SMAD6 is required for homeostatic laminar flow-induced endothelial cell alignment and polarization
To begin to understand how endothelial cell flow-mediated responses that characterize vascular homeostasis are maintained, we asked whether an inhibitory SMAD, SMAD6, regulates endothelial cell responses to extended periods of laminar shear stress. Smad6−/− mutant embryos and pups with a knock-in lacZ reporter in the Smad6 locus strongly express lacZ in endothelial cells of embryonic and early post-natal arteries, but not veins of similar diameter and stage, after the onset of blood flow [42, 43]. Since shear stress induced by laminar flow is higher in arteries, this suggests that endothelial SMAD6 expression is induced by homeostatic laminar flow. Thus, we chose conditions of 15 dynes/cm2 (d/cm2) for 72 h for analysis of SMAD6 function in homeostatic arterial flow responses. Under these conditions, both human umbilical vein endothelial cells (HUVEC) and human arterial endothelial cells (HAEC) showed a two to threefold increase in Smad6 RNA (Supp. Fig 1A, B), consistent with a previous report .
We next examined the function of SMAD6 in endothelial cell responses to homeostatic laminar flow, by interrogating HUVEC and HAEC depleted for Smad6 RNA via knockdown (KD) and exposed to flow. We found that both venous and arterial endothelial cells with reduced Smad6 RNA levels failed to align under homeostatic laminar flow, as indicated by cell axis ratio and nuclear displacement angle measured parallel to the direction of flow (Fig. 1a, b, Supp. Fig. 1C–F). Endothelial cell misalignment in response to homeostatic laminar flow occurred with multiple siRNAs targeting Smad6 (Smad6-1 and Smad6-2), including a Smad6 siRNA pool (Smad6-3) (Supp. Fig. 1G, H). Both venous and arterial endothelial cells depleted for Smad6 also had mis-positioned Golgi and centrosomes under flow compared to controls (Fig. 1c, d, Supp. Fig. I–K, not shown), indicating that SMAD6 is important for endothelial cell polarization in response to homeostatic laminar flow. These results show that endothelial cell SMAD6 expression is flow-regulated and induced by homeostatic laminar flow, and that SMAD6 is functionally required for homeostatic flow-mediated endothelial cell alignment and polarization.
SMAD6 is required downstream of Notch for flow-mediated alignment of endothelial cells
SMAD6 is both a negative regulator of BMP signaling and a transcriptional target of the pathway [38, 44,45,46]. Consistent with the importance of canonical BMP signaling for endothelial cell flow alignment , endothelial cells were misaligned under homeostatic flow when incubated with the BMP inhibitor Crossveinless-2 (CV2) (Supp. Fig. 2A, B). However, since BMP receptors are not identified as direct mechanotransducers of flow-mediated signals, we searched for another pathway more directly linked to mechanotransduction that regulates SMAD6. Notch1 is a direct mechanotransducer of flow-mediated signaling [21, 22], and Notch regulates SMAD6 expression under static (non-flow) conditions , leading us to hypothesize that SMAD6 functions downstream of Notch in endothelial cell responses to homeostatic laminar flow. Reduced Notch signaling, via siRNA depletion of Notch1 or by treatment with the γ-secretase inhibitor DAPT which blocks Notch signaling, prevented endothelial cell alignment under homeostatic laminar flow (Fig. 2a, b, Supp. Fig. 2C, D). The misalignment induced by reduced Notch was accompanied by reduced expression of Smad6 RNA relative to controls under flow (Fig. 2c, f), indicating that SMAD6 expression levels are regulated downstream of Notch in endothelial cells and are important for homeostatic flow alignment. Consistent with this idea, ectopic SMAD6 expression rescued the flow-mediated misalignment of endothelial cells downstream of reduced Notch. Upon Notch blockade via DAPT, control endothelial cells transfected with empty vector remained misaligned when exposed to homeostatic laminar flow and did not differ from nearby untransfected cells; in contrast, similarly treated endothelial cells expressing SMAD6 aligned in response to homeostatic laminar flow while nearby untransfected cells remained misaligned (Fig. 2d, e). Endothelial cells over-expressing SMAD6 after Notch1 KD also aligned to homeostatic laminar flow while nearby cells remained misaligned (Fig. 2g, h). Thus SMAD6 over-expression rescued Notch loss-induced endothelial cell misalignment, indicating that SMAD6 is a functional effector of Notch-mediated homeostatic flow alignment in endothelial cells.
We next asked which Notch signaling components are required for homeostatic endothelial cell flow alignment. RPBJ is a transcriptional co-activator required for canonical downstream Notch signaling, and endothelial cells with reduced levels of RPBJ had reduced expression of Smad6 RNA (Supp. Fig. 2E). Moreover, RPBJ knockdown led to misalignment of both arterial and venous endothelial cells in response to homeostatic laminar flow (Fig. 2a, b; Supp. Figure 2F–I). Several Notch ligands are expressed in endothelial cells and implicated in Notch responses to flow. Reduced levels of Dll4, but not Jagged1 or Jagged2, resulted in endothelial cells misaligned in response to homeostatic laminar flow (Fig. 2a, b; Supp. Fig. 2F, G), indicating that Dll4-mediated activation of Notch1 signaling is important for homeostatic endothelial cell flow alignment via canonical Notch signaling.
The misalignment induced by RPBJ or Dll4 knockdown was rescued by expression of SMAD6 in HUVEC (Fig. 3a, b), and SMAD6 expression also rescued alignment of arterial endothelial cells with homeostatic laminar flow after RPBJ KD (Supp. Fig. 2H, I), suggesting that Notch regulation of SMAD6 expression is an important component of endothelial cell responses to homeostatic flow. The SMAD6 protein consists of two major domains connected by a linker (Supp Fig. 2J); the N-terminal portion includes several arginine residues that are methylated to regulate SMAD6 activity , while the C-terminal portion contains the MH2 protein-interacting domain . Since both major domains are required for the regulatory role of SMAD6 , we hypothesized that SMAD6 rescue of homeostatic endothelial cell alignment downstream of Notch required full-length SMAD6. Reduced Notch signaling, either via Notch1 or RPBJ depletion, led to endothelial cell misalignment in response to homeostatic flow that was not rescued by expression of constructs encoding only either the N-terminal or C-terminal portion of SMAD6 (Fig. 3c–f). These data indicate that full-length SMAD6 is required to mediate the effects of Notch signaling on homeostatic endothelial cell flow alignment.
SMAD6 regulates endothelial cell proliferation
To better understand the effects of reduced SMAD6 function on flow-mediated endothelial cell responses, we examined the transcriptome of endothelial cells under homeostatic laminar flow relative to non-flow conditions, and with depleted Smad6 levels, via RNA-seq analysis. Pearson Correlation Analysis revealed good correlation between experimental replicates of each condition (Supp. Fig. 3A). Principle Component Analysis (PCA) distinguished transcriptomes of control (non-targeting siRNA) and SMAD6 depleted endothelial cells, and transcriptomes also clustered by flow status (Supp. Fig. 3B). Overall comparisons (33,694 genes) showed that, when binned by flow status, only 1.2% of transcripts were significantly up- or down-regulated with reduced Smad6 levels under non-flow (static) conditions, while 6.9% of transcripts changed with reduced Smad6 levels under homeostatic flow conditions (Supp. Table 1). When binned by depletion condition, control endothelial cells under static vs. flow conditions had 6.7% of transcripts showing expression level changes, while Smad6 KD cells had 8.9% of transcripts changed in static vs. flow conditions. These numbers suggest that the magnitude of flow-mediated changes is greatest in endothelial cells with reduced Smad6 levels, consistent with a role for SMAD6 in endothelial cell flow responses.
Despite the fact that flowed EC with Smad6 KD morphologically resembled non-flowed EC [NT (static) vs. Smad6 KD (flow)], 7.9% of analyzed genes (2655/33,694) differed in relative expression between these gene sets, even more than the 6.7% that differed between static and flow without Smad6 manipulation (Supp. Table 1). Finally, the greatest change in gene expression (11.2%) was seen when comparing flowed normal EC to non-flowed EC with Smad6 KD (Supp. Table 1). We next looked at individual gene sets, and found that a significant number of genes normally flow-responsive (up or down) became non-responsive with Smad6 KD (1041/2262, 46%), and that an even larger group of genes that were normally non-responsive to flow became flow-responsive (up or down) with Smad 6 KD (1782) (Supp. Fig. 3C–F shows top 50 genes/category; Supp Fig. 3G shows overlap). Finally, many genes that appeared concordant in direction of flow-responsiveness between control and Smad6 KD had baseline (non-flow) changes (up or down) that led to expression changes under flow in Smad6 KD relative to control (data not shown). We conclude that reduced SMAD6 levels have multiple impacts on the EC transcriptome under homeostatic laminar flow: (1) dampened expression changes for some flow-responsive genes; (2) about half of flow-responsive genes becoming non-responsive; and (3) aberrant up- and down-regulation of genes usually not responsive to flow. This analysis supports that SMAD6 is a key regulator of endothelial cell flow responses.
Smad6 expression is upregulated in lung EC of adult mice relative to infants , suggesting a role in vascular quiescence. Consistent with this idea, Gene Ontogeny (GO) Analysis to identify cellular processes significantly affected by Smad6 depletion revealed that transcripts associated with the cell cycle were enriched in endothelial cells with reduced Smad6 levels, independent of flow status (Supp Fig. 3H). We hypothesized that endothelial cell proliferation is negatively regulated by SMAD6, and we found that expression of the proliferation marker Ki67 was increased with Smad6 depletion under both static and homeostatic flow conditions (Supp Fig. 3I, J). BrdU incorporation, which labels S-phase cells, was increased upon Smad6 depletion in both static conditions and trending upward under flow conditions (Supp Fig. 3K, L). Thus, repression of endothelial cell proliferation that is a prerequisite of vascular quiescence requires SMAD6.
SMAD6 regulates endothelial cell barrier function and junctions
GO Analysis also indicated that expression of genes associated with cell–cell junctions was down-regulated in endothelial cells with reduced Smad6 levels (Fig. 4a). Since SMAD6 regulates junction morphology in the absence of flow , we hypothesized that the barrier formed by endothelial cell–cell junctions and important for proper vascular function was compromised by SMAD6 depletion. Endothelial cells with depleted Smad6 levels had reduced barrier function relative to controls, as measured by trans-endothelial electrical impedance (Fig. 4b), and the defective resistance downstream of reduced Smad6 levels was also significant under flow conditions (Fig. 4c).
To further examine how SMAD6 manipulations affect endothelial cell–cell junctions, we focused on expression differences in cell–cell adhesion and junction genes between control and Smad6 KD endothelial cells under both static and flow conditions. Overlapping genes from these two genelists were identified, and a small subset with significant overall expression levels was examined further. Most junction genes were upregulated with homeostatic laminar flow in controls, and relative expression of a subset of these genes was reduced in endothelial cells depleted for Smad6, regardless of flow status. The net effect was that a group of cell junction genes had reduced expression in Smad6-depleted endothelial cells relative to contols when both were exposed to homeostatic laminar flow (Fig. 4d, compare Flow-NT to Flow-Smad6 KD). Of these, the protocadherin PCDH12 was chosen for further analysis, since PCDH12 is selectively expressed in arterial endothelial cells, and its deletion leads to changes in murine arterial blood pressure, while human mutations are associated with brain arterial calcification [52,53,54]. Validation of PCDH12 expression changes via qRT-PCR showed significant upregulation of PCDH12 expression with homeostatic laminar flow, and this increase was blunted to 60% of static control levels in flowed endothelial cells with reduced Smad6 levels (Fig. 4e). We next asked whether PCDH12 functions in flow alignment responses, by subjecting HUVEC treated with PCDH12 siRNA to homeostatic laminar flow, and found that PCDH12 depleted endothelial cells failed to align (Fig. 4f, g; Supp Fig. 3M). These results show that SMAD6 is required for full PCDH12 expression in response to flow, and that PCDH12 expression is necessary for proper endothelial cell alignment under homeostatic flow conditions. These findings suggest that SMAD6 regulation of PCDH12 contributes to endothelial cell homeostatic flow-mediated responses downstream of Notch-induced mechanotransduction.