The possibility that an immune response to pathogens may have also deleterious effects on the host homeostasis has been the focus of several studies. For example, the hyperinduction of cytokines following avian influenza virus infection has been implicated in the severity of the disease (47
), and infection of cells by antibody-dependent enhancement has been known to occur for several viral diseases (59
). We demonstrate here that anti-Spike antibody potentiates infection of immune cells by SARS Spike-pseudotyped lentiviral particles and replication-competent SARS-coronavirus. Antibody-mediated infection is dependent on one subtype of Fcγ receptor, FcγRII (CD32). However, ADE mediated by FcγRII does not utilize the endosomal/lysosomal pathway utilized by ACE2-bound virus, i.e., the common cellular receptor for SARS-CoV in susceptible cells. Finally, we have found evidence that this ADE phenomenon is dependent on the immunization strategy used for candidate vaccines.
We and others (14
) reasoned that SARS-CoV Spike glycoprotein, which has been shown to be responsible for viral attachment and entry, was a prime target for the development of SARS vaccines. Because of earlier reports of immune-mediated enhancement of coronavirus infection both in vitro
) and in vivo
), we investigated this possibility and reported preliminary data that a neutralizing, anti-Spike immune-serum induced ADE of SARS-CoVpp (30
). The effect of SARS-CoV antiviral antibodies on ADE has been studied in susceptible cells, but conflicting data have appeared. One group demonstrated antibody-dependent enhancement of lentiviral particles pseudotyped with SARS-CoV Spike variants detected in palm civets when incubated with rodent (vaccine-elicited) and convalescent SARS patient-derived monoclonal antibodies (73
). In contrast, another group, using panels of isogenic SARS-CoV, did not find any evidence of infection enhancing potential with a series of monoclonal antibodies (50
Our experiments conclusively demonstrate with several lines of evidence that both SARS-CoVpp and replication-competent SARS-coronavirus infect certain immune cells only in the presence of anti-Spike immune serum and not in its absence. Furthermore, we found that homologous anti-Spike immune-serum never potentiated infection of the permissive epithelial cell line VeroE6.
In cells infected via the ADE pathway we detected both SARS-CoV genomic and subgenomic material, as well as the expression of two viral proteins, N and M. It is believed that an antibody-mediated infection pathway could provide SARS-CoV additional entry routes, allowing the virus to broaden its tropism. Interestingly, early after ADE infection, we detected far more (ca. 50- to 100-fold) SARS-CoV-positive RNAs compared to its negative strand, a situation found during the productive replication of coronaviruses (55
) and distinct from the abortive replication of SARS-CoV reported after direct infection of macrophages (7
). Although we unambiguously illustrated ongoing infection thanks to the detection of de novo
synthesis of the structural viral proteins N and M, the ADE-infected Raji cells did not support productive replication of SARS-CoV. After the initiation of viral gene transcription and viral protein synthesis, a block appears to occur in the replication process, ultimately ending in an abortive viral cycle without the detectable release of progeny virus (12
). Whether ADE occurs in other immune cells and whether such infection will be abortive remains to be investigated.
Despite not leading to productive infection, it remains of interest to investigate the functionality and cell fate of ADE infected cells. Thus, ADE infection induced the initiation of viral gene transcription—with production of viral gene intermediate species—and viral protein synthesis. Given the ability of intracellular innate immune sensors, such as the pattern recognition receptor families TLR and RLH, to detect viral gene species (2
) and the disturbance to the cell homeostasis caused by SARS-CoV viral proteins (such as 3a protein and SARS-unique domain) (5
), a possible participation of immune-mediated enhanced disease during SARS pathogenesis cannot be ruled out. Of note, clinical observations have already reported poor disease outcomes in early seroconverted SARS patients (22
The observation of a change of tropism of SARS-CoV in the presence of antiviral immune serum distinguishes the ADE of SARS-CoV infection from many other examples of antibody-mediated viral infection, where the presence of antiviral antibodies increases the viral load due to infection of a higher number of already-susceptible cells (59
). Of note, a shift of tropism from epithelial to immune cells, albeit mediated by a different mechanism, was associated with an increased severity of disease during the course of lethal coronavirus-induced feline infectious peritonitis (52
FcγR blocking experiments clearly demonstrate the involvement of FcγRII in the occurrence of ADE infection of Raji and Daudi cells. Since both anti-FcγRI and anti-FcγRII treatments, however, had an impact on the occurrence of ADE infection of THP-1 cells, no firm conclusion could be drawn as for the possible contribution of FcγRI. Due to the coexpression of the two FcγR in this cell line (cf. B, where ~100% of the cells expressed FcγRI and FcγRII), blocking antibodies could indeed bind not only to the targeted FcγR via their Fab portions but also to other FcγR via their Fc portions. Thus, for example, mouse IgG1 anti-FcγRI antibody could bind specifically to FcγRI via its Fab portions and to FcγRII via its Fc portion, making both FcγR unavailable for interacting with opsonized pseudoparticles.
On the basis of these results, we decided to undertake a systematic analysis of the individual role of the human Fcγ receptors subfamilies. Our experiments demonstrate that only FcγRIIA and, to a lesser extent, FcγRIIB1 triggered infection by SARS-CoVpp in the presence of anti-Spike serum. These results extend our observations with FcγR-blocking antibodies in Raji cells (30
) because only FcγRII, but not FcγRI or FcγRIII, is found in that cell line. However, in light of the IgG composition of the anti-Spike immune serum used in the present study and of the inability of FcγRI and FcγRIII to bind mouse IgG1 (10
), further investigations are required to formally rule out the involvement of these receptors in triggering ADE of SARS.
Despite these results suggesting an important role of the FcγRII family in the occurrence of ADE of SARS-CoV, our conclusion has to be qualified by the observation that other human immune cell lines (such as K-562, U-937, MT4-R5, and 721.221) that also have detectable cell surface expression of FcγRII by flow cytometry are refractory to ADE infection. The ability of those cell lines to bind SARS-CoVpp/anti-Spike immune complexes is now under investigation. If binding and/or internalization occurs, it would also be of interest to ensure that failure of infection is not due to inability of the viral particles to uncoat/initiate replication in those cells. Finally, in order to define more precisely the mechanism underlying the susceptibility of immune cells to antibody-mediated SARS-CoV infection, a better understanding of the events downstream binding of the immune complexes to the Fc receptors is needed.
Viral glycoproteins typically mediate attachment, fusion, and entry by one of two mechanisms. Some viruses, such as HIV, infect mainly through a pH-independent entry process (fusion occurring at the plasma membrane), whereas other viruses, such as influenza, utilize a pH-dependent endocytic pathway (31
). It has been demonstrated that, following binding to its ACE2 receptor, SARS-CoV entry into susceptible cells occurs via pH-dependent endocytosis (24
), although Spike protein can also induce cell fusion at a neutral pH (40
). More recently a crucial activity of the lysosomal cysteine protease cathepsin L for an efficient infection by SARS-CoV has been uncovered (26
). In contrast, our results indicate that Fc receptor-mediated infection is not dependent on either endosomal acidic pH or the activity of cysteine proteases.
A striking observation was the opposite outcome of the ACE2- versus FcγR-mediated infection subsequent to neutralization of the endosomal acidic pH. Whereas infection of VeroE6 cells was completely inhibited by NH4
Cl, as previously described (24
), the same treatment triggered an approximately 2- to 4-fold increase in ADE infection. The exacerbation of the antibody-mediated infection in the presence of NH4
Cl is not due to a higher cell surface expression of the FcγRII, which was actually slightly decreased. A likely explanation for our results would be that most of the viral particles that enter cells through ADE pathway remain trapped in an acidic compartment and are eventually degraded. The inhibition of endosomal acidification prevents their degradation and grants them opportunity to accumulate and more efficiently infect cells upon removal of the drug at 7 h postinfection. Alternatively, the ability of Spike to mediate cell membrane fusion at neutral pH (40
), could allow more antibody-opsonized SARS-CoVpp to fuse with the endosomal membranes. Such a mechanism has already been observed with a poorly infectious HIV strain, which exhibited greatly enhanced infectivity under conditions where endosomal pH was neutralized (16
). Taken together, our results demonstrate that the biochemical requirements of FcγR-mediated infection are markedly distinct from those underlying ACE2-dependent entry and illustrate a novel mechanism by which SARS-CoV can enter into target cells.
The emergence of SARS in late 2002, which resulted in significant human and economic losses, highlighted the lack of antiviral strategies to control coronavirus infections. Given the continuous presence of SARS-CoV related agents in the environmental reservoir, vaccination remains a major option for preventing resurgence of SARS in humans, especially individuals at highest risk in the case of an outbreak (i.e., healthcare workers). Some vaccines against animal coronaviruses have been successfully generated, but their development has proven very difficult due to immune enhancement of disease in vaccinated recipients (28
We have evaluated, therefore, the neutralizing versus enhancing capabilities of 5 different SARS vaccine formulations. Our results show that ADE was dependent on the immunization strategy with two of five candidate vaccines displaying obvious neutralizing capabilities without triggering SARS-CoVpp entry into immune cells. It is not clear why the other three formulations induced ADE while showing similar neutralizing potencies. We can rule out the lack of recognition/binding to SARS-CoVpp, since all tested sera displayed a clear dose-dependent neutralization of the viral particles over at least four serial dilutions. Furthermore, the total IgG content was comparable among the immunized cohorts, except for the control group that, as expected, had a lower amount (data not shown).
In the FCoV model, ADE has been linked to both neutralizing and non-neutralizing antibodies (8
). In our experiments we have observed that, among sera causing ADE there was a direct correlation between the neutralization titer and the enhancing capability. Because anti-SARS-CoV Spike sera were able to both neutralize SARS-CoVpp infection and bind native recombinant full-length Spike protein with similar potency (i.e., Ssol ≈ WKV
S1 > S ≥ S.ECD), we speculate that vaccination with a given immunogen did not induce a marked bias in eliciting non-neutralizing anti-Spike IgGs. However, we cannot exclude that variations in the ratio of neutralizing versus non-neutralizing epitopes between vaccine preparations could have an impact on the occurrence of ADE.
Because a single class of anti-FCoV Spike antibodies (i.e., IgG2a) has been demonstrated to play a crucial role in initiation of ADE (8
), we have investigated the relative amount of anti-Spike IgG1 and IgG2a in our mouse sera. Interestingly, none of the ADE-inducing sera contained anti-Spike IgG2a antibodies, whereas the two neutralizing/nonenhancing sera did. Further experiments are needed to understand why a markedly distinct polarization of the immune response was achieved with highly related SARS-CoV Spike immunogens and whether absence of IgG2a is causally related to the occurrence of ADE infection.
Recently, allelic polymorphism of the human FcγRIIA has been incriminated as a risk factor for developing severe SARS pathology (75
). The authors of that study found that individuals harboring the R/R131 genotype were more likely to develop severe SARS pathology than were their H/H131 counterparts. The R/R131 form of FcγRIIA can bind to both IgG1 and IgG2a, whereas H/H131 can only bind to IgG2a (48
). However, in our transduction experiments using B cell clones stably expressing a single polymorphic FcγRIIA, both allelic variants were able to trigger ADE of SARS-CoVpp.
It is usually agreed that infection by SARS-CoV is not only confined to the respiratory tract but also involves the gastrointestinal tract and other organ systems. Several reports have highlighted the direct infection of hematopoietic cells by SARS-CoV (19
); however, it is not yet clear how SARS-CoV gets a foothold into immune cells, which do not express the SARS-CoV receptor ACE2 (20
). Antibody-mediated infection, in addition to other alternative entry routes, such as L-SIGN or DC-SIGN (29
), may provide SARS-CoV with an opportunity to broaden its target options. Interestingly, antibody-mediated infection has been shown to elicit markedly distinct responses compared to viral entry through the natural host receptor (23
). Indeed, antibody-mediated infection of feline coronavirus not only enhances viral progeny but also dramatically increases production of tumor necrosis factor alpha by infected cells. The hyperinduction of the proinflammatory cytokine is believed to account for the exacerbated apoptosis of the neighboring leukocytes (61
). Similarly, ADE of Ross river virus has been shown to facilitate viral replication by disrupting transcription of antiviral type I interferons (58
). Because it has been hypothesized that deregulation of host cytokine/chemokine responses is a hallmark of SARS (7
), it would be interesting to investigate how ADE of SARS-CoV infection may contribute to the disease pathogenesis.
Altogether, our results report a novel mechanism by which SARS-CoV can enter into target cells that do not express the conventional virus receptor and are otherwise refractory to the virus. This may have implications for understanding the tropism and pathogenesis of the virus and highlight potential pitfalls associated with immunization against this coronavirus. These findings should prompt further investigations for a better understanding of the molecular and cellular mechanisms underlying ADE of SARS-CoV infection. The consequences of this alternative infection pathway on the functionality and/or homeostasis of the target cells also need to be resolved.