Promyelocytic leukemia protein promotes the phenotypic switch of smooth muscle cells in atherosclerotic plaques of human coronary arteries

Weronika Karle1, Samuel Becker1, Philipp Stenzel2, Christoph Knosalla3,4, Günter Siegel1, Oliver Baum1, Andreas Zakrzewicz1 and Janine Berkholz1,4 1Institute of Physiology, Charité – Universitätsmedizin, Berlin, Germany; 2Institute of Pathology, University Medicine Mainz, Mainz 55131, Germany; 3Department of Cardiothoracic and Vascular Surgery, German Heart Center Berlin, Berlin 13353, Germany; 4DZHK (German Centre for Cardiovascular Research), Partner Site Berlin, Berlin 13353, Germany


Introduction
Atherosclerotic heart disease is one of the most frequent causes of death in humans worldwide [1]. Despite the ever-growing understanding of the molecular processes underlying atherosclerosis, the molecular factors that initiate blood vessel degeneration, i.e. the formation of plaques and their sudden breakage are still widely unknown. This lack of knowledge leads to limited treatment options and slows down the development of new drugs and therapies.
Morphological changes in atherosclerotic blood vessels are caused by inflammatory and fibroproliferative processes that facilitate the proliferation and migration of vascular smooth muscle cells (VSMCs),

Table 1 Clinical anthropometric information of the patients involved in the present study Comorbidities Sex
Ischemic heart disease (all were found to be atherosclerotic) Dilated cardiomyopathy> (all were found to be non-atherosclerotic) endothelial cells (ECs) and white blood cells (macrophages/monocytes), which, in addition to the deposition of subendothelial lipids, contribute to typical atherosclerotic lesions known as plaques [2,3]. During the formation of atherosclerotic plaques, the VSMCs from the arterial media undergo an extensive, not yet fully understood dedifferentiation process towards a proliferative phenotype and migrate from the media into the intima [3]. Especially inflammatory processes appear to play an important role in the development of advanced atherosclerotic lesions which are associated with the secretion of pro-inflammatory cytokines such as tumor necrosis factor α (TNF-α), transforming growth factor β (TGF-β), interferon γ (IFN-γ) and interleukins, preferably released by activated T cells and macrophages [4,5]. Interestingly, the expression of these cytokines is also related to the intracellular expression and signaling of the promyelocytic leukemia protein (PML) [6]. PML was originally identified in acute promyelocytic leukemia and was therefore primarily considered to act as a tumor suppressor [7]. Subsequently, PML was shown to play important roles in stress response, cell cycle regulation, apoptosis, senescence, transcriptional and post-transcriptional regulation, DNA repair, inflammatory responses and intermediary metabolism [8,9]. Remarkably, many of these cellular functions are involved in or altered during plaque formation in atherosclerotic vessels. PML expression itself is also modulated under various stress conditions such as inflammation [9,10]. In addition, a microarray study identified PML target genes that play a role in atherogenesis [11].
Together with a wide variety of over 100 functionally diverse proteins, PML forms dynamic aggregates inside the nucleus known as 'PML nuclear bodies' (PML-NBs), which are potent modifiers of proteins and their functions [8,12]. Almost all PML-associated proteins are modified by small ubiquitin-like modifier (SUMO), and SUMOylation of PML itself is essential for the integrity of PML-NBs [12]. SUMOylation is a post-translational modification which is characterized by reversible covalent binding of SUMO to target proteins and is involved in the regulation of protein-protein interactions, subcellular nuclear localization, protein-DNA interactions and enzymatic activity [13].
Based on its role in promoting inflammation and its established biochemical impact on SUMOylation, we hypothesized that PML plays a hitherto unknown function in the pathogenesis of atherosclerosis. To address this issue, we investigated whether PML is expressed in human coronary arteries, especially in association with atherosclerotic plaques, and assessed in a cell culture system if and how PML affects atherogenesis.

Human artery specimens
Coronary arteries were obtained from 16 patients receiving heart transplantation at the German Heart Institute Berlin. The patients with failing hearts had previously suffered from either dilated cardiomyopathy or ischemic heart disease ( Table 1). Samples were examined by means of conventional microscopy and were classified into atherosclerotic vessels with plaques or non-atherosclerotic vessels. Specimens of human coronary arteries from explanted hearts were analyzed under the written informed consent of the patients to their HTx.
Aorta tissue samples were provided by the tissue bank of the University Medical Center Mainz in accordance with the regulations of the tissue biobank and the approval of the ethics committee of the University Medical Center Mainz.

Transfection with small interfering RNA
Transfection of dHCASMCs was performed using a mixture of four unrelated small interfering RNA (siRNA) species (25 nM final) against the PML nucleotide sequence (Dharmacon, Lafayette, CO, U.S.A.; Cat. No: LQ-006547-00-0010). A non-gene-specific 'scrambled' siRNA was used as a negative control. HCASMCs were transfected using the transfection reagent Lipofectamine 2000 (Thermo Fisher Scientific, Waltham, MA, U.S.A.; Cat. No: 11668027) according to the manufacturer's protocol. Knockdown of the target mRNA was monitored 24 h after transfection of siRNAs by RT-qPCR and immunoblotting. dHCASMCs were transfected with 25 nM PML-specific siRNAs or scrambled siRNA, respectively, for 4 h to be then incubated with IFN-γ in a final concentration of 100 ng/ml for additional 24 h.

Immunohistochemistry with chromogenic substrates
Hematoxylin-Eosin (HE) and Elastica van Gieson (EvG) staining were performed on cross-sections (4 μm-thick) of formalin-fixed and paraffin-embedded vessel tissue. For immunohistochemistry, human coronary artery sections were incubated with the chromogenic substrates 3,3 -diaminobenzidine (DAB), Fast Red and Vector Blue.

Immunocytochemistry
Cellular localization of PML, SUMO-1 and α-smooth muscle actin (α-SMA) was assessed on cross-sections of human coronary arteries or fixed dHCASMCs by immunocytochemistry evaluated by confocal laser scanning microscopy, as previously described [23]. In brief, after blockage with 10% milk powder (AppliChem, Darmstadt, Germany; Cat. Sections and cells were assessed with a confocal laser microscope (Leica DMI 6000, Wetzlar, Germany) equipped with 20× and 63× oil immersion lens, a single photon argon laser, a solid-state laser and a helium-neon laser. At higher magnification, the microscopic setting was adjusted to the fluorescence signal. Digital images were processed using the Leica LAS AF Lite software.

RNA isolation, reverse transcription and real time PCR
Isolation of total RNA, reverse transcription and realtime-PCR (RT-qPCR) using the primers listed in Table 2 were performed, as previously reported [24]. RNA was extracted from dHCASMCs using the GeneMATRIX Universal RNA Purification kit from EURx (Gdansk, Poland) according to the manufacturer's instructions. mRNA of 1 μg total   Table 2. In each experiment, melting curve analysis was performed to verify that a single transcript was produced. RT-qPCR relative mRNA levels were calculated using the comparative C T (2 − C T ) method, with GAPDH as a reference. Non-RT and non-template controls were run for all reactions.

Scratch wound assay
Equal numbers of HCASMCs were grown to 70-80% confluence in small Petri dishes (Ø 4 cm) and then transferred to smooth muscle cell differentiation medium for 5 days. The monolayer was scratched with a pipette tip along a ruler 48 h after transfection. After 24 and 48 h of incubation at 37 • C in a humidified atmosphere (5% CO 2 ), images of the scratch were taken with a digital camera (Kappa, DX4-285fW, Turin, Italy). The size of the cell-free (open) area was calculated using analysis algorithm-based T-scratch software [25].

Proliferation assay
Two proliferation assays were performed. For both proliferation assays, HCASMCs were seeded at the same density on 96-well plates. The MTS dye (CellTiter96 proliferation assay, Promega Corporation, Madison, WI, U.S.A.; Cat. No: G3582) was added to the culture medium of dHCASMCs (48 h after transfection of full-length PML-IV or the corresponding control vector) and incubated for 2 h at 37 • C. The extinction was measured in triplicates by microplate reading (Tecan Sunrise, Männedorf, Switzerland) at 490 nm. Alternatively, the proliferation rate was measured using a BrdU ELISA kit (Abcam, Cambridge, U.K.; Cat. No: ab126556) according to the manufacturer's instructions.

Statistical analysis
Data were analyzed with SigmaPlot 13.0 software and presented as mean + − SD (n≥3). All datasets were tested by Shapiro-Wilk for their normality of distribution prior to statistical analysis and the Brown-Forsythe test was used to test the equality of variance. Comparisons between two groups were performed by Student's t test (two-tailed unpaired), between more than two groups by one-way ANOVA followed by Dunnett's test or by two-way ANOVA followed by Bonferroni's post hoc test. In case of non-normally distributed datasets or data without equal variance the Welch's t test or the Mann-Whitney U-test were performed. The differences were considered significant if P<0.05 in all statistical tests.

PML protein levels in atherosclerotic human coronary arteries
Using a specific ELISA, PML was quantitatively assessed in coronary arteries isolated from sixteen human hearts-seven vessels with plaques from patients with atherosclerosis and nine vessels without plaques from patients showing no signs of atherosclerosis when undergoing heart transplantation ( Table 1); 2.8-fold more (P<0.001) PML was found in lysates of the atherosclerotic arteries than in those of the non-atherosclerotic arteries ( Figure 1A). Six of the atherosclerotic arteries were divided in two segments of which both were subjected to the PML-specific ELISA: 3.1-fold more (P<0.01) PML was measured in the plaque-containing segments than the non-plaque segments of these arteries ( Figure 1B). Lysates of the coronary arteries that had been subjected to the ELISA were subsequently assessed by immunoblotting followed by densitometric analysis showing a 2.2-fold higher (P<0.001) level of PML protein in the six atherosclerotic coronary arteries than the six non-atherosclerotic ones ( Figure 1C). Furthermore, 3.3-fold more (P<0.001) PML protein was demonstrated in plaque segments than non-plaque segments of six coronary arteries isolated from the hearts of patients with atherosclerosis ( Figure 1D).

Immunohistochemical demonstration of cellular and subcellular PML expression in human arteries
To determine and compare the cellular PML localization in coronary arteries isolated from the transplanted hearts of the two patient groups, immunohistochemistry was conducted ( Figure 2; Supplementary Figure S1). In non-atherosclerotic arteries, moderate PML-immunoreactivity was confined to most cells present in the intima and the adventitia as well as weak PML-immunoreactivity in many cells of the media (Supplementary Figure  S1A2,A3,A5,A6). In atherosclerotic arteries, PML-immunoreactivity in the non-plaque regions corresponded to that observed in the non-atherosclerotic arteries (Supplementary Figure S1B2,B3,B5,B6). Most striking, however, was the PML-immunoreactivity in the thickened intima-media area of the plaque, with most of the cells there being strongly immunoreactive and showing the highest signal-density at the vulnerable shoulder region of the plaque cap (Supplementary Figure S1B5). Quantification in the cross-sections revealed a 1.7-fold higher (P<0.05) number of PML-immunoreactive cells in the intima-media area of atherosclerotic arteries than in non-atherosclerotic arteries (Supplementary Figure S1D).
Double immunohistochemistry with antibodies against PML and markers specific for the three cell types most frequently present in plaques (ECs, smooth muscle cells (SMCs) and macrophages) was performed to determine the cell type-specific distribution of PML in atherosclerotic coronary arteries in more detail (Figure 2). In the non-plaque area, ECs of the intima as well as many SMCs of the media were clearly immunoreactive (Figure 2C,E,G). In the plaque area, α-SMA-positive cells, which were the majority of the cells found in the plaque area (Supplementary Figure S1B2) and macrophages were strongly PML-positive ( Figure 2B,D,F). If the signal density of PML-immunoreactivity was compared between the atherosclerotic and the non-atherosclerotic regions, α-SMA-immunoreactive cells showed the greatest differences in PML-signal intensities: α-SMA-immunoreactive cells of the intact media contained less PML-immunoreactivity than the α-SMA-immunoreactive cells present in the thickened intima-media of the plaque area ( Figure 2).
Higher microscopic magnification revealed most of the PML-immunoreactivity in the three cell types to be consistently localized in intranuclear aggregates likely representing PML-NBs ( Figure 2H-K). Strikingly, in the α-SMA-immunoreactive cells in the thickened intima-media of the plaque area, more and larger PML-NBs were visible than in the α-SMA-immunoreactive cells of the media ( Figure 2J,K).
Taken together, the immunohistochemical analyses revealed that 1. PML is expressed in the three major cell types present in atherosclerotic coronary arteries and 2. significantly higher PML levels were detected in α-SMA-positive cells of the plaque than in α-SMA-positive cells of the media.
To examine the cell type-specific distribution pattern of PML in another type of atherosclerotic vessel besides the coronary artery, a section of a human aorta was subjected to immunohistochemistry. All three layers of the blood vessel contained cells with strong PML-immunoreactivity in the nuclei, such as ECs (data not shown), SMCs of the media (Supplementary Figure S2C) and macrophages (Supplementary Figure S2D). Thus, the cellular expression pattern of PML in the atherosclerotic aorta corresponds to that in atherosclerotic coronary arteries.

PML promotes the phenotypic switch of coronary artery SMCs
To investigate whether PML plays a role in the phenotypic switch in which SMCs of the media transform from a contractile (or differentiated) phenotype to a synthetic (or dedifferentiated) phenotype, a cell culture model with primary VSMCs isolated from human coronary arteries (HCASMCs) was implemented. Therefore, the primary HCASMCs  (pHCASMCs) were cultivated for 5 days in differentiation medium to obtain populations of differentiated HCASMCs (dHCASMCs), as shown in Supplementary Figure S3A. Compared with the pHCASMCs, dHCASMCs actually expressed higher concentrations of marker genes specific for cell differentiation such as smooth muscle 22 α (SM22α; ∼480%, P<0.01), α-SMA (∼650%, P<0.01) and smooth muscle myosin heavy chain (SM-MyHC; ∼340%, P<0.01) at the mRNA level as well as α-SMA (∼1050%, P<0.001) and calponin (∼1300%, P<0.01) at the protein level (Supplementary Figure S3B,C). Next, dHCASMCs were transfected with the expression plasmid pEGFP-C1 containing either the complete PML-IV gene or lacking a specific gene insert as control (Figure 3, Supplementary Figure S4). The PML-overexpressing dHCASMCs did not show the SMC-characteristic spindle-like, elongated shape and were not arranged in parallel as the control-transfected dHCASMCs. They furthermore exhibited less pronounced α-SMA immunoreactivity compared with the control cells ( Figure 3B). Accordingly, lower levels of α-SMA (∼50%, P<0.01; Figure 3D) and the contractile protein calponin 1 (∼60%, P<0.05; Figure 3E) were found in the cells overexpressing PML versus control-transfected dHCASMCs. RT-qPCR showed that mRNA levels of markers for SMC differentiation (α-SMA, SM22α, CNN1, SM-MyHC) were lower (∼10, ∼15 ∼20, ∼30%; P<0.05) and the connexin 43 (Cx43) mRNA levels were higher (∼420%, P<0.01) in the PML-overexpressing dHCASMCs than the control-transfected cells ( Figure 3C). Furthermore, in dHCASMCs transfected with the PML gene, proliferation was increased to ∼150% ( Figure 3F, P<0.01). In contrast, knockdown of PML with specific siRNAs resulted in lower proliferation of dHCASMCs ( Figure 3F, ∼25%, P<0.05). In addition, scratch closure (i.e. migration) was increased to ∼300% 48 h after scratching ( Figure 3G, P<0.01). In summary, these experiments revealed that characteristic markers for differentiation are lower and markers for proliferation and migration are higher in dHCASMCs transfected with the PML-vector than in those transfected with the empty vector.

PML increases the SUMO-1-dependent SUMOylation of proteins in cultured coronary artery SMCs
In the next step, dHCASMCs overexpressing the PML-IV gene were subjected to co-immunocytochemistry with PML and SUMO-1 antibodies. A co-localization of PML and SUMO-1 was observed within many PML-NBs of these cells ( Figure 4A). Interestingly, overexpression of PML by transfection with the PML-IV gene did not change the SUMO-1 mRNA levels (∼11% increase, P≥0.05, Figure 4B). On the other hand, immunoblotting with anti-SUMO-1 antibodies revealed more SUMO-1-dependent protein SUMOylation in PML-IV-transfected dHCASMCs and less SUMO-1-dependent protein SUMOylation in PML-siRNA transfected dHCASMCs compared with control-transfected dHCASMCs ( Figure 4C). Together, these findings show that SUMO-1-dependent protein SUMOylation in dHCASMCs is related to PML expression.
To investigate whether dedifferentiation, proliferation and migration of dHCASMCs depend on the SUMOylation state and SUMOylation activity of PML, dHCASMCs were transfected with either a vector without a specific gene insert or a mutated PML-IV gene that is translated into a PML-IV protein with three altered lysines within the SUMOylation consensus motif (K65, K160, K490) and thus lacking SUMOylation activity. The qualitative and quantitative SUMO-1-dependent patterns of SUMOylated proteins were similar in the dHCASMC populations subjected to transfection with the vector containing the gene for the mutated SUMOylation-defective PML-IV form or with the vector lacking a specific gene insert (Supplementary Figure S5A). Furthermore, neither the expression levels of the differentiation marker α-SMA protein ( Figure 4D) nor the proliferation nor the migration activity of the cells ( Figure 4E, Supplementary Figure S5B) differed (P≥0.05) between the two dHCASMCs transfections. These results imply that the PML protein, increased by overexpression, must undergo SUMOylation to induce dedifferentiation, proliferation and migration of dHCASMCs.

SUMO-1 is highly expressed and co-localizes with PML in atherosclerotic human arteries
As for PML, SUMO-1 levels in atherosclerotic arteries exceeded SUMO-1 levels in non-atherosclerotic arteries by ∼5-fold (P<0.001) in immunoblots ( Figure 5A,B).
In non-atherosclerotic arteries, SUMO-1 immunoreactivity was mainly located at the luminal side of the internal elastic lamina (IEL), i.e. in the intima ( Figure 5C, upper row). In the case of an intima-media thickening, a high signal-density of SUMO-1 immunoreactivity was found in the plaque region of atherosclerotic arteries, especially at the plaque cap ( Figure 5C, lower row).  In addition, immunofluorescence double labeling demonstrated that PML and SUMO-1 are co-localized in a variety of cell types that are present in atherosclerotic and non-atherosclerotic arteries as well as in adventitial capillaries ( Figure 5D).

IFN-γ increases expression of PML and SUMO-1 in dHCASMCs
Since PML expression is induced by different cytokines in a variety of cells, the effect of three cytokines on dHCASMCs phenotype was evaluated. For this purpose, dHCASMCs were exposed to TNF-α, TGF-β or IFN-γ for 24 h to be subsequently analyzed by immunocytochemistry, RT-PCR and immunoblotting. Immunocytochemistry revealed higher signal-density for nuclear PML-NBs and more co-localization of SUMO-1 after incubation especially with IFN-γ compared with untreated control cells ( Figure 6A). Counting of the PML-NB numbers in the dHCASMCs exposed to the various cytokines showed higher numbers (P<0.05) of PML-NBs in the cells incubated with TNF-α (∼25%) and IFN-γ (∼40%) while stimulation with TGF-β (∼10%, P≥0.05) did not cause any significant changes ( Figure 6B). Immunoblotting (∼280%, Figure 6D) and RT-PCR analyses (∼320%, Figure 6C) likewise revealed a significant increase in PML after IFN-γ stimulation, which was accompanied by suppression of α-SMA transcription (∼20%, P<0.05, Figure 6E,F), while TGF-β stimulation resulted in higher (220%; P<0.05) α-SMA mRNA levels ( Figure 6E). The down-regulation of α-SMA mRNA levels observed after IFN-γ incubation was abolished if PML mRNA expression was knocked-down with specific siRNAs (∼40%, P<0.01, Figure 6F). A similar effect could also be observed for the mRNA level of SM-MyHC (∼45%, P<0.01, Figure 6G). Remarkably, PML expression in dHCASMCs did not have significant effects (P≥0.05) on mRNA and protein secretion levels of TNF-α, IFN-γ or TGF-β (Supplementary Figure S6). We hypothesize that the transfection-induced up-regulation of PML expression resulted in more PML-NB aggregates into which soluble SUMO-1 was recruited. As a result, higher SUMOylation rates were achieved in dHCASMCs even without induction of SUMO-1 mRNA expression. In addition, since PML itself exhibits SUMO E3 ligase activity [35] which causes covalent SUMO-binding to protein substrates, the higher availability of PML after transfection should lead to more SUMOylation activity even without changes in SUMO-1 mRNA expression levels. Accordingly, we interpret the results of the experiments with PML siRNA, in which the PML availability revealed to be down-regulated, as a reduction in SUMOylation platform activity. If dHCASMCs were transfected with a gene coding for a SUMOylation defective PML-IV mutant that lost its ability to recruit specific interacting proteins despite the formation of morphological normal PML-NBs [12,36], neither α-SMA protein levels nor cellular proliferation and migration were changed. These results indicate that SUMO-1-dependent SUMOylation of PML is a prerequisite for the PML-induced phenotypic switch of VSMCs. The SUMOylation site of PML could thus be a new target to prevent the progress of atherosclerosis.
The expression profile of marker genes for the contractile phenotype is regulated in VSMCs by a network of transcription factors and cofactors. Interestingly, some of these transcription factors such as serum response factor (SRF), Kruppel-like factor 4 (KLF4), myocardin and protein Elk-1 (ELK1) containing ETS domains are known to be modified by SUMO-1 [37][38][39][40]. In addition, some of these proteins have been shown to colocalize with PML-NBs, such as SRF whose activity is inactivated by its SUMOyation [37]. Another factor that is important for the regulation of differentiation is PIAS1, which plays a role in the control of SMC-selective gene expression, when it interacts with PML-NBs to act as SUMO E3 ligase within the SUMOylation cascade [41,42]. The disruption of the SUMO E3 ligase activity of PIAS1 abolishes its ability to activate the α-SMA promoter [43].
PML-NBs and SUMO-1 co-localized not only in cultivated VSMCs but also in human coronary arteries in situ. This observation underlines a possible role of PML-NBs in controlling post-translational SUMOylation of proteins during the formation of atherosclerotic plaques.
Since PML expression is induced by different cytokines [6,44] and cytokines influence the growth of SMCs [45], dHCASMCs were exposed to TNF-α, TGF-β and IFN-γ. Stimulation with IFN-γ increased the expression of PML as well as number and size of PML-NBs, which was accompanied with enhanced co-localization of SUMO-1 to PML-NBs.
The pro-inflammatory cytokine IFN-γ is secreted by activated T-cells and therefore present at high levels in atherosclerotic lesions. A higher IFN-γ availability could cause the up-regulation of PML in VSMCs and thus induce or at least promote their phenotype change. Interestingly, it has been previously shown that IFN-γ suppresses the expression of α-SMA, at least in arterial SMCs and myofibroblasts [46,47], with the signaling pathway being unidentified so far. As our study demonstrated, the mRNA transcription level of α-SMA in dHCASMCs was reduced after stimulation with IFN-γ. On the contrary, as already shown in the literature before, treatment of SMCs with TGF-β elevated their level of α-SMA [48,49]. To elucidate whether the known IFN-γ-mediated suppression of α-SMA was due to the effect of IFN-γ on PML, PML was knocked-down by the use of specific siRNAs prior to stimulating the cells with IFN-γ. This showed that the effect of IFN-γ on α-SMA was mediated by PML, at least in the model tested. A similar influence of PML was also demonstrated on the IFN-γ-mediated suppression of SM-MyHC. Thus, our molecular studies suggest that PML is critically involved in IFN-γ-stimulated pathways of VSMCs dedifferentiation, as it occurs during atherogenesis.
Taken together, the findings of the current study support the hypothesis that PML promotes atherogenesis by promotion of the phenotypic switch of VSMCs from the 'contractile' to the 'synthetic' state, which included dedifferentiation as well as cellular migration and proliferation. PML thereby served as a mediator of IFN-γ-induced upstream effects. Co-localization of PML-NBs with SUMO-1, as known from other tissues, supported the speculation that SUMOylation is required as downstream signal in this PML-dependent pathway. Remarkably, neither TNF-α nor IFN-γ nor TGF-β were influenced by PML in dHCASMCs suggesting that these cytokines are only acting as regulators but not as downstream mediators of PML expression.
In the present study, we combined the in vivo analysis of PML expression in atherosclerotic coronary artery samples of patients with the phenotypic characterization of dHCASMC batches that were transfected with genes coding differently modified PML variants. To validate these two sets of findings, it would be consequent to carry out studies on transgenic animals. In fact, a knockout-mouse strain has been developed that lacks total PML expression in all tissues and cells, which did not result in striking phenotypic changes of the animals [50]. However, since we have investigated the special issue of whether PML-IV is involved in the phenotypic switch of SMC-derived cells in atherosclerotic plaques, it would not be sufficient to use total PML knockout-mice for the validation of the key observations of the present study. Instead, it would be necessary to work experimentally with conditional PML knockout-mice, PML-sensitive atherosclerosis-prone mice (e.g. PML/ApoE double knockout-mice) or PML-isoform-specific knockout-mice, which, to the best of our knowledge, have not yet been generated. And even if such genetically engineered mice were available, they would not necessarily be useful tools to adequately mimic the hemodynamic conditions in human physiology (e.g. mice do not develop plaque ruptures or coronary lesions in their vasculature [51][52][53]). We therefore decided not to extend the study to the analysis of transgenic mice.
The present study is focused on the biochemical analysis of PML-transfected dHCASMCs, which are originally derived from VSMCs and represent an experimental model for the cell type most commonly found in atherosclerotic plaques. However, significant levels of other cell types, such as ECs or macrophages, also occur in these plaques. PML in macrophages selectively activates the NLRP3 inflammasome [54], while PML in ECs influences angiogenesis and cell migration [55,56]. Inflammation, angiogenesis and cell migration are cellular processes associated with atherosclerosis [3,4]. We therefore speculate that the pathogenesis of atherosclerosis is exacerbated if PML is also up-regulated in these cells being present in atherosclerotic plaques besides VSMCs.
Some biological and methodological constraints may limit the significance of our findings: (1) although the effects described here were statistically significant already with coronary artery specimens of 16 patients, a higher number of samples/subjects would further increase the validity of the quantitative data. (2) The coronary artery specimens analyzed in the present study were derived from patients suffering from complex advanced stages of atherosclerosis. Thus, the plaques may have grown over decades. It is therefore difficult to make a valid statement about the role of PML and PML-NBs in the etiology and course/progression of atherosclerosis. (3) Furthermore, PML-NBs are very complex and highly dynamic subnuclear structures and, thus, may differ in composition/function in plaque cells at different stages of atherosclerosis. Future studies should therefore focus on a more detailed analysis of such changes in the PML interactome during the time course of atherogenesis.
Our study clearly demonstrated that PML is a hitherto unknown actor in the development of atherosclerosis, which may be important for a better understanding of the progression of this disease and the development of additional treatment options.

Clinical perspectives
• Since a growing body of evidence suggests that PML is a key regulator of inflammatory responses, it has been investigated whether PML plays a role in atherogenesis.
• In human atherosclerotic coronary arteries, expression of PML was significantly higher, especially in α-SMA-immunoreactive cells of plaques where it was co-localized with SUMO-1. In cultured dHCASMCs PML increased global SUMOylation pattern as downstream event and triggered a phenotypic switch by promoting dedifferentiation as well as proliferation and migration of the cells.
• Identification of the PML-dependent SUMOylation pathway as a so far unrecognized modulator system for the phenotypic switch of VSMCs in atherosclerotic plaques that may facilitate the development of new tools for the biochemical characterization and histological diagnosis of atherosclerosis.
In addition, our results may offer new therapeutic options to treat atherosclerosis using specific drugs, e.g. already existing SUMOylation inhibitors.

Data Availability
All data generated or analysed during the present study are included in this published article and its supplementary files.