Abortive Infection of Animal Cells: What Goes Wrong

Abstract

Even if a virus successfully binds to a cell, defects in any of the downstream steps of the viral life cycle can preclude the production of infectious virus particles. Such “abortive infections” are likely common in nature and can provide fundamental insights into the cell and host tropism of viral pathogens. Research over the past 60 years has revealed an incredible diversity of abortive infections by DNA and RNA viruses in various animal cell types. Here, we discuss the general causes of abortive infections and provide specific examples from the literature to illustrate the range of abortive infections that have been reported. We also discuss how abortive infections can have critical roles in shaping host immune responses and in the development of virus-induced cancers. Finally, we describe how abortive infections can be applied to basic and clinical research, underscoring the importance of understanding these fascinating aspects of virus biology.

Introduction

During a productive infection, a virus must bind and enter the cell, uncoat, undergo gene expression and genome replication, assemble new virions, and release them to infect other cells. A cell that supports productive infection is termed “permissive” (1). The availability of suitable host cell receptors is a key determinant for a virus to successfully infect a cell. However, even if a virus binds effectively to a host cell, downstream steps in its life cycle can still go awry. Non-productive infections that initiate but ultimately fail to release infectious particles are termed “abortive infections” (1). Cell types in which a virus can either not bind or aborts its life cycle after binding are termed “non-permissive” (Figure 1) (1). Although cell types are often categorized as permissive or non-permissive, a mixture of productive and abortive infections may occur within a single cell population. Indeed, single-cell analyses of infected cell populations suggest that abortive infections can be common even within permissive cell cultures, with upwards of 40% cells experiencing an abortive infection (2–5). Thus, the terms permissive and non-permissive may only reflect the overall phenotype of most cells in a population and do not necessarily reflect a situation where every cell is strictly permissive or non-permissive to viral replication.

Figure 1.

Possible infection outcomes. Viral infection of permissive cells can result in productive infections that generate new virus particles (A) or non-productive latent infections, typified by little or no viral gene expression, but can reactivate later to become productive (B). Initiation of infection in non-permissive cells results in abortive infections that fail at one or more steps of the life cycle and do not produce virus particles (C). Cell survival or death may ensue in A-C. However, abnormal integration of oncogenic DNA virus genomes into cellular DNA during abortive infection can result in sustained oncogenic viral gene expression and cellular transformation when these cells are not eliminated by the immune response (D). Figure was created with biorender.com.

Abortive infections are distinct from latent infections wherein a virus has established a non-productive, dormant infection in a permissive cell type but still retains the capacity to reactivate and become productive (1) (Figure 1). In contrast, abortive infections do not have the intrinsic capacity to become productive. However, as discussed below, some abortive infections can be “rescued” and become productive, when additional factors (e.g. a co-infecting virus) immunocompromise the host cell and remove a block to the normally restricted virus. Although the term “abortive infection” has also been used to describe the clearance of subclinical infections before seroconversion or detection of viral antigen through typical diagnostic testing (6), here we use the term to exclusively refer to host cell infections in which the viral life cycle initiates but does not complete.

The development of cell culture systems in the mid-1900s has facilitated the study of virus replication in diverse host cell types, leading to the identification of abortive infections and investigation of their molecular causes. Such information has been fundamental to understanding the cellular and viral factors that determine virus host range. This review aims to explore the causes, diversity, consequences, and potential applications of abortive infections. While primarily focusing on cell culture studies, we also briefly touch upon examples of abortive infections in vivo and their implications for viral disease and cancer formation (Figure 1). Finally, we showcase how abortive infections can have applications for both the lab and the clinic.

General Causes of Abortive Infection

Abortive infections can be complex phenomena influenced by factors intrinsic or extrinsic to the virus-cell system. Here we discuss four general causes of abortive infections: defective viral genomes, deficiencies in host factors/activities required for viral replication, host immune responses, and viral interference. However, additional therapeutic [e.g. antiviral drugs (7)] or environmental [e.g. temperature/fever (8)] factors can also cause abortive infections during clinical viral disease.

Defective Viral Genomes

The packaging of genomes that are incompletely synthesized or encode lethal mutations renders newly assembled virions non-infectious. These genomic defects can be introduced from errors in viral genome synthesis or arise from the upregulation of reactive oxygen species that is often associated with host responses to infection (9, 10). The importance of genome replication fidelity for productive infection is underscored by the potent antiviral activity of the nucleotide analog ribavirin that reduces RNA virus fitness by increasing the error rate of viral genome synthesis (11). Inappropriate integration of viral genomes into cellular DNA can render them incapable of directing proper viral gene expression or becoming packaged into newly assembled virions, resulting in abortive infection. Importantly, as discussed below, these abnormal viral genome integration events are a hallmark of cancers caused by abortive oncogenic virus infections (10).

Deficiencies in Host Factors/Activities Needed to Support Productive Infection

In some cases, cells may be non-permissive to a virus because they lack the necessary surface receptors or co-receptors. However, deficiency of, or inaccessibility to, host factors/activities essential to support virus replication post-binding can also underlie abortive infections. For instance, interaction between influenza virus RNA polymerase PB2 subunits and host ANP32 proteins is critical for proper viral RNA polymerase activity needed for viral replication (12–14). ANP32 proteins differ between mammals and avian species. In mammals, ANP32A and ANP32B can serve essential, but redundant roles in supporting influenza polymerase activity during infection with mammalian-adapted strains that encode a PB2 E627K adaptive substitution that promotes interaction with mammalian ANP32 proteins (15, 16). However, avian influenza species may undergo abortive infection in mammalian cells if they do not acquire adaptive substitutions that permit interaction with mammalian ANP32 proteins (14). In avian cells, only ANP32A proteins can support influenza polymerase activity and depletion of this factor restricts avian influenza virus replication (17). The main difference between avian and mammalian ANP32A proteins lies in a 33 amino acid motif present in avian ANP32A that modulates interactions with PB2 proteins encoded by avian strains (14). Insertion of this motif into human ANP32 proteins can rescue avian strain polymerase activity in human cells (14). This example illustrates how the inability of a virus to co-opt necessary host factors can result in abortive infection and host range restriction.

Even large DNA viruses can be reliant on host activities for their genome replication and productive replication if they cannot supply their own replication-related factors. For example, certain mutant strains of the poxvirus, vaccinia virus (VV), undergo abortive infections in human cell lines that have low levels of cellular ribonucleotide reductase (RR) subunits (18). Mammalian RR enzymes catalyze the rate-limiting step in dNTP synthesis and require interaction between both large (R1) and small (R2) subunits for activity. VV encodes its own R1 and R2 subunits, allowing it to produce sufficient dNTPs during replication in non-cycling cells that have low RR levels (18). However, mutant VV strains encoding catalytically-inactive viral R2 subunits undergo abortive infection in human cell lines with low cellular RR levels and only productively replicate in cells with high RR expression (18), suggesting host RR activity can influence infection outcomes. The narrow host range of poxviruses that lack RR genes, such as molluscum contagiosum virus and parapoxviruses, may partly result from a greater reliance on host RR machinery compared to orthopoxviruses like VV that typically encode their own RR and have broad host ranges (18).

Large-scale screens using RNA interference (RNAi), retroviral mutagenesis, and CRISPR-Cas9 approaches have greatly aided in the rapid identification of host factors essential for productive virus infection at various steps of the viral life cycle and have been reviewed in detail elsewhere (19–22). These screens utilize recombinant viruses expressing fluorescent markers and measure the loss of marker gene expression or prevention of virus-induced cell death as indicators of host gene knockdown/knockout conditions that block viral replication. Identifying host factors required for viral replication may allow for their targeting by “host-directed therapies” that are less susceptible to resistance through viral mutation than classical antiviral drugs that target viral factors (23).

Restrictive Host Immune Responses

Virus-host coevolution generates diverse host antiviral responses and viral countermeasures. The outcome of this evolutionary arms race can determine whether infections will be productive or abortive. The type I IFN response [rev. in (24, 25)] is a critical antiviral defense pathway in mammals. Central to the antiviral function of the IFN response is the induction of IFN-stimulated genes (ISGs) that encode antiviral factors that can directly target various steps of the viral life cycle (24). Here, we provide two specific examples of how host ISGs can cause abortive infection in the absence of effective viral antagonism at different stages of the viral life cycle. However, there are numerous other examples of host restriction factors that can block viral replication reviewed elsewhere (24–26).

Restriction After Entry: TRIM5α & HIV-1

Human immunodeficiency virus-1 (HIV-1) can replicate in human cells but was found to abort in cells derived from closely-related Old World monkeys shortly after entry, but prior to reverse transcription (27). Subsequent studies found this restriction to depend on the E3 ubiquitin ligase and ISG, TRIM5α, which interacts with incoming HIV-1 capsids and forms hexagonal “cages” around them (28–31). This leads to two major outcomes. The first is proteasome- or autophagy-dependent capsid degradation and premature uncoating, which blocks infection by preventing reverse transcription (28). The second outcome is TRIM5α-dependent production of K63-linked polyubiquitin chains that activate IFN signaling through AP-1 and/or NF-kB pathways (28–30). Although the activation of IFN signaling by TRIM5α is not required for HIV-1 restriction in cell culture, it likely has important restrictive effects in vivo (28). The lack of HIV-1 restriction by human TRIM5α may be due to reduced affinity of the human TRIM5α C-terminal SPRY domain for the HIV-1 capsid compared to SPRY domains encoded by Old World monkey TRIM5α proteins (31). Interestingly, human TRIM5α can inhibit HIV-1 nearly as effectively as Old World monkey TRIM5α in human T-cells if human TRIM5α levels are stabilized by fusion to mCherry, suggesting that the more rapid turnover of human TRIM5α may also contribute to human susceptibility to HIV-1 (32). TRIM5α can also target capsid proteins from DNA viruses, including Epstein-Barr virus (EBV) (33) and VV (34), for degradation in a manner dependent on the C-terminal SPRY domain. Thus, characterization of abortive HIV-1 infection in primate cells led to the discovery of TRIM5α as a restriction factor for diverse viruses.

Restriction of Viral Gene Expression: SAMD9/SAMD9L & Poxviruses

Roles for mammalian sterile alpha motif domain containing 9 (SAMD9) proteins in promoting abortive viral infection were unraveled through poxvirus studies involving VV mutants lacking two host range factors, K1 and C7, and myxoma virus (MYXV) strains lacking the host range factor M062. In the absence of VV K1L and C7L genes (encoding K1 and C7, respectively), the VV ΔK1L/ΔC7L strain undergoes an abortive infection in most mammalian cell types at the stage of post-replicative gene expression (35, 36). However, the abortive infection of the ΔK1L/ΔC7L strain in human cells can be rescued by introduction of the MYXV C7L homolog, M062R (encoding M062), into the ΔK1L/ΔC7L strain (37). MYXV naturally infects rabbits and has limited replication ability in primary human cells, but it can replicate in certain human cancer cell lines (38). However, ΔM062R MYXV strains undergo abortive replication in most human cell lines, similar to the restricted ΔK1L/ΔC7L VV strain phenotype (38). Proteomic studies revealed that M062 proteins bind and inhibit human SAMD9, and depletion of SAMD9 through RNAi rescues the replication of ΔM062R MYXV strains and ΔK1L/ΔC7L VV strains in human cells (38). SAMD9 was also identified independently through RNAi screening as a host restriction factor for ΔK1L/ΔC7L strain (39). Interestingly, despite their lack of structural similarity, VV K1 and C7 both bind SAMD9 proteins (40–42) (Figure 2).

Figure 2.

Restriction of VV by human SAMD9 and its antagonism by K1 and C7. VV virion release into the cytoplasm results in expression of early genes (including K1L and C7L), uncoating, viral genome replication, post-replicative (intermediate/late) gene expression, and assembly. SAMD9 is activated by VV infection to bind and cleave phenylalanine tRNA (tRNA^Phe), resulting in a global arrest of mRNA translation that predominantly affects post-replicative viral mRNA translation. However, VV K1 and C7 counter this response by binding and inhibiting SAMD9 to ensure productive replication. SAMD9L, a paralog of SAMD9 may also function similarly and can also be targeted by K1 and C7 (not shown) (40, 48, 49). Cowpox virus CP77 and myxoma virus M062 proteins can also bind and inhibit SAMD9 (38, 44). Figure was created with biorender.com.

SAMD9 also has a lesser-known paralog called SAMD9L (43). Interestingly, while humans express both SAMD9 and SAMD9L, mice only express SAMD9L (40). This led Meng et al. to determine if mouse SAMD9L could inhibit ΔK1L/ΔC7L strain replication (40). The ΔK1L/ΔC7L strain was restricted in mouse cells and animals, but this restriction was relieved in SAMD9L-deficient cells or animals (40). SAMD9 and SAMD9L genes are both ISGs, with the latter being more highly induced by IFN treatment (40). Knockout of SAMD9 alone was sufficient to abolish restriction in human cell types, but in the presence of IFN treatment, disruption of both SAMD9 and SAMD9L expression was necessary to rescue ΔK1L/ΔC7L strain replication (40). These observations suggest the basal level of SAMD9 in human cells is sufficient for restricting ΔK1L/ΔC7L strains but SAMD9L must be induced by IFN to have a similar antiviral effect. Importantly, this study also revealed VV K1 and C7 to both bind human SAMD9 and SAMD9L (40).

A third poxvirus protein, cowpox virus CP77, was also recently shown to bind to SAMD9L proteins in Chinese Hamster Ovary (CHO) cells (44). This finding began to explain why the abortive infection of CHO cells by wild-type VV is rescued by CP77 expression (45, 46). Zhang et al. showed SAMD9L knockout to relieve VV restriction in CHO cells and that CP77, but not VV K1 or C7, could interact with SAMD9L encoded by these cells (44). Given these observations, combined with the significant divergence of SAMD9/SAMD9L sequences among mammals (43, 47), the ability of poxvirus antagonists to effectively target species-specific SAMD9/SAMD9L factors may greatly influence their host range (43).

Recent studies suggest that SAMD9 may inhibit both poxviral and cellular mRNA translation through an unusual mechanism involving the degradation of specific tRNAs (48) (Figure 2). Zhang et al. showed that human SAMD9 encodes endoribonuclease activity that is latent in uninfected cells but activated upon VV infection (48). Strikingly, SAMD9 uses a double-stranded nucleic acid-binding domain to specifically bind phenylalanine tRNA (tRNA^Phe) and cleave it (48, 49). Both the SAMD9 nucleic acid-binding domain and its endoribonuclease activity are required for ΔK1L/ΔC7L strain restriction, and overexpression of tRNA^Phe can rescue both global protein synthesis and viral replication (48, 49), suggesting that SAMD9 blocks viral and cellular mRNA translation during infection with ΔK1L/ΔC7L strains by cleavage of tRNA^Phe pools (Figure 2). How SAMD9 is activated after infection, and whether SAMD9L functions similarly, is still unclear. Interestingly, SAMD9 has also been reported to potentiate proinflammatory responses triggered by the DNA sensor cGAS during abortive ΔM062R MYXV infection of human macrophages, although it is not yet clear if these SAMD9-mediated responses contribute to viral restriction (50). These studies highlight how investigating the host factors responsible for the replication deficits of certain poxvirus strains can uncover both host range restriction factors (SAMD9/SAMD9L) and their corresponding viral antagonists (K1, C7, M062).

Whether infections will abort due to host immune responses may depend on several factors, including the timing, intensity, and type of immune response in a specific cell type, the ability of a virus to counteract the host response, and the inoculum size (51–53). For instance, virus replication in normally permissive cells can result in abortive infection if cells are treated with recombinant IFN before the virus has a chance to express IFN antagonists (25, 54, 55). Indeed, the induction of an antiviral state in uninfected cells through paracrine IFN signaling from infected cells plays a crucial role in blocking viral spread in vivo (56). Therefore, the occurrence of abortive infections is highly dependent on the context of host cell responses.

Viral Interference

During virus co-infection one virus may block the replication of the other. Such “viral interference” can occur between viruses within the same family (homologous interference) or in different families (heterologous interference) (57–59). Viral interference can result from one virus outcompeting the other for essential resources needed for replication. For example, respiratory syncytial virus replication is blocked in canine cells during co-infection with influenza virus due to competition for resources required for viral protein synthesis and budding (60). Viral interference can also occur when one virus strongly stimulates antiviral (e.g. IFN) responses that block the replication of the other (57, 58). This has been widely observed among mammalian respiratory viruses and in flavivirus co-infection of insect cells (57, 58). For example, IFN responses induced by human rhinovirus (HRV) causes abortive coronavirus replication in co-infections of primary human airway epithelial cells (61, 62). HRV-induced IFN responses can also block influenza virus replication in primary human airway epithelial cells (63), and it has been suggested that infections with common cold viruses such as HRV may protect against more severe infections caused by heterologous respiratory viruses (61).

Interestingly, even the production of virions encoding defective viral genomes can also interfere with the replication of wild-type viruses when both defective particles and normal particles infect the same cell. This can also occur due to competition for replication resources or stimulation of restrictive host responses (64, 65). Such “defective-interfering particles” have been observed during infection with diverse viruses and their impact on viral replication and host immune responses has been recently reviewed elsewhere (64, 65).

Diversity of Abortive Infections

There have been many reports of abortive infections by various viruses over the past 60 years (Table 1). However, the defective step in the viral life cycle and the underlying cause of the defect remain unclear in most cases. In this section, we highlight examples from baculoviruses, poxviruses, and influenza viruses, where the defective step in the viral life cycle is known, to demonstrate the diversity of reported abortive infections.

Table 1.

Selected examples of reported abortive virus infections.

Virus	Cell typea	Putative defective step of viral life cycle	References
Bombyx mori nucleopolyhedrovirus	High Five cells	Endosome escape/trafficking	(69)
Molluscum contagiosum virus	Human keratinocytes, fibroblasts and HeLa cells	Uncoating	(150)
Vesicular stomatitis virus	Spongy moth LD652 cells	Gene expression	(140)
Herpes simplex virus-1	Mouse peritoneal macrophages	Early gene expression or Genome synthesis	(151)
Influenza virus	Primary human CD8+ T cells	Gene expression (abnormal splicing)	(152)
Vesicular stomatitis virus	Rabbit cornea (RC-60) cells	Gene expression (viral mRNA translation)	(153)
Vesicular stomatitis virus	Mouse L cells	Gene expression (viral mRNA translation)	(54)
Human immunodeficiency virus-1	Resting human CD4+ T-cells	Reverse transcription	(85, 154)
Human Cytomegalovirus	Rabbit RK13 cells	Genome replication or late gene expression	(155)
Vaccinia virus	Drosophila S2 cells	Genome replication	(75, 76)
Influenza virus	Mouse L929 cells	Genome replication	(156)
Amsacta moorei entomopoxvirus	Monkey CV-1 cells	Genome replication	(157)
Hirame novirhabdovirus	Flounder B cells	Genome replication	(158)
Vaccinia virus	Chinese hamster ovary (CHO) cells	Intermediate gene expression	(45, 46)
Vaccinia virus	Primary human B cells	Late gene expression	(92)
Middle East respiratory syndrome coronavirus	Human plasmacytoid dendritic cells	Post-gene expression	(90)
Respiratory syncytial virus	Mouse alveolar macrophages	Assembly or budding	(159)
Modified vaccinia virus Ankara	Human A549 cells	Assembly	(160)
Vaccinia virus	Spongy moth LD652 cells	Assembly	(78)
Influenza virus	Mouse bone marrow-derived dendritic cells	Assembly	(96)

^aSome cell types tested in referenced publications are not listed due to space limitations.

Defects Prior to Genome Replication

After binding to the host cell, viruses must enter and traffic to appropriate intracellular locations to uncoat to facilitate genome replication. Premature uncoating or improper trafficking can abort infection (66, 67). Intracellular trafficking is crucial for large DNA viruses like baculoviruses, which require import into the nucleus for replication. Baculoviruses exhibit preferences for specific insect orders and usually target a particular species or a limited set of species within the same order (68). For example, Bombyx mori nucleopolyhedrovirus (BmNPV) can productively infect B. mori-derived BmN cells but aborts in cells from other lepidopteran species, such as Trichoplusia ni-derived High Five cells due to defective nuclear import (69). However, co-infection with Autographa californica multiple nucleopolyhedrovirus (AcMNPV), a closely related baculovirus, can rescue abortive BmNPV infection of High Five cells (69). This rescue is facilitated by the AcMNPV GP64 fusion protein, which promotes BmNPV escape from endosomes and restores BmNPV nuclear import (69). Other abortive baculovirus infections can be rescued by co-infection or by introducing genes from replication-competent strains into defective strains, making baculoviruses excellent models for examining virus host range mechanisms (68).

Defects in Gene Expression and Genome Replication

AcMNPV is widely used for recombinant protein expression in insect cells and has been adapted for transduction of mammalian cells for gene therapy, vaccine, and recombinant protein expression applications (70). AcMNPV virions efficiently enter mammalian cells and traffic to the nucleus, but the infection aborts due to limited viral gene expression, likely owing to poor recognition of viral promoters by host transcription machinery (71). However, incorporation of mammalian promoters into the AcMNPV genome enables robust expression of foreign genes, making them valuable tools for mammalian cell transduction (70).

Accumulating evidence suggests that influenza virus interactions with host splicing machinery can greatly affect viral replication, leading to abortive infection and host range restrictions (26, 72–74). For example, Bogdanow et al. found a cis-regulatory element in avian influenza virus M1 pre-mRNA that facilitates splicing during infection of human cells, but is absent in human-adapted strains (72). Splicing of M1 pre-mRNA in avian strains leads to decreased production of M1 protein, which is responsible for exporting viral ribonucleoprotein complexes from the nucleus, and is required for productive infection (72). Therefore, the limited replication of avian strains in human cells may not only involve incompatible PB2-ANP32 protein interactions mentioned earlier, but may also result from defective gene expression caused by improper splicing of viral pre-mRNA.

Despite being a mammalian poxvirus, VV can enter a wide range of vertebrate and invertebrate cells in culture. However, in invertebrate cells, VV replication is usually abortive. For example, VV can enter Drosophila S2 cells and express all early genes, yet it fails to replicate its genome (75, 76). It is unclear if this is due to defective uncoating or a specific block to viral DNA synthesis. Interestingly, when plasmid DNA is transfected into VV-infected mammalian cells, it undergoes replication by VV proteins (77). However, VV fails to replicate transfected plasmids in S2 cells, suggesting there is a general block to DNA replication by VV machinery, although additional uncoating defects cannot be ruled out (76).

Defects in Virion Assembly

Defects in late stages of infection, such as virion assembly, can also abort infection. This is evident in VV infection of LD652 insect cells derived from Lymantria dispar. Unlike in Drosophila cells, VV undergoes all stages of gene expression and replicates its genome but still fails to produce infectious particles in LD652 cells (78). Electron microscopy and pulse-labeling studies suggest that this abortive infection is due to defective virion assembly caused by a lack of necessary host protease activity required for maturation of VV structural proteins (78). Interestingly, VV also aborts shortly after late gene expression in Xenopus melanophores (79), suggesting that non-mammalian hosts either lack factors needed for VV assembly or the virus cannot overcome restrictive antiviral responses in these unnatural hosts.

Abortive virion assembly has also been observed in certain influenza strains during macrophage infection. Highly pathogenic avian influenza strains can productively infect murine macrophages, while seasonal influenza A viruses undergo abortive virion assembly (80). Similar findings have been reported for seasonal and pathogenic avian strains in human alveolar macrophages (81) and may contribute to differences in strain virulence (82, 83). In a recent study, defective influenza virus assembly in primary monocyte-derived human macrophages was attributed to the failure of viral hemagglutinin and M2 proteins to interact at the plasma membrane surface (84). Interestingly, treatment with the actin polymerization inhibitor, cytochalasin D, partially restored virus release, suggesting cytoskeleton dynamics can regulate virion assembly (84). Since late-stage abortive infections still produce viral proteins and undergo genome replication, they can have important effects on host immune responses (discussed below).

Host Immune Response Modulation by Abortive Infections

Even in permissive hosts, viruses can undergo both productive and abortive infections in different cells and tissues. Abortive infections, though unable to produce infectious viruses, can still affect the host immune response and influence viral disease outcomes. One particularly striking example is the effect of abortive HIV-1 infections on CD4+ T-cell death in lymphoid tissue. While activated human CD4+ T-cells are permissive to productive HIV-1 infection, resting CD4+ T-cells are non-permissive (85). However, the massive CD4+ T-cell death observed during untreated HIV-1 disease cannot be accounted for by the relatively small population of activated CD4+ T-cells, suggesting that resting CD4+ T-cells undergo some form of “bystander” cell death. Seminal work by Doitsh et al. and Monroe et al. subsequently found these resting CD4+ T-cells likely die as a result of abortive HIV-1 infections that produce incomplete reverse transcription products that are recognized by the DNA sensor IFI16 that, in turn, activates inflammasome-dependent pyroptosis (85, 86). Interestingly, the severe lymphopenia associated with Ebola virus infection (87) may also be caused by abortive viral transcription in T-cells that triggers necrotic and apoptotic cell death (88). Abortive infection of other immune cells such as B cells, monocytes, macrophages, and dendritic cells (DCs) have been widely reported for many viruses and can have contrasting effects on immune cell function (89–93). For example, VV aborts infection in classical human DCs shortly after early viral gene expression but this infection still negatively impacts the expression of DC costimulatory molecules and cytokines and impairs their ability to activate T-cells (94, 95). In contrast, seasonal influenza virus strains undergo abortive virion assembly in bone marrow-derived murine DCs, yet these infected DCs still generate robust cytokine responses and activate T-cells (96, 97). Thus, despite incomplete viral replication, abortive infections can have both negative and positive effects on immune cell function.

Abortive infections in non-immune cell types can also influence immune responses in vivo. Astrocytes are the most abundant cell type in the central nervous system where they function in synaptogenesis and maintain the blood-brain barrier (98). However, abortively-infected astrocytes appear to be the major IFN-β-producing cell type in murine brains during infection with a variety of neurotropic viruses (99, 100). Astrocytes have high basal levels of ISGs, possibly explaining why so many viruses undergo abortive infections in these cells (101). In contrast to wild isolates, lab-adapted strains of rabies virus (RABV) undergo abortive infections in astrocytes, triggering strong IFN-β and proinflammatory cytokine responses that enhances blood-brain barrier permeability, immune cell and antibody infiltration into the CNS, and viral clearance (102). Thus, despite normal replication of both lab-adapted and wild isolates of RABV in neurons, lab-adapted strain infections can ultimately be cleared due to the robust immune responses initiated from abortively-infected astrocytes (102). However, it should be noted that abortive infections can also drive pathological inflammatory states or “cytokine storms” which may be detrimental to the host (89, 103, 104).

Abortive Infections and Cancer

Approximately 15–20% of human cancers globally are caused by viruses, with seven tumor or oncogenic viruses afflicting humans identified thus far (105). Most virus-associated cancers arise from latent or abortive infections that persist due to an ineffective host immune response. Consequently, these cancers are primarily observed in immunocompromised individuals (105, 106). In these non-productive infections, viral oncogene expression, inactivation of host tumor suppressors by viral antagonists, and/or dysregulation of cellular gene expression from viral DNA integration can lead to cellular transformation over time. Two oncogenic viruses that cause cancer through abortive infections in humans are the small DNA viruses human papillomavirus (HPV) and Merkel cell polyomavirus (MCPyV).

Infection with high-risk HPV strains (e.g. HPV-16, HPV-18) is associated with increased incidence of cervical, anal, and oropharyngeal cancers (107). During productive infection, the circular HPV DNA genome exists in a non-integrated, episomal state. However, in HPV-associated cancers, the viral DNA integrates into the cellular genome, preventing productive infection (108). Reactive oxygen species that arise from infection-associated inflammatory responses may facilitate integration by generating double-stranded DNA breaks (109). HPV integration can activate nearby proto-oncogenes or inactivate tumor suppressor genes, contributing to oncogenesis (110, 111). However, the key requirement for cellular transformation is elevated expression of two viral oncoproteins, E6 and E7, in infected, dividing cells (107). In 75–85% of cases, abnormal E6 and E7 expression occurs due to the integration of HPV DNA in a manner that deletes parts of the viral genome but retains the E6 and E7 genes and their P97 promoter (107). Elevated E6 and E7 expression can also be supported by cellular enhancer sequences near HPV integration sties (112). Integration events typically disrupt expression of HPV E2, a viral transcription factor that normally represses E6 and E7 expression during productive infection, further contributing to their unusually elevated expression (108, 111).

Together, E6 and E7 promote oncogenesis by stimulating cell proliferation and inducing chromosomal instability through a variety of mechanisms. For instance, E6 targets the human tumor suppressor p53 for degradation, while E7 targets Rb, preventing them from blocking cell proliferation and inducing cell death (107, 108). Inhibition of p53 function in DNA damage responses can also increase genomic instability (107, 108). In a normal infection, E6 and E7 drive the infected cell into S-phase, producing host factors such as dNTPs and DNA polymerases, that HPV utilizes for its replication cycle (107, 113). Thus, HPV does not encode these factors to promote cellular transformation but rather normally uses E6/E7 to create an intracellular environment that favors productive replication. However, in abortively-infected cells that are not cleared by the immune response, sustained E6/E7 expression can sensitize cells to transformation over time. Indeed, E6 and E7, while required to maintain transformed cell phenotypes, are not sufficient on their own to transform normal cells (107), and thus additional host genetic and environmental factors are needed for cellular transformation.

MCPyV is linked to the development of Merkel cell carcinoma (MCC), a rare and aggressive form of skin cancer usually found in elderly and immunocompromised individuals (114). MCC cells express markers of specialized neuroendocrine skin cells called “Merkel” cells, but it’s uncertain whether MCC originates directly from transformed Merkel cells or other skin cells that express these markers after transformation (106, 114). Similar to HPV, the MCPyV genome usually exists as an episome in infected cells but can integrate into the cellular genome in some cases, preventing productive infection. Around 80% of MCC cases involve cells with integrated MCPyV genomes, suggesting that most arise from MCPyV infection (106, 115). As with HPV, MCPyV integration events typically observed in transformed cells retain sequences encoding viral oncogenes while other regions of the viral genome are lost (106, 114, 116). MCPyV integration events in MCC cells typically retain the early promoter that drives expression of a single viral transcript that is alternatively spliced to produce various T antigens, including small T (sT) and large T (LT) antigens (10, 106).

MCPyV sT is the major driver of MCPyV-mediated oncogenesis, as it can transform cells in culture and is essential for the survival of MCC cells (106, 117). During productive MCPyV infection, sT stabilizes LT protein levels, inhibits the tumor suppressor phosphatase PP2A (118), and antagonizes 4E-BP1, a negative regulator of cap-dependent translation (119). The C-terminal sequences of sT, which keep 4E-BP1 in an inactive, hyperphosphorylated state are necessary for cellular transformation (119), suggesting that dysregulation of cellular translation is critical in MCPyV-driven MCC.

In productive MCPyV replication, the LT C-terminal helicase domain mediates viral genome replication and activates p53-mediated DNA damage response (DDR) to promote recruitment of DDR host factors that MCPyV co-opts for viral genome replication (106). However, MCPyV genomes integrated into MCC tumors typically lose the helicase domain, resulting in truncated LT proteins that cannot activate p53 responses (10, 120), which would be incongruent with oncogenesis. Despite this, the N-terminal sequences of LT that interact with Rb remain in transformed cells, compromising tumor suppressor function (120, 121). Though LT cannot transform cells in culture, its Rb-binding domain is required for MCC cell survival (122), suggesting it plays a secondary role in MCC pathogenesis. Thus, as with HPV, the unchecked expression of MCPyV oncoproteins inhibits tumor suppressor function and promotes oncogenesis. Abortive infections by other polyomaviruses, like simian virus 40, can also promote tumor formation in animals due to abnormal integration events that retain viral oncoprotein expression (123). Viral gene expression during abortive and latent infections by oncogenic herpesviruses can also promote cellular transformation (124). Collectively, these observations illustrate how abnormal integration events and/or viral gene expression in non-productively infected cells can drive transformation if the immune system fails to eliminate infected cells.

Applications of Abortive Infections

Identifying and characterizing abortive infections can provide invaluable tools applicable to both basic research and the treatment of viral disease. Here we discuss selected examples of how natural or experimentally-contrived abortive infections have been employed in various areas of basic and clinical research.

Abortive infections by temperature-sensitive mutants

The isolation of temperature-sensitive (ts) virus strains has been important for understanding viral protein function and identifying promising antiviral drug targets (125–128). These strains have specific mutations in essential genes that cause the production of non-functional or unstable viral proteins at higher (non-permissive) temperatures, while they function properly at lower (permissive) temperatures (128, 129). By incubating infected cultures at non-permissive temperatures, ts mutants can be used to study the consequences of inactivating a specific protein function/step in the viral life cycle on infection outcomes (128, 130). For example, the identification of VV strains with ts mutations in the B1R gene revealed the essential role of the B1 viral kinase in promoting VV replication by inactivating the host protein barrier to autointegration factor (BAF) (131, 132). BAF was originally identified as a pro-viral factor that prevented “suicidal” autointegration of retroviral DNA, but has since been shown to function in nuclear reassembly, cellular gene expression, and DNA damage responses in the nucleus of uninfected cells (133). However, during VV infection, BAF localizes to the cytoplasm and, in the absence of functional VV B1, binds viral DNA to prevent viral genome replication and intermediate gene expression (132, 134). However, functional VV B1 can overcome this block to replication by promoting BAF hyperphosphorylation, which inhibits its binding to viral DNA (135).

The characterization of abortive infections by ts strains has also provided valuable tools for cell biology. For example, VSV ts strains that fail to assemble at non-permissive temperatures have shown the importance of VSV glycoprotein trafficking to the plasma membrane for virion assembly (136, 137). These glycoprotein mutants are used as tools for studying intracellular trafficking pathways as they get stuck in the ER at non-permissive temperatures (137, 138). These examples illustrate how understanding the molecular reasons behind abortive ts strain infections can uncover essential aspects of virus-host interactions and provide tools for investigating cellular processes.

Virus-Host Interaction Screens

Another application of abortive infections is their use in screens for viral immune evasion proteins and host restriction factors (139). These screens use recombinant viruses with reporter genes (e.g. GFP, luciferase) that undergo naturally abortive infections in a specific cell type. If marker gene expression becomes detectable, it indicates the “rescue” of these abortive infections. Conversely, if an abortive infection occurs at a late stage and viral gene expression is not defective, rescue is indicated by increased spread of the rescued virus under low multiplicity of infection conditions. This strategy has the advantage of low signal-to-noise ratios due to initially abortive infections.

For example, the abortive VSV infection of LD652 moth cells has created a screening platform to discover virally-encoded immunomodulators and eukaryotic host factors that restrict arbovirus replication (139). VSV productively infects dipteran insects (flies and mosquitoes) but not cells derived from lepidopterans (moths and butterflies) (140). In LD652 cells, VSV infection aborts after limited viral gene expression (140). However, inhibiting host transcription rescues VSV replication, suggesting antiviral responses in LD652 cells block VSV replication (140). This restriction enables screening for conditions that rescue VSV strains encoding reporter genes by co-infecting cells with immunosuppressive viruses or expressing candidate viral immune evasion factors (Figure 3). This approach identified poxvirus A51R proteins as novel immune evasion factors that rescue VSV replication in LD652 cells during either VV co-infection or when expressed from plasmids (139, 140). Once virus-encoded factors that rescue VSV replication are identified, they can be used as molecular bait to identify cellular antiviral factors they target. Depletion of host factors by RNAi- or CRISPR-Cas9-based techniques can also identify immunity factors restricting arbovirus replication in LD652 cells, and has identified roles for RNAi-, NF-κB-, and proteasome-related pathways in arbovirus restriction (139, 140) (Figure 3).

Figure 3.

Using abortive VSV infections in LD652 cells in virus-host interaction screens. Abortive infection of VSV strains encoding GFP (VSV-GFP) can be used to screen for immunosuppressive co-infecting viruses or immune evasion factors expressed from plasmids that rescue GFP expression. Host antiviral factors restricting VSV replication can also be identified by transecting dsRNAs (for RNAi) targeting host transcripts. Figure was created with biorender.com.

Instead of using naturally abortive infections by wild-type strains, host range-restricted mutant viruses can be employed. For example, the restricted host range of modified VV Ankara (MVA) is due to repeated passage of the parental chorioallantois VV Ankara (CVA) strain in chicken embryo fibroblasts that resulted in genetic changes that cause MVA to undergo abortive virion assembly in human cells (141). To identify human factors contributing to MVA restriction, Peng et al. conducted a genome-wide RNAi screen using A549 cells infected with MVA encoding GFP and looked for conditions that allowed viral spread in cell monolayers (141). This screen identified human zinc finger antiviral protein (ZAP) as a key restriction factor as its knockdown rescued MVA assembly and productive replication (141). Interestingly, the CVA-encoded C16 protein normally inactivates ZAP by sequestering it to specific sites in the cytosol (141). However, genes encoding C16 are disrupted in MVA, rendering it susceptible to ZAP restriction (141). This is just one example of how combining host range-restricted mutant viruses with non-permissive cell types has been fruitful in identifying host factors restricting viral replication (39, 41, 142–144).

Clinical Applications: Vaccines

Abortive infections that still produce significant viral gene expression can be an attractive method of vaccination because they can still invoke strong innate and adaptive responses while having improved safety profiles over replication-competent vaccine strains. Indeed, ts strains of respiratory viruses have made effective, yet safe, vaccines given they can replicate in the lower temperatures of the upper respiratory tract (32–34°C) and stimulate immune responses but cannot replicate at the higher temperatures (37°C) of the lower respiratory tract (145). Additionally, replication-defective MVA strains are safer vaccine alternatives to smallpox vaccines based on replication-competent VV strains and are now also licensed for mpox prophylaxis (146). Recombinant MVA strains expressing antigens from other pathogens are also being studied as potential vaccines, given their robust gene expression and amenability to engineering (147). The use of assembly-defective strains as vaccines is also being investigated with RNA viruses (148, 149). Shan et al. generated a dengue virus (DENV) strain encoding an NS2A mutant virus defective in virion assembly by supplying wild-type NS2A protein in trans in a helper cell (149). The resulting DENV mutant can initiate a single round of infection but ultimately aborts during assembly (149). Even a single dose of this DENV mutant could protect mice from wild-type DENV infection, suggesting that this may be an effective strategy for the development of vaccines for other RNA viruses (149).

Concluding Remarks

Abortive infections can have various underlying causes including inactivating viral mutations, deficiencies in pro-viral host factors, ineffective viral antagonism of host antiviral responses, and interference from co-infecting viruses. Although these infections do not produce infectious particles, the virus keeps trying to replicate itself in the host cell. In these persistent attempts, viral gene expression can have immunosuppressive or immunostimulatory effects on the host cell, and in rare cases, contribute to cellular transformation. Research on abortive infections will need to move beyond cell culture studies to fully appreciate the prevalence and consequences of abortively-infected cells during viral disease in animals. Additionally, identifying the molecular defects underlying abortive infections could help identify targets for antiviral drugs. Abortive infections will undoubtedly continue to provide valuable tools for identifying critical virus-host interactions, recombinant protein expression, gene therapy, and vaccine development.