Unlike the prevalent synergistic pathology discussed above, pathogens may form an antagonistic relationship, called ‘interference.’ Interference studies have been conducted using defective interfering (DI) virus particles that cannot replicate as they are missing essential genes (18). DI particles were recognized as having greater clinical importance in recent years when they were found to have interfering potential in vivo (19–21). Early studies also focused on the competition between two replication-competent viruses, especially between mutant and their parental wild-type (wt) viruses (22). As mutants largely maintain the original genetic sequences, these viruses can efficiently coinfect the same cell with wt virus and compete for the host’s mechanisms required for their replication. In general, mutant viruses generated by spontaneous or artificial mutation showed superior growth compared to parental wt virus during coinfection (22). The interference was observed in other RNA viruses including poliovirus (22, 23). These studies suggest that interference occurs at earlier steps of viral RNA synthesis. Interference was also reported in DNA viruses, including the temperature-sensitive herpes simplex virus type 1 (HSV-1) mutant that showed interference against a wt strain by delaying viral genome synthesis (24).

The early pioneering studies contributed to establishing the concept of viral interference. Homologous viral interference can be considered ‘genetic interference’ because proliferation of virulent strains is hindered by introduction of suboptimal mutant genomes. Unfortunately, interference at the genetic level based on an in vitro cell culture system has limitations for human disease studies because it is difficult to assess immune-mediated interference.

Living multicellular organisms provide a niche for growth of both intracellular and extracellular microorganisms. For this reason, heterologous interference is common in humans and animals. Even harmless microorganisms compete with each other for energy sources and sometimes provide ‘colonization resistance’ against pathogenic infection (25, 26). Heterologous interference in living organisms is complicated because they induce counteracting host factors related to innate and adaptive immunity. Despite its complexity, study of heterologous interference is important to understand the pathophysiology of superinfection by unrelated pathogens.

IAV and influenza B virus (IBV) are antigenically distinct but both have significant clinical importance (27). IAV and IBV, like other RNA viruses, frequently yield variant strains that have mutations in antigenic sites. Antibodies against IAV or IBV may cross-react to homologous virus types. For example, IBV hemagglutinin-specific antibodies can provide protection against multiple IBV lineages (28). In contrast, IAV-specific antibodies induced by vaccination are not reactive to heterotypic (i.e., IBV) antigens (29). Thus, antibody-mediated immune response to IAV is not protective against IBV and vice versa. However, attenuated IAV infection provided shot-term protection against IBV in a mouse model (29). Moreover, IAV infection induced protection against IBV in the ferret model and it was achieved within short intervals of less than 7 days (30). Observed immediate protection was associated with decreased lung viral burden of secondarily infected virus. As antigen-specific adaptive immunity takes at least a week to reach the peak level, heterotypic interference should employ other mechanisms, such as innate immunity, to mitigate secondary viral propagation. Another study showed that surviving club cells, a type of lung epithelial cell that secret anti-inflammatory protein (31), are reprogrammed after IAV infection to elicit elevated antiviral responses (32). This antigenically unrelated interference between IAV and IBV was temporary, and animals eventually became susceptible again.

There have been other efforts to prove that IAV infection can provoke non-specific protection against unrelated pathogens. IAV-infected ferrets showed reduced viral burden after secondary respiratory syncytial virus (RSV) infection (33). When 3-, 7-, and 11-day intervals were tested, IAV infection suppressed RSV better during shorter intervals. The heterologous interference was not mediated by adaptive immunity because cross-reactive cellular response was minimal between IAV and RSV. In a mouse model, immediate protection (up to 7~14 days) against RSV was achieved by attenuated infection with IAV. Another study with IAV infection showed interference against the RSV challenge for a longer period (34). In that study, mice were challenged with RSV 5 weeks after IAV infection and showed reduced weight loss and histopathology. As in the ferret model, IAV-infected mice did not develop RSV-specific cellular immunity but transfer of splenocytes from IAV-infected mice resulted in reduced pathology after RSV infection. It was proposed that IAV-specific CD8⁺ T cells are recruited to RSV-infected lungs to exert bystander activity. This type of protection, called ‘heterologous immunity,’ will be discussed later in this review.

Systemic superinfection of lymphocytic choriomeningitis virus (LCMV) and ectromelia virus (ECTV) in mice also generated heterologous interference (35). When mice were infected with LCMV 1 or 2 days prior to ECTV infection, the coinfection extended survival of mice in an IFN-I–dependent manner; however, infection of LCMV after ECTV infection did not interfere with ECTV infection, indicating that timing of LCMV administration affects the degree of interference. In this case, the CD8⁺ T cell count in the spleen was comparable between coinfected and ECTV-only infected mice, suggesting immediate heterologous interference was mediated by innate immunity, such as IFN-I, rather than by long-lasting heterologous immunity.

Chronic infection of murine gammaherpesvirus 68 (MHV68) is an animal model for human gammaherpesviruses. MHV68-infected mice showed significantly lower mortality after lethal IAV infection compared to mock-infected control mice (36). The interference lasted for 60 days after the MHV68 infection but had disappeared by day 120. Adoptive transfer of alveolar macrophages isolated from MHV68-infected mice protected recipient mice from IAV challenge. In addition, MHV68 infection induced protection against another virus, bacteria, and malaria infections (37–39). Broad-spectrum heterologous interference achieved by MHV68 further suggests that protection is mediated by non-specific immune responses rather than by antigen-specific immune responses.

BCG vaccine is an attenuated Mycobacterium bovis strain developed to protect against tuberculosis. BCG is known for its interference potential against viral infections, including ECTV, encephalomyocarditis virus, HSV-1, HSV-2, human papillomavirus, IAV, Japanese encephalitis virus, RSV, and vaccinia virus (41). Epidemiological studies have shown that all-cause mortality, including non-tuberculous infections, is reduced by BCG vaccination in early life (58–60). BCG also has antifungal effects against Candida albicans and Cryptococcus neoformans (61, 62). Most of the antifungal effects were assessed using immune-disrupted SCID or IFN-γ knock-out mice. Thus, further validation is required.

Similarly, interference against influenza infection by attenuated bacteria is achieved by oral and nasal administration of Salmonella LAV (56). As Salmonella LAV has been widely studied as a mucosal vaccine platform that expresses foreign antigen and is usually introduced orally (14), the potential of interference against infections in different mucosal sites need further study. In other attempts for LAV development, live attenuated Bordetella pertussis vaccine showed protection against influenza and RSV infections (51, 52). Of note, most interference induced by bacterial LAVs was against viral infections.

Genetic interference to based on the replication dominance of attenuated virus. For this reason, genetic interference by LAV is observed in interrelated pathogens. In addition to genetic dominance studies using A/Ann Arbor/6/60 ca LAIV, mechanistic analysis of genetic interference was performed with another LAIV backbone strain, X-31 ca (H3N2). When X-31 ca was administered to mice 1 to 4 days before virulent challenge with heterosubtypic A/New Caledonia/20/99 (H1N1), these mice showed reduced weight loss and mortality compared to non-vaccinated control mice (29). Even simultaneous introduction of X-31 ca with virulent virus showed protective effects. As this early protection was achieved without specific antibody responses, which started to increase 5 days after vaccination, interference was analyzed at the genetic level. IAV has eight segmented genomes and coinfection of two viruses enables them to interchange their genes to generate various combinations of reassortant progenies. Attenuated, temperature-sensitive, and ca phenotypes are the consequence of accumulated mutations dispersed in the segmented RNA genome (65, 82). Thus, an increased ratio of reassortant progenies that contain ca-originated gene(s) contributes toward overall attenuation of virulence. The ratio of reassortant virus was 10% to 46% in mice that received both wt and ca viruses at different intervals, indicating active genetic exchanges occur in vivo. Clearly, LAIV can maximize the genetic interference based on the advantage of segmented gene structure.

As shown in interference between virulent IAV and IBV (30), X-31 ca induced protection against heterotypic IBV challenge in mice (29). Basically, genetic interference can be best achieved when genes of two viruses are compatible. Although IBV contains segmented genomes with similarity in terms of protein functions they encode, natural reassortment between influenza A and B viruses has not yet been reported. However, the polymerase complex of IAV and IBV recognize the non-coding terminal sequence of heterotypic genes and the polymerase complex that contains heterotypic subunits are still functional albeit at a low level (83–85). Genetic interference seems clearly to contribute interference between homotypic IAV infections. Although the possibility of heterotypic genetic interference between IAV and IBV were suggested by other studies, it seems other mechanisms, including immediate induction of innate immune response, play a major role.

Innate immunity provides general barrier function because it is designed to respond to a wide range of microbial infections. Therefore, innate immunity induced by certain microorganisms can elicit broad protection against other pathogens. Vaccination with LAIV induces innate immunity similar to natural infection as the vaccine contains every element required for self-replication and actually ‘infects’ the host (86). Although LAIV administration is generally asymptomatic, it elicits sufficient innate immune response (87).

As a prototype LAIV, X-31 ca induced the production of multiple inflammatory cytokines and IFN-I (29, 53). Such inflammatory and IFN responses are initiated via pattern recognition receptors (PRRs). Toll-like receptor (TLR) 3 and 7 are major intracellular sensors for viral RNA genomes and recognize influenza virus infections (88, 89). Mice with defects in TLR signaling were highly susceptible to infection, confirming the crucial role of TLR-mediated innate immunity in protection against IAV (90). When TLR3 and 7 agonists were administered, they elicited immediate protection against IAV challenge as seen with LAIV vaccination (29). Other studies also proved that agonists for TLRs, including TLR2, 3, 4, and 7, induced therapeutic effects on influenza infection (91). When tested for prophylactic effects, TLR agonists protected mice against different strains including H5 avian influenza and 2009 pandemic strains within several days (92–96). Agonists targeting other PRRs, including RIG-I and NOD2, also induced antiviral innate immune responses and protected mice from IAV infection (97, 98). Based on extensive studies that proved the protective innate immune response via PRR stimulation, LAIVs are expected to be sufficiently immunogenic to mitigate subsequent viral infection, even antigenically unrelated pathogens. Indeed, when X-31 ca was administered 2 days before RSV challenge, it significantly suppressed RSV propagation in a TLR3- and TLR7-dependent manner (53).

IFN-I triggers the expression of multiple IFN-stimulated genes (ISGs) that cooperate to inhibit the virus replication cycle in infected cells (99). For example, IFN-induced transmembrane (IFITM) family proteins restrict entry (100), Mx GTPases block early replication of viral genomes (101), and 2’-5’-oligoadenylate synthetase (OAS) family proteins degrade viral genomes by activating the ribonuclease (RNase) L (102). Virus infection is the main stimulator that potentiates cells to produce IFN-I via PRR signaling (88, 103, 104). Likewise, LAIV induces IFN-I in the lung shortly after vaccination (29, 53). Even non-viral stimulation can increase the host’s antiviral status by inducing IFN-I production (105). IFN-I induced by host DNA in a mouse lung fibrosis model attenuated IAV infection (106). It is important to note that ISGs, which can be produced by LAIV and other LAVs, also induce antiviral effects to unrelated viral infections. Furthermore, IFN-I not only provides direct protection to infected cells but also increases the antiviral status of non-infected cells, which together mitigate overall virus propagation in infected hosts.

IFN-γ, another subtype of IFN, is involved in both innate and adaptive immune responses. Vaccination with BCG allows the host to induce prolonged IFN-γ responses that contribute non-specific protection against ECTV and vaccinia virus (40, 49).

New viral infectious diseases, including highly pathogenic avian influenza, 2009 pandemic influenza, SARS, and Middle East respiratory syndrome, have evolved during the last two decades. More recently, COVID-19 spread rapidly and yielded more than 481 million confirmed cases as of 29 March 2022. SARS-CoV-2 vaccines were developed in an exceptionally short time period and applied to human use under provisional approval beginning December 2020. COVID-19 caused significant social and health disruptions worldwide in the early stages of the pandemic and options to minimize the disease burden were discussed.

Because of the well-known heterologous protection of BCG vaccination, its potential use for control of COVID-19 was suggested (119, 120). Multiple clinical studies have assessed the impact of BCG vaccination against COVID-19. These fall into two categories (1): the effect of childhood BCG vaccination on COVID-19 and (2) the effect of re-vaccination on COVID-19 in adults. BCG vaccination in childhood was associated with decreased SARS-CoV-2 prevalence or mortality among health care workers (44, 45); however, in clinical studies, the protective effect of re-vaccination with BCG in adults was controversial. One randomized clinical trial showed that BCG vaccination in the elderly was correlated with lower incidence of SARS-CoV-2 infection (46). Yet other studies found no significant protective effect with BCG vaccination (121–123). Similarly, studies of COVID-19 in animal models also produced contradictory results. In studies using mice expressing the human SARS-CoV-2 receptor, administration of BCG vaccine reduced morbidity and mortality upon SARS-CoV-2 infection (47, 48). But in studies with mice, hamsters, and rhesus macaques, BCG vaccination did not protect animals from SARS-CoV-2 infection (124, 125). The results need to be compared in terms of animals used, study methods and administration routes, and BCG strains. Despite the disparate findings, the positive results support the possible use of BCG vaccine for NSE-mediated control of the current pandemic. In addition, one study reported that peptides derived from BCG vaccine induced SARS-CoV-2-specific T cells based on antigen homology (126). That study showed that heterologous immunity can be induced by BCG vaccination.

OPV may also induce heterologous protection (119). Mothers who were passively exposed to OPV by their children had a decreased incidence of COVID-19 infection (55). This suggests that OPV exposure may be one means of lowering the viral load of unrelated viruses in a population. Another attempt to induce heterologous protection using DI poliovirus particles showed promising prophylactic and therapeutic effects against various RNA virus infections including SARS-CoV-2 (127). When mice received the defective poliovirus genome intranasally, DI particles were generated to induce systemic IFN-I–dependent antiviral responses. This successful example of IFN-I–mediated protection suggests LAV-mediated IFN-I production, which is also induced by LAIV, will similarly protect hosts from SARS-CoV-2 infection.

During the influenza season, there were concerns that people infected with COVID-19 might experience comorbidity. Although coinfection with SARS-CoV-2 and influenza did not affect the overall mortality rate, some studies suggested significant increases in morbidity and mortality (128, 129). In a mouse model, coinfection with SARS-CoV-2 and IAV resulted in increased mortality and severe histopathology (130). However, coinfection with SARS-CoV-2 and LAIV did not enhance the histological pathology in the ferret model (54). In the latter study, prior administration of LAIV 3 days before challenge reduced the replication of SARS-CoV-2 from throat swabs, indicating the potential of LAIV as an immediate prophylactic measure for highly pathogenic coronavirus infection.

BCG vaccine has been used for more than 100 years and administered to billions of children (131). As planned, BCG vaccination significantly lowered the incidence of tuberculosis while simultaneously providing unintended beneficial effects. The COVID-19 outbreak provided an opportunity to study the NSEs of BCG and other vaccines. Although the effectiveness of LAV-induced NSEs against SARS-CoV-2 needs further study, the evidence of BCG and OPV vaccination-induced interference suggests promising potential therapeutic use of LAV in future pandemics, especially during early stages when vaccines and drug therapies have not been developed.

Results of animal and human studies found that NSE duration differs by type of LAV ( Figure 2 ). For example, BCG vaccines led to long-lasting NSEs while LAIVs typically were associated with short-lived effects. Therefore, two distinct types of NSE-based vaccine programs need to be considered. Long-lasting NSEs induced by BGC and OPV vaccination can be considered as passive outcomes. Epidemiological studies showed that effectiveness of vaccination correlates with population-wide coverage (132). Thus, NSE-based protection will be more effective in communities with higher LAV coverage (133). The beneficial effects associated with long-lasting NSEs will be most striking when sufficient vaccines are provided to countries with poor medical infrastructures ( Figure 2A ). Pending the distribution of vaccines, we suggest that LAIVs might be a solution for immediate protection against newly emerging infectious pathogens. However, this approach has been proven mostly at the research level and studies are urgently needed to address safety issues related to NSEs resulting from superinfection of the LAIV and the pathogen. There needs to be an assessment of the risk from superinfection and the expected benefits from NSEs. For example, before the availability of vaccines, most COVID-19 fatalities occurred in the elderly. Assuming this scenario is likely in future pandemics, it seems reasonable to selectively administer LAIV to the elderly in communities in which the infection is actively spreading. Another hurdle for immediate use of an NSE strategy is vaccine supply. To solve this problem, planning might focus on alternative uses of stockpiled pre-pandemic and seasonal LAIV ( Figure 2B ).