Abstract
Processing bodies (PBs) are ribonucleoprotein granules important for cytokine mRNA decay that are targeted for disassembly by many viruses. Kaposi’s sarcoma-associated herpesvirus is the etiological agent of the inflammatory endothelial cancer, Kaposi’s sarcoma, and a PB-regulating virus. The virus encodes Kaposin B (KapB), which induces actin stress fibres (SFs) and cell spindling as well as PB disassembly. We now show that KapB-mediated PB disassembly requires actin rearrangements, RhoA effectors and the mechanoresponsive transcription activator, YAP. Moreover, ectopic expression of active YAP or exposure of ECs to mechanical forces caused PB disassembly in the absence of KapB. We propose that the viral protein KapB activates a novel mechanoresponsive signaling axis and links changes in cell shape and cytoskeletal structures to enhanced inflammatory molecule expression using PB disassembly. Our work implies that cytoskeletal changes in other pathologies will similarly impact the inflammatory environment.
Importance For the first time, we demonstrate that processing bodies (PBs), cytoplasmic sites of RNA decay, are regulated by mechanical signaling events that alter actin dynamics. Using the overexpression of a viral protein called KapB, known previously to mediate PB disassembly, we show that actin stress fibers (SFs) and the mechanoresponsive transcription factor, YAP, are required for PB loss. We also show that other established mechanical signals (shear stress or stiff extracellular matrix) that lead to the formation of SFs and activate YAP also cause PB disassembly. This is important because it means that KapB activates, from the inside out, a pathway that links cell shape to post-transcriptional gene regulation via cytoplasmic PBs.
Introduction
Cells are exposed to a variety of environments and they respond to changes in external force by adjusting internal tension. These mechanical cues can be transmitted to the cell through changes to extracellular contact nodes (focal adhesions) and contractile actomyosin structures to maintain tension homeostasis (Friedland, Lee, and Boettiger 2009; Kong et al. 2009; del Rio et al. 2009; Grashoff et al. 2010; reviewed in Finch-Edmondson and Sudol 2016). Actin stress fibres (SFs) are cytoskeletal structures composed of thick actin bundles, often associated with focal adhesions (Vallenius 2013), that are force-responsive, maintaining cytoskeletal integrity in changing mechanical environments (Burridge and Guilluy 2016). SF formation is coordinated by the GTPase, RhoA; it activates the formin, mammalian diaphanous protein-1 (mDia1) to promote actin filament growth and Rho-associated coiled-coil kinase (ROCK) to promote actomyosin contractility through non-muscle myosin II (Watanabe et al. 1997; Amano et al. 1997; Kimura et al. 1996). These RhoA-effectors act together to promote the formation of contractile and stable actin filaments in response to mechanical and chemical stimuli (Watanabe et al. 1999).
External forces elicit a cascade of signals using actin as force transducers to alter gene expression. Activated serum response factor (SRF) transcription responds to actin polymerization (reviewed in Chai and Tarnawski 2002). SRF activation is negatively regulated by the cytoplasmic concentration of monomeric G-actin (Sotiropoulos et al. 1999). However, inducers of filamentous actin (e.g. active RhoA) deplete G-actin levels leading to SRF nuclear translocation and transcription (Sotiropoulos et al. 1999). A more recent example is the mechanoresponsive transcriptional coactivator Yes-associated protein (YAP), whose activity can be controlled by cell shape and cytoskeletal structure (Dupont et al. 2011; Wada et al. 2011; Halder, Dupont, and Piccolo 2012; Yu et al. 2012). YAP is nuclear and active in response to low cell-cell contact (Zhao et al. 2007), high stiffness of the extracellular matrix (ECM) (Dupont et al. 2011), in shear stress due to fluid flow (K.-C. Wang et al. 2016; Nakajima et al. 2017; Lai and Stainier 2017; H. J. Lee et al. 2017; Huang et al. 2016), or after G-protein coupled receptor (GPCR) activation (Yu et al. 2012). Most of these signals induce the activity of RhoA and promote the formation of SFs, (Noria et al. 2004; Lee and Kumar 2016), implicating actin cytoskeletal structures as requisite intermediates for YAP activation.
Nuclear YAP associates with its coactivators to mediate transcription of genes involved in cell proliferation, differentiation, survival and migration (Halder, Dupont, and Piccolo 2012). Consistent with this, nuclear YAP is often pro-tumourigenic and drives progression of many oncogenic traits in a variety of cancers. These include the induction of cell stemness (Panciera et al. 2016), altered metabolism (C. Yang et al. 2018), cancer cell invasion/vascular remodeling (Calvo et al. 2013; Liu et al. 2018; Kimura et al. 2020), and altered growth and proliferation (Kapoor et al. 2014; Zanconato et al. 2015; Jang et al. 2017). Kaposi’s sarcoma (KS) is an endothelial cell (EC) cancer that is strongly linked to Kaposi’s sarcoma-associated herpesvirus (KSHV) (Chang et al. 1994; Russo et al. 1996; Zhong et al. 1996; Ganem 1997). KSHV establishes persistent, life-long infection of its human host, and displays two types of infection, latent and lytic. In KS, the majority of the tumour ECs are latently infected while lytic replication is rare; in part, because these cells die as a result of viral replication (Boshoff et al. 1995; Staskus et al. 1997; Umbach et al. 2010; Speck and Ganem 2010; Arias et al. 2014). That said, during their short lifetime lytic cells expel progeny virus and secrete large quantities of pro-inflammatory and angiogenic molecules, making even infrequent lytic replication an important driver of KS. A key contributor to this secretory phenotype is the constitutively active viral G protein-coupled receptor (vGPCR), a lytic viral protein (Montaner et al. 2006; Corcoran et al. 2012). Despite the paracrine contributors like vGPCR, the few gene products that are expressed during the KSHV latent cycle are central for viral tumourigenesis. Many features of in vivo KS are recapitulated by in vitro latent infection of primary ECs, or ectopic expression of individual KSHV latent genes, including enhanced proliferation and an elongated or ‘spindled’ morphology characteristic of KS. Spindling is induced by two KSHV latent genes, vFLIP (Grossmann et al. 2006) and Kaposin B (KapB; (Corcoran, Johnston, and McCormick 2015)). Spindled cells also secrete a variety of proinflammatory cytokines and angiogenic factors, to further promote tumour progression through inflammatory cytokine production (Ensoli 1998; Ciufo et al. 2001; Naranatt et al. 2003; Grossmann et al. 2006; Ojala and Schulz 2014). However, no information exists to demonstrate pro-tumourigenic YAP activation in KSHV latency, despite the fact that the vGPCR has been shown to activate YAP during KSHV lytic infection (Liu et al. 2015).
One way that KSHV latency promotes the pro-inflammatory and pro-tumourigenic KS microenvironment is via KapB-mediated disassembly of cytoplasmic ribonucleoprotein granules called processing bodies (PBs) (Corcoran, Johnston, and McCormick 2015). PBs are involved in many RNA regulatory processes such as RNA silencing, nonsense-mediated decay and mRNA decay (Eulalio, Behm-Ansmant, and Izaurralde 2007). We and others have shown that PBs are the major site for the translational suppression or constitutive decay of human mRNAs that code for potent regulatory molecules such as proinflammatory cytokines (Franks and Lykke-Andersen 2007; Corcoran, Johnston, and McCormick 2015; Vindry et al. 2017; Blanco et al. 2014). There are ~4500 of these transcripts, all of which bear destabilizing AU-rich elements (AREs) in their 3’-untranslated regions (3’-UTRs) (Shaw and Kamen 1986; Shyu, Greenberg, and Belasco 1989; Chen and Shyu 1995; Winzen et al. 1999; Stoecklin, Mayo, and Anderson 2006; Franks and Lykke-Andersen 2007; Bakheet, Williams, and Khabar 2006; Bakheet, Hitti, and Khabar 2017). PB abundance and composition is extremely dynamic and responds to cellular stress (Sheth 2003; Kedersha and Anderson 2007; Aizer et al. 2008; Takahashi et al. 2011). Specifically, activation of the stress-responsive p38/MK2 MAP kinase pathway by KapB elicits PB disassembly and prevents constitutive ARE-mRNA turnover (Winzen et al. 1999; Docena et al. 2010; Corcoran et al. 2012; Corcoran and McCormick 2015; Corcoran, Johnston, and McCormick 2015). This is an important yet underappreciated regulatory mechanism that fine tunes the production of potent proinflammatory cytokines and angiogenic factors in KS.
Though PBs are dynamic and stress-responsive, the precise signaling events that lead to PB assembly or disassembly are not well understood. We showed previously that KapB binds and activates MK2, which then phosphorylates hsp27, complexes with p115RhoGEF, and activates RhoA to elicit PB disassembly (Corcoran, Johnston, and McCormick 2015; Garcia et al. 2009; McCormick and Ganem 2005). While it is well-established that RhoA coordinates SF formation (Ridley and Hall 1992; Watanabe et al. 1999; Schmitz et al. 2000; Hotulainen and Lappalainen 2006), the precise mechanism of how RhoA promotes PB disassembly is not appreciated (Corcoran, Johnston, and McCormick 2015; Takahashi et al. 2011). In an effort to better understand the regulation of PB disassembly by KapB and RhoA, we began by targeting downstream RhoA effectors reported to promote SF formation to determine if the proteins known to mediate cytoskeletal remodeling were also necessary for PB disassembly. We reasoned that at some point we would be able to uncouple the signaling events that led to SFs from those that led to PB disassembly. We were not. We now present data that conclusively shows KapB-mediated PB disassembly is dependent not only on RhoA, but on cytoskeletal structures, actomyosin contractility and the presence of the mechanoresponsive transcription transactivator, YAP. We also present the first evidence of elevated YAP levels in response to expression of a KSHV latent gene, KapB. We also extend these studies beyond their impact on viral tumourigenesis, by determining the mechanical regulation of PB dynamics in the absence of KapB expression, and show that induced cell contractility, cytoskeletal structures and active YAP all precede PB disassembly. Using a viral protein from an oncogenic virus, we have discovered a novel mechanoresponsive signaling pathway that transduces signals from cell shape and cytoskeletal structures to YAP to control PBs, post-transcriptional regulators of cellular gene expression.
Results
RhoA effectors controlling SF formation are required for PB disassembly
We previously showed that KapB-mediated PB disassembly required RhoA (Corcoran, Johnston, and McCormick 2015). In this work, we investigated whether downstream RhoA effectors known to control SF formation also control PB disassembly. Mammalian diaphanous protein 1 (mDia1) and Rho-associated coiled-coil kinase (ROCK) are considered the main coordinators of RhoA-mediated SF formation (Watanabe et al. 1999; Tojkander, Gateva, and Lappalainen 2012). mDia1 is a formin that promotes actin filament polymerization (Watanabe et al. 1999). To examine whether mDia1 was required for KapB-mediated PB disassembly, we designed short hairpin RNAs (shRNAs) to silence mDia1 mRNA. KapB- and vector-expressing human umbilical vein endothelial cells (HUVECs) were transduced with mDia1-targeting shRNAs and selected. Silencing efficacy was confirmed with immunoblotting (Fig 1A). PB analysis was performed using CellProfiler to quantify immunofluorescence images stained for the hallmark PB-resident protein, Hedls, as described in detail in the methods (J. H. Yu et al. 2005; Kedersha et al. 2008). Knockdown of mDia1 increased PBs in KapB-expressing cells (Fig 1B, D). mDia1-sh1 showed a greater increase in PBs in comparison to mDia1-sh2 (Fig 1B), likely because mDia1-sh1 reduced protein expression by 90% whereas mDia1-sh2 reduced it by 40-50% (Fig 1A). To ensure that the loss of mDia1 did not increase PBs globally but rather that mDia1 contributed specifically to KapB-mediated PB disassembly, we calculated the ratio of PBs per cell in KapB-expressing cells and normalized to PBs per cell in vector controls. This is important because this calculation shows whether KapB is still able to disassemble PBs, relative to vector, in the context of mDia silencing. If the ratio is ≥1 after sh-mDia treatment, it indicates that KapB is no longer able to disassemble PBs in comparison to the vector control, and that mDia contributes directly to KapB-mediated PB disassembly. Conversely, if the ratio is ~ 0.4 to 0.6, it indicates that KapB can still disassemble PBs even in the context of sh-mDia treatment. In this case, we determined that silencing using both mDia1-sh1 and mDia1-sh2 restored the PB ratio in KapB:Vector cells to ~1, indicating that the ability of KapB to disassemble PBs is lost after mDia silencing and that this is a specific effect (Fig 1C). We note that this ratio will be reported in subsequent figures for every RNA silencing or drug treatment applied to test KapB-mediated PB disassembly. We also observed that mDia1 silencing did not eliminate SF formation (Fig 1D) but, instead, increased elongated cells with visible actin SFs across the cell in both vector and KapB conditions. The visible actin structures may represent different SF subtypes or actin bundles that compensate for the loss of mDia1 (Hotulainen and Lappalainen 2006).
ROCK promotes SF formation by increasing actin contractility and inhibiting actin severing activity (Julian and Olson 2014). Chemical inhibition of both isoforms of ROCK, ROCK1 and ROCK2, with Y-27632 (Ishizaki et al. 2000) restored PBs in KapB-expressing cells and increased the ratio of KapB:Vector PBs (Fig 2A-C). To determine whether PB disassembly is dependent on a single ROCK isoform, both ROCK1 and ROCK2 were knocked down with isoform-specific shRNAs. Knockdown efficacy was confirmed with immunoblotting (Fig S1). Independent knockdown of ROCK1 and 2 increased PBs counts in KapB-expressing cells (Fig 2D, F) and restored the ratio of KapB:Vector PBs counts (Fig 2E). This indicated that both ROCK1 and ROCK2 can contribute to KapB-mediated PB disassembly. ROCK2 knockdown showed more robust PB restoration, both in terms of PB counts and PB size, than that seen with ROCK1 knockdown (Fig 2D, F). Quantification of PB counts in control cells for both pan-ROCK inhibition and ROCK knockdown experiments is reported in Figure S1. While pan-ROCK inhibition and ROCK1 knockdown treatments both eliminate SFs, ROCK2 knockdown retains pronounced actin fibres in the cells (Fig 2F). Similar to mDia1 knockdown, this may indicate a compensatory mechanism to retain cell shape and suggests that only a subset of SFs may be required for PB disassembly. Taken together, these data show that inhibition of RhoA effectors that mediate SF formation can reverse KapB-mediated PB disassembly. Put another way, we have been unable to uncouple KapB-mediated SF formation from KapB-mediated PB disassembly.
ROCK phosphorylates and activates LimK, which then phosphorylates and inactivates cofilin, an actin-severing protein (Ohashi et al. 2000). In this way, ROCK promote SF formation. To investigate the role of cofilin in KapB-mediated PB disassembly, shRNAs to knockdown cofilin expression were used (Fig S2A). Since ROCK activation results in less cofilin activity and reduced actin severing, we hypothesized that knockdown of cofilin in KapB-expressing cells would augment KapB-mediated PB disassembly. Knockdown of cofilin resulted in elongated cells with more SFs (Fig S2D). Cofilin knockdown also augmented PBs disassembly in KapB-expressing cells (Fig S2B, C). This indicates that inhibition of cofilin elicits PB disassembly and supports the hypothesis that by reducing cofilin activity to promote KapB-mediated SF formation, PB disassembly is enhanced.
G-actin concentration does not influence PB disassembly
Since we could not uncouple the signalling controlling SF formation from PB disassembly, we investigated whether changes in the concentration of monomeric G-actin, known to control cellular stress and SRF transcriptional responses (Sotiropoulos et al. 1999; Chambers et al. 2015), could be controlling PBs. Several studies have shown that increasing the proportion of filamentous actin decreases the cytoplasmic concentration of monomeric G-actin (Rasmussen et al. 2010; Bunnell et al. 2011; Chambers et al. 2015). We investigated if our phenotype, PB disassembly, was controlled by changes in the proportion of monomeric G-actin. To determine this, cells were treated with drugs known to either decrease or increase the proportion of monomeric G-actin. Jasplakinolide (Jasp) treatment decreases the G-actin fraction by facilitating actin nucleation and aberrant polymerization of actin (Bubb et al. 1999). Conversely, the actin polymerization inhibitor Cytochalasin D (CytD) caps the barbed end of actin filaments, preventing further elongation of the actin filament and increasing the free G-actin concentration (Wakatsuki et al. 2001). If the level of G-actin is the signal, we hypothesized that jasplakinolide, which decreases G-actin levels, would mediate PB disassembly, while cytochalasin D would do the opposite, and promote PB assembly. However, both treatments increased the PB count per cell (Fig S3A-C); these data indicate that the concentration of G-actin does not influence PB disassembly, and this is not the mechanism by which actin SF formation or enhanced activity of RhoA alters PB dynamics. These data are congruent with our mDia1 and ROCK knockdown experiments that show retention of visible F-actin bundles despite PB restoration.
α –actinin-1 activity promotes PB disassembly
The actinins are primarily known for their role in bundling actin fibres, though in non-muscle cells, α -actinin-1 and 4 do not mediate actin bundling to the same extent (Pellegrin and Mellor 2007). α -actinin-4 can, at times, localize to dorsal SFs, but it primarily mediates focal adhesion turnover and can act as a transcriptional regulator of genes associated with cell proliferation and differentiation (Honda et al. 1998; Kovac 2010; Honda 2015). α -actinin-1 primarily mediates SF bundling and formation, as well as focal adhesion maturation (Honda et al. 1998; Kovac 2010). Using immunofluorescence, we observed that the localization of the two isoforms seen in HUVECs (Fig S4A, B) was consistent with the reported localization and function, as α -actinin-1 was localized to actin fibres and α -actinin-4 was more diffusely cytoplasmic and nuclear, with some actin fibre localization (Honda et al. 1998; Kovac 2010). Since α -actinin-1 associated with SFs in HUVECs and overexpression of alpha-actinin-GFP has been shown to localize and reinforce SFs (Edlund, Lotano, and Otey 2001; Jackson et al. 2008), we asked whether its overexpression would promote PB disassembly. This was indeed the case, suggesting that enhancing SF bundling and focal adhesion maturation positively regulates PB disassembly (Fig S4C, D).
Changes in cytoskeletal contractility control PB disassembly
One of the downstream activities of the kinase, ROCK, is to phosphorylate myosin light chain to induce non-muscle myosin II (NMII)-mediated actomyosin contraction (Mutsuki Amano et al. 1996). Since ROCK is required for KapB-mediated PB disassembly, we determined whether functional actomyosin contractility is also required. KapB-expressing cells were treated with blebbistatin, which inhibits NMII-mediated actomyosin contractility by maintaining NMII in a conformation that is unable to bind actin filaments (Kovacs et al. 2004). Treatment of KapB-expressing cells with blebbistatin restored both PBs levels in KapB-expressing cells, as well as the KapB:Vector ratio of PBs (Fig 3A-C), indicating that contractility is required for KapB-induced PB disassembly. To determine if contraction would elicit the same phenotype in the absence of KapB, cells were treated with Calyculin A (CalA), an inhibitor of myosin light chain phosphatase that promotes NMII phosphorylation and actomyosin contraction (Asano and Mabuchi 2001). Inducing contraction with CalA decreased counts of PBs (Fig 3D, E), again consistent with the hypothesis that actomyosin contractility controls PB disassembly.
Actomyosin contractility impacts cytoskeletal tension in adherent cells with SFs (Katoh et al. 1998; Tan et al. 2003). Additionally, both Jasp and CytD interfere with cytoskeletal tension (Rotsch and Radmacher 2000), and both increased PB counts (Fig S3). Since the mechanoresponsive transcription activator, YAP, is activated by increases to cytoskeletal tension via actomyosin contractility (Dupont et al. 2011), we predicted the following: 1) KapB expression increases cytoskeletal tension, 2) KapB expression will activate YAP and 3) both cytoskeletal tension and YAP will be required for PB disassembly. Though unable to directly test the first prediction, we now consider the role of YAP in KapB-mediated PB disassembly.
YAP activation induces PB disassembly
We investigated the cellular localization of YAP in KapB-expressing cells. KapB-transduced human umbilical vein endothelial cells (HUVECs) showed increased levels of nuclear YAP, as well as increased total YAP intensity by immunofluorescence, though the ratio of nuclear:cytoplasmic YAP was not markedly increased (Fig 4A). When YAP is phosphorylated by LATS, it is sequestered in the cytoplasm and transcriptionally inactive (Zhao et al. 2007). While YAP has multiple phosphorylation sites, phosphorylation at serine 127 is the most potent LATS-mediated phosphorylation site that promotes cytoplasmic distribution of YAP (Zhao et al. 2007). To investigate the phosphorylation status of YAP in KapB-expressing cells, levels of P(S127)-YAP and total YAP were measured by immunoblot. In KapB-expressing cells, there was a decrease in the ratio of P(S127)-YAP to total YAP suggesting that YAP is active when KapB is expressed (Fig 4B). We also observed an increase in total steady-state levels of YAP by immunoblotting, corroborating the increase in total YAP intensity seen by microscopy (Fig 4A, B). Taken together, these observations are the first evidence of enhanced YAP activity in response to expression of a KSHV latent gene. We next asked if active YAP in KapB-expressing cells can interact with TEAD and other transcription factors to elicit changes in gene expression (Vassilev et al. 2001). We used a TEAD-element luciferase assay to assess if canonical YAP transcription was activated. As a positive control, we used YAP 5SA, a mutant version of YAP that is unable to be phosphorylated and inactivated by the inhibitory kinase LATS (Zhao et al. 2007) and is thus considered constitutively active. YAP 5SA robustly activated the TEAD element-containing firefly luciferase reporter (Fig S5A). Despite our observations of increased nuclear and total YAP, KapB did not induce the transcription of the TEAD element-containing firefly luciferase reporter (TEAD-Fluc; Fig S5A). Further, KapB did not increase steady-state mRNA levels of common YAP target genes CTGF, CYR61 and ANKRD1 by RT-qPCR, although these genes were elevated by YAP 5SA (Fig S5B). These data indicate despite the observation that YAP appears more abundant and nuclear in KapB-expressing cells, it is not activating transcription of its canonical gene targets.
We expressed shRNAs targeting YAP in KapB-expressing HUVECs to assess whether the altered levels of YAP impacted PB disassembly. Immunoblotting confirmed knockdown efficiency (Fig 4C). Knockdown of YAP increased PBs in KapB-expressing cells (Fig 4D-F). In the context of YAP knockdown, the KapB:Vector ratio of PBs counts was restored, indicating that YAP is required for KapB-mediated PB disassembly (Fig 4E) and suggesting that KapB is activating a mechanoresponsive signalling axis to elicit PB disassembly via YAP. We wondered if YAP was central to PB disassembly in the absence of KapB expression. To this end, we examined PBs after YAP 5SA expression. These cells displayed decreased number of PBs per cell, indicating that YAP 5SA elicited disassembly of PBs (Fig 5A, B). KapB-mediated PB disassembly correlates with increases in stability and levels of ARE-mRNA (Corcoran, Johnston, and McCormick 2015; McCormick and Ganem 2005). To examine whether YAP 5SA-mediated PB disassembly elicits the same changes in ARE-mRNAs, we used a luciferase assay previously established to measure the stability of ARE-mRNAs by measuring luminescence of an ARE-containing firefly luciferase reporter (Corcoran, Khaperskyy, and McCormick 2011). In this assay, as previously shown in Corcoran, Johnston, and McCormick (2015), KapB increased level of firefly luminescence indicating enhanced stability of its RNA transcript (Fig 5C). However, despite also inducing pronounced PB disassembly, YAP 5SA does not increase Fluc luminescence significantly more than the control construct (Fig 5C). This points to a divergence of KapB and active YAP outcome. Although PB disassembly is induced by the expression of both constitutively active YAP and KapB, active YAP increases the transcriptional activation of genes CTGF, CYR61 and ANKRD1 while KapB does not; conversely, KapB enhances the stability of ARE-mRNAs while active YAP does not.
YAP activators disassemble PBs
Since overexpression of constitutively active YAP leads to disassembly of PBs, we wanted to determine whether activation of endogenous YAP could do the same in the absence of KapB. We tested various upstream mechanical signals described to activate YAP for their ability to elicit PB disassembly: shear stress, low cell confluence and high ECM stiffness (Nakajima et al. 2017; Lee et al. 2017; Noria et al. 2004; Zhao et al. 2007; Dupont et al. 2011). For the first, we subjected HUVECs to shear stress by fluid flow (shear forces of 2 and 10 dyn/cm2) and PBs were examined via immunofluorescence. Both treatments showed prominent cell elongation and resulted in robust PB disassembly (Fig 6A, B). To test if cell confluence regulates PB levels, HUVECs were seeded at low, medium and high densities. Cells at low confluence are reported to have active YAP and we predicted PBs would disassembly; however, the low-density monolayer displayed more PBs per cell then those at medium and high densities (Fig 6C, D). To test the impact of collagen stiffness on PB disassembly, HUVECs were plated on coverslips coated with increasing densities of collagen (0 to 64 μg/cm2). While collagen density does not perfectly reproduce matrix stiffness as it does not eliminate effects from increasing collagen-binding sites, increasing collagen densities correlate with increases in matrix stiffness (Yang, Leone, and Kaufman 2009; Lee et al. 2014; Joshi, Mahajan, and Kothapalli 2018). As collagen density increased, PBs decreased (Fig 6E, F). Taken together, these data indicate that PB disassembly occurred in response to mechanical stimuli known to require RhoA and altered cytoskeletal structures to activate YAP (shear stress and increased ECM concentration) (Zhao et al. 2012; Huang et al. 2016; Lee and Kumar 2016; Moreno-Vicente et al. 2018). Again, our model points to the importance of actin SF formation as a requisite precursor to PB disassembly irrespective of YAP activation status.
Shear stress mediated PB disassembly requires YAP
YAP responds to external forces that induce active RhoA, actin SFs, and pronounced cell elongation; in short, the typical behaviour of ECs in response to fluid flow. However, how YAP responds to shear stress is controversial (Wang et al. 2016; Huang et al. 2016; Lee et al. 2017; Nakajima and Mochizuki 2017). To verify YAP activation by continuous, unidirectional fluid flow in our system, HUVECs subjected to 2 and 10 dyn/cm2 of shear stress were lysed and used for immunoblotting for P(S127)-YAP and total YAP. Shear stress the ratio of phosphor-YAP/YAP in both conditions, suggesting a higher proportion of active YAP (Fig 7A). To assess if YAP was required for PB disassembly in response to shear stress, HUVECs transduced with YAP-targeting shRNA were subjected to 10 dyn/cm2 shear stress. PBs disassembled in cells treated with a non-targeting shRNA when subjected to shear stress (Fig 7B, C), consistent with earlier experiments (Fig 6A, B). When YAP was reduced by shRNA expression, ECs exposed to shear stress had more PBs than control cells without shear (Fig 7B, C). Therefore, YAP is required to disassemble PBs in response to shear stress. Taken together with our analysis of KapB-mediated PB disassembly, these data suggest that when KapB is expressed, it turns on the same mechanoresponsive signals that endothelial cells use to withstand mechanical forces like shear, in the absence of an external stimulus. The outcome of both scenarios is YAP-dependent disassembly of cytoplasmic PBs.
Discussion
In this manuscript, we have used a viral protein from an oncogenic virus to uncover the relationship between cytoplasmic PBs and the mechanical regulation of actin SF formation. We present data to support the existence of a novel mechanoresponsive pathway that links actin SFs, actomyosin contractility, and the transcription transactivator YAP to the disassembly of PBs and show that this pathway is hijacked by KapB during KSHV latency. Our major findings are as follows. i) KapB-mediated PB disassembly requires actin SF effectors ROCK1/2 /mDia1 and is enhanced by loss of the actin-severing protein, cofilin. ii) KapB-mediated PB disassembly is reversed when blebbistatin is used to inhibit actomyosin contractility or after knockdown of the mechanoresponsive transcription transactivator, YAP. iii) In the absence of KapB, we can induce PB disassembly when we promote the formation of actin SFs, actomyosin contractility, and YAP activity using overexpression of α-actinin-1 (promotes actin bundling into SFs and increases cytoskeletal tension (Jackson et al. 2008)), Calyculin A (inhibits myosin light chain phosphatase to promote actomyosin contraction (Asano and Mabuchi 2001)), or overexpression of active YAP (YAP 5SA). Exposure of endothelial cells to the external forces created by shear stress or a stiff extracellular matrix also induces PB disassembly in the absence of KapB. Together, these data show for the first time, that PBs disassemble in response to mechanical signals that transduce external forces from outside the cell to the actin cytoskeleton and that this is a pathway used by endothelial cells to regulate gene expression in response to diverse stimuli. Moreover, this work also highlights the remarkable pizzazz used by viruses to hijack cellular pathways. In this case, we reveal that the viral protein KapB taps into this mechanoresponsive pathway to trigger mechanical changes to cytoskeletal structures and downstream effectors that would normally respond to force, thereby inducing PB disassembly from within the cell, rather than from without.
During the process of actin polymerization, the monomeric form of actin, globular actin (G-actin), aggregates in groups of three subunits or more to nucleate an actin filament, which extends into filaments via addition of further G-actin monomers through ATP-dependent polymerization (reviewed in Pollard 2016). 10 to 30 actin filaments (F-actin) bundle together into SFs, primarily using the α-actinin family for crosslinking (Small et al. 1998; Cramer, Siebert, and Mitchison 1997; Lazarides and Burridge 1975; Pellegrin and Mellor 2007). SFs with periodic distribution of actin-crosslinking proteins and non-muscle myosin II (NMII) are contractile structures (Katoh et al. 1998; Tan et al. 2003), but not all actin SFs function equally in this regard. For any structure to be able to generate tension, it must be tethered at the ends. Of the types of SFs (ventral SFs, dorsal SFs and transverse arcs; (Small et al. 1998)), ventral SFs are attached at both termini to the extracellular matrix (ECM) through focal adhesions and contain NMII, which imparts a contractile phenotype (Hotulainen and Lappalainen 2006; Small et al. 1998; Vallenius 2013). Dorsal SFs are attached through focal adhesions but do not contain NMII, and thus are not contractile (Small et al. 1998; Vallenius 2013). However, dorsal SFs are thought to work in concert with transverse arcs, which contain NMII but are not attached to focal adhesions, to mediate cellular contractility.
In this work, while we did not directly determine the subtype of actin SF structures that form in response to KapB-mediated RhoA activation, several features of our data suggest that the structures that are important for PB disassembly must be contractile and cytoskeletal tension. When both mDia1 and ROCK2 were silenced in KapB-expressing cells (Fig 1, 2), visible actin bundles are still apparent despite PB restoration in both contexts. This suggests that not all SF subtypes are required for our phenotype. In addition, blebbistatin treatment of KapB-expressing cells dramatically restored PBs; these data suggest that PB disassembly requires actin-mediated contractility rather than merely structural support (Fig 3). Furthermore, overexpression of α-actinin and shear stress increase cell stiffness (Lee et al. 2006; Jackson et al. 2008). Both treatments induced PB disassembly (Fig S4, 6), reinforcing the correlation between increasing cell tension and PB disassembly. Finally, our data show that YAP is required for PB disassembly (Figs 4, 5, 7). YAP is mechanoresponsive; it becomes active when tension-forming actin structures are induced by external forces e.g. focal adhesion engagement by stiff ECM (Dupont et al. 2011; Sugimoto et al. 2018). As YAP activation and PB disassembly both rely on RhoA-induced cytoskeletal contractility, any activator of YAP that induces cytoskeletal tension through RhoA should mediate PB disassembly. Our data supports this notion, as shear stress forces and increasing collagen density both cause PB disassembly in the absence of KapB, while low confluence does not (Fig 6). We also know that GPCRs (G11/12 and Gq/11) activate YAP in a RhoA-dependent manner (Yu et al. 2012) and LPA treatment or overexpression of KSHV-derived constitutively active vGPCR (both activate G11/12) both induce PB disassembly (Corcoran et al. 2012; Corcoran and McCormick 2015), these findings support the conclusion that PB disassembly requires the formation of contractile actin structures like those associated with YAP transactivation responses.
KSHV is an oncogenic virus associated with the endothelial neoplasm, Kaposi’s sarcoma (KS). Cells within the KS lesion display latent KSHV infection, proliferate abnormally, spindle, and release many pro-inflammatory and pro-tumourigenic mediators into the microenvironment. KapB expression alone recapitulates two of these key features, cell spindling and pro-inflammatory mediator production that results from enhanced stability of ARE-containing cytokine mRNAs that would normally shuttle to PBs for constitutive turnover (Corcoran, Johnston, and McCormick 2015). Our previous work showed that both phenotypes require KapB activation of the stress-responsive kinase, MK2, and the downstream activation of the GTPase RhoA (Corcoran, Johnston, and McCormick 2015; Corcoran and McCormick 2015). We also showed that the lytic vGPCR protein mediates PB disassembly and the concomitant stabilization of ARE-mRNAs; more recently ORF57 has been also reported to disrupt PBs (Corcoran et al. 2012; Sharma et al. 2019). The observation that KSHV encodes at least three separate gene products sufficient to drive PB disassembly suggests that PB disassembly is beneficial for some aspect of the infectious cycle. Further research is required to definitively address how PBs influence the KSHV infectious cycle and the fate of infected cells.
We and others observed that the presence or absence of PB punctae visible by microscopy directly correlates with the stability of ARE-mRNAs (Corcoran, Johnston, and McCormick 2015; Vindry et al. 2017; Blanco et al. 2014). We predicted that YAP-mediated PB disassembly would also promote ARE-mRNA stability. Indeed, several YAP-target genes contain ARE elements in their 3’UTR, including CTGF and ANKRD1 (Shen and Stanger 2015; Bakheet, Hitti, and Khabar 2017). Shear forces also cause YAP-dependent PB disassembly and have previously been shown to upregulate many genes containing ARE-mRNAs (Vozzi et al. 2018; Bakheet, Hitti, and Khabar 2017). Comparison of the transcriptomic data from HUVECs subjected to shear stress from Vozzi et al (2018) (Accession: GEO, GSE45225) to entries in the ARE-mRNA database (Bakheet, Hitti, and Khabar 2017) showed a 20% enrichment in the proportion of genes that contained AREs in those transcripts that were upregulated by shear stress. This suggests that PB disassembly enables efficient translation of YAP targets by preventing recruitment of the ARE-containing transcripts to PBs. That said, overexpression of constitutively active YAP (YAP 5SA) disassembles PBs but does not increase stability of an ARE-containing firefly luciferase reporter (Fig 5) (Corcoran, Khaperskyy, and McCormick 2011)). This discrepancy may be due to different functional responses for different classes of AU-rich elements. Our ARE-containing luciferase reporter contains the AU-rich sequence derived from the 3’-UTR of GM-CSF, categorized in Cluster 5, whereas canonical YAP genes CTGF and ANKDR1 are in Clusters 1 and 2, respectively (Bakheet, Hitti, and Khabar 2017).
Data presented herein clearly implicate a requirement for YAP in the PB disassembly phenotype that is induced by KapB and by the external force, shear stress (Fig 4,7). However, the precise connection between YAP and PB disassembly is unclear. What we do know is that despite the clear reliance on YAP for PB disassembly, KapB does not increase expression of canonical YAP-regulated transcripts (Fig S5). Our data also show increases in total YAP, decreases in phosphorylated YAP; however, the ratio of nuclear:cytoplasmic YAP is not markedly altered (Fig 4). Taken together, these data suggest that PB disassembly is independent of YAP’s role as a gene transactivator and may also be independent of YAP nuclear translocation. In the discussion that follows, we explore two possible models for how YAP may promote PB disassembly that are independent of its transactivation of canonical genes. i) Cytoplasmic YAP promotes autophagic flux to promote PB catabolism. Several studies link YAP with the regulation of the catabolic process of autophagy, though many of these are contradictory and suggest YAP-mediated autophagy control is cell type and context-dependent (Song et al. 2015; Liu et al. 2017; Pei et al. 2019; Totaro et al. 2019). Totaro et al. provided strong evidence to support that YAP promotes autophagic flux by promoting the expression of Armus, a Rab7-GAP that is required to mediate the fusion of autophagosomes with lysosomes in the final degradative step of autophagy (Totaro et al. 2019). Their data also show that autophagic flux is a mechanoresponsive process; this supported by other studies wherein endothelium exposed to unidirectional shear stress upregulates autophagy (Liu et al. 2015; Yao et al. 2015; Wang et al. 2018; Das et al. 2018). These data are also consistent with preliminary data from our group that suggests that PB disassembly mediated by KapB requires autophagy (knockdown of Atg5 restores PBs [Robinson, Singh, Corcoran, unpublished data]) and the work of others (Hardy et al. 2017). In this model, we propose that the intermediate step(s) linking the requirement of YAP to the disappearance of PBs is the upregulation of autophagic flux, which results in the autophagic degradation of PB granules or PB components. ii) YAP and PBs are antiviral PBs are sites where innate immune factors congregate that are disrupted by most viruses during infection (Burdick et al. 2010; Li et al. 2012; Ostareck, Naarmann-de Vries, and Ostareck-Lederer 2014; Burgess and Mohr 2015; Cuevas et al. 2016; H. Wang et al. 2016; Lumb et al. 2017; Balinsky et al. 2017; Núñez et al. 2018; Ng et al. 2020). Indeed, KSHV encodes three separate proteins that all induce PB disassembly (Corcoran et al. 2012; Corcoran, Johnston, and McCormick 2015; Sharma et al. 2019). PBs are likely playing an as yet undefined and underappreciated role in regulating innate antiviral responses. YAP is also a novel and unappreciated negative regulator of innate immune signaling pathways. YAP blocks the ability of the innate immune kinase, TBK1, a downstream effector for several innate signaling pathways, to associate and activate its substrates (Zhang et al 2017). In so doing, YAP blocks downstream induction of interferons and increases viral replication (Zhang et al. 2017). This feature of YAP is independent of its ability to act as a transcriptional transactivator (Zhang et al. 2017). We speculate that KapB-induced PB disassembly, like active YAP, favours viral replication and survival and is promoted by KSHV in order to reshape subsequent antiviral innate immune responses.
In this manuscript, we describe the surprising convergence of two previously unrelated yet essential regulators of cellular gene expression – the mechanoresponsive transactivator YAP and cytoplasmic PBs, known regulators of AU-rich mRNA decay. We show that PB disassembly is mechanoresponsive; external forces that change cell shape and tension-forming cytoskeletal structures cause PB disassembly in a YAP-dependent manner. This discovery was made courtesy of the unique KSHV protein, KapB, and provides yet another example of how viruses have evolved surprising ways to manipulate their host and ensure their survival. In this case, KapB induces, from the inside out, a mechanoresponsive pathway to cause PB disassembly and elevated YAP. Future study will untangle how these related mechanoresponsive events are induced by KSHV to better promote viral replication.
Materials and Methods
Antibodies, Plasmids and Reagents
The antibodies used in this study can be found in Table S1. The plasmids used in this study can be found in Table S2. Forward and reverse shRNA sequences were selected from the TRC Library Database in the Broad Institute RNAi consortium. YAP target shRNAs in pLKO.1 were obtained from Dr. C. McCormick (Dalhousie University, Halifax, Canada). Sequences for all shRNA oligonucleotides used for cloning are listed in Table S3. Cloning of shRNAs was conducted according to the pLKO.1 protocol (Addgene 2006). The chemical inhibitors used in this study can be found in Table S4.
Cell Culture
Human embryonic kidney 293T and 293A cells (HEK-293T/A, ATCC, Manassas, Virginia, US) and human cervical adenocarcinoma cells expressing a tetracycline-inducible repressor (HeLa Tet-Off, Clontech, Mountain View, California, US) were cultured in Dulbecco’s Modified Eagle Medium (DMEM, Gibco, Carlsbad, California, US) supplemented with 10% heat-inactivated fetal bovine serum (Gibco), 100 U/mL penicillin, 100 μg/mL streptomycin, and 2 mM L-glutamine (Gibco). Pooled human umbilical vein endothelial cells (HUVECs, Lonza, Basel, Switzerland) were cultured in endothelial cell growth medium 2 (EGM-2, Lonza)). For HUVEC passaging, tissue culture plates were precoated for 30 min at 37°C with 0.1% (w/v) porcine gelatin (Sigma, St. Louis, Missouri, US) in 1X PBS (Gibco).
Transfection for Lentivirus Production
HEK-293T cells at 70-80% confluence were transfected using 3.3 μg of the target lentiviral construct, 2 μg pSPAX2 and 1 μg pMD2.G with 1 mg/mL polyethyenimine (PEI, Sigma) in serum-free DMEM. After 5 to 6 h, media was replaced with antibiotic-free DMEM containing 10% FBS and 2 mM L-glutamine (Gibco). Transfected cells were incubated for 48 h at 37°C to allow lentivirus production. The supernatant media containing viral particles was harvested and filtered through a 0.45 μm polyethersulfone (PES) filter (VWR, Randor, Pennsylvania, US) and aliquoted. Virus was stored at −80°C until use.
Lentiviral Transduction
Lentivirus was supplied into wells of plated HUVECs in EGM-2 supplemented with 5 μg/mL hexadimethrine bromide (polybrene). After 24 h incubation, cells were selected with either 5 μg/mL blasticidin (Sigma) for 96 h, replacing the media and antibiotic at 48 h, or 1 μg/mL puromycin (Sigma) for 48 h. Following selection, HUVEC medium was replaced with EGM-2 without selection for at least 24 h recovery before further use.
Immunofluorescence
Immunofluorescence was performed as described previously (Corcoran, Johnston, and McCormick 2015). Briefly, cells were grown on coverslips (no. 1.5, Electron Microscopy Sciences, Hatfield, Pennsylvania, US). Following treatment, coverslips were fixed in 4% paraformaldehyde (PFA, Electron Microscopy Sciences) in PBS at 37°C for 10 min, permeabilized with 0.1% Triton-X100 (Sigma) in 1X PBS for 10 min at RT, and blocked in 1% Human AB serum (blocking buffer, Sigma) in 1X PBS for 1 h at RT. Coverslips were then incubated with diluted primary antibody in blocking buffer overnight at 4°C in a humidified chamber. After primary antibody incubation, coverslips were washed with 1X PBS and then incubated in fluorescently-tagged secondary antibody diluted in blocking buffer for 1 h at RT. If applicable, coverslips were stained with Phalloidin-conjugated Alexa-Fluor 647 (Invitrogen, 1:100) in 1X PBS for 1.5 h. Coverslips were mounted onto microscope slides (FisherBrand, Pittsburgh, Pennsylvania, US) using Prolong Gold Antifade Mounting Media (Invitrogen, Carlsbad, California, US). For coverslips that were used for Hedls puncta quantification, the following modifications to immunofluorescence were made: (1) Prior to permeabilization, coverslips were stained with wheat germ agglutinin (WGA) Alexa-647 conjugate (Invitrogen, 1:400) in 1X PBS for 10 min at RT. (2) Following secondary antibody incubation, coverslips were stained with 4’,6-Diamidino-2-Phenylindole (DAPI, Invitrogen, 1:10,000) in 1X PBS for 5 min.
Confocal imaging was performed on the Zeiss LSM 880 Confocal Microscope (Charbonneau Microscopy Facility, University of Calgary, Calgary, Canada) at the 63X oil objective. CellProfiler imaging was performed on the Zeiss AxioImager Z2 (CORES facility, Dalhousie University, Halifax, Canada) or Zeiss AxioObserver (Charbonneau Microscopy Facility, University of Calgary) at the 40X oil objective.
Quantification of Processing Bodies Using CellProfiler Analysis
CellProfiler (cellprofiler.org) is an open source software for high-content image analysis (Kamentsky et al. 2011) and was used to develop an unbiased method for quantifying changes to PB dynamics. The pipeline used for quantifying PBs was structured as follows: To detect nuclei, the DAPI image was thresholded into a binary image. In the binary image, primary objects between 30 to 200 pixels in diameter were detected and defined as nuclei. Cells were identified as secondary objects in the WGA image using a propagation function from the identified nuclei, which determined the cell’s outer edge. Using the parameters of a defined nucleus and cell border, the cytoplasm was then defined as a tertiary object. The Hedls channel image was enhanced using an “Enhance Speckles” function to identify distinct puncta and eliminate background staining. The cytoplasm image was then applied as a mask to the enhanced puncta image to ensure quantitation of only cytoplasmic puncta. Hedls puncta were measured in the cytoplasm of cells using a ‘global thresholding with robust background adjustments’ function as defined by the program. The threshold cut-off for identified Hedls puncta remained constant between all experiments with identical staining parameters. Puncta number per cell, intensity and locations with respect to the nucleus were measured and exported as .csv files and analyzed in RStudio. A template of the RStudio analysis pipeline is attached in Appendix A. Data was represented as fold change in Hedls puncta count per cell normalized to the vector puncta count. ‘Relative Hedls Puncta/Cell (KapB/Vector)’ demonstrates the KapB puncta count divided by vector puncta count, a ratio that was calculated within each treatment group for each biological replicate.
Protein Electrophoresis and Immunoblotting
Cells were lysed in 2X Laemmli buffer (20% glycerol, 4% SDS, 120 mM Tris-HCl), between 150 to 300 μL, depending on cell density. Lysates were homogenized with a 0.21-gauge needle, and supplemented to contain 0.02% (w/v) bromophenol blue (Sigma) and 0.05 M dithiothreitol (DTT, Sigma), then heated at 95°C for 5 min. 7.5 or 12% TGX Stain-Free SDS-polyacrylamide gels (BioRad) were cast according to the instructions of the manufacturer and 5 to 15 μg of total protein were subjected to SDS gel electrophoresis using 1X SDS running buffer (25 mM Tris, 192 mM Glycine, 0.1% SDS). Precision Plus Protein All Blue Prestained Protein Standards (BioRad, Hercules, California, US) was used as a molecular weight marker. After electrophoresis, gels were UV-activated using the ChemiDocTouch (BioRad) Stain-Free Gel setting with automated exposure for 45 s. The protein was transferred to low-fluorescence polyvinylidene difluoride (PVDF) membranes (BioRad) on the Trans-Blot Turbo Transfer System (BioRad) according to the instructions of the manufacturer. Following transfer, total protein amounts on membranes were imaged on the ChemiDocTouch using the Stain-Free Membrane setting with automated exposure. Membranes were then blocked using 5% BSA (Sigma) in 1X TBS-T (150 nM NaCl, 10 mM Tris, pH 7.8, 0.01% Tween-20) for 1 h at RT. Primary antibody was diluted in 2.5% BSA in 1X TBS-T. Membranes were incubated in primary antibody solution overnight at 4°C with rocking. The following day, membranes were washed 3 times for 5 min in 1X TBS-T. Membranes were incubated with the appropriate secondary antibody, conjugated to horseradish peroxidase (HRP) for 1 h at RT. Membranes were washed 4 times for 5 min in 1X TBS-T. Clarity™ Western ECL Blotting Substrate (BioRad) was mixed at a 1:1 ratio and applied to the membrane for 5 min. Chemiluminescent signal was imaged on ChemiDocTouch Chemiluminescence setting. Band intensity was quantified using ImageLab software (BioRad), normalizing to total protein.
Quantitative Reverse-Transcriptase Polymerase Chain Reaction (qRT-PCR)
Cells were lysed in 250 μL RLT buffer (Qiagen, Hilden, Germany) and RNA was extracted using the RNeasy Plus Mini kit (Qiagen) according to the manufacturer’s instructions. Complementary DNA (cDNA) was synthesized from 1 μg of total RNA using the qScript cDNA SuperMix (QuantaBio, Hilden, Germany) according to the manufacturer’s instructions. Real-time quantitative PCR with SsoFast EvaGreen qPCR MasterMix (BioRad) was used to quantify the fold-change in mRNA abundance. Relative fluorescence was quantified using CFX Connect (BioRad). All qRT-PCR primers efficiencies were between 90-110% in HUVECs and sequences are found in Table S5.
Luciferase Assay for TEAD Transcriptional Activity
HEK-293A cells were seeded in antibiotic-free DMEM at 75,000 cells/well. Mixtures of 500 ng of the target construct (pcDNA (Vector), pcDNA-KapB (KapB), p1XFLAG or p2XFLAG-YAP 5SA), 450 ng 8X-GTIIC luciferase reporter, 50 ng TREX-Renilla luciferase reporter and 3 μL FuGENE HD Transfection Reagent (Promega, Madison, Wisconsin, US) were transfected into HEK-293A cells. After 36 h, DMEM containing only 2 mM L-glutamine (starvation media) was supplied to the cells. Twelve hours after addition of starvation media, cells were lysed in 200 μL passive lysis buffer (Promega) and luciferase activity was assayed using the Dual-Luciferase Reporter Assay System (Promega) according to the manufacturer’s instructions. Luminescence was measured using the GloMax Luminometer (Promega).
Luciferase Assay for Stability of mRNA with AU-Rich Elements
This technique is described in Corcoran, Khaperskyy, and McCormick (2011). Briefly, Hela-Tet Off cells were seeded in antibiotic-free DMEM at 100,000 cells/well. Mixtures of 900 ng of the target construct (pcDNA (Vector), pcDNA-KapB (KapB), p1XFLAG or p2XFLAG-YAP 5SA), 90 ng TREX-Firefly ARE luciferase, 10 ng TREX-Renilla luciferase and 3 μL FuGENE HD Transfection Reagent (Promega) were transfected into Hela Tet-Off cells. After 36 h, 1 μg/mL doxycycline was supplied to the cells to inhibit further transcription of each reporter. Twelve hours after addition of doxycycline, cells were lysed in 200 μL passive lysis buffer (Promega) and luciferase activity was assayed using the Dual-Luciferase Reporter Assay System (Promega) according to the manufacturer’s instructions. Luminescence was measured using the GloMax Luminometer (Promega).
Collagen-Coating for Altering Matrix Stiffness
Coverslips (no. 1.5, Electron Microscopy Sciences) were coated with a dilution series (0 to 64 μg/cm2) of rat-tail collagen-1 (Gibco) in 0.02 M acetic acid for 4 h at RT. Slides were sterilized with UV irradiation and washed with 2 times with sterile 1X PBS prior to seeding cells.
Unidirectional Fluid Flow for Endothelial Cell Shear Stress
A parallel-plate flow chamber was used to expose ECs to shear stress. The system was described in detail in Gomez-Garcia et al. (2018). Briefly, cleaned, unfrosted microscope slides (Cole-Parmer, Vernon Hills, Illinois, US) were coated for 4 h at RT with rat-tail collagen-1 (Gibco) in 0.02 M acetic acid for a resultant 8.3 μg/cm2 collagen density. Slides were sterilized with UV irradiation and washed 2 times with sterile 1X PBS. HUVECs were seeded at a density of 350,000 cells/slide and cultured for 24 h. Forty-five mL of EGM-2 supplemented with dextran (Spectrum Chemical, New Brunswick, New Jersey, US) for a resultant 3 cP viscosity was added to the stock media bottle. The stock media bottle was connected with the associated tubing and pulse dampener. Slides with seeded cells were inserted onto the flow chamber, a gasket (Specialty Manufacturing, Calgary, Canada) was added, and the system was sealed shut and attached to the flow loop following the outlet of a pulse dampener. The rate of fluid flow was started at 0.3 L/min and doubled every 15 min until final flow rates of 0.6 L/min and 2.7 L/min were reached, corresponding to shear stress rates of 2 and 10 dyn/cm2. Following 21 h, cells were removed and immediately fixed for immunofluorescence or lysed for immunoblotting.
Statistical Analysis
Statistical analysis was performed in GraphPad Prism 8.0 software. Significance was determined using a ratio paired t-test or repeated measures analysis of variance (ANOVA). One-tailed ratio paired t-tests were applied in comparisons specifically examining PB restoration in KapB-expressing cells as a directional hypothesis. In all other comparisons, two-tailed ratio paired t-tests were applied. Significance was determined at p = 0.05. Each biological replicate for CellProfiler quantification consisted of 6 images of each treatment in a given experiment, counting approximately 100 to 200 cells per treatment.
Competing Interests
The authors have no competing interests to declare.
Author Contributions
Elizabeth L. Castle: Conceptualization, Experimentation, Analysis, Paper Writing
Dr. Pauline Douglas: Experimentation
Dr. Kristina Rinker: Conceptualization
Dr. Jennifer A. Corcoran: Conceptualization, Experimentation, Supervision, Funding Acquisition, Project Administration, Paper Writing
Supplementary Information
Acknowledgements
We would like to sincerely thank the members of the Corcoran lab for helpful discussions about this work, notably Carolyn-Ann Robinson. We would like to thank Dr. Craig McCormick (Dalhousie University) and his lab for plasmids, expertise and invaluable advice. ELC was supported by a Killam predoctoral scholarship, an NSERC CGS-M scholarship, and a Nova Scotia Graduate scholarship. Operating funds to support this work derive from an NSERC Discovery grant RGPIN-2015-04882 to JAC.