AAV2/8-hSMAD3 gene delivery attenuates aortic atherogenesis, enhances Th2 response without fibrosis, in LDLR-KO mice on high cholesterol diet

TGFβ1 is an important anti-inflammatory cytokine and is protective against various cardiac pathologies. It has been identified to be protective against myocardial ischemia-reperfusion injury [1],[2], apoptosis of cardiomyocytes [3], pro-inflammatory adhesion molecule expression on the vascular endothelial cells [4], and foam cell formation [5]. It also inhibits ox-LDL-induced expression of adhesion molecules in endothelial cells [6]. Furthermore, it is well established that TGFβ1 is a strong inhibitor of immune response [7],[8]. Taken together, TGFβ1 would seem to be a logical candidate for an effective anti-atherogenesis gene. In a previous study TGFβ1^ACT gene delivery has shown efficacy against atherosclerosis in the low density lipoprotein receptor knockout (LDLR-KO) mice on high-cholesterol diet (HCD) [9]. While we did not observe significant adverse effects upon TGFβ1^ACT–gene delivery into LDLR-KO/HCD animals, we did find certain higher cytokine levels. Moreover, higher TGFβ1 expression is well known to be linked to adverse reactions such as fibrosis, infection, cancer and increased infections, coupled with unwanted immune effects (reviewed in [10]).

Another approach to use TGFβ1 therapeutically might be to utilize the genes down-stream from TGFβ1 in its signal transduction pathway(s), through which TGFβ1 acts. Previously, for example, we have shown that STAT3 gene delivery, down-stream of interleukin 10 (IL-10), can substitute for IL-10 gene delivery in the inhibition of atherosclerosis in a mouse model [11]. These downstream genes would then code for intracellular proteins, usually transcription factors, and have a more limited effect than a systemically secreted protein. The mechanism would be that such genes might enhance the effects of low level endogenous TGFβ1, and take the place of direct TGFβ1^ACT gene delivery [9]. Thus, hopefully, by going through the most desirable TGFβ1^ACT signal transduction pathway, undesirable side effects (eg. fibrosis) would be avoided. The anti-inflammatory abilities of TGFβ1 work through a number of signal transduction pathways, including Ras-ERK, TAK1-JNK Rho-Rac-cdc42, mothers against decapentaplegic homologs (SMADs) 2, 3, 4, and others [9]. While SMAD2, SMAD3 or SMAD4 might all be proposed as substitute agents for TGFβ1, perhaps the most interesting and appropriate of these agents may be SMAD3 [12]. SMAD3 knockout (KO) mice have a phenotype that closely mimics that of the TGFβ1 KO mouse [13]. Moreover, SMAD3 KO mice display much higher and systemic inflammatory cell infiltration [12]. While many studies suggest TGFβ1 acts through SMAD3/SMAD4, other studies suggest SMAD3 can act without SMAD4, and can even compete with SMAD4 for binding both protein and DNA targets [14]. This is likely due to SMAD3’s ability to either homodimerize or heterodimerize with SMAD4.

Thus we hypothesized that SMAD3 might be a reasonable intracellular substitute for TGFβ1 with a more specific signal transduction effect. It is unclear at what level SMAD3 is basally expressed in endothelial and smooth muscle cells, but it is clear that SMAD3 can be induced in these cells [15],[16]. Thus, by inducing higher constitutive SMAD3 levels in cardiovascular tissues, in particular, aortic smooth muscle cells (the main target of adeno-associated virus in arteries, AAV) [17], we should be able to enhance the effects of endogenous secreted TGFβ1. Adeno-associated virus (AAV), first investigated in 1984, is a useful tool for gene delivery to study gene function / therapeutic effect [18]-[21], and its expression is known to last at least 10 years in clinical trials [22]. The predicted amino acid (aa) sequence homology of mouse and human (h) SMAD3 is 99%, thus, hSMAD3 was the choice as a therapeutic gene. In this study we delivered the hSMAD3 gene using AAV type 8 (AAV8) capsid, which has been shown to be effective in gene delivery into cardiovascular tissues by ourselves and others [23]-[27]. Here we observed that AAV/hSMAD3 delivery resulted in efficacy in inhibiting atherosclerosis in low density lipoprotein receptor knockout (LDLR-KO) mice on high cholesterol diet (HCD), but without the concomitant fibrosis associated with TGFβ1.

The TGFβ1/SMAD signaling pathway has been shown to mediate immunosuppressive responses. This fact suggests that SMAD3, one member of these down-stream genes, might limit atherogenesis through its anti-inflammatory effects. However, TGFβ1/SMAD3 signaling is also pro-fibrotic [10]. Using AAV-based gene delivery, we studied which of these effects was dominant in vivo and if hSMAD3 might be an intra-cellular, non-systemic substitute for TGFβ1 in treating/inhibiting atherogenesis. As defined by larger aortic lumen, thinner aortic walls, by lower systolic blood velocity, and lower macrophage burden, hSMAD3-gene delivery clearly resulted in a significant anti-atherogenic effect. While AAV8-delivered hSMAD3 expression was only approximately 0.3% mRNA expression compared to endogenous βactin expression, assuming that the average aortic cell (smooth muscle) translates the βactin and the hSMAD3 mRNAs at roughly the same efficiency and are roughly equivalent to the average cultured fibroblast [31], then a 0.3% expression level appeared to result in, approximately 400 hSMAD3 molecules per minute per cell produced. However, this analysis doesn’t address that the transduction level of AAV8 after 20 weeks (less than 100%), nor the likelihood that smooth muscle cells express more βactin than fibroblasts. Thus the level of hSMAD3 being produced specifically in AAV-transduced cells is likely to be multiples higher than the 400 molecules per minute per cell.

The level of hSMAD3 expression clearly affected intra-aortic immune response. We observed that IL-4 was significantly up, as expected, suggesting that Th2 response is favored over Th1 response. IL-10, another Th2 cytokine, also trended higher, in agreement with IL-4. Correspondingly, IL-12, a Th1 cytokine trended lower in the hSMAD3-treated liver, consistent with higher Th2 cytokines. The expression of IL-7, was technically higher in hSMAD3-treated aortas, but actually levels were little changed. Thus, overall, the delivery of hSMAD3 resulted in a pro-Th2 intra-aortic response, as is known for TGFβ1 [32], and is believed beneficial for countering atherogenesis [33].

While SMAD3 is most often linked with TGFβ1 signaling, there are other ligands which also involve SMAD3 in their signal transduction pathways. One example is activin, another member of the TGFβ super family. Its effects are on the reproductive system, insulin and muscle metabolism, and enhancement of fibrosis [34]. However, these pathways are not as extensively studied as TGFβ1. Additionally, angiotensin II (Ang II), a well known blood pressure regulator of the renin-angiotensin system, also signals though SMAD3, but in a less direct manner. Ang II signaling through AT1R, down-regulates SMAD7, a negative regulator of SMAD3, and results in higher SMAD3 signaling. Yet, again, fibrosis is enhanced by Ang II [35]. Other ligands, such as inhibin may also signal through SMAD3. Thus, SMAD3’s role in signal transduction has been described in the title of one review article as having “smaddening complexity” [36].

The purpose of using hSMAD3 gene delivery was to provide the therapeutic effect of TGFβ1, but with lower adverse effects of fibrosis, cancer and infections. While we can only study cancer and infection levels with great difficulty in our model, we can more readily study fibrosis. A major finding of this study was that collagen and connective tissue growth factor (CTGF) expression was unaffected or lower in the aorta, and significantly lower in the liver, after AAV/hSMAD3 delivery (Figure 6). We were surprised to find collagen 1A2 expression was the same in Neo-HCD and hSMAD3-HCD-treated aortas. Supporting these data, collagen 2A1 was significantly higher in Neo-HCD than in hSMAD3-HCD-treated aortas. Consistent with these data, CTGF, known to induce SMAD3-associated fibrosis, was unchanged throughout all groups. Even more telling, collagen 1A2, collagen 2A1, and CTGF expression were significantly lower in hSMAD3-HCD-treated livers. There is evidence that SMAD3, as it lowers fibrosis, may also lower cancer. SMAD3 knockout mice develop various cancers and SMAD3 expression is known to be cell cycle regulated by ras [37]. Thus, a simple explanation may be that our gene delivery of constitutive SMAD3 expression gives a constant anti-proliferative effect, with fibrosis being down-regulated along with cell proliferation. Leivonen et al. [38] found that SMAD3 is specifically required (not SMAD2 or SMAD4) for the induction of matrix metalloproteinase-13 (MMP-13) by TGFβ1 [38]. Higher MMP-13 levels are also associated with lower fibrosis [39]. This latter signaling pathway and phenotype is also more consistent with what we observe.

It must also be mentioned that our results are in contrast with those of Kundi et al., [40], who found increased fibrosis after adenovirus-based gene delivery of SMAD3, following carotid injury in rats [40]. However, the adenovirus vector used in the Kundi study is well known to cause inflammation, NFkB induction and fibrosis during gene therapy experiments as well as clinically [41]-[44]. Thus, when issues of inflammation and fibrosis are possible the use of adenoviral vectors would not seem to be ideal. It should also be mentioned our results are consistent with those of Meng et al. 2012 [13], who found that SMAD3-dimers could translocate to the nucleus without SMAD4, and that SMAD3 were defective in activating the COL1A2 promoter (and perhaps others). As SMAD3 and SMAD4 recognize and bind the identical palindromic promoter sequence during transcriptional regulation (14), this suggests that the effects we see on COL1A2 expression may simply be due to SMAD3 homodimers inhibiting the formation of transcriptionally active SMAD3/SMAD4 heterodimers.

In conclusion, AAV-based hSMAD3 gene therapy exceeded our expectations in providing the therapeutic “good face” of TGFβ1 over that of the bad (no fibrosis). hSMAD3-therapeutic gene delivery was successful in reducing the pathology of HCD, a prototype Western diet in LDLR KO mice, both in reducing atherogenesis and enhancing Th2 response. Yet, hSMAD3 delivery did this without inducing the serious TGFβ1-associated adverse side effect of fibrosis as measured by CTGF and collagen expression. As of now, our main hypothesis is that the effects of AAV/hSMAD3 gene delivery are driven by an increase in nuclear SMAD3 homodimers and their resulting changes in transcriptional regulation. Overall effects of hSMAD3 gene delivery were without documented or noticed side effects. However, atherosclerosis was not fully inhibited. Perhaps increasing the extent of hSMAD3 gene delivery, or improving it’s level of expression can result in further down-regulation of the disease state. Thus, further studies into the use of AAV-based hSMAD3 gene delivery are warranted.

This animal study focused on the gene therapy manipulation of the very powerful TGFβ1 signal transduction pathway for inhibiting atherosclerosis, by the delivery of the human SMAD3 gene. SMAD3 is one of the transcription factors through which TGFβ1 acts. We found that AAV2/8-hSMAD3 delivery did give efficacy in inhibiting HCD-induced atherosclerosis in LDLR KO mice. Moreover, significantly increased fibrosis was not observed as is usually the case when direct, primary TGFβ1-signalling is stimulated. Previously, we have shown that, analogous to this study, AAV/STAT3 gene delivery, with STAT3 being down-stream of interleukin 10 (IL10), is similarly able to substitute for IL10, again, for inhibiting atherosclerosis. Thus, through this strategy of using downstream signal transduction genes in place of their powerful primary chemokines, we might be able to effect superior treatment results.

HZ and MC carried out the experiments and edited the manuscript, MCI, JAF, EC, BFU edited the manuscript and data/conclusions, PLH conceived of and designed this study, interpreted the data, and wrote the manuscript. All authors have read and approved the final manuscript.