Here we have shown that infectivity of primary isolates from individuals with early HIV infection can be enhanced over 350-fold when opsonised with autologous antibodies and complement, compared with infection in the presence of complement alone. The range of increases in infection are comparable to those reported for dengue virus, a disease in which ADE contributes to pathogenesis during secondary infection by a different serotype and in infants carrying sub-neutralising levels of maternal antibodies [44, 46, 47, 69, 70]. Previous studies using X4-tropic TCLA strains of HIV, or primary isolates opsonised with complement alone, typically showed up to 10-fold increases in infection [25, 28, 36, 41, 42, 66], as we have also confirmed when performing similar experiments. A well-characterised enhancing mAb, 246-D, enhanced infection of the TCLA HIV strain IIIB by up to 3.7-fold in our assay system, whereas the same mAb showed a very modest enhancement of the patient primary isolate MM38.29. The same patient primary isolate was enhanced up to 46-fold by autologous serum. We attribute the high level of C'-ADE shown in this study to the use of primary clinical isolates of HIV and their autologous antibodies on T cells able to support infection by R5-tropic viruses. The high levels reported here suggest that C'-ADE may be an important infection-enhancing mechanism in vivo. Furthermore, access to a cohort of recently infected individuals allowed us to perform longitudinal studies of C'-ADE and provided us with a unique perspective of the development of this response over time. Later virus isolates were enhanced more than early isolates from the same individuals, indicating that the C'-ADE observed in our model system reflects a replicative advantage for the enhanced viruses in vivo.
Most HIV antibody assays in vitro are carried out in the absence of complement, yet, in vivo, cell-free HIV is likely to be opsonised with antibodies and complement at all stages of disease post-seroconversion . Given that complement can modulate antibody activity, it is logical to consider the effects of complement alongside any study of antibodies. Furthermore, HIV replicates predominantly in lymphoid tissues, which are rich in immune cells expressing CRs, including macrophages, DCs, and some T cells and B cells [11, 71]. We have shown that the presence of a CR on an HIV target cell can drastically alter the outcome of virus opsonisation with antibodies and complement. We have shown that sera with apparently no activity when assayed on CR-negative cells (Table 2 and Figure 3), or with virus-inhibitory activity when assayed on CR-negative cells in the presence of complement (Additional File 2
; Figure S2), can enhance infection by several orders of magnitude on CR-positive cells in the presence of complement (both on T cells naturally expressing CR2, and on unrelated cells engineered to express CR2; Figure 2, Figure 3, Additional File 2
; Figure S2).
The C'-ADE observed in our model system could represent a variety of processes mediated through CRs in vivo. Through analogy with previous reports carried out with virus opsonised with complement alone, these could include direct infection-enhancement of CR+ target cells, enhanced B cell-to-T cell or DC-to-T cell trans -infection, or enhanced FDC trapping of complement- and antibody-opsonised virions [35–38, 72–77], all of which could influence HIV pathogenesis. For example, CR2 has been implicated in HIV trapping, archiving and trans -infection of T cells mediated by B cells and FDCs [73–78]. The high levels of enhancement seen here through CR2 may reflect an important role for non-neutralising, enhancing antibodies in these processes in vivo, and may be particularly relevant to early B cell-mediated viral dissemination and subsequent FDC trapping. In addition to B cells and FDCs, CR2 expression has been reported on T cell subsets [79–84], thymocytes [82, 85, 86] and astrocytes , which may support directly enhanced infection by HIV.
The principal complement component ligands of CR2 are the C3 fragments C3dg and C3 d, and to a lesser extent iC3b [71, 88]. The enhancement reported here through CR2 might also occur through other receptors for C3 complement fragments, namely CR1, CR3 and CR4 . Complement alone has been shown to substantially increase infection of primary monocytes/macrophages and DCs expressing CRs [35–38], and enhancing antibodies may augment this process by increasing the deposition of complement on the virion. However, effects mediated through CRs may be counteracted by reported virus-inhibitory effects mediated through FcRs on macrophages and DCs [37, 89], and the net outcome may be influenced by the receptor balance on the target cell and the nature of the antibodies [36, 37]. Future evaluation of CR and FcR expression on primary cell targets for HIV, and investigation of the role of these receptors in enhancing and inhibiting HIV infection, is warranted.
In agreement with results obtained by Montefiori and colleagues in a macaque model of acute SIV infection , the presence or emergence of neutralising activity masks pre-existing enhancing activity. This is shown here by the temporal disappearance of enhancing activity upon the detection of a potent neutralising response (Figure 2
and Table 2), and by detection of enhancement upon dilution of neutralising IgG (Figure 6). Mechanistically, this could be because abundant non-neutralising antibodies enhance infection through increasing attachment to the target cell, while the neutralising antibodies are ultimately able to block entry of the virus if present in sufficient quantities, whether it is through enhanced or direct infection.
Our studies of heterologous virus-serum pairs show that enhancing, but not neutralising, activity can be transferred to heterologous viruses, perhaps because the enhancing antibodies are directed to more conserved, non-neutralising, immunodominant epitopes. Yet, this does not equate with any antibody having the potential to enhance - it is likely that antibodies must be of a reasonable affinity and concentration for enhancement to occur. Early studies of C'-ADE of HIV demonstrated an association between gp41 antibodies and C'-ADE activity [91, 92]. A key "enhancing domain" was identified within the PID of gp41, in a region primarily accessible on disassembled or post-fusion envelope spikes . The association between C'-ADE and gp41 antibodies may be of particular relevance to the acute stage of HIV infection, when the first antibodies produced in response to infection are directed against gp41 and are non-neutralising .
Laboratory preparations of HIV, and cell-free HIV present in plasma of infected individuals, are often attributed with low infectivity relative to the number of physical virus particles present, with discrepancies between infectious units and physical virus particles estimated to be as high as 1 to 60,000 [94–96]. The traditional explanation for this is a high level of "defective" viral particles, produced as a consequence of the error-prone nature of reverse transcriptase and the labile nature of the viral envelope proteins. An alternative explanation is that, due to a relatively low efficiency of initial cellular receptor engagement , coupled with the low number of trimeric envelope spikes on the viral surface , the infectivity of HIV is underestimated because the vast majority of virus particles in a preparation lack the opportunity to infect the target cells . This is supported by other demonstrations of dramatic enhancement of HIV infectivity in vitro, through coating of HIV with semen amyloid fibrils , incorporation of host attachment proteins into the viral membrane , and direct interactions between gp120 and C-type lectin receptors on target or bystander cells [102, 103]; all are mechanisms that stabilise the virus on the target cell and improve the efficiency of receptor interaction. It is unclear to what extent these mechanisms would function in vivo when the virus is coated with antibodies and complement. The possibility of C'-ADE also drawing on the so-called "non-infectious" pool of virions, as opposed to increasing the replication of readily infectious virions, is supported by our studies of the mechanism of C'-ADE, which show that a version of CR2 lacking a cytoplasmic tail (and consequently lacking signalling ability) supports high-level C'-ADE. Under these conditions, increased attachment to the target cell is the most likely mechanism of C'-ADE.
In conclusion, our study highlights the possibility that antibodies with seemingly no activity in standard cell-based neutralisation assays may have other activities when assayed under conditions that accommodate effector functions, and may even contribute to viral pathogenesis.