In view of the limited number of BR-Nabs that have been identified to date, better characterization of antibody responses in HIV-1-infected patients and identification of new and potent BR-Nabs is critical in designing vaccine immunogens. As a first step in our efforts toward achieving these goals, we have systematically generated and purified five soluble fusion proteins containing the C-terminal 30, 64, 100, 142 or 172 amino acids of gp41 ectodomain from an M group consensus sequence. These recombinant proteins appear to have intact antigenic structures based on immunoprobing analyses with BR-Nabs 2F5 and 4E10, 98-6, and antisera from HIV-1-infected patients. Characterization of antibody responses against gp41 had been largely limited to short peptides because the protein has been difficult to produce in soluble forms. Therefore, the set of gp41 fragments we have generated will be invaluable reagents for characterizing antibody responses against HIV-1 gp41.
Expression of eukaryotic or viral proteins in bacteria typically results in misfolding and aggregation of proteins, which accumulate in inclusion bodies. Our gp41 protein fragments were no exception despite the fact that they were fused to GST. However, we were successful in denaturing the proteins, refolding, and renaturing them into soluble forms in the absence of any detergent, although we experienced greater difficulties in solubilizing the two larger fragments and obtained substantially lower yields. Although we did not attempt to express gp41 protein fragments without GST, we are quite certain that GST played a critical role in allowing the proteins to renature into soluble forms. This conclusion is based in part from our observation that attempts to remove GST by cleaving between the two fusion partners with thrombin resulted in immediate precipitation of gp41 (data not shown). We speculate that GST could be functioning in a way analogous to gp120 by masking some of the hydrophobic patches on gp41 away from aqueous solvent to prevent non-specific aggregation.
Since GST-gp41 fusion proteins are antigenically intact, shown by their ability to bind not only BR-Nabs 4E10 and 2F5, but also conformational antibodies that bind to cluster III, they could be potential candidates for vaccine development. However, these proteins are not ideal as antigens for a vaccine as the large GST portion is likely to be immunodominant over smaller gp41 fragments. On the other hand, these proteins are highly suitable for structural studies, especially GST-gp41-64 since it is well recognized by both 2F5 and 4E10 and we can easily produce up to 40 mg of the protein from one liter of bacterial culture. To date, the only solved crystal structures of gp41 are HR1/HR2 coiled-coil core and peptide epitopes that bind to 2F5 and 4E10 (Cardoso et al., 2005
; Chan et al., 1997
; Ofek et al., 2004
; Tan et al., 1997
; Weissenhorn et al., 1997b
). Determining the structure of the entire MPER with the HR2 domain would facilitate better understanding of gp41 function and designing antigens that can elicit antibodies such as 2F5 or 4E10.
All five GST-gp41 fusion proteins were equally well recognized by 2F5, indicating that the epitope is conformationally identical and similarly exposed amongst all proteins (). Although all five proteins were also recognized by 4E10, GST-gp41-100 and -30 were about 5-and 20-fold less reactive, respectively, when compared to the other three proteins (). This result was somewhat unexpected taking into account that the 4E10 epitope is considered to be linear and is only four amino acids downstream of the 2F5 epitope. One likely explanation is that although the epitope is linear, it is highly conformational, requiring other regions of gp41 for proper formation of the antigenic epitope structure. In this regard, it has previously been reported that the 4E10 epitope assumes a helical conformation (Cardoso et al., 2005
) and modifications that enhance helical properties increase antibody-binding affinity (Brunel et al., 2006
; Cardoso et al., 2007
). Based on these findings, our current hypotheses are (1) sequences within GST-gp41-64 upstream of the MPER (i.e.
between C-terminal residues 64 and 30) provide constraints on the helical conformation of 4E10 epitope, thereby enhancing antibody binding compared to GST-gp41-30; (2) additional sequences between HR1 and HR2 within GST-gp41-100 are either disrupting or masking the epitope partially to reduce 4E10 binding; and (3) this inhibition is reversed when the HR2 domain forms coiled-coil structure with HR1, as demonstrated by reactivity to mAb 98-6 (), in GST-gp41-142 and -172. More detailed biochemical and structural analyses are needed to test these hypotheses.
We observed tremendous variation in antibody responses to gp41 amongst different patients, not only with respect to the magnitude, but also the pattern of antibody reactivity against different gp41 fragments and peptides. One parameter that could affect the magnitude of antibody responses is plasma viral load. Greater antigenic stimulation might be expected to induce stronger antibody responses. In this regard, the duration of infection since seroconversion and possible treatment of patients with anti-retroviral therapy could potentially influence antibody levels (Binley et al., 2000
; Lafeuillade et al., 1997
; Morris et al., 2001
). An important contribution may also stem from the overall responsiveness of the patients’ immune system to antigenic stimulation; higher viral load could also mean greater deterioration of helper T cell immunity, and therefore weaker B cell response. In the present study, the patients were chosen randomly, without considerations to these parameters, because we were primarily interested in studying differential epitope targeting in individual patients and identifying those who mounted strong antibody responses against potential neutralization epitopes (viz.
The variation in antibody responses against different gp41 regions or peptides in different patients could be attributable to differences in patients’ immune system (e.g. immunoglobulin gene repertoire) and/or HIV-1 isolates the patients are infected with (e.g. envelope protein sequence). We saw that many patients who mounted strong antibody responses against the MPER (i.e. reactivity against GST-gp41-30) exhibited broader, more potent neutralizing activity compared to those who did not. It should be emphasized, however, that we presently do not know the epitope(s) targeted by Nabs. They could be directed against the MPER, other regions of gp41, or epitopes within gp120. Furthermore, the breadth of neutralizing activity could be polyclonal with multiple Nabs targeting different epitopes. In any event, further characterization of B cell repertoire, virus isolates from these patients, and mAbs generated from the patients could provide clues as to how BR-Nabs could be elicited.
Considering that plasma samples from four of six patients that showed strong antibody reactivity against GST-gp41-30 had potent neutralizing activity, this fusion protein could be used as a tool for rapidly screening patient sera to identify those who might have BR-Nabs. In this regard, we were somewhat disappointed with the fact that patients CWRU-1 and -2, whose antisera reacted most strongly against the protein, did not exhibit potent neutralizing activity as we had hoped. One possible explanation is that there might be epitopes in the MPER that elicit non-neutralizing antibodies. It is conceivable that these antibodies could prevent binding of Nabs that target epitopes in close proximity due to steric hindrance (e.g. 2F5 or 4E10). In this case, observed neutralizing activity would be determined by relative antibody titers and affinity between neutralizing and non-neutralizing antibodies that compete for the adjacent epitopes. If this hypothesis is true, then it would have strong implication in terms of vaccine design; not only does the antigen have to have correct neutralization epitopes, the antigen should not have competing non-neutralizing epitopes nearby.
Attempts to elicit antibodies with similar properties to 2F5 and 4E10 in animals immunized with antigens containing the gp41 MPER have not been successful. It has been proposed that this difficulty is due to autoantigen mimicry by HIV-1 based on observations that 2F5 and 4E10 cross-react with phospholipid cardiolipin (Alam et al., 2007
; Haynes et al., 2005
; Sanchez-Martinez et al., 2006
). This view that 2F5 and 4E10 have properties of autoantibodies is quite controversial since they have been used in passive immunization studies without any complication (Ferrantelli et al., 2004
; Joos et al., 2006
; Trkola et al., 2005
). In addition, a more recently published study reports that 2F5 fails to exhibit any cardiolipin reactivity under their set of experimental conditions (Scherer et al., 2007
). Also, while 4E10 does have general affinity to lipids, this reactivity resembles that of anti-phospholipid antibodies elicited during many infections rather than that of autoimmune antiphospholipid syndrome. Thus, the inability to elicit antibodies with similar properties to 2F5 and 4E10 might not be attributable to immune tolerance mechanisms. In this study, we have identified many patients who mounted antibodies against the same peptides recognized by 2F5 and, to a lesser extent, by 4E10 (). Although we have yet to determine the precise epitopes and specificities of these antibodies, the results of our study suggest that patients who mount antibodies against epitopes that are near, or overlap with, those targeted by 2F5 or 4E10 may not be as rare as has previously been thought. In agreement with our findings, Gray et al.
(Gray et al., 2007
) have recently reported up to one-third of HIV-1-infected patients mount Nabs against the MPER. Additional studies with a larger panel of patient samples and detailed biochemical analyses of purified antibodies that target the MPER could provide more definitive answers. In this regard, the fusion proteins we generated could be ideal reagents for rapid assessment of antibody responses against gp41 and for affinity purification of MPER-directed antibodies.