The Host Cytoskeleton: A Key Regulator of Early HIV-1 Infection

Abstract

Due to its central role in cell biology, the cytoskeleton is a key regulator of viral infection, influencing nearly every step of the viral life cycle. In this review, we will discuss the role of two key components of the cytoskeleton, namely the actin and microtubule networks in early HIV-1 infection. We will discuss key contributions to processes ranging from the attachment and entry of viral particles at the cell surface to their arrival and import into the nucleus and identify areas where further research into this complex relationship may yield new insights into HIV-1 pathogenesis.

Graphical Abstract

The host cytoskeleton plays a crucial role in replication of diverse viral families. In this review, we provide an overview of the roles of the key components of the cytoskeleton, namely the actin and microtubule networks in the early events of HIV-1 infection, encompassing attachment and entry of viral particles at the cell surface, passage through the cortical actin barrier and cytosol, and arrival at the nuclear membrane.

Introduction

The Cytoskeleton and Viral Infection

As an important regulator of cellular morphology and function, the cytoskeleton presents both challenges and opportunities during viral infection. Cytoskeletal elements are the primary structural determinants of cellular morphology, and play crucial roles in processes such as intracellular cargo transport, membrane dynamics, as well as cell division and motility [1, 2]. As a result, cytoskeletal elements regulate nearly every aspect of viral infection, from attachment and entry to virus particle assembly and egress [3]. In recent years, considerable work has been undertaken to examine the role of the cytoskeleton and its associated proteins in the replication of numerous viruses with considerable implications to human health. In this review, we will discuss the role of the actin and microtubule components of the cytoskeleton in HIV-1 infection.

The Mammalian Cytoskeleton

The eukaryotic cytoskeleton is composed of dense and interconnected networks of filamentous proteins arranged throughout the cell. These can be delineated into three classes of filament by size and composition; actin filaments (or microfilaments), intermediate filaments, and microtubules (Figure 1). Microfilaments represent the thinnest and most abundant element of the cytoskeleton, composed of actin homodimers organized head to tail into a double helical filament. Actin filaments are highly dynamic polar structures, composed of plus and minus ends with distinct assembly kinetics and growth rates. Growth occurs more rapidly at the plus ends, which are generally pointed towards the plasma membrane and into cellular protrusions. Actin filament assembly and disassembly are tightly regulated by a large and complex network of actin interacting proteins, including various cellular factors promoting actin polymerization, depolymerization, and branching, as well as the Rho family of GTPases [1].

Figure 1. Organization of the host cytoskeleton.

The arrangement of cytoskeletal elements within the cell are shown. The cell membrane is bordered by a dense cap of actin filaments (or microfilaments) underneath the plasma membrane (termed the cortical actin) that regulates membrane shape, cell motility, and organizes receptors and signaling complexes on the cell surface. Microtubules are typically arranged with their growing plus ends oriented outward, while their less dynamic minus ends are typically anchored at a microtubule organizing center (MTOC), such as the centrosome or the Golgi. The cortical actin network is linked to the microtubule cytoskeleton through a variety of mechanisms including actin-microtubule crosslinking factors or microtubule plus-end tracking protein (+TIP) complexes, which can result in local stabilization of a subset of otherwise dynamic microtubules. These subsets of stable microtubules can serve as platforms for long range cargo transport to specific regions of the cell to generate cellular polarity. Transport along microtubules is driven by kinesin (plus-end directed) and dynein (minus-end directed) motors. Intermediate filaments are dispersed throughout the cytosol.

Microtubules are the largest component of the cytoskeleton by diameter, composed of alpha and beta tubulin heterodimers arranged into hollow tubes [4]. Similar to actin filaments, microtubules are polarized and highly dynamic. The less-dynamic minus ends of microtubules are anchored at and nucleate the formation of new filaments from various microtubule organizing centers (MTOCs) (Figure 1). In most cell types, the centrosome, a perinuclear structure central to cell division, is the primary MTOC, while other MTOCs also exist at regions such as the Golgi apparatus and plasma or nuclear membrane [2]. Microtubules grow from these MTOCs and often exhibit rapid phases of growth and disassembly (catastrophe) through the addition and loss of α-/β-tubulin homodimers at the plus-end. This dynamic behavior allows the microtubule plus end to move throughout the cytosol in a “search and capture” process to encounter, bind and initiate the transport of cargos such as cellular organelles. The stability of individual microtubules can be influenced by microtubule interacting proteins and post-translational modifications, and the proportion of stabilized and dynamic microtubules can vary considerably between cell types [4]. Stabilized microtubules are characterized by a distinct profile of post-translational modifications, including acetylation and detyrosination, and are specifically recognized by motor proteins to serve as platforms for long-range transport of specific cargoes. Microtubules are recognized by a vast network of microtubule-associated proteins (MAPs) that include motor proteins, nucleating, stabilizing and disassembly factors, along with a highly specialized subset of MAPs known as the plus-end tracking proteins (+TIPs)[5]. +TIPs act as key regulators of microtubule dynamics and function, serving as bridging factors between microtubules, the cell membrane, and the actin network, and controlling microtubule growth and assembly. These +TIP networks are in turn organized by end-binding proteins (EBs), key regulatory members of the +TIP family which bind to the GTP-tubulin cap that transiently exists at the microtubule plus end and which facilitate the recruitment of other +TIP family members to execute different effects and functions of the microtubule plus-end, such as cargo capture or engagement with the plasma membrane [6].

The polarity of actin and microtubule filaments coupled with their polarity-directed motors is central to their role as ‘transport highways’, both locally and long-range. Transport along microfilaments is typically short-range and largely limited to the cell periphery, while long range transport to and from the nucleus is mediated by microtubule associated motors [7, 8]. Myosin motors regulate the transport of cargoes along actin filaments, with distinct myosin families serving as plus- and minus end directed motors [8] (Figure 2). Microtubule-directed transport is mediated by two families of motor proteins; members of the kinesin family largely direct plus-end or anterograde movement while dynein controls minus-end directed or retrograde transport to the nucleus [7].

Figure 2. Exploitation of the host actin cytoskeleton during early HIV-1 infection.

1. HIV-1 particles consist of a conical capsid encasing two copies of the viral single-stranded RNA genome, surrounded by a matrix protein shell within a lipid envelope studded with viral glycoproteins (gp120/gp41). Incoming HIV-1 particles engage their receptor, CD4 and coreceptors (CXCR4 or CCR5) at receptor and coreceptor-enriched domains in the cell membrane. Binding of the HIV-1 envelope to CD4/coreceptor triggers a signal transduction cascade that promotes actin polymerization through activation of the WAVE2, Abl, ARP2/3 complex and LIMK1-mediated phosphorylation and inactivation of cofilin. 2. Increased actin dynamics results in the formation of a dense “cap” of actin at the entry site that facilitates HIV-1 interactions with receptors and coreceptors and promotes fusion. 3. Following internalization of the HIV-1 core, dephosphorylation and activation of cofilin results in localized actin depolymerization at the entry site, forming a pore-like gap in the actin network that enables the particles to pass through to the cytoplasm. 4. HIV-1 cores at the cell periphery exploit myosin motors for short-range transport on actin filaments prior to their transition onto microtubules through actin-microtubule crosslinking proteins and microtubule plus-end tracking proteins (+TIPs) to initiate retrograde transport to the nucleus.

Intermediate filaments are intermediate in diameter between actin filaments and microtubules and comprise a stable and static network of structural support proteins (Figure 1). Intermediate filaments lack polarity and are intrinsically stable, composed of antiparallel helices of disulfide-linked component proteins [9]. As a result, intermediate filaments relocalize to and organize cytosolic assembly compartments during the replication of diverse viral groups [10–13]. However, they have not been reported to play an active role in early HIV-1 infection and as such are not discussed in this review.

HIV-1 and the Actin Cytoskeleton

Although actin filaments are found throughout the cell, including transiently forming inside the nucleus itself, they are abundant at the cell membrane and play a crucial role in the very earliest stages of HIV-1 infection [14, 15] (Figure 2). This includes influencing the organization of receptors used by viruses for cell attachment and entry. HIV-1 entry occurs through the engagement of the HIV-1 receptor, CD4, and one of two chemokine coreceptors CXCR4 (T-cell tropic) and CCR5 (macrophage-tropic) by the viral envelope proteins gp120 and gp41, triggering conformational changes that expose the gp41 fusion peptide and bring it into contact with the host membrane to initiate membrane fusion [16]. A number of studies have revealed that HIV-1 entry occurs specifically at CD4 and coreceptor-enriched microdomains at the cell surface [17–21]. While some studies have pinpointed cholesterol-enriched lipid rafts as HIV-1 entry sites [18, 22], other studies have suggested that HIV-1 binding and entry are not raft-dependent [23, 24]. This likely reflects both cell-type dependent entry mechanisms and the ability of HIV-1 to use more than one route of entry. Indeed, beyond membrane fusion, HIV-1 can also enter cells by endocytosis [25]. Regardless of the precise nature and composition of HIV-1 entry sites, actin-dependent clustering of CD4 and its coreceptors at discrete membrane sites is required for HIV-1 receptor engagement and entry. Several studies have shown that CD4 and CXCR4 colocalize in membrane microdomains in the absence of HIV-1 envelope in natural target cell types [20, 26, 27]. Some early studies have suggested that the presence of HIV-1 envelope may promote CD4/CXCR4 clustering to facilitate fusion and entry [17, 22, 28]. This clustering mechanism appears to be partially dependent on actin, as treatment with the actin polymerization inhibitor cytochalasin D decreased the recruitment of CXCR4 at CD4 clusters induced by treatment of Jurkat and primary T cells with anti-CD4-coated beads [22]. However, other studies do not observe envelope-induced receptor/coreceptor clustering above the levels observed at baseline, suggesting that this mechanism may also play a cell-type dependent role in HIV-1 infection [26, 29].

Actin filaments continue to play important roles directly after entry into the cytosol as viral particles must pass through the cell cortex, a dense network of actin filaments and associated proteins underneath the cell membrane [14, 30] (Figure 2). Indeed, cortical actin represents a significant barrier to viral entry and successful replication, and viruses have developed a diverse array of strategies to control local actin dynamics. These include encoding proteins that directly bind to and modulate actin assembly kinetics or binding and coopting cellular actin regulators [3, 14, 15]. On the other hand, an intact actin cytoskeleton is also required for HIV-1 entry and post-entry processes [17, 31]. Advances in live cell imaging have provided insights into the role of actin in HIV-1 entry through direct, real-time visualization of actin dynamics in cells infected with fluorescently labeled HIV-1 particles. Using live cell imaging of quantum-dot labeled HIV-1 particles showed that HIV-1 binding and entry into resting CD4+ T cells is associated with a series of actin rearrangement events. Binding of labeled HIV-1 particles to the cell surface was associated with the formation of a thick actin cap at the site of entry, followed by the formation of a pore-like gap in the actin layer through which the viral particle passes to enter the cytosol [32]. These results suggest that HIV-1 binding induces actin polymerization and depolymerization events to facilitate cell entry and passage through the cortical actin barrier. Indeed, several studies have shown that both actin polymerization and depolymerization are required for HIV-1 entry and productive infection. The formation of an actin ‘cap’ through actin polymerization downstream of HIV-1 receptor binding has been suggested to facilitate infection by stabilizing interactions between HIV-1 envelope and its coreceptors and receptors during entry [15, 17, 32–34]. This appears to occur primarily through the activation of intracellular signal transduction cascades triggered by HIV-1 receptor and coreceptor binding. Chemokine receptor signaling has been shown to modulate actin dynamics during chemotaxis, and treatment of primary CD4+ T cells with gp120 or inactivated virus promotes actin polymerization and chemotaxis through CXCR4 activation [35, 36]. Induction of chemotactic responses is also seen with CCR5-tropic viruses, suggesting that this mechanism promotes HIV-1 entry regardless of receptor tropism [37]. Transient actin polymerization is reported to be regulated by a number of cellular kinases including LIMK1 (a kinase that phosphorylates and inactivates cofilin), which lies downstream of both the CD4 and CXCR4 signaling axes that is triggered by HIV-1 attachment [34] (Figure 2). Knockdown of LIMK1 decreases actin polymerization and increases the rate of membrane-associated CD4 turnover, suggesting that LIMK1 activation may be required to stabilize HIV-1 coreceptor complexes at the membrane to enable completion of entry prior to receptor recycling [34]. Moreover, fusion of the HIV-1 envelope with the plasma membrane activates the RhoA-ROCK signaling pathway resulting in actin polymerization at sites of viral entry through phosphorylation and inactivation of cofilin. This signaling event was found to be dependent on Filamin A, an actin binding protein involved in CD4 and coreceptor clustering [38]. Membrane-fusion mediated HIV-1 entry has also been suggested to induce Rho GTPase signaling pathways to promote actin polymerization through additional actin regulatory proteins such as Abl, WAVE2 and Arp2/3 complex, as well as moesin, a member of the ERM (ezrin-radixin-moesin) family of proteins that cross-link actin filaments with the plasma membrane [38–40]. Following entry, the newly internalized HIV-1 particle must pass through the cortical actin barrier for transport to the nucleus. This appears to be coupled with the actin remodeling events that facilitate receptor clustering and includes the activation of cofilin downstream of CXCR4 in an envelope-dependent manner, resulting in reduced actin polymerization[33]. The actin-interacting protein alpha-actinin also appears to play a key role in this process, as aggregation of alpha-actinin at sites of HIV-1 entry coincides spatially and temporally with the formation of a pore in the cortical actin cap in live imaging studies [32].

Beyond these early signaling events, late in infection both actin and actin regulatory proteins are incorporated into newly assembled HIV-1 particles or virions [41, 42]. This raises the possibility that these cellular proteins may then function to facilitate early steps in infection of neighboring cells. However, this remains unclear as treatment of HIV-1 particles with the actin depolymerizing agent cytochalasin D prior to infection does not affect particle infectivity [17], and additional work has suggested that actin loading in virions may occur passively during egress [43]. Moreover, several HIV-1 proteins have been implicated in regulating several of these early events. Remodeling of the cortical actin barrier to allow HIV-1 penetration is actively induced by the viral accessory protein Nef, which has been shown to enhance infectivity at least in part by promoting membrane-fusion mediated HIV-1 entry [44, 45]. In addition, the viral reverse transcriptase and integrase proteins as well as the matrix and nucleocapsid structural components of the virion have all been shown to interact directly with actin, although the precise roles of these interactions in early HIV-1 infection remain to be elucidated [46–50]. Finally, studies suggest that actin also plays a role in the transport of incoming HIV-1 particles both at the cell periphery and at the nucleus. Specifically, live cell imaging revealed that HIV-1 particles undergo movement consistent with myosin-directed transport in both the periphery and perinuclear region of infected cells [51], suggesting that actin-based transport at these subcellular sites may be involved in the loading and unloading of HIV-1 particles onto and off of microtubules that generally mediate their longer-range transport.

HIV-1 and the Microtubule Cytoskeleton

Like actin filaments, microtubules play a critical and multifaceted role in early HIV infection [52]. Similar to many other viruses, HIV-1 switches from initial short-range movement on actin at the cell periphery to microtubule tracks to facilitate long-range movement of to the nucleus [53] (Figure 2). Several host actin-microtubule crosslinking factors associate with viral cores and appear to both regulate actin-microtubule crosslinking and positively or negatively affect early infection [54–57]. The focal adhesion proteins talin and vinculin negatively regulate HIV-1 infection by inhibiting the phosphorylation of paxillin, a key scaffolding protein in focal adhesions, which is associated with both actin and the microtubule cytoskeleton [58]. Moreover, some of these actin-microtubule cross-linking proteins, such as the ERM family members, moesin and ezrin function as negative regulators of stable microtubule formation and early HIV-1 infection [54–56]. By contrast, the moesin interacting factor, PDZ domain-containing protein 8 (PDZD8) positively regulates microtubule stabilization and facilitates early infection [59]. This illustrates the importance of stable microtubules during early HIV-1 infection (discussed further below), with the effects of ERM or PDZD8 proteins on early infection correlating with their effects on these microtubule networks. In addition, other actin-microtubule cross-linking factors, such as the diaphanous-related formin family members Dia1 and Dia2, have more recently been shown to facilitate HIV-1-induced stabilization of microtubule subsets and also promote infection [57]. This occurs in a complex manner and involves HIV-1 particles binding to their actin regulatory domains, in effect mimicking actin in the actin-microtubule crosslinking process to induce their microtubule stabilizing properties. This in turn highlights both the complex interplay between actin and microtubules at the cell periphery, and the complex ways in which HIV-1 particles themselves intersect into and exploit these activities.

Beyond these actin-microtubule cross-linking proteins, HIV-1 further interacts with and exploits highly specific microtubule regulatory proteins that enable the virus to stabilize microtubule filaments that in turn facilitate long-range transport across the cytoplasm to the nucleus (Figure 3). While early studies using microtubule depolymerizing agents appeared to suggest a negligible role for microtubules in HIV-1 infection in human T cells [48, 60], a significant body of subsequent research has implicated a number of microtubule regulators, microtubule-associated proteins, and microtubule-based motors as essential host factors in HIV-1 infection [3, 52]. There are a number of explanations for this apparent discrepancy with earlier studies using tubulin-targeting drugs. For example, some studies removed the drug during infection but these inhibitors are rapidly reversible and microtubules can quickly reassemble. Moreover, stable microtubules, which have emerged as key players in early HIV-1 infection, are highly resistant to many depolymerizing drugs such as nocodazole[52, 61–63]. As such, it is possible that the extent of microtubule depolymerization in these studies was insufficient to prevent HIV-1 infection. Alternatively, complete depolymerization of microtubules can dramatically alter cell shape and may enable viral particles to instead use alternate routes to the nucleus, such as actin filaments. Indeed, it is now clear that productive HIV-1 infection is dependent on stabilized microtubule subsets and several factors that regulate their formation have significant impacts on infection. The first evidence of this emerged from studies of the central microtubule plus-end regulatory protein, EB1, whose depletion prevents HIV-1 from stabilizing microtubules [61]. HIV-1-mediated stable microtubule formation was shown to be regulated by the matrix protein of incoming viral particles targeting a novel EB1-associated +TIP, Kif4, soon after viral entry into the cell. These stabilized microtubules subsets become acetylated and detryosinated, and are required for efficient transport of HIV-1 particles to the nucleus [61] (Figure 3). Since this initial discovery, a range of MAPs and +TIPs have been found to contribute to this process in different ways and combined, targeting various MAPs and +TIPs in multiple ways likely allows HIV-1 to amplify the formation of stable MT networks during the early stages of infection. Depletion of the microtubule stabilizing protein suppressor of G2 allele of skp1 (SUGT1) results in increased instability of microtubule plus-ends at the cellular cortex and a significant impairment in early HIV-1 infection in lymphocytes and macrophages [64]. The mechanistic target of rapamycin (mTOR) signaling also appears to regulate HIV-1 infection in part through its effects on microtubule acetylation [65]. Pharmacological inhibition of mTOR depletes cellular acetyl-CoA pools through its effects on glucose and glutamine metabolism, inhibiting HIV-1 induced microtubule acetylation and nuclear import in T cells [65]. In addition, knockdown of the microtubule associated proteins MAP1A and MAP1S results in a post-fusion reduction in HIV-1 replication, characterized by defective retrograde trafficking of HIV-1 particles and loss of nuclear accumulation and import [66]. This was accompanied by a reduction in capsid-microtubule colocalization. As both MAP proteins were shown to interact with HIV-1 capsid-nucleocapsid (CA-NC) assemblies, these proteins likely represent important tethering factors required for maintaining HIV-1 association with microtubules during transport [66]. Interestingly, two other MAPs, MAP4 and DNAL1 promote HIV-1 infection prior to nuclear import, though their precise role in infection remains unclear [67].

Figure 3. Exploitation of the host microtubule cytoskeleton during early HIV-1 infection. Microtubule capture and stabilization.

Incoming HIV-1 particles engage and stabilize microtubules very early in infection, which is regulated through interactions with actin-microtubule cross-linking proteins and microtubule plus-end tracking proteins (+TIPs). Upon fusion of HIV-1 particles into the cytosol, the matrix protein shell (pink) is released and stimulates MT stabilization by binding the EB1-associated protein, KIF4. Microtubule stabilization is further induced by incoming capsids that engage proteins such as Dia1/2, CLASP2 and MAP1A/S. Of note, viral cores also bind the SxIP motif containing +TIP, CLIP170, which does not facilitate MT stabilization but plays a role in loading HIV-1 cores onto MTs for transport and regulates core uncoating. This activity of CLIP170 is negatively regulated by dynactin-1 (DCTN1), which may have driven HIV-1 choice of BICD2 as its dynein adaptor. Other factors that don’t directly associate with incoming viral capsids, such as SUGT1, also maintain the stability of microtubule plus-ends to facilitate trafficking of viral cores towards the nucleus. Bidirectional transport. Long-range bi-directional transport of the HIV-1 cores towards the nucleus on stable microtubules is facilitated by capsid binding to the kinesin adaptor, FEZ1 and the dynein adaptor, BICD2. Transfer to nuclear pore. It is unclear how HIV-1 particles migrate from microtubules to the nuclear membrane, but this may involve interactions between viral particles and centrosomal components and/or actinomyosin based transport from the perinuclear MT organizing center (MTOC) to the nuclear pore. Whether HIV-1 capsids entering the nucleus are fully intact or are partially uncoated remains a hotly debated topic in the field, and we therefore depict both uncoated and coated states in this illustration of viral transport. Ultimately, it is possible that HIV-1 can utilize both processes.

A subset of MAPs, namely the microtubule plus-end regulatory +TIPs, are emerging as particularly important and multifunctional players in early HIV-1 infection [57, 61, 68, 69]. This includes mediating the capture and loading of viral particles onto microtubules, stabilization of microtubules, and control of the subsequent transport and uncoating (disassembly of capsid) of viral particles as they transit the cytosol. Recent studies have revealed various steps in these processes and the roles played by different +TIPs. The central +TIP protein, EB1, does not bind HIV-1 particles themselves but plays a critical role in delivering other +TIPs that do bind viral cores and execute different functions [69]. Binding to some +TIPs, such as CLASP2 [68], enables HIV-1 to stabilize microtubules while binding to CLIP170 does not [69] (Figure 3). Instead, binding to CLIP170 facilitates the capture of HIV-1 particles by microtubules for subsequent transport, and also regulates uncoating of the viral core. The ability of HIV-1 particles to bind several +TIPs to facilitate these processes was found to lie in a motif within the viral capsid protein that structurally and functionally mimics a key +TIP binding motif in EB1 [69]. In effect, once EB1 delivers +TIPs to the cell periphery, incoming viral particles mimic EB1 to then engage these +TIPs and control several of their activities. EB1 binds directly to microtubules through its N-terminal calponin homology domain and recruits other +TIP proteins through interaction domains in its disordered C-terminus [70, 71]. This C-terminal domain in EB1 contains two distinct motifs that facilitate the recruitment of two distinct families of +TIP proteins. +TIPs containing what is termed an SxIP motif bind to the EBH domain of EB1, while +TIPs containing cytoskeleton-associated protein glycine rich (CAP-Gly) domains are recruited to an EEY motif in the EB1 C-terminus [70]. While EB1’s CAP-Gly binding EEY motif is required for it to deliver +TIPs to the cell periphery and promote early HIV-1 infection, HIV-1 capsid protein mimics EB1’s EBH domain to engage specific subsets of +TIPs that contain SxIP motifs which include CLASP2 and CLIP170 [68, 69]. The EB1 mimetic motif in the HIV-1 capsid protein lies in its major homology region (MHR)[69]. Interestingly, the MHR domain is oriented inward, away from the cytosolic face of incoming viral particles. However, +TIPs such as CLIP170 bind to both native cores isolated from infected cells and to in vitro assembled capsid-nucleocapsid tubes or capsid like particles (CLPs). Super resolution imaging further showed that unlike other capsid binding co-factors such as cyclophilin A (CypA), which bind hexamers of capsid along the lattice of cores or CLPs, CLIP170 binds to the distal ends of CLPs and has a unique ability to bind and stabilize mutant CLPs that form pentamer rich assemblies [69]. These CLPs often contain breaks at their ends that likely make the MHR accessible. In the context of native cores, the MHR likely becomes accessible upon uncoating. Indeed, many studies suggest that early after entry, HIV-1 cores undergo partial uncoating to accommodate reverse transcription of the viral genome (reviewed in [72, 73]), while live cell imaging approaches show partial uncoating of infectious HIV-1 soon after entry into the cytoplasm [74]. This in turn may expose the MHR which is then bound by CLIP170 and related +TIPs that regulate the metastable state of viral core after entry.

Intriguingly, the importance of CLIP170 in these early events may also shape how the HIV-1 uses adaptors to engage motor proteins. A central subunit of the dynein adaptor complex, dynactin-1 (DCTN1) also acts as +TIP and dual-action regulator of CLIP170 functionality (Figure 3) [75]. DCTN1 both binds to CLIP170 and competes with it for binding to EB1 [76]. Similar to EB1, due to viral mimicry of EB1’s +TIP binding activity, DCTN1 negatively regulates HIV-1 infection by restricting CLIP-170 availability, both through direct competition for binding to the capsid protein EBH-like motif in the MHR of incoming capsids and through sequestration of CLIP170 from viral particles through direct binding to its zinc knuckle domain [77]. Given that these +TIPs and motor adaptors are predominantly localized in the cytoplasm, like several other cellular factors that regulate capsid stability, these findings support the notion that partial uncoating in the cytoplasm is functionally important for early HIV-1 infection. However, it must be noted that some recent studies suggest that uncoating occurs only after entry of viral cores into the nucleus (reviewed in [78]). However, it remains unclear whether these particles are infectious and whether they are truly fully intact, as several of these studies report “near intact” capsids that still align with the concept of partial uncoating before entry into the nucleus. As such, while cytoplasmic versus nuclear uncoating remains controversial in the HIV filed, it would seem most likely that incoming viral cores do at least undergo some level of uncoating in the cytoplasm, or are simply capable of undergoing both processes of either cytoplasmic or nuclear uncoating.

Microtubule-based Motors in HIV-1 Infection

Due to the constraints imposed by their size and macromolecular crowding within the cytosol, HIV-1 particles cannot undergo efficient diffusion within the cytosol [3, 79]. Instead, HIV-1 particles must recruit cytoskeleton-associated motors for directional transport to the nucleus. Several live cell imaging studies have shown that internalized HIV-1 particles undergo directional motion with velocities consistent with both actin and microtubule-based transport, suggesting that HIV-1 engages host motors within the cytosol [51, 80]. Further research has identified these motors as well as the adaptor proteins responsible for their recruitment to incoming HIV-1 particles. Both cytoplasmic dynein and kinesin-1 are required for HIV-1 trafficking, uncoating and nuclear import [81–84]. While the requirement for an anterograde motor for the retrograde trafficking of HIV-1 particles may appear surprising at first, bidirectional transport is a common feature of the transport of viruses and other intracellular cargoes, enhancing processivity and avoiding roadblocks by allowing viruses to reverse course and switch to other microtubule filaments [7, 85]. Interaction between HIV-1 and microtubule-based motors are facilitated by capsid-binding adaptor proteins. Retrograde transport of HIV-1 particles is mediated by the cytoplasmic dynein cargo adaptor bicaudal D2 (BICD2) (Figure 3). BICD2 associates directly with the HIV-1 capsid in in vitro binding assays, and loss of BICD2 results in almost complete abrogation of microtubule-based transport of HIV-1 particles in infected cells [86, 87]. By contrast, the dynactin adaptor subunit, DCTN1 is either not required or is in fact inhibitory to infection [77, 87]. As discussed above, this is due to its secondary role as a +TIP and its negative effects on CLIP170 availability to HIV-1 particles. As such, this may have driven HIV-1 particles to use BICD2 as their key adaptor to engage dynein.

Prior to the discovery of BICD2 as the adaptor for dynein, FEZ1 was identified as the kinesin heavy chain adaptor for HIV-1 (Figure 3). FEZ1 binds to HIV-1 capsid-nucleocapsid assemblies and is required for HIV-1 infection in a variety of cell lines and natural target cell types [83]. Knockdown of FEZ1 results in loss of HIV-1 nuclear accumulation and a shift from retrograde to anterograde transport, suggesting it plays a key role in regulating motor activity balance at the HIV-1 capsid surface [83]. Structural studies have shown that FEZ1 binds HIV-1 capsid hexamers through a series of strong electrostatic interactions between negatively charged glutamate stretches in FEZ1 and a ring of positively charged arginine residues within the central hexamer pore [88]. FEZ1’s role in cargo transport and kinesin-binding activity is also dependent on its phosphorylation at serine 58 by cellular microtubule associated regulatory kinases (MARKs) [89, 90]. Following viral entry, FEZ1 and MARK2 are both recruited to incoming viral capsids, resulting in localized phosphorylation of FEZ1 at the capsid surface, which is required to control kinesin-1-based transport of viral particles to the nucleus [91]. Interestingly, depletion of FEZ1 and BICD2 produces strikingly different transport phenotypes, with FEZ1 depletion resulting in HIV-1 particles undergoing longer anterograde runs toward the periphery, while BICD2 depletion appears to nearly eliminate HIV-1 trafficking [83, 86]. This is in line with our broader understanding of how opposing motors function to control the bi-directional movement of cargos [7, 92].

The issue of HIV-1 microtubule-based trafficking is further complicated by the interaction and co-trafficking of HIV-1 with known microtubule motor substrates. The HIV-1 capsid serves as a binding site for a diverse array of host factors that promote or restrict infection by regulating HIV-1 capsid stability, nuclear import and other viral processes [93, 94]. Several of these host factors have been shown to co-traffic with HIV-1 particles along microtubules. Sec24C, a component of the COPII adaptor complex [95] binds directly to HIV-1 capsids and regulates HIV-1 uncoating, reverse transcription, and nuclear import [96]. mCherry-tagged Sec24C co-traffics with labeled HIV-1 particles undergoing directional motion in the cytoplasm, suggesting that the Sec24C-capsid interaction occurs early in HIV-1 infection [96]. The HIV-1 host factor cleavage and polyadenylation specific factor 6 (CPSF6) is a major regulator of infection in target cell types and is believed to facilitate HIV-1 nuclear import by binding to the viral capsid and recruiting transportin-3 (TNPO3) to facilitate passage of incoming HIV-1 particles through the nuclear pore [97, 98]. While the CPSF6-capsid interaction has been extensively studied, a recent study has suggested that CPSF6 may play a role in HIV-1 trafficking as well. While CPSF6 expression is mostly nuclear, fluorescently tagged CPSF6 forms discrete puncta within the cytosol that traffic along microtubules in the perinuclear region. Fluorescently labeled HIV-1 particles colocalized with cytoplasmic CPSF6 puncta in infected HeLa cells and co-trafficked with CPSF6 and TNPO3 in a microtubule-dependent manner. Furthermore, mutations in the CPSF6 RS domain and modulation of CPSF6-capsid interactions through altered CypA binding altered the trafficking behavior of HIV-1 particles in infected cells, suggesting that CPSF6 may directly regulate HIV-1 microtubule-based transport [99]. The nuclear pore complex component NUP358 is another well-defined HIV-1 capsid interactor implicated in HIV-1 nuclear import, facilitating the docking of HIV-1 particles at nuclear pores and their subsequent passage into the nucleus [84, 97]. Nup358 has been shown to associate with kinesin and dynein motors and serves to facilitate proper nuclear positioning through microtubule-based transport [100, 101]. HIV-1 infection results in the KIF5B (kinesin-1 heavy chain subunit)-dependent relocalization of Nup358 from the nuclear pore to the cytosol, where it binds HIV-1 capsids in a CPSF6-dependent manner and accelerates uncoating [84]. This further implies that uncoating begins in the cytosol and is coupled to microtubule-based transport mechanisms. Taken together, these findings indicate that several HIV-1 capsid interactors regulate the microtubule-based trafficking of HIV-1 particles alongside known HIV-1 motor adaptors and direct microtubule regulatory proteins.

Transfer of HIV-1 particles from the MTOC to the nuclear pore

While incoming HIV-1 particles are known to undergo kinesin-1 and dynein-mediated bidirectional transport to the nucleus, the mechanism by which incoming cores are transferred from microtubules to the nuclear pore remains poorly understood. As discussed above, this involves coupling between kinesin-1 and the nuclear pore protein, Nup358 [84]. But before they reach the nuclear envelope, fluorescently tagged HIV-1 particles appear to accumulate around centrosomes, suggesting that this region may serve as a stop-gap and point at which the virus switches from inward movement to the centrosome to seemingly paradoxical anterograde transport to the nuclear membrane [80] (Figure 3). This likely explains some of the importance of kinesins in early infection, similar to how viruses such as herpes simplex virus type 1 co-opt kinesin 1 functions to control the switch from retrograde transport to the centrosome to the final jump to back out to the nucleus [102]. Several studies in yeast models have also suggested a putative role for the HIV-1 integrase in this transfer step through interactions with centrosomal proteins. A yeast-two hybrid screen using HIV-1 integrase as bait identified several yeast microtubule-associated proteins as integrase interactors, including DYN2p, the yeast dynein light chain, and STU2p, a microtubule-binding component of the spindle pole body (SPB), an MTOC analogous to the centrosome in yeast [103]. Further work using GFP-tagged integrase showed that integrase-GFP puncta colocalized with STU2p at the SPB prior to nuclear import in a microtubule and Dyn2P dependent manner and co-pelleted with the SPB fraction in ultracentrifugation experiments [104]. The HIV-1 accessory protein Vpr, another component of the viral pre-integration complex (PIC), may also promote the transport of viral particles to the nuclear pore [97, 105]. When expressed alone, Vpr localizes to the nuclear membrane and undergoes nuclear translocation in an importin alpha-dependent manner, and Vpr is required for HIV-1 nuclear import in macrophages [105]. In the cytosol, Vpr interacts with the dynein light chain, and a clinical isolate-derived Vpr mutant lacking efficient dynein binding activity shows reduced perinuclear accumulation and nuclear import in COS-7 cells [106]. Interestingly, a role for integrase, PIC components or Vpr in these processes would further suggest that viral cores must have undergone at least some degree of partial uncoating at this stage of infection to allow these proteins or core components to function. Finally, perinuclear actin may serve as a bridging factor between the centrosome and nuclear pores for HIV-1 transport, as live cell microscopy analysis of HIV-1 particle motion shows a transition from fast bidirectional motion to slower movements consistent with actin-based transport as the particle nears the nuclear membrane [51] (Figure 3). While the transition from microtubule-based movement to nuclear docking is one of the most poorly understood aspects of HIV-1 replication, this process may be clinically relevant due to its relationship with pre-integration latency. Studies have shown that HIV-1 particles accumulate at the centrosomes of resting CD4+ T cells, where they can remain for several weeks until replication is resumed following T cell activation [107]. Understanding the molecular basis of this reactivation mechanism may provide new insights into the formation and maintenance of the HIV-1 latent reservoir.

Conclusions and Future Perspectives

The host cytoskeleton plays a crucial role in viral infection, with effects on nearly every aspect of the viral life cycle from entry to egress. As we have discussed here, both the actin and microtubule cytoskeletons are crucial for HIV-1 early infection, promoting receptor attachment and fusion, entry into the cytosol, and motor-driven transport from the cell periphery to the nucleus. However, many outstanding questions remain to be addressed, including the mechanisms by which HIV-1 particles are transferred between actin and microtubule networks, and the key players involved in regulating HIV-1-associated motor activity during microtubule-associated transport. Answering these questions may not only provide new insights into HIV-1 replication and virus-cytoskeletal interactions, but also could provide key insights into cytoskeletal biology in other cellular contexts. Finally, studies of how cytoskeletal regulators control HIV-1 uncoating will likely provide important insights in this somewhat contentious topic. While it is well established that HIV-1 cores are metastable and it has long been thought that HIV-1 undergoes partial uncoating in the cytosol prior to completion of the process in the nucleus, recent studies have suggested that uncoating only occurs in the nucleus. Yet this concept is difficult to reconcile with the broad body of literature describing cytosolic factors, many of which are microtubule regulators, that control HIV-1 uncoating. Indeed, partial uncoating of infectious virus in the cytosol has been directly visualized using live cell imaging approaches. Further studies of how cytosolic MAPs, +TIPs and motors regulate uncoating may help to reconcile these divergent views that currently exist in the field.

Abbreviations:

Abl

Abelson tyrosine-protein kinase

Arp2/3 complex

actin related protein 2/3 complex

BICD2

bicaudal D2

CAP-Gly Domain

cytoskeleton-associated protein glycine rich domain

CA-NC

capsid-nucleocapsid

CCR5

CC motif chemokine receptor 5

CD4

cluster of differentiation 4

CLASP2

cytoplasmic linker associated protein 2

CLIP170

cytoplasmic linker protein CLIP-170

CLP

capsid-like particle

CXCR4

CXC motif chemokine receptor 4

COPII

coat protein complex II

CPSF6

cleavage and polyadenylation specific factor 6

CypA

cyclophilin A

DNAL1

dynein axonemal light chain 1

DCTN1

dynactin-1

Dia1

diaphanous related formin 1

Dia2

diaphanous related formin 2

DYN2p

dynein light chain protein (S. cerevisiae)

EBs

end binding proteins

EBH Domain

end binding homology domain

ERM

ezrin-radixin-moesin Family

FEZ1

fasciculation and elongation factor zeta 1

GFP

green fluorescent protein

GTP

guanosine triphosphate

HIV-1

human immunodeficiency virus type 1

gp120

HIV-1 envelope glycoprotein gp120

gp41

HIV-1 envelope glycoprotein gp41

Kif4

kinesin family member 4

KIF5B

kinesin family member 5B

LIMK1

LIM domain kinase 1

MAPs

microtubule associated proteins

MAP1A

microtubule associated protein 1A

MAP1S

microtubule associated protein 1S

MAP4

microtubule associated protein 4

MARK

microtubule associated regulatory kinase

MARK2

microtubule associated regulatory kinase 2

MHR

major homology region

MTOC

microtubule organizing center

mTOR

mechanistic target of rapamycin

Nef

HIV-1 negative factor

NUP358

nuclear pore complex protein NUP358, RAN binding protein 2

PDZD8

PDZ domain-containing protein 8

PIC

pre-integration complex

RhoA

Ras homolog gene family member A

ROCK

Rho associated protein kinase

RS Domain

CPSF6 arginine/serine-rich domain

SPB

spindle pole body

Sec24C

SEC24 related gene family, member C

SUGT1

suppressor of G2 allele of skp1

+TIPS

plus-end tracking proteins

TNPO3

transportin-3

Vpr

HIV viral protein R

WAVE

WASP-family verprolin-homologous protein