The adenoviral protein E4orf4: a probing tool to decipher mechanical stress-induced nuclear envelope remodeling in tumor cells

ABSTRACT

The human adenovirus (Ad) type 2/5 early region 4 (E4) ORF4 protein (E4orf4) exerts a remarkable tumor cell-selective killing activity in mammalian cells. This indicates that E4orf4 can target tumor cell-defining features and is a unique tool to probe cancer cell vulnerabilities. Recently, we found that E4orf4, through an interaction with the polarity protein PAR3, subverts nuclear envelope (NE) remodeling processes in a tumor cell-selective manner. In this Perspective, we outline mechanical signals that modify nuclear dynamics and tumor cell behavior to highlight potential mechanisms for E4orf4’s tumoricidal activity. Through an analysis of E4orf4’s cellular targets, we define a protein subnetwork that comprises phosphatase systems interconnected to polarity protein hubs, which could contribute to enhanced NE plasticity. We infer that elucidating E4orf4’s protein network at a functional level could uncover key mechanisms of NE remodeling that define the tumor cell phenotype.

KEYWORDS:

• Nuclear envelope
• nuclear rupture and repair
• actin
• cellular contractility
• polarity proteins
• cancer vulnerability

Altered cell mechanics in cancer

Tumor progression is driven by a limited subset of genetic and epigenetic alterations that promote cell-autonomous capabilities and facilitating properties that form the various hallmarks of cancer[1]. However, progression to a malignant cell phenotype is fueled by multiple microenvironmental cues, including mechanical signals that are largely generated by increased cell density and extracellular matrix rigidity [2–4]. To cope with the highly selective pressures imposed by these microenvironment cues, tumor cells exploit cellular plasticity programs[5]. This enables them to adapt to their microenvironment, in part, by modifying their mechanical properties. Epithelial-mesenchymal plasticity is a prototypal cellular plasticity program associated with tumor cell metastasis and high-grade carcinoma[6]. Reversible remodeling of cell-cell junctions and cell-matrix signaling complexes enables tumor cells to spread (epithelial-to-mesenchymal transition, EMT) and to grow at distant sites (mesenchymal-to-epithelial transition, MET). Additionally, tumor cells typically display a deregulated mechanical response. While epithelial cell stiffness is linked to tissue rigidity, in rigid tumors, tumor cells appear “softer” than normal cells[7]. Indeed, evidence indicates that in several tumor cell lines, cell stiffness is inversely correlated with the malignant phenotype [8,9]. However, tumor cells can exert higher traction forces on the microenvironment, due to increased cellular contractility [4,9]. Oncogenic signaling, which empowers cellular contractility, increases cellular mechanoperception and exacerbates cellular responses to subtle changes in the mechanical microenvironment[10]. Thus, dynamic reciprocity of cell-tissue interactions, involving changes in tumor cell mechanical properties, has a crucial impact on tumor progression.

The mechanisms underlying these paradoxical changes in tumor cell mechanics are still poorly understood. However, actin network and nuclear envelope dynamics both play dominant roles in determining cell mechanical properties. Actomyosin structure remodeling, controlled by small GTPase signaling networks downstream of cell surface receptor activation, govern intracellular forces [11,12]. The spatial control of these contractile forces is crucial to determine cell behavior. The polarity protein signaling network acts to segregate actomyosin subnetworks, to control their attachment to membranes, and to locally restrict biochemical signaling that transduces intracellular forces[13]. Since cancer cells typically lose some aspects of cell polarity, polarity proteins were assumed to function as tumor suppressors. However, it appears that they have a more complex relationship with cancer [14,15]. Changes in the polarity protein signaling network can promote new interactions of these proteins with the actin assembly machinery and enhance cell invasiveness and tumor progression[16]. Moreover, these proteins show multifaceted interactions with components of the Hippo pathway, a pathway that controls epithelial cell proliferation and thereby influences intracellular forces[15].

Mechanical signals also modify nuclear envelope (NE) dynamics. The nucleus is the largest and stiffest organelle limiting cell shape plasticity, particularly during cell migration in 3D environments [17–19]. The nucleus can sense and trigger adaptive responses to mechanical forces that modify nuclear shape and deformability, along with cell polarization and migration [20,21]. Indeed, nuclear shape and stability are typically deregulated in tumor cells by multiple alterations of nuclear stiffness, notably, changes in lamin levels[22–24]. These tumor cell alterations profoundly modify NE dynamics and correlate with increased metastatic potential. During metastatic migration, a reduction in nuclear stiffness is expected to facilitate cell shape adaptability in the face of physical constraints. However, enhanced nuclear plasticity is also correlated with nuclear fragility. In line with this, NE deformation and rupture are observed during migration in a restricted environment (confined cell migration) in vitro and in vivo Figure 1A [25–27]. Moreover, tumor cells are prone to spontaneous NE ruptures, presumably because of NE component alterations[28]. While most tumor cells survive to NE ruptures by NE repair, a transient loss of NE barrier function could promote innate immune signaling, genomic instability, tumor heterogeneity, and even resistance to first-line therapies [29–31]. It is important to note that NE rupture is rare in most healthy cells during interphase [28,32–34]. Thus, NE rupture and repair are currently considered as novel hallmarks of cancer. Strategies to either inhibit NE rupture or NE repair could selectively target NE rupture-prone tumor cells. Impairing NE repair efficiency alone does not reduce tumor cell viability; however, simultaneously reducing both NE repair and DNA repair does increase cell death after NE rupture [25,26]. This suggests that NE rupture and repair could be exploited as synthetic lethal targets. Loss of function for p53 or RB tumor suppressors impairs genome stability and can also promote NE rupture via undefined mechanisms [33,35,36]. Thus, targeting NE rupture and repair mechanisms could provide a novel entry point for combination therapies in a wide range of cancers bearing common alterations in p53 or RB pathways.

Figure 1. Migration-induced NE rupture and mechanisms of force transmission to the nucleus by different types of NE-associated actin cables

A) A tumor cell migrating through confined spaces in tissues experiences pronounced NE deformation that can cause nuclear bleb formation and rupture at the leading tip. B) LINC complex proteins SUN1/2 and Nesprins form a physical bridge between the actin cytoskeleton and the nuclear lamina. Stiffening of the nuclear lamina in response to force reinforces the coupling between the lamina and LINC complex proteins. Strong forces exerted by actin cables (red arrows) cause NE indentations. The permeability of NPCs, as well as chromatin organization can also be altered by the propagation of strong forces through the LINC complex. BAF, a protein associated with lamina-associated chromatin domains and LEM-domain proteins, has an important role in NE remodeling processes[157]. C) Representation of the retrograde actin flow (blue cables) moving toward the TAN lines (grey dashed lines). TAN lines were originally described as actomyosin-dependent structures composed of linear arrays of Nesprin-2G and SUN2 [59,158]. TAN lines can couple the nucleus to the moving acting cables resulting in the nucleus being pushed at the cell rear (top, blue arrows)[75]. D) Representation of the perinuclear actin cap cables (red cables) originally described as a dome-like actin network composed of thick actomyosin cables that cover the apical surface of the nucleus[60]. Actin cap fibers interact with the NE via LINC complex connections and terminate at actin cap associated focal adhesions (ACAFA)[159]. Being under high tension, actin cap fibers may apply pressure (compressive forces) on the nucleus (top, red arrows).

Despite important advances, studying these events at the molecular level is very challenging as they are stochastic. In this Perspective, we outline knowledge of mechanical stress-induced NE remodeling during interphase. We also cover relevant aspects of human adenovirus (Ad) type 2/5 early region 4 (E4) ORF4 protein (E4orf4) biology. In a novel paradigm, E4orf4 was found to exert tumor cell-selective killing by causing a high incidence of repetitive NE rupture-repair events.

Nuclear envelope remodeling during interphase

NE structure, nuclear mechanosensation, and nuclear mechanotransduction

The NE ensures nuclear compartmentalization and is composed of two specialized lipid bilayers. These include the outer nuclear membrane (ONM) contiguous with the endoplasmic reticulum and the inner nuclear membrane (INM)[37]. NE membranes fuse at multiple sites to form donut-shaped channels (nuclear pore complexes, NPC) that control the bidirectional transport of macromolecules across the NE[38]. In multicellular organisms, a dense fibrillar network of intermediate filaments (AC-type and B-type lamins), located between the INM and chromatin, provides a rigid scaffold that mechanically supports the NE and spatially organizes the genome by controlling heterochromatin domain organization[39]. A-type lamins are believed to increase nuclear stiffness, while B-type lamins support nuclear elasticity [40,41]. The NE serves as an important signaling hub. This is due to interactions between intracellular signaling pathways and NE transmembrane (NET) proteins, including LEM (LAP2, Emerin, MAN1) domain proteins at the INM[42]. Therefore, the NE functions as a scaffold for the regulation of cell cycle progression, mitosis, DNA repair, cell senescence, and cell migration. However, signal transduction across the NE, and the contribution of most NET proteins, remain poorly explored.

The NE undergoes multiple morphological changes [43,44]. In human cells, the most dramatic change is undoubtedly the rapid NE disassembly and reassembly during mitosis, a process that needs strict spatiotemporal coordination to prevent NE damages and aneuploidy. NE remodeling also occurs during interphase. For instance, NE remodeling is present in growing cells to increase NE surface and insert NPC. Additionally, NE remodeling occurs in response to force, for nuclear structure adaptation to mechanical stress, and to prepare cells for confined migration [43–45]. Moreover, vesicular transport across the NE enables large RNP particles and viral particles to exit the nucleus via a NPC-independent process called nuclear budding, exemplifying unforeseen NE plasticity [46,47]. Therefore, NE remodeling uses conserved molecular mechanisms that may be exploited by tumor cells to increase NE plasticity and cell invasiveness.

Cytoplasmic mechanosensitive structures include ion channels, adhesion complexes, and intercellular junction complexes [3,48]. They also include cytoskeletal elements, and in particular, actin filaments. Forces are mainly sensed through the binding and conformation of proteins residing in mechanosensitive structures. These proteins have motifs that can change conformation over mechanical forces to expose cryptic sites and promote events, such as protein complex assembly and posttranslational modifications[49]. These events, in turn, initiate biochemical responses that widely implicate focal adhesion kinase (FAK), oncogenic tyrosine kinase Src, mitogen-activated protein (MAP) kinases, and Rho family of small GTPases. Interestingly, all of these responses are generally upregulated in tumor cells and enable cells to adjust shape, contractility, and stiffness in response to extracellular matrix rigidity [3,48]. Upon sustained mechanical stress, tumor cells engage in cellular adaptive responses that have pro-proliferative and pro-migratory functions, such as Hippo pathway effectors YAP and TAZ[50]. Downstream of these signaling cascades also, nuclear lamin A dephosphorylation can stabilize lamin filaments, remodel the NE, and adjust nuclear stiffness [40,41,51]. However, lamin A/C can directly respond to mechanical stress since changes in lamin A/C epitope expression have been reported[52].

Indeed, the forces generated at cell-surface mechanosensitive structures can be transmitted across the NE via the linker of nucleoskeleton and cytoskeleton (LINC) complex [53–55]. This complex consists of KASH domain proteins (Nesprins) that span the ONM, and SUN domain proteins that span the INM Figure 1B[56]. Interactions between the KASH and SUN domains in the perinuclear space form a physical bridge between the ONM and the INM. This bridge connects chromatin to the microenvironment through SUN protein binding to nuclear lamin A/C and Nesprin binding to elements of the cytoskeleton. Increased intracellular forces, through myosin-mediated stress fiber assembly, can remodel the LINC complex by reinforcing the coupling of LINC proteins to lamin A/C [57,58]. Various LINC complex connections act by anchoring actomyosin filament networks to the apical NE surface and regulating their spatial dynamics. For instance, LINC complex clusters, composed of Nesprin-2G and SUN2, that colocalize with actin cables were named transmembrane actin-associated nuclear (TAN) lines Figure 1C[59]. TAN lines are functionally linked to nuclear positioning and cell migration. In addition, distinct actomyosin cables, which are aligned with the cell axis and connected to large peripheral focal adhesion sites, can form an actin cap at the apical cell surface Figure 1D[60]. Actomyosin cables of the actin cap appear to provide a bridge for the rapid transmission to the nucleus of changes in external forces, such as in substrate compliance or fluid shear stress, which are relevant mechanical stimuli for circulating tumor cells [57,61]. By exerting both compressive forces to the NE and tension to the LINC complex, actin cap structures can also shape the cell’s nucleus. Thus, various types of actin-dependent cytoskeletal forces are generated in different cellular contexts and can remodel LINC complex connections in a regulated manner[55].

Through these connections, LINC complex-mediated signal transduction could enable the cells to respond to strong forces, in part, by changing NE structure and chromatin organization [20,41,55]. Applying local stresses to integrins can trigger, via the LINC complex, chromatin deformation and transcription[62]. Through the LINC complex, forces associated with significant nuclear deformation can also alter NPC permeability and enhance the nuclear transport of large molecules, such as YAP[63]. Moreover, applying direct force to the LINC complex stiffens isolated nuclei. This nuclear response is triggered by Src-induced phosphorylation of the INM protein Emerin/LEMD5[64]. Remarkably, Emerin can mediate myosin II recruitment to the NE, perinuclear actin polymerization, cell polarization, confined cell migration, and YAP nuclear localization [64,65].

Thus, direct evidence supports a role for mechanical stress-induced NE remodeling in mechanotransduction. However, it is difficult to determine which nuclear event directly responds to forces or is downstream of cytoplasmic signaling cascades, since the pathways involved are tightly intertwined. Interestingly, NE rupture and repair is an extreme case of NE remodeling during interphase. Additionally, the mechanisms involved in NE rupture and repair may be particularly relevant for highly motile metastatic cancer cells.

Mechanisms of NE rupture and repair

When strong forces exceed nuclear mechanical resilience, NE deformation and remodeling can generate intranuclear pressure which causes the appearance of fragile NE sites. These fragile NE sites are believed to promote nuclear bleb formation and rupture [28,32,66,67]. NE rupture susceptibility, along with a tumor’s cell metastatic potential, depends on lamin levels. It was shown that NE rupture and confined cell migration are promoted by the downregulation of either A- and/or B-type lamin levels, whereas they are impeded by the upregulation of lamin A/C levels [18,28,68,69]. The primary mechanical forces that cause NE rupture are generated by actomyosin filaments[66]. In line with this, loss of myosin phosphatase subunit MYPT1/PPP1R12A, which antagonizes actomyosin contractility, promotes cancer cell nuclear dysmorphia and NE rupture[34]. Conversely, inhibition of myosin II or actin dynamics usually impedes NE rupture [25,66,70,71]. While forces applied to the nucleus during confined cell migration can be independent of the LINC complex[72], they generally rely on an intact LINC complex to trigger efficient NE remodeling and rupture.

As discussed above, the LINC complex may provide a dynamic signal-responsive interface between the nucleus and the cytoplasm, similar to cellular adhesion complexes at the cell surface[55]. Different LINC complexes appear to play specific roles in the organization of actin networks[73]. However, little is known on the LINC complex-associated signal transduction mechanisms that have an impact on NE rupture events. Several proteins regulating TAN line assembly were identified, including Formin homology 2 domain-containing 1 FHOD1, the actin-bundling protein Fascin, the ATP-binding protein Torsin 1A, and Spindle-associated membrane protein 1 Samp1/TMEM201 [74–79]. However, only Fascin has been tested for its ability to regulate nuclear deformation during confined cell migration and the role of other TAN line proteins in this process remains unclear[76]. While TAN line assembly relies on the LINC complex protein SUN2, spontaneous NE rupture is specifically hindered by SUN1 depletion, suggesting different modes of NE-actomyosin filament tethering[66]. Regarding actin cap assembly, an integrated signaling platform, using RhoA and Rac1 GTPases, appears to be essential [60,80,81]. Intriguingly, SUN1 and SUN2 can have opposing effects on RhoA and actin organization[80]. Therefore, future studies will be important to clarify LINC complex-associated signaling pathways that enhance NE rupture susceptibility in tumor cells.

Increased NE rupture can eventually cause cell death[82]. However, NE rupture-susceptible tumor cells can survive by the rapid recruitment of molecular repair machinery[83]. Studies have shown that NE repair after rupture has several molecular effectors in common with post-mitotic NE reassembly. Initial work on migration-associated NE rupture uncovered a role for components of the endosomal sorting complex required for transport III (ESCRT-III), which mirrored their function in post-mitotic NE reassembly and plasma membrane repair [25,26,84–87]. Recently, the recruitment of cytosolic barrier to autointegration factor (BAF) was shown to be critical to initiate NE repair, as required for postmitotic NE reassembly [88–91]. BAF is an early-acting factor that recruits LEM domain proteins and endoplasmic reticulum membranes to NE rupture sites. This effect is dependent on BAF’s ability to bind DNA, suggesting that NE sealing is initiated at the exposed chromatin surface [89,92,93]. Finally, NE membrane composition and lipid synthesis also impact NE repair [82,83].

However, the factors that determine NE repair efficiency (e.g. whether the repaired NE site might be more susceptible to future rupture events) remain unknown. In this regard, it will be important to determine the role of the phosphatases that control post-mitotic NE reassembly[94]. Another key question that remains is whether an integrated signaling interface exists between NE rupture and repair machinery to ensure a tight spatiotemporal coupling of these important events. In the following sections, evidence supporting the value of exploiting the adenoviral protein E4orf4 to address these issues in tumor cell-defining mechanics will be discussed.

Targeting tumor cell mechanics with the adenoviral protein E4orf4

E4orf4, perinuclear actomyosin dynamics, repetitive NE rupture, and tumor cell-selective killing

Adenoviral type 2/5 early region 4 (E4) open reading frame 4 (ORF4) protein (E4orf4) is an early viral gene product which, although not essential, plays several functions to facilitate viral replication. A recent comprehensive review on E4orf4 is available[95]. The interest in the biology of E4orf4 stems from the discovery of its unusual intrinsic toxicity in a variety of tumor cells, which is typified by unique changes in actomyosin structure organization [96–100]. E4orf4’s cell-killing ability is highly correlated with the assembly of robust actomyosin fibers that wrap around the nucleus at the apical cell surface and connects to enlarged focal adhesions at the basal cell surface Figure 2A [101,102]. The induction of polarized cell blebbing generally follows the assembly of the perinuclear contractile structure and is correlated with intense nuclear deformations and chromatin condensation. Mechanistically, E4orf4 deregulates focal adhesions stability and several components of mechanoresponsive signaling pathways, including tyrosine kinase Src, MAP kinases ERKs and JNKs, and Paxillin [99,102–104]. The E4orf4 phenotype also depends on the small GTPases RhoA, Rac1, Cdc42, and Rab11, along with myosin II [101,105,106]. Importantly, the downregulation of myosin II motor activity inhibits E4orf4’s phenotypic hallmarks and impedes tumor cell killing [101,107]. Most intriguingly, myosin II inhibition also hinders E4orf4 nuclear exit, thereby inhibiting actomyosin filament remodeling. These E4orf4 hallmarks argue that it functions by deregulating tumor cell mechanics.

Figure 2. E4orf4's tumoricidal hallmark features

A) Schematic model of the implication of PAR3 in E4orf4-induced NE rupture. E4orf4 promotes PAR3 clustering at the NE and recruitment of actin cables that are correlated with pronounced nuclear deformation and NE ruptures. The depletion of LINC complex proteins or PAR3 can recapitulate the same inhibition of E4orf4’s phenotypic hallmarks and toxicity in tumorigenic cells[107]. How PAR3 is engaged with the NE remains to be determined. B) E4orf4-induced PAR3 engagement with the NE. Single slice Airyscan confocal images of a nucleus from a E4orf4-expressing U2OS cell (apical view), showing endogenous PAR3 clusters (in green) decorating lamina grooves that are delineated by lamin A/C staining (in magenta); the PAR3 signal specificity was confirmed in cells transfected with PAR3-specific siRNAs[107]; DNA was stained with Hoechst, revealing a nuclear bleb proximal to NE indentation sites (yellow asterisk). Scale Bar: 10 µm. Images are from J.N. Lavoie and M.A. Rodrigue and were taken using a 63×/1.4 C-Plan Apo oil immersion lens on a Zeiss LSM 800 microscope in Airyscan mode. Image acquisition and processing of Airyscan datasets was performed with the Zen Blue 2.6 software.

Recently, the deregulation of cell mechanics by E4orf4 was shown to be tumor cell-selective, since various nontumorigenic epithelial cell lines resist to E4orf4-induced cytoskeletal remodeling and killing[107]. While these findings corroborate a longstanding premise that E4orf4 targets tumor cell-defining features, they provide a first hint on how E4orf4’s actions are dictated [107,108]. E4orf4-sensitive cells display striking nuclear lamina disorganization that forms folds at sites of NE-actomyosin filaments tethering, as typically observed when mechanical stress is applied to the nucleus[109]. Live-cell imaging revealed that E4orf4 induces a high frequency of nuclear blebs that often burst, thereby provoking a sudden efflux of E4orf4 into the cytoplasm Figure 2A. Nuclear bleb formation is dependent on myosin II motor activity. Nuclear bleb formation and bursting were also shown to reflect repetitive NE rupture and repair events, often at related NE sites, suggesting that NE repair is not efficient. Ultimately, these events are associated with a sustained loss of nuclear compartmentalization that correlates with the onset of cell blebbing, nuclear condensation, and cellular integrity loss[107].

Overall, these results argue that the disruption of NE mechanical regulation is a primary lesion induced by E4orf4 in tumorigenic cells. E4orf4-induced NE rupture appears to establish a vicious lethal cycle empowered by E4orf4 nuclear efflux. Cytoplasmic E4orf4, by promoting actomyosin filament assembly, may exacerbate the mechanical stress applied to the nucleus and exceed the cell’s NE repair capacity. E4orf4’s potential effects on DNA repair efficiency could also contribute to cell death in this context, considering that NE rupture and transient nuclear integrity loss can promote DNA damage due to the uncoordinated exchange of nuclear and cytoplasmic material [29,110]. In line with this notion, E4orf4 causes the accumulation of DNA damage foci in tumor cells showing NE rupture increases.¹ However, E4orf4’s effects on DNA repair appear to be complex, as the protein can either inhibit or promote DNA damage signaling in the context of virally infected cells[95]. In cells challenged with DNA-damaging compounds, E4orf4 can inhibit DNA damage signaling and presumably DNA repair. It will be informative to clarify E4orf4’s impact on both NE repair efficiency and DNA damage signaling in the context of NE rupture-associated tumor cell-selective killing, in the absence of externally inflicted DNA damage.

Nonetheless, our recent findings indicate that by identifying E4orf4’s targets in tumor cells, it will be possible to decipher conserved elements of molecular NE remodeling mechanisms underlying the tumor cell phenotype. Since it was shown that E4orf4 inhibits the development of highly aggressive tumors in Drosophila without causing significant damage to healthy tissues, the mechanisms identified will be relevant to the understanding of tumorigenesis[111].

Deciphering E4orf4’s protein network

E4orf4, like many viral proteins, exerts its cellular activities by disrupting normal protein complex functions in cells. Studies have sought to exploit large scale affinity purification followed by mass spectrometry (AP-MS) to identify E4orf4’s protein targets. Three main E4orf4 proteomic datasets have been obtained using diverse experimental approaches and model cell lines [107,112,113]. We combined these proteomic datasets to perform a comprehensive bioinformatic analysis and defined a set of 90 putative E4orf4’s cellular partners Figure 3. Strong enrichments were found for subunits of two major phosphatase systems (PP1 and PP2), and proteins related to intercellular junctions and polarity signaling, mainly PAR3/PARD3 (2/3 studies; Figure 3, asterisks). PP1 and PP2A are believed to account for ~90% of all phosphoserine/phosphothreonine phosphatase activity in mammalian cells [114,115]. They mainly consist of heterodimers and heterotrimers, which contain a catalytic subunit (C), a structural subunit (A, PP2A), and a variable regulatory subunit (R, usually designated B subunit for PP2A) that dictates substrate specificity[115]. Remarkably, E4orf4 shows highly selective binding to a limited subset of PP1- and PP2A-regulatory proteins Figures 3 and 4. PAR3/PARD3 is a signaling scaffold that functions within the conserved polarity PAR complex, consisting of PAR3, PAR6, and atypical PKC (aPKC) (PKCɩ/PRKCi, Figure 3) [116,117]. Interestingly, PAR3 has both tumor suppressor and oncogenic activities and interacts with the Hippo signaling pathway[15]. In the following sections, we discuss the role of PAR3 in E4orf4-induced NE remodeling and explore potential contributions for PP1- and PP2A-regulatory proteins.

Figure 3. E4orf4’s protein network identified by AP-MS analyses

Comparative analysis of proteomic datasets from three independent AP-MS studies using E4orf4 as a bait, expressed in MDA-MB-231 (purple, high throughput).[107], 293T (grey, high throughput)[112], and H1299 cells (light blue, low throughput)[113]. High throughput proteomic data were analyzed using Significance Analysis of INTeractome (SAINT) algorithm with a threshold of 0.5 based on background filtering (Dziengeleski et al. dataset: PXD014329)[107], or 0.95 (Mui et al, dataset: PXD001956). [160,161] High confidence E4orf4’s interactors were manually classified in relevant biological processes by verifying the main associated GO: Biological process identified with the Uniprot database for each protein (Update: October 15, 2019)[162].

Figure 4. Schematics of E4orf4-associated PPP-regulatory proteins, and their relationships to the Hippo pathway and PAR complex network

Left: E4orf4 targets a limited subset of PPP-regulatory proteins that may impact mechanotransduction mechanisms. Center: Several proteins related to the Hippo pathway were detected in E4orf4 complexes, including CCDC85C and RASSF8, two proteins that affect cell junction remodeling with opposing effects on Hippo signaling. While CCDC85C can promote YAP nuclear export, RASSF proteins can inhibit the MST/Hippo kinase [128,130]. How E4orf4 influences Hippo components and PP1 catalytic activity remains to be defined. Right: Analysis of proteomic datasets was performed using Cytoscape (V3.8.0) and the ReactomeFIVIz plugin (app6), revealing a putative PPP phosphatase relay that may be controlled by polarity proteins and hijack by E4orf4 [163,164].

PAR3 and the E4orf4-dependent induction of NE rupture

The PAR complex regulates epithelial polarity, apical domain formation, and cell junction remodeling [117,118]. PAR3 also plays a crucial role in the segregation of active Cdc42/aPKC to facilitate the spatial control of actomyosin filament organization [119,120]. Therefore, we postulated that E4orf4 subverts PAR3’s scaffolding function to deregulate perinuclear actomyosin filament dynamics and NE mechanical stability.

Our studies revealed that E4orf4’s main tumoricidal phenotypes, including NE rupture, are regulated by PAR3[107]. Functional dissection of the E4orf4-PAR3 interaction using integrated protein biochemistry and cell imaging approaches indicates that these phenotypes are highly correlated with a proximal interaction between E4orf4 and PAR3 at perinuclear sites. Moreover, E4orf4 can associate with other members of the PAR complex (PAR6/PARD6 and PKCɩ), in a PAR3-dependent manner. While the E4orf4-PAR3 interaction is observed also in nontumorigenic cells, the functional outcome is different. Rather than the appearance of tumoricidal phenotypes, E4orf4 promotes the assembly of cell-cell junctions in nontumorigenic cells, suggesting that E4orf4 reinforces polarity signaling. These results argue that differences in cell mechanics enable E4orf4 to exploit an undescribed PAR complex activity in tumor cells. Mechanistically, E4orf4 induces PAR3 clustering at the NE in a tumor cell-selective manner, thereby triggering the recruitment of actin fibers, nuclear deformation, and transient NE rupture Figure 2A. High-resolution confocal microscopy revealed that the PAR3 clusters are recruited to sites of nuclear indentation, showing striking alignment with NE folds Figure 2B. Since apical actin cables also localize along nuclear surface invaginations or NE folds of lamin A/C in E4orf4-expressing cells, PAR3 clusters may define sites of actin-dependent mechanical stress at the NE[107]. Importantly, disrupting either myosin II, PAR3, or the LINC complex, can all recapitulate the same inhibition of E4orf4-induced NE rupture, cell blebbing, and nuclear condensation[107]. These results suggest that E4orf4 acts by corrupting PAR3’s scaffolding function within the NE, which may control actomyosin filament tethering and thereby, force transmission to the nucleus. Since we observed that PAR3 regulates a retrograde flowing of actin fibers towards the nucleus, the PAR complex could provide the spatial information to polarize actomyosin filaments and nucleo-cytoskeletal coupling. However, future work is needed to determine whether PAR3 is engaged with the NE via an interaction with the LINC complex or the nuclear lamina.

Based on current knowledge, PAR proteins could also affect nuclear stiffness via the phosphorylation of lamina proteins, as reported during NE remodeling. This enables the nuclear exit of large RNP particles via NPC-independent nuclear budding in Drosophila[46]. By acting on both lamina proteins and perinuclear actin organization, PAR complex proteins could represent key regulators of nucleo-cytoskeletal connections in response to mechanical forces.

PP1-regulatory proteins targeted by E4orf4 in cellular mechanotransduction

PP1, through its catalytic subunits, interacts with hundreds of putative regulatory proteins that control substrate access and activity on different pathways[115]. Proteomic datasets indicate that E4orf4 interacts with the three PP1 catalytic subunits (PPP1C) but only a few PP1-regulatory proteins (PPP1R; Figures 3 and 4) [101,112]. Remarkably, these PP1-regulatory proteins show functional relationships with cellular contractility networks and polarity protein signaling.

Three PP1-regulatory proteins, MYPT1, MYPT2, and MBS85, are myosin phosphatase targeting subunit (MYPT) family members (Figure 4, PPP1R12A-C). These subunits share homology and function in targeting PP1 catalytic activity to dephosphorylate myosin light chain, thereby inhibiting cellular contractile forces [121,122]. These interactions are likely to be relevant for E4orf4’s tumoricidal action, as myosin phosphatase complexes act to safeguard nuclear integrity in cancer cells. Indeed, the loss of either MYPT1 or PPP1CB subunits was shown to cause nuclear dysmorphia, NE rupture, and genome instability in cancer cells[34]. Notably, the presence of actin filaments at NE rupture sites was correlated with nuclear damage. Thus, it is reasonable to speculate that E4orf4 may inhibit MYPT-PP1 catalytic activity to enhance NE-actomyosin filament tethering. In line with this, E4orf4 was found to stimulate MYPT1 phosphorylation on residues that inhibit myosin phosphatase complex activity, a process that is mainly catalyzed by ROCK [101,121]. Moreover, E4orf4 recruits the Rho-ROCK signaling axis to the perinuclear region, promoting actomyosin filament assembly [101,102]. Of note, E4orf4-interacting MYPT proteins share the ROCK-targeted residue[121]. Additionally, myosin phosphatase complexes are suggested to play other non-contractile functions[122]. Among potentially relevant substrates, MYPT1 can interact with ezrin-radixin-moesin (ERM) proteins that organize specialized actin filament-membrane domains in a phosphorylation-dependent manner[123]. The MYPT1-PP1 complex also appears to modulate the tumor suppressor RB and could, in principle, affect its ill-defined function in NE rupture [33,124].

Intriguingly, other E4orf4- and PP1-associated proteins are members of the apoptosis stimulating proteins of p53 (ASPP) family, notably ASPP1 and ASPP2, and inhibitor of ASPP protein (iASPP) (Figures 3 and 4; PPP1R13A, 13B, 13 L). ASPP proteins were initially characterized as modulators of apoptosis mediated by the p53 family of tumor suppressors[125]. However, E4orf4-induced cell killing is p53-independent [96,98,108]. Therefore, E4orf4 might target other functions of ASPP proteins.

Current knowledge of PP1 regulation by ASPP proteins is limited. Interestingly, ASPP1/2-PP1 complexes were shown to dephosphorylate and activate the Hippo pathway transcriptional coactivator YAP Figure 4 [126,127]. As aforementioned, the Hippo pathway is deregulated in cancer and YAP regulates cytoskeletal remodeling and cell proliferation in response to high density or mechanosensory stimulation[50]. It is important to note that E4orf4-induced cell killing was correlated with YAP phosphorylation and downregulation[112]. Since YAP silencing can enhance cell death in response to E4orf4 expression, this further argues that E4orf4 susceptibility is regulated by cell mechanics. Moreover, several additional proteins linked to Hippo signaling were found in E4orf4 complexes Figures 3 and 4. This includes Ras-associated domain family proteins 7 and 8 (RASSF7/8) and coiled-coil domain-containing protein 85 C (CCDC85C), which were shown to differentially affect Hippo signaling [128–131]. It will be important to dissect the exact impact of E4orf4 on these Hippo pathway components and PP1-catalytic activity. Since enhanced YAP phosphorylation seems independent of the Hippo pathway Large tumor suppressor kinase 1/2 (LATS1/2), YAP inhibition could rather be a result of E4orf4-dependent cellular mechanotransduction deregulation[112]. Of note, we observed that NE rupture correlates with YAP nuclear exit in response to E4orf4 expression.²

Finally, ASPP2 function in cellular junction remodeling is particularly interesting for mechanical signal transduction to the nucleus. As mentioned earlier, intercellular adhesions, like cell-matrix adhesions, are connected to the nuclear lamina via the LINC complex that transmits actomyosin forces to the NE [54,55]. ASPP2, by restraining Src kinase activity and recruiting the polarity protein PAR3 and RASSF7/8 to cell junctions, positively regulates apical domain formation and epithelial cell polarity [132–134]. Furthermore, ASPP2 was shown to act as a molecular switch in epithelial plasticity, favoring a mesenchymal-to-epithelial transition (MET) in tumors via a pathway that implicates cell junction remodeling[135]. These findings suggest that an ASPP2 function in intercellular force transmission is relevant for tumorigenesis. A role for PP1 catalytic activity was not explored in this specific context. However, recent work indicates that Drosophila ASPP protein shares most of its biological functions with PP1 catalytic activity and the PP1-associated proteins RASFF7/8 and CCDC85C[136]. It is tempting to speculate that E4orf4 could hijack components of the ASPP2-PP1 complex to exploit their function in the control of mechanical forces and segregation of specialized actomyosin filament-membrane domains. Recent finding for a dynamic between E4orf4 and PAR3, another component of epithelial cell junction that interacts with ASPP2, provides support to this hypothesis [107,132,133].

E4orf4-associated PP2A targeting subunits in nuclear dynamics

PP2A phosphatase is the first identified E4orf4 target with important roles in both viral replication and cell killing [95,137], While there is still no consensus regarding how E4orf4 impacts PP2A catalytic activity, it is interesting to speculate on a potential function for PP2A in E4orf4-induced NE remodeling. Strong evidence supports a crucial role for B55α, PP2A regulatory subunit, in mediating efficient cell killing by E4orf4 in mammalian cell lines and Drosophila [97,111,138]. Initial work revealed that E4orf4 selectively associates with the trimeric PP2A holoenzyme via interaction with B55α, consistent with proteomic datasets showing strong enrichment for B55α, but also for B55γ and δ Figures 3 and 4 [137–139]. Since the E4orf4-PP2A complex is associated with PP2A-like phosphatase activity, it is proposed that E4orf4 can target the B55-PP2A complexes to dephosphorylate a subset of substrates required for cell death. In the context of virally infected cells, E4orf4 can target PP2A to dephosphorylate serine/arginine-rich proteins (SR proteins) along with the NPC protein NUP205 [113,140,141]. Nevertheless, in the context of tumor cell killing, no substrate for the B55-PP2A complexes has been formally identified.

Dissecting the role of B55 subunits in E4orf4-induced NE remodeling could help uncover relevant substrates. It is interesting to note that PP2A is a major regulator of cell junctions remodeling and polarity signaling, which could, therefore, affect force transmission to the NE[142]. Moreover, B55α was identified as the main regulatory subunit to target PP2A to dephosphorylate BAF during post-mitotic NE reassembly [143–146]. We confirmed that BAF is recruited to NE rupture sites in E4orf4-expressing cells.³ This points to the possibility that molecular mechanisms are induced to couple both NE rupture and repair in response to E4orf4 expression. Therefore, the role of B55α-PP2A complex in these events merits continued investigation.

Presently, it is not known if E4orf4 affects NE stiffness directly and decreases nuclear mechanical resilience to enhanced perinuclear actomyosin contractility. Chromatin compaction, which is linked to the nuclear lamina, is considered as a potential source of intranuclear forces that influence nuclear rigidity [147,148]. Since E4orf4 promotes PP2A targeting to the chromatin remodeler complex ACF, it could theoretically alter chromatin structure and thereby affect nuclear stiffness [149–151]. Also, the ability of E4orf4 to modulate B55-PP2A-mediated dephosphorylation of NPC components, notably NUP205, could impact NE stability, as several connections between the nucleopore complex, lamina, and chromatin have been identified [21,113]. Finally, diversion of B55-PP2A complexes could indirectly alter the phosphorylation of NE components and affect NE mechanical stability. For instance, lamin A phosphorylation by constitutive kinases can stimulate its degradation, and, theoretically, reduce nuclear stiffness [40,51,152]. If B55-PP2A complexes contribute to E4orf4-induced NE remodeling, the identification of relevant B55-PP2A complex interactions will have broad significance for cell biology.

Deciphering bona fide regulators of NE plasticity using the adenoviral protein E4orf4

The work using E4orf4 to model mechanical stress-induced NE rupture suggests the existence of unappreciated relationships between polarity protein signaling and NE remodeling in tumor cells. Since several of E4orf4’s targets share a function in cell junction remodeling, they could contribute to LINC complex-associated signal transduction mechanisms that may be rewired in tumor cells to promote NE plasticity. This hypothesis is in line with our finding that PAR3 can regulate spontaneous NE rupture in susceptible tumor cells, in the absence of E4orf4. Indeed, PAR3 depletion was shown to hinder spontaneous NE rupture facilitated by the downregulation of nuclear lamins[107]. These results suggest that PAR3 is a bona fide regulator of the actin-dependent forces that remodel the NE in tumor cells that are harnessed by E4orf4. Cell-cell adhesion and polarity signaling mechanisms are typically deregulated in tumor cells [3,13,15]. This can trigger ectopic signaling by these proteins in other cell compartments, which might render these networks prone to be hijacked by E4orf4 [153,154].

We suggest that E4orf4 subverts bona fide regulators of NE rupture and repair that enable tumor cells to cope with physical constraints during metastatic migration. Dissecting the biology of E4orf4’s targets will be crucial. Intriguingly, our analysis of proteomic datasets revealed that the B55-PP2A and MYPT-PP1 complexes can potentially interact with each other via the polarity signaling proteins aPKC-PAR3, and ASPP2, respectively Figure 4[155]. There is precedent for a PP1-PP2A phosphatase relay in yeast, and the molecular mechanisms appear to be conserved in mammalian cells[156]. In this paradigm, PP1 is recruited to B55-PP2A at the mitotic exit to coordinate the sequential re-activation of these phosphatase systems that contribute to post-mitotic NE reassembly[94]. It is tempting to speculate that polarity signaling hubs could couple actomyosin filament tethering at the NE and the recruitment of the NE repair machinery during NE remodeling, to rapidly cope with possible NE damage.

In conclusion, our understanding of E4orf4’s protein network, as it relates to NE rupture and repair is rather incomplete. Identifying highly dynamic and short-lived interactions that are likely to involve cytoskeletal and membrane-bound insoluble proteins will require using alternative methods. Since NE rupture and repair are tightly coupled, the rapid biotinylation of proximal proteins could capture bona fide molecular effectors that are near to E4orf4 as it leaks out via local NE rupturing. Future work using the adenovirus protein E4orf4 will undoubtedly decipher NE plasticity mechanisms that broadly define tumor cells and identify novel entry points for combination therapies.

Acknowledgments

We apologize to those whose work was not discussed, due to space limitations. The authors wish to acknowledge the crucial contribution of Claire Dziengelewski for the initial discovery of a role for PAR3 in NE rupture. This work was funded by the Cancer Research Society Inc (CRS) under Operating Grant to J.N.L. (NIP: 23395), and by the Natural Sciences and Engineering Research Council (NSERC) under Discovery Grant to J.N.L. (RGPIN/05849). M-A Rodrigue is supported by a doctoral award from the Fonds de la Recherche en Santé du Québec (FRQS).

Disclosure statement

No potential conflict of interest was reported by the authors.

Additional information

Funding

This work was funded by the Cancer Research Society Inc (CRS) under Operating Grant to J.N.L. (NIP: 23395), and by the Natural Sciences and Engineering Research Council (NSERC) under Discovery Grant to J.N.L. (RGPIN/05849). M-A Rodrigue is supported by a doctoral award from the Fonds de la Recherche en Santé du Québec (FRQS).

Notes

1. Jacquet K. and Lavoie J.N., unpublished observations.

2. Dziengelewski C. and Lavoie J.N., unpublished observations.

3. Jacquet K. and Lavoie J.N., unpublished observations.