficient VSMC may promote a more robust T cell response that drives medial loss. To determine the effect of smooth muscle TET2 depletion in a model of vascular remodeling in which T cells do not play a major role, we employed the femoral artery endothelial denudation injury model. Notably, significant early medial VSMC apoptosis is known to precede intimal hyperplasia in this model²⁸. We hypothesized that TET2-deficient grafts would exhibit increased intimal hyperplasia in this restenosis model due to the VSMC dedifferentiation that occurs with loss of TET2 (Figure VI in the Supplement and²³). We performed femoral artery injuries on the ROSA-26 mTmG control and Tet2-deficient mice (Fig. 4, Figure VIII in the Supplement). 21 days after surgery we observed GFP⁺ neointima in the control animals, as expected. Remarkably, in the majority of TET2-deficient littermates, there was no neointima present, the medial layer had thinned, and few GFP⁺ VSMCs remained. In the control animals, the intima:media ratio was significantly higher and an average of 25% of the intima-media combined area was GFP⁺ as compared to only 3.8% GFP⁺ in the TET2-deficient animals. Marked apoptosis (cleaved caspase 3) was found in the injured TET2-deficient femorals (Figure VIII in the Supplement). These data confirm that TET2 is important for VSMC survival in pathologic contexts where apoptosis is involved in the etiology. Importantly, these data provide evidence for the intrinsic increased susceptibility of TET2-deficient VSMCs to apoptosis in another model of vascular pathology.

Figure 4: TET2-deficient femoral arteries exhibit medial thinning, VSMC loss, and limited intimal hyperplasia in response to endothelial denudation injury.

Guidewire-injured right femoral arteries (Inj-RFA), and contralateral control left femoral arteries (Contra-LFA), of littermate Tet2⁺⁺Myh11-Cre (Control N=6; Control mTmG-labeled N=6) and Tet2^FFMyh11-Cre (iKO N=5; iKO mTmG-labeled N=6) were harvested 21days after surgery. A) Representative micrographs of uninjured (LFA) and injured (Inj-RFA) vessels. Scale bar= 100μm. Below: high power insets (indicated by white dashed line in lower power images). Yellow dashed line indicates the internal elastic lamina in the control Inj-RFA sample. B) Quantification of intima:media ratio from both mTmG-labeled and unlabeled animals (Control N=12, iKO N=11). C) Quantification of media area from both mTmG-labeled and unlabeled animals (Control N=12, iKO N=11). D) Quantification of average media layer thickness from both mTmG-labeled and unlabeled animals (Control N=12, iKO N=11). Medial thickness was measured at eight points (0°, 45°, 90°, 135°, 180°, etc.) around the circumference of each vessel and averaged. E) Quantification of intima area from both mTmG-labeled and unlabeled animals (Control N=12, iKO N=11). F) Quantification of %GFP⁺ area in the media and intima from mTmG-labeled animals only (Control N=6, iKO N=6). B-D Student’s t test, E-F Welch’s t test: **P<0.01, ***P<0.001.

TET2 deficiency sensitizes VSMCs to apoptosis in vivo and in response to IFNγ in vitro:

To determine whether more TET2-deficient donor VSMCs undergo apoptosis early in GA progression, as compared to TET2-competent VSMCs, we evaluated an earlier time point post-transplant. There was a 1.75-fold increase in cleaved caspase 3 positive area, in the combined media and intima area, in donor derived (GFP⁺) cells in TET2-deficient grafts versus controls at 10 days after transplant (Fig. 5A). Interestingly, we observed GFP+ donor-derived cells in the sub-endothelial space in some control but not TET2-deficient grafts at this time point (Fig. 5A). These data demonstrate that more donor VSMCs undergo apoptosis in TET2-deficient grafts versus controls. Concordantly, we found that there was an inverse correlation between cleaved caspase 3 staining and nuclear 5-hmC staining in human CAV coronaries versus age-matched controls (Figure IX in the Supplement). In vitro, IFNγ treatment was sufficient to induce hVSMC apoptosis in culture (Figure X in the Supplement). Further, we found that TET2-deficient mVSMCs were sensitized to apoptosis induced by IFNγ treatment (Fig. 5B–D). Specifically, we observed a decrease in cell viability as assessed by crystal violet staining (Fig. 5B), increased cleaved caspase3 (Fig. 5C), and increased cytochrome c release into the medium (Fig. 5D) from TET2-deficient cells treated with IFNγ as compared to control cells. An apoptosis antibody array showed that TET2-deficient mVSMCs exposed to IFNγ for 5 days expressed significantly higher levels of Fas, TNF-R1, and p53, with a trend toward increased p27 compared to control cells (Fig. 5E–F). These data reveal that TET2 deficient mVSMCs are sensitized to extrinsic apoptotic signals that could account for increased apoptosis in response to IFNγ or GA.

Figure 5: Depletion of Tet2 increases VSMC apoptosis in response to GA and IFNγ through upregulation of mediators of extrinsic apoptosis.

Control and TET2-deficient thoracic aorta segments from male donor mice were transplanted into sex-mismatched recipients of the same strain. A) Micrographs of GFP fluorescence and cleaved caspase 3 (clCASP3) immunostaining in littermate Tet2⁺⁺mTmG^mut+Myh11-Cre (Ctrl; N=5) and Tet2^FFmTmG^mut+Myh11-Cre (iKO; N=4) grafts 10 days after surgery; Right) quantitation of cl.CASP3-positive area within GFP-positive cells (% clCASP3) in the media-intima combined region. Red bars indicate the medial layer. B-F) mVSMCs isolated from Tet2⁺⁺mTmG^mut+Myh11-Cre (++) and Tet2^FFmTmG^mut+Myh11-Cre (FF) mice were treated with Cre recombinase-encoding adenovirus in vitro. ++ and FF cells were then cultured under control conditions (Ctrl or “-“) or treated with 100ng/ml rm-IFNγ (IFNγ or “+”) at day 0. B) Viability was assessed by crystal violet staining from 0– 11 days after IFNγ treatment (N=6). Western blot of VSMC C) whole-cell lysates or D) conditioned medium treated from mVSMCs treated with IFNу for 7days as in A. Ø indicates medium conditioned for 7 days in the absence of cells. In C), total STAT1 upregulation (long or short exposure) serves as a positive control known response to IFNγ treatment. E) Apoptosis antibody array of VSMC lysates treated with IFNу for 5 days. F) Quantitation of apoptosis array (N=3). A) Student’s t test: *P<0.05. B, D,F) Two-way ANOVA *P<0.05, **P<0.01.

TET2 loss induces a signature of TNFR signaling in TET2-deficient VSMCs, and sensitizes VSMCs to TNFα-induced apoptosis:

The late kinetics of IFNу-induced apoptosis observed in Figure 5B suggests that IFNу acts through an indirect mechanism to drive VSMC apoptosis in the context of TET2 depletion. We performed bulk RNA-sequencing on cultured control and TET2-deficient VSMCs treated with or without IFNу for 5 days to assess transcriptomic changes that could account for this effect (Figure XI in the Supplement). We noted efficient deletion of floxed exon3 of TET2 (Figure XI-A in the Supplement), and principal component analysis confirmed that the data appropriately clustered into four distinct groups (Figure XI-B in the Supplement). We filtered significantly regulated genes (P<0.01, FDR<0.1) based on those induced in TET2-deficient cells by IFNу that were not similarly induced in control cells by IFNу. We found 60 genes met these criteria (Table III in the Supplement). Ingenuity Pathway analysis revealed that TNF receptor signaling was significantly induced in this gene set (Figure XI-C, and Table III in the Supplement). We therefore assessed whether TNFα treatment induces more apoptosis in TET2-deficient VMSCs versus controls. 48hr treatment with 15ng/ml TNFα resulted in both more caspase 3 cleavage and cytochrome C release into the medium, in TET2-deficient cells versus controls (Figure XI-D in the Supplement). These data suggest that induction of TNF signaling pathway components in TET2-deficient cells functionally sensitizes the cells to TNFα-induced apoptosis.

Decreased TET activity promotes apoptosis and proliferation in vivo:

Medial VSMC apoptosis has been proposed to induce proliferation of neighboring VSMCs and to contribute to neointima formation in GA^{11, 12}. Time course analysis in vivo revealed that apoptosis (cleaved CASP3) and VSMC proliferation (Ki67) are tightly correlated (Fig. 6A–B). Further, decreased TET activity (nuclear 5-hmC) correlated with increased apoptosis and proliferation (Fig. 6A–B). We hypothesized that increased apoptosis in the TET2 deficient grafts results in more proliferative surviving VSMCs. In our 10day GA samples (see Fig. 5A), we found that there were more proliferative donor VSMCs (Ki67⁺, GFP⁺) in the media/intima region in the TET2-deficient grafts versus controls (Fig. 6C, left graph). We also assessed proliferation of intimal and medial cells that are non-donor VSMCs and non-T cells (CD3^–GFP^–), as recruited host-derived cells may contribute to intimal hyperplasia in human CAV and GA models. There was a trend toward increased proliferation of these cells in the TET2-deficient grafts (Fig. 6C, right graph). Proliferative donor VSMCs exhibit low 5-hmC and are found near apoptotic cells in grafts (Fig. 6A, Fig. 6D). These data show that TET2 repression predisposes VSMCs to apoptosis which correlates with proliferation of neighboring or surviving VSMCs or potentially of other cells recruited from the recipient. CXCL12 (SDF1) is a known mitogen for VSMCs¹² and has been implicated in recruiting mesenchymal stem cells to potentiate neotintima formation in GA¹¹. We further noted increased expression of CXCL12 mRNA in TET2-deficient versus control cells (Figure XI-E in the Supplement), suggesting a mechanism by which loss of TET2 may contribute to neointima formation.

Figure 6: TET activity is negatively correlated with apoptosis and proliferation in GA.

Thoracic aorta segments from male donor mice were transplanted into sex-mismatched recipients of the same strain. A) Micrographs of immunostained sections from donor (N=5) and graft vessels at 3 (N=3),7 (N=3),10 (N=3),14 (N=2),21 (N=4), and 28days (N=4) after surgery. Upper panels) Ki67 and cleaved caspase 3 (clCASP3), and lower panels) 5-hmC. Representative images shown. B) Quantification of Ki67 and clCASP3 % positive area (left y-axis) from micrographs as described in A. Quantification of nuclear 5-hmC signal (right y-axis) from the same tissues as described in A. C) 10day Control and Tet2-deficient graft sections were immunostained for Ki67, CD3, and GFP. Left graph) Ki67 positive (Ki67+) donor derived VSMCs (GFP⁺) and right graph) Ki67 positive non-donor derived VSMC (GFP^–), non-T cells (CD3^–) were quantified in the media-intima combined region. Student’s t test: *P<0.05. D) Donor and graft sections were immunostained for 5-hmC or GFP, Ki67, cleaved caspase 3 (clCASP3), and DAPI. Yellow arrows indicate 5-hmC^lo Ki67⁺ cells, orange arrows indicate GFP⁺Ki67⁺ cells. Representative images of 4 independent samples are shown.

Promotion of TET2 activity by Ascorbic Acid increases donor VSMC survival in GA in a TET2-dependent fashion:

Ascorbic acid (Vitamin C) is a co-factor for TET2 that can promote its catalytic activity, enhancing levels of 5hmC and DNA demethylation^{29, 30}. We hypothesized that increasing TET2 activity with ascorbic acid (AA)³¹ would protect the cells from apoptosis in GA. AA treatment prevented repression of genomic 5-hmC in response to IFNγ in cultured hVSMCs but did not affect 5-hmC in hVSMCs after TET2 knockdown (Figure XII-A in the Supplement). AA treatment did not change basal TET2 mRNA levels or prevent their repression by IFNу (Figure XII-B in the Supplement). As high dose AA has been successfully employed to rescue TET2-dependent functions in hematopoietic stem cells and suppress leukemia progression in vivo³¹, we hypothesized that AA treatment may increase TET2 activity and rescue donor VSMC apoptosis in GA. We transplanted control and TET2-deficient grafts, both with SMC labeled GFP⁺ via Myh11CreER^T2-mTmG, into C57bl/6J (B6) recipients. There is a 5% background mismatch (genome scanning data not shown) between the donor animals and B6 recipients; this greater mismatch lessens the difference between the control and TET2-deficient animals at baseline, allowing for better comparison of the effects of AA. Intraperitoneal AA administration (4g/kg/day)³¹ dramatically protected donor medial VSMCs from GA-induced apoptosis, as shown by retention of GFP⁺ medial cells, and this effect was TET2-dependent (Fig. 7A–B). Furthermore, AA treatment significantly reduced neointima formation in control, but not in TET2-deficient, grafts (Fig. 7B). There was no significant change in media area any of the groups (Fig. 7B). The trend toward a decrease in medial area in the TET2-deficient saline-treated grafts versus the control saline-treated grafts (Fig. 7B) was not as pronounced as in Fig. 3, likely due to differences in the degree of MHC mismatch between these experiments. Finally, there was a significant increase in nuclear 5-hmC staining in the control-transplanted animals treated with AA, but not in the TET2-deficient-transplanted animals, confirming a rescue of Tet2 activity in the medial layer (Fig. 7B). We assessed whether high-dose AA prevents T cell recruitment or IFNγ signaling in vivo. AA did not reduce T cell recruitment to the grafts of either genotype, and did not decrease IFNγ signaling (Figure XII-C-D in the Supplement). As in culture (Figure XII-B in the Supplement), in vivo AA treatment did not increase TET2 expression in either the control or TET2-deficient grafts (Figure XII-E in the Supplement). The anti-TET2 antibody used in S12E (ab124297) recognizes a C-terminal epitope of the protein, mRNA for which is still transcribed after locus recombination (see Figure XI-A in the Supplement). Notably, there was an inverse correlation between GFP⁺ medial cell survival and intima:media ratio (Fig. 7B). At a 10day timepoint, before significant loss of GFP⁺ donor VSMCs, AA treatment significantly reduced apoptosis in a TET-dependent manner (Fig. 7C). These data reveal that preventing donor medial cell apoptosis by restoring TET2 activity reduces neointima formation.

Figure 7: Promoting TET2 activity with ascorbic acid rescues GA-induced donor-VSMC apoptosis and decreases intimal hyperplasia.

A) Micrographs of GFP fluorescence and DAPI, 28 days after surgery, in littermate Tet2⁺⁺mTmG^mut+Myh11-Cre (Control) and Tet2^FFmTmG^mut+Myh11-Cre (iKO) grafts in C57bl/6J recipients receiving either saline or 4mg/kg/day ascorbic acid (AA). Right) High-power insets of areas in panel A denoted by dashed boxes; micrographs of 5-hmC, CD3, and GFP immunostaining. Yellow arrows indicate GFP⁺ VSMCs with high nuclear 5-hmC. B) Quantification of intima:media ratio, media area, intima area, relative GFP positive (GFP⁺) area in the media/intima, and relative nuclear 5-hmC intensity in the media/intima, in grafts from control and TET2-deficient animals treated with saline or AA. C) Left) Micrographs of graft tissues -performed as in 7A-B but for 10 days – were immunostained for cleaved caspase 3 (clCASP3). Right) quantification of clCASP3⁺ area in GFP⁺ cells, relative to ++ saline. A-B) ++ saline N=5, ++ AA N=5, FF saline N=5, FF AA N=4. C) ++ saline N=6, ++ AA N=5, FF saline N=5, FF AA N=4. Two-way ANOVA: *P<0.05, **P< 0.01, ***P<0.001.

Discussion:

The mechanisms underlying graft arteriopathy leading to heart transplant failure are not fully understood. Our work reveals that TET2 is downregulated in both human CAV and in mouse models and identifies this loss of TET2 function as central to medial VSMC apoptosis and subsequent hyperproliferative responses (See model, Fig. 8). Further, we report that increasing activity of the residual TET2, in control grafts, with systemic high-dose ascorbic acid prevents both medial VSMC loss and intimal hyperplasia. These findings provide new mechanistic insights into early events that drive GA and suggest novel therapeutic approaches to prevent luminal narrowing and graft ischemia.

Figure 8: Model of TET2’s role in limiting intimal hyperplasia through preventing VSMC apoptosis.

Alloantigen-primed effector T cells infiltrate into the vessel wall and elaborate IFNу, resulting in repression of VSMC TET2. TET2 repression sensitizes VSMCs to extrinsic apoptosis, leading to pathogenic crosstalk that promotes proliferation of surviving VSMCs. This crosstalk may potentially extend to resident progenitor cells, and/or extravasated circulating recipient-derived progenitor cells. These effects lead to exacerbated intimal hyperplasia.

Tet2 regulation by IFNγ

Our previous work reported that TET2 is an epigenetic master regulator of VSMC phenotype that promotes expression of contractile genes and pro-differentiation transcription factors while repressing expression of dedifferentiation genes, including KLF4²³. We found high levels of TET2 and 5hmC in differentiated VSMC in vivo, but repression of TET2 mRNA and protein levels by PDGF-BB, a growth factor that promotes VSMC dedifferentiation²³. By modifying patterns of DNA methylation, TET enzymes can exert broad control over programs of gene expression in cells undergoing phenotypic transitions²⁹. We now report that TET2 protein and its epigenetic mark, 5-hmC, are reduced in both human CAV, GA, and murine GA; and this repression correlates temporally with T cell infiltration and canonical IFNу signaling. In vitro, we find that IFNу is sufficient to repress TET2 and 5-hmC and to induce a dedifferentiated VSMC phenotype, including induction of KLF4 and PDGF receptors. TET2 knockdown recapitulates many elements of the IFNу-induced phenotype in VSMCs (Liu et. al.²³), and TET2 over-expression abrogates IFNу-induced phenotypic modulation. Importantly, IFNу alone is sufficient to induce GA in the absence of leukocytes¹⁰. We now implicate VSMC TET2 repression as a key event by which the T cell-derived cytokine IFNу drives this pathology.

TET2 depletion promotes VSMC apoptosis and medial thinning

Given its role as a master regulator of differentiation²³, we hypothesized that genetic depletion of VSMC TET2 in vivo would exacerbate dedifferentiation and neointima formation in GA. We found that the intima:media ratio was indeed increased in TET2-deficient grafts, but, surprisingly, this effect was largely due to apoptosis of donor medial VSMCs rather than an increase in intimal thickness. Analysis of human CAV samples revealed an inverse correlation between TET activity (5-hmC) and apoptosis. Notably, the increased susceptibility of TET2-deficient VSMCs to apoptosis was also observed in the femoral artery denudation injury model of restenosis. Indeed, smooth muscle-specific TET2 deficiency resulted in profound loss of medial VSMCs post-injury and an inability of these vessels to give rise to a neointima. These results were in contrast to prior work from our laboratory showing that periadventitial transduction of femoral artery with TET2-shRNA-expressing virus exacerbated intimal hyperplasia²³. Given the ectopic viral delivery by extravascular biopolymer²³, we cannot rule out that TET2 knockdown in infiltrating leukocytes or adventitial cells could have significantly contributed to the overall phenotype observed. We did not assess apoptosis in those studies. We speculate that pervasive apoptosis that occurs early post-injury may preclude neointima formation, while a partial apoptotic response allows surviving cells to contribute to the neointima.

Although several studies have suggested a relationship between TET enzymes and apoptosis^{32, 33}, there is evidence for both pro- and anti-apoptotic functions of TETs in different contexts and cell types, and little understanding of the underlying mechanisms. To date, our study is the first to identify an anti-apoptotic role for TET2 in VSMC. Our transcriptomic data reveal that the combination of IFNγ and TET2 depletion induces TNFR pathway signaling. Using a proteomic approach, we noted increases in several pro-apoptotic proteins in TET2 deficient VSMCs, including p53, TNF-R1, Fas, p27/Kip. These were further induced by IFNу treatment, likely explaining the increased susceptibility to apoptosis in TET2-deficient VSMCs. In the transplant setting, increased VSMC expression of extrinsic apoptotic pathway mediators like Fas or TNF-R1 could result in enhanced apoptosis induced by FasL-expressing T cells³⁴, or TNFα-expressing macrophages³⁵. TET2-deficient mVSMCs were sensitized to TNFα-mediated apoptosis in vitro. The upregulation of p53 in TET2-deficient cells, particularly in the presence of IFNу, might be responsible for Fas and TNF-R1 induction (reviewed in³⁶). Upregulation of p27/Kip may elicit an anti-proliferative effect due to inhibition of CDK-cyclin complexes³⁷, but may also influence apoptosis³⁸. Our RNAseq also noted an induction of the H2-Dmb2 gene encoding the class II histocompatibility locus Mb2 (Table III in the Supplement), suggesting the intriguing hypothesis that the combination of IFNу and TET2 depletion may enhance antigen presentation by VSMC, leading to enhanced T cell activation to promote progression of graft arteriopathy.

It is known that apoptosis and/or thinning of the medial layer is associated with severity of coronary allograft vasculopathy (CAV) in human heart transplants^18–22. In animal models this medial thinning has been shown to precede intimal thickening and to compromise vessel reactivity¹⁵. Additionally, abrogation of apoptosis in GA by over-expression of BCL-xL has been shown to reduce VSMC apoptosis, CXCL12 production, and intimal hyperplasia¹¹. We now note that reduction in TET2 expression and activity precedes, and correlates with, VSMC apoptosis and proliferation in transplants. 5-hmC repression in murine aorta grafts occurred rapidly within the first week after surgery, with apoptosis and proliferation following by approximately three days. Strikingly, there was a tight correlation between apoptosis and proliferation. While the close relationship between apoptosis and proliferation may seem counterintuitive, there is precedent in vascular injury models demonstrating that CXCL12 released from apoptotic VSMCs may induce paracrine proliferation of non-apoptotic VSMCs¹¹. Additionally, VSMC-derived CXCL12 is implicated in recruitment of circulating cells to vascular lesions³⁹. IFNу alone promotes intimal hyperplasia in GA, but not proliferation of VMSC in vitro¹⁰. Our data suggest that IFNγ promotes intimal hyperplasia in GA by promoting apoptosis and subsequent paracrine effects on surviving VSMCs and other cells. Indeed, we note increased CXCL12 expression in TET2-deficient VSMCs compared to control as a potential mediator of VSMC proliferation¹² or host cell recruitment¹¹. Interestingly, it has been suggested that Fas^highFasL^low VSMCs may undergo apoptosis while Fas^lowFasL^high VSMCs may contribute to intimal hyperplasia¹². We noted that surviving GFP⁺ donor VSMCs were more proliferative in TET2-deficient grafts versus in control grafts, and that TET2-deficient cells in culture express higher Fas than control VSMCs.

In the mouse aorta graft model, mature neointimal lesions are ultimately comprised of recipient-derived cells which are likely recruited progenitor cells and/or invading recipient-derived dedifferentiated VSMCs. We identified a trend toward increased proliferation of GFP^–CD3^– cells in the neointima, suggesting the possibility that, in addition to proliferation of medial cells at these relatively early time points, there may also be increased recruitment of other recipient-derived cell populations in the TET2-deficient context (Fig. 8). It remains unknown what percent of the ACTA2-expressing intimal cells in this model are derived from resident vascular progenitor cells, infiltrating recipient VSMCs, and circulating progenitors. Interestingly, at the 10day timepoint, we observed GFP⁺ cells in the sub-endothelial space in control, but not TET2-deficient grafts. This likely reflects the reduced susceptibility of control VSMCs to apoptosis compared to TET2-deficient cells. These donor-derived cells do not persist in the intima at the 28day timepoint, suggesting that even the control VSMCs are ultimately subject to alloimmune-mediated cytolysis in the long-term.

Ascorbic acid restoration of TET2 activity as a potential therapeutic strategy

Since we implicated the reduction in TET2 expression as an early event that promotes GA, we hypothesized that increasing TET2 activity would be a potential therapeutic strategy. We have successfully transduced mouse femoral artery with periadventitial delivery of viruses to enhance or inhibit TET2 expression in a vascular injury model²³, however, this approach was not viable in the mouse aorta graft model due to diffusion of the virus into the peritoneal cavity and clearance to the neighboring lymph nodes (data not shown). The ability of ascorbic acid to enhance Tet2 activity^29–31 provided an intriguing alternative. While this approach does not prevent downregulation of Tet2 expression by IFNγ, it does enhance enzymatic activity and 5-hmC modification of DNA, as well as cell viability in vitro and in vivo, in a TET2-dependent manner. Most notably, high-dose ascorbic acid significantly rescued donor medial VSMC loss compared to saline in a TET2-dependent manner. This effect was accompanied by a significant decrease in intimal area in TET2-competent grafts only. Because AA has been shown to be potentially toxic to T cells at supraphysiologic concentrations (reviewed in⁴⁰), we assessed whether T cell recruitment or IFNу signaling was inhibited in these tissues. We found that AA neither repressed T cell recruitment nor reduced IFNу signaling. While it is possible that systemic treatment with AA exerts pleiotropic beneficial effects on multiple cell types in vivo, the dependence of the phenotype on smooth muscle-specific expression of Tet2 further suggests that VSMC Tet2 is a critical target for this therapeutic efficacy.

Coronary allograft vasculopathy is a devastating pathology in human heart transplant; retransplantation or modification of immunosuppressive regimen are the only available strategies for treating this complication. Better targeted therapies are needed to prevent CAV development and decrease disease-free survival in transplant recipients. It has previously been suggested that dietary antioxidant supplementation with vitamins C and E may retard intimal thickening in transplant patients⁴¹. Our data suggest that the mechanism of this effect may include bolstering TET2 activity to prevent donor VSMC apoptosis and subsequent pathologic cross talk with the mosaic of cells that give rise to the neointima. Importantly, high dose injection of ascorbic acid is emerging as a safe and effective strategy for treatment of malignancies, especially those which may stem from Tet2 loss of function^{42, 43}. Additionally, inducing TET2 activity with AA may stabilize T regulatory cell populations (reviewed in⁴⁴) to repress the alloresponse in CAV. Taken together with the clinical data from Fang et. al.⁴¹, we propose that supraphysiologic ascorbic acid may be a safe and cost-effective way to prevent CAV in heart transplant recipients.

Non-standard Abbreviations and Acronyms:

AA

ascorbic acid

CAV

Coronary allograft vasculopathy

clCASP3

cleaved caspase 3

CM

conditioned media

Contra

contralateral (uninjured control)

DAPI − 4′,

6-diamidino-2-phenylindole

DMEM

Dulbecco’s Modified Eagle Medium

DOC

deoxycholic acid

EEL

external elastic lamina

ELISA

enzyme-linked immunosorbent assay

EVG

Elastin Van Geison

FBS

fetal bovine serum

FDR

false discovery rate

FF

Tet2^flox/flox genotype

FPKM

fragments per kilobase of transcript per million mapped reads

GA

graft arteriopathy

GSA

gene set analysis

HBSS

Hanks’ balanced salt solution

5-hmC

5-hydroxymethylcytosine

hMe-DIP

hydroxymethylated DNA immunoprecipitation

HRP

horseradish peroxidase

hVSMCs

human aortic vascular smooth muscle cells

IEL

internal elastic lamina

IFNg

Interferon-gamma

iKO

inducible knockout

Inj-RFA

injured right femoral artery

IP

intraperitoneal

IPA

Ingenuity Pathway Analysis

KD

knockdown

LFA

left femoral artery

MOI

multiplicity of infection

mVSMCs

murine aortic vascular smooth muscle cells

NT

non-targeting

PBS

phosphate buffered saline

PCA

principal component analysis

qPCR

quantitative polymerase chain reaction

rh

recombinant human

RIPA

radioimmunoprecipitation assay buffer

rm

recombinant mouse

SDS

sodium dodecyl sulfate

TCA

trichloroacetic acid

VSMC

vascular smooth muscle cell

Xplant

Transplant