Enhanced Neutralization of HIV by Antibodies Displayed on the S-Layer of Caulobacter crescentus

Abstract

Innovative methods of prevention are needed to stop the more than two million new HIV-1 infections annually, particularly in women. Local application of anti-HIV antibodies has been shown to be effective at preventing infection in nonhuman primates; however, the concentrations needed are cost prohibitive. Display of antibodies on a particulate platform will likely prolong effectiveness of these anti-HIV agents and lower the cost of goods. Here, we demonstrate that the bacterium Caulobacter crescentus and its highly expressed surface-layer (S-layer) protein can provide this antibody display platform. Caulobacters displaying protein G, alone or with CD4 codisplay, successfully captured HIV-1-specific antibodies and demonstrated functional neutralization. Compared to soluble antibodies, a neutralizing anti-HIV antibody displayed on Caulobacter was as effective or more effective at neutralizing diverse HIV-1 isolates. Moreover, when an antibody reactive with an epitope induced by CD4 binding (CD4i) was codisplayed with CD4, there was significant enhancement in HIV-1 neutralization. These results suggest that caulobacters displaying anti-HIV antibodies offer a distinct improvement in the use of antibodies as microbicides. Furthermore, these reagents can specifically evaluate anti-HIV antibodies in concert with other HIV-1 blocking agents to assess the most suitable tools for conversion to scFvs, allowing for direct display within the S-layer protein and further reducing cost of goods. In summary, C. crescentus, which can be easily produced and chemically stabilized at low cost, is well suited for engineering as an effective platform, offering an inexpensive way to produce and deliver HIV-1-specific microbicides.

INTRODUCTION

Despite the success of some prevention programs, transmission of HIV infection remains a significant problem. Women account for over half of all adults living with HIV worldwide, and given the increasing prevalence of new infections in women, the development of prevention alternatives which can be controlled by females is imperative. A number of agents have been proposed and/or tested as a topical therapy to prevent HIV infection, including nonspecific inhibitors and specific anti-HIV inhibitors (1, 8, 9, 18, 31, 32). Though there has been modest success, additional inhibitors and delivery systems are required for sustained prevention of infection. Systemic administration of specific human monoclonal antibodies has been shown to prevent infection (2, 10, 11, 21–23). Given that the practical application of passive immunotherapy is limited by the quantities of antibodies required for systemic therapeutic levels and timing, passive immunotherapy by topical application of antibodies to at-risk areas may be attainable. Topical delivery of small-molecule inhibitors of HIV, as well as human monoclonal antibodies, has been shown to prevent transmission of HIV in nonhuman primate models (19, 37, 38). The therapeutic half-life of these inhibitors remains to be determined and may limit the effectiveness of antibody or protein inhibitors.

Expression of antibodies or protein inhibitors in vivo by engineered lactobacillus has been proposed as one means to improve the availability of HIV inhibitors. It is reasoned that, being a normal component of the vaginal flora, engineered lactobacilli will provide long-term stable expression of anti-HIV inhibitors. Engineered lactobacilli have been shown to inhibit HIV infection when tested in vitro (5, 6, 36); however, it is unclear that stable expression of inhibitors by the engineered lactobacilli will be maintained in vivo given the abundant and dynamic normal flora of the female genital tract. We propose to develop an alternative bacterial expression system using a nonpathogenic bacterium that presents HIV-blocking agents efficiently, is inexpensive to produce as a killed, stabilized agent, and is likely to be compatible with exposure to human mucosal tissue.

Caulobacter crescentus is a harmless soil and water bacterium that elaborates a protein surface layer (S layer) composed of a 98-kDa monomer protein (RsaA). It is secreted at high levels (20 to 25% of cell protein [17]) and self-assembles on the surface of the bacterium in a hexagonal pattern. A typical cell has approximately 40,000 copies of this protein monomer (25, 33, 34). We have demonstrated that it is feasible to insert large genetic segments within the S-layer gene while maintaining all functional aspects: secretion, assembly (crystallization), and cell surface attachment of the resulting recombinant protein, as well as consequent high density presentation of the inserted peptide (12, 17, 25, 26). S-layer expression/display of CD4 (domain 1) and MIP1α has successfully generated recombinant caulobacters that can inhibit HIV infection (27). Protein G from Peptostreptococcus, which binds to the Fc region in IgG of many species, has also been expressed from within the S-layer protein (26). We reasoned that expression of protein G on the surfaces of caulobacter cells would enable us to capture HIV-specific antibodies, generating a more stable physiological platform that might prolong the half-life and effectiveness of these antibodies as preventative reagents or increase their effectiveness by coordinate, dense display.

To test this concept, we employed caulobacters displaying protein G to capture anti-HIV antibodies, and we tested their ability to retain the attached antibodies and then to neutralize HIV. With the eventual goal of expression of antibodies as scFvs on the surfaces of caulobacter cells, we first needed to demonstrate that antibodies displayed on the bacterium retained HIV neutralization activity. Next, we asked whether dense, multimeric display of an anti-HIV antibody (via a bacterial “particle”) is more effective at virus inhibition than soluble antibody, perhaps based on an increased avidity to HIV. We also examined the use of combinations of microbicide agents displayed simultaneously on a bacterium as synergistic blockers of HIV infection.

MATERIALS AND METHODS

Bacterial strains, plasmids, and growth conditions.

C. crescentus strains were grown at 30°C in PYE medium (0.2% peptone, 0.1% yeast extract, 0.01% CaCl₂, 0.02% MgSO₄) with 1.2% agar for plates. Escherichia coli DH5α was used for cloning manipulations and was grown at 37°C in Luria broth (1% tryptone, 0.5% NaCl, 0.5% yeast extract) with 1.3% agar for plates. Ampicillin (Amp) and kanamycin (Km) were used at 50 μg/ml, and chloramphenicol (Cm) was used at 20 μg/ml for E. coli and 2 μg/ml for C. crescentus. Isolation of plasmid DNA was performed using the QIAprep spin mini prep system (Qiagen). Restriction enzyme digestions were visualized by electrophoresis on 0.9% Tris-borate-EDTA (TBE) agarose gels. DNA fragments were excised and purified using the QIAEX II gel extraction kit (Qiagen) following the manufacturer's protocol. Electroporation to introduce plasmids into C. crescentus or E. coli was performed as previously described (13).

The C. crescentus strains used were modifications of strain CB2A, which has a spontaneous amber mutation in rsaA, resulting in the loss of S-layer production. JS4022 (26) is further modified by mutation of sapA, which eliminates degradation of heterologous insertions in RsaA, and the introduction of repBAC, which allows replication of the p4A shuttle plasmid vector in Caulobacter (12). p4A containing a variant of rsaA with a BamHI site inserted at a position corresponding to amino acid 723 and a version further derived to contain domain 1 of CD4 have been described (27). Strain constructs maintaining these vectors will be referred to as C-C (control) and C-CD4, respectively.

A C. crescentus construct displaying a version of protein G were constructed by gene replacement of the chromosomal gene rsaA of JS4022 with a modified rsaA gene containing a protein G construct (three GB1 IgG Fc binding domains, flanked and separated by 20 amino acid spacers) positioned at the amino acid 723 site (26). This was done by subcloning the rsaA/protein G construct as an EcoRI-HindIII segment into pMOBSacB and performing gene replacement procedures as previously described (12). Protein G display was confirmed by immunofluorescence microscopy using Alexa Fluor 488-coupled secondary antibody. This strain construct (JS4028) is referred to herein as C-PG.

C. crescentus displaying both protein G and CD4 was constructed by ligating the rsaA/protein G segment into pTZ18U (Stratagene) as an EcoRI-HindIII fragment. The resulting plasmid was fused with the p4A-rsaA/CD4 construct by HindIII digestion of both ligation and eventual introduction into JS4022. Herein, this construct is referred to as C-PG/CD4.

To confirm display and to estimate the amounts of chimeric RsaA/CD4 and RsaA/protein G constructs, S-layer present on the cell surface was isolated by low-pH extraction as previously described (39). Extracted protein samples from equal amounts of cells were examined by SDS-PAGE with 10% separating gels and Coomassie brilliant blue R-250 staining.

Antibody capture by protein G-displaying caulobacters.

C. crescentus cells displaying protein G (C-PG) or protein G and CD4 (C-PG/CD4) were grown in peptone-yeast extract (PYE) medium overnight at 30°C. After cells were washed by centrifugation and suspension and the optical density at 600 nm was measured, the cell density was adjusted to 5 × 10⁸ bacteria/ml in PYE. To capture antibody, 400 μl of cells was combined with 100 μl of human monoclonal antibody at 20 μg/ml. Antibodies and specificities are listed in Table 1, and all were purified from tissue culture supernatant using protein G chromatography. Purified antibodies were quantitated by enzyme-linked immunosorbent assay (ELISA) prior to use. Caulobacters were incubated with antibody for 30 min at 4°C, unbound antibody was collected after centrifugation, and the pelleted bacteria were used in neutralization assays as described below. The unbound antibody concentration was determined by capture ELISA using plates coated with goat anti-human IgG (or kappa), and detection was determined by using horseradish peroxidase-labeled goat anti-human IgG (or kappa) with isotype controls used as standards to determine concentration.

Table 1.

Capture of antibody by protein G displayed on caulobacter cells^a

Antibody	Specificity	Amt of antibody (μg/ml)
IgG1
F71	Isotype control	19.7 ± 0.4b
b12	gp120 (CD4BS)	19.7 ± 0.4
F240	gp41	15.5 ± 3.6
F425 A1g8	gp120 (CD4i)	16.4 ± 3.0
F285	env	19.7 ± 0.5
F759 A2c2	gp41	19.0 ± 2.0
F424	env	19.7 ± 0.5
F759B1c7	gp41	18.6 ± 1.7
F105	gp120 (CD4BS)	18.8 ± 1.9
F425B4a1	gp120 (V3)	10.2 ± 6.3
F530	gp120 (V3)	18.5 ± 1.7
F756A2f9	gp41	19.9 ± 0.1
IgG2
B4e8	gp120 (V3)	16.5 ± 4.0
IgG3
F464	gp120	14.9 ± 8.6
F223	gp120	10.7 ± 10.8

^aQuantity of antibody captured by caulobacter cells displaying protein G (C-PG), as determined by unbound antibody. Isotypes of individual antibodies are shown.

^bDerived from four independent experiments.

HIV neutralization assay.

Caulobacters with bound antibody were suspended in phosphate-buffered saline (PBS) and tested for neutralization of virus using a standard TZM-bl assay. TZM-bl cells were obtained from the NIH AIDS Research and Reference Reagent Program (ARRRP) and grown in Dulbecco's modified Eagle's medium (DMEM) containing 10% fetal bovine serum. HIV-1 clade B primary isolates BaL (macrophage tropic), JR-FL (R5 tropic), SF162 (R5 tropic), 67970 (X4 tropic), and 89.6 (R5X4 tropic) and clade C isolate 93MW960 (R5 tropic) were obtained through the NIH ARRRP. Primary viruses were propagated in phytohemagglutinin (PHA)-stimulated peripheral blood mononuclear cells (PBMCs). Peripheral blood was collected from normal healthy donors after informed consent was obtained, and the study was approved by the Beth Israel Deaconess Medical Center Institutional Review Board. C-PG and C-PG/CD4 with a known amount of antibody captured by displayed protein G were serially diluted 2-fold in a plate in a volume of 50 μl. The number of Caulobacter cells used for each experiment was a function of the antibody concentration tested. That is, assuming 100% efficiency in capture, if 20 μg/ml antibody was required, 2 × 10⁸ caulobacters were used. Virus (100 50% tissue culture infective doses [TCID₅₀]) in 50 μl was added, and plates were incubated for 1 h at 37°C. TZM-bl cells were resuspended to 1 × 10⁵ cells/ml in DMEM containing DEAE-dextran, and 100 μl was added to each well. Plates were incubated for 48 h at 37°C and 5% CO₂, and β-galactosidase activity was measured using a β-galactosidase assay reagent according to the manufacturer's specifications (Pierce). For each test antibody, the 50% inhibitory concentration (IC₅₀) or IC₉₀ was determined by linear regression. A control consisting of caulobacter cells without added antibody but expressing protein G or protein G and CD4 as well as caulobacters not expressing either moiety was run in each experiment. At the concentration of caulobacter cells used, there consistently was no effect on viral infection (data not shown).

Statistical analysis.

Student's t test was used to determine significance.

RESULTS

Display of protein G and CD4 by Caulobacter.

In previous studies, we have shown that CD4 and protein G can be displayed on caulobacter cells by genetic insertion into the S-layer protein (26, 27). Here we extended this approach and produced a construct displaying both moieties on the same cell. The low-pH extraction method has proven to be a reliable way to rapidly estimate the amount of S-layer protein that resides on the surface and so can be used to compare the proficiency of export for the S-layer chimeras. Both CD4 and protein G insertions lead to some loss of secretion efficiency (Fig. 1), but we estimate that 10 to 20% of the normal 40,000 copies per cell was exported. In addition, there was little reduction in the level of CD4 display and only a modest reduction of protein G display level when both were displayed simultaneously.

Fig. 1.

SDS-PAGE results of recombinant variants of RsaA prepared from the strains described in Materials and Methods. Equal amounts of low-pH-extracted protein samples, prepared from cells normalized to the same density, were loaded on the gel. Lane 1, C-C (control); lane 2, C-CD4; lane 3, C-PG; lane 4, C-PG/CD4. Molecular mass standards are shown to the left.

Antibody capture by Caulobacter.

To demonstrate that antibodies retained functional activity when displayed on the surfaces of caulobacter cells, caulobacters expressing protein G or protein G and CD4 were tested for antibody capture and virus neutralization. Therefore, a panel of human monoclonal antibodies, both directed at HIV and control non-HIV antibodies, was tested for capture by C-PG and C-PG/CD4. Antibodies were loaded onto the caulobacters at a concentration of 20 μg/ml and unbound antibody was quantitated by ELISA to yield an estimate of absorbed antibody. As shown in Table 1, for the majority of antibodies, more than 90% of the antibody is captured, regardless of isotype. Some antibodies were captured less efficiently, such as F425B4a1 or F223. It is unclear whether this was a function of accessibility of the protein G binding site on the displayed protein G or the Fc region itself. Similar results were also obtained using caulobacters expressing protein G and CD4 (Table 2). More importantly, even after 48 h incubation at 37°C, all of the antibody remains bound by the protein G on the caulobacters and there is no dissociation. That is, duplicate wells of antibody captured on C-PG or C-PG/CD4 were set up in parallel to the neutralization assay with just the addition of medium. At 24 and 48 h postplating, supernatant was removed and tested for antibody. No antibody was detected, indicating that dissociation did not occur (Table 3).

Table 2.

Comparison of antibody capture between caulobacter cells displaying protein G or those displaying protein G and CD4

Antibody	Amt of antibody (μg/ml) displaying indicated proteina
	C-PG	C-PG/CD4
F71	19.7 ± 0.4b	19.3 ± 0.6b
b12	19.7 ± 0.4	19.7 ± 0.3
F240	15.5 ± 3.6	14.4 ± 3.8
F425A1g8	16.4 ± 3.0	14.9 ± 3.0
F425B4e8	16.5 ± 4.0	13.7 ± 4.3

^aQuantity of antibody captured by caulobacter cells displaying protein G (C-PG) or those displaying protein G and CD4 (C-PG/CD4), as determined by unbound antibody.

^bDerived from five independent experiments.

Table 3.

Stability of monoclonal antibody captured by caulobacter cells^a

Stability of antibody	Amt of antibody (μg/ml)
	F71	F240	b12	F425A1g8	F425B4e8
Capturedb	36	25	38	20	20
Dissociatedc	0.02	0.63	0.02	0.01	0.02

^aResults represent a single experiment and are representative of two independent experiments.

^bQuantity of antibody captured by caulobacter cells displaying protein G (C-PG), as determined by unbound antibody.

^cQuantity of antibody released into the supernatant of C-PG incubated for 48 h in media in parallel to samples in the neutralization assay.

Neutralization of HIV by antibody captured on caulobacter cells.

In order to more directly compare soluble antibodies to caulobacter-displayed antibodies, samples were tested for HIV neutralization. For each neutralization assay, unbound antibody was quantitated such that it was known how much antibody was captured by the caulobacters and used in the neutralization assay. In initial studies, antibody absorbed caulobacters were tested for the ability to reduce virus stock infectivity. Antibody absorbed caulobacters were incubated with SF162 virus stock for 1 h at 37°C. Caulobacters were removed by centrifugation, and the titer of absorbed virus stock on TZM-bl cells was determined. As controls, control caulobacters (C-C, no display of protein G or CD4) were used to absorb virus stock in addition to titration of unabsorbed virus stock. Consistent with what was observed with soluble antibody, b12 and F425 B4e8 reduced infectivity by more than 95%. F425 A1g8 partially reduced infectivity, which is consistent with that seen with soluble antibody. Given those results, studies were broadened to include more isolates and antibodies, and results are shown in Table 4.

Table 4.

Neutralization of SF162 by caulobacter-bound anti-HIV antibody^a

Antibody	IC90b (μg/ml) for SF162
	Soluble	Displaying protein G	Displaying protein G/CD4
F71	>20	>20	>20
b12	6.7 ± 3.9	12.0 ± 0.9	6.7 ± 4.1
F240	>20	>20	>20
F425 A1g8	>20	>20	9.8 ± 3.9c
F425 B4e8	17.7 ± 4.0	9.5 ± 0.9	7.2 ± 2.1d

^aNeutralization of SF162 was tested using known concentrations of soluble antibody or antibody captured onto caulobacters expressing protein G or those displaying protein G and CD4 in a TZM-bl assay.

^bIC₉₀, concentration of antibody required to inhibit 90% of SF162.

^cP < 0.003.

^dP < 0.04.

F71 is an irrelevant antibody and does not react with HIV. Consistent with previous studies, F240, an antibody reactive with the immunodominant domain of gp41, fails to neutralize infection in this assay. Of the remaining antibodies listed in Table 1, only four demonstrated consistent neutralization. The CD4 binding site antibody, b12 (30), neutralized infection well as soluble antibody or when presented on caulobacters. The F425B4e8 (B4e8) antibody reacts with the V3 loop and has potent neutralization activity (4, 28). F425 B4e8 has also been shown to inhibit gp120 binding to syndecans, which are broadly expressed on epithelial cells (7). It is shown here that neutralization activity was markedly increased when the B4e8 antibody was presented on caulobacters. For example, 17.7 μg/ml soluble F425B4e8 antibody is required to neutralize 90% of SF162, whereas only 9.5 μg/ml or 7.3 μg/ml is required for antibody presented using caulobacters displaying protein G or protein G and CD4, respectively. We hypothesize that the multiple copy display on a single “particle” (C-PG) may be responsible for this enhanced activity. Furthermore, the displayed antibody cooperated with CD4 in even more effective neutralization (C-PG/CD4; P < 0.04).

The F425A1g8 (A1g8) antibody reacts with an epitope on gp120 exposed by CD4 binding (CD4i). As reported by others and observed in our laboratory, this antibody is limited in neutralization activity in the absence of CD4 engagement. However, when presented on caulobacter cells in codisplay with CD4, significant neutralization was observed (Table 5). To further explore the ability of a CD4i antibody to neutralize when coexpressed with CD4, neutralization of a variety of isolates was tested using soluble F425A1g8, and caulobacter cells displayed F425A1g8 alone and in the presence of either soluble or displayed CD4. As predicted, soluble F425A1g8 had minimal neutralizing HIV activity, with the exception of the SF162 isolate. Addition of a low, subneutralizing concentration of sCD4 enabled enhanced neutralization of this isolate. However, when CD4 was codisplayed on caulobacters, there was significant neutralization of all the isolates, including those generally resistant to neutralization. With the exception of the SF162 isolate, which is readily neutralized by soluble antibody, neutralization was significantly enhanced by codisplay with CD4. Increased neutralization was observed at all concentrations of antibody, as demonstrated for isolate 89.6 in Fig. 2. Capture of antibody by protein G expressed on caulobacter did not necessarily always lead to increased neutralization activity, presumably due to differences in the mechanism of antibody neutralization and/or the expression of the particular epitope. However, these results clearly demonstrate that display of HIV binding components and antibody on the surfaces of caulobacter cells can effectively inhibit infectivity and strongly support the further design and combinatorial use of caulobacter-based HIV-specific microbicides.

Table 5.

Neutralization of HIV by CD4i antibody displayed on caulobacter cells^a

Antibody	IC50 (μg/ml) for indicated antibodyb
	JR-FL	67970	93MW960	SF162	89.6
Soluble F425A1g8	>40	>40	>40	0.3 ± 0.03	>40
Soluble F425A1g8 + sCD4	>40	>40	>40	0.1 ± 0.06	>40
F425A1g8 displayed on C-PG	15.8 ± 6.8	14.3 ± 0.1	>40	1.0 ± 0.6	24.6
F425A1g8 displayed on C-PG/CD4	8.9 ± 0.9	7.7 ± 0.2	24.1 ± 1.9	0.9 ± 0.4	5.3

^aNeutralization of five isolates of HIV was tested using known concentrations of soluble antibody or antibody captured onto caulobacters expressing protein G (C-PG) or those expressing protein G and CD4 (C-PG/CD4) in a TZM-bl assay. Results are derived from three independent experiments.

^bIC₅₀, concentration of antibody required to inhibit 50% of virus.

Fig. 2.

Enhanced neutralization of isolate 89.6 by codisplay of CD4i antibody F425A1g8 and CD4. F425A1g8 was captured by protein G displayed on caulobacter cells (C-PG) and codisplayed with CD4 (C-PG/CD4) and incubated with HIV (89.6; 100 TCID₅₀) prior to addition to TZM-bl cells. After 48 h, viral infection was measured as β-galactosidase activity, and the percent neutralization was determined. Results are representative of three experiments.

DISCUSSION

While the female genital tract is covered by epithelial cells that normally form a tight barrier to exclude pathogens, HIV is able to penetrate or be translocated across the surface and establish infection. Once the mucosal barrier has been crossed, HIV targets resting T cells in the lamina propria and virus disseminates to the lymphoid compartment (24). Therefore, a microbicide needs to work not only by preventing entry and transgression through the mucosa but also by preventing infection of local cells. A number of protein anti-HIV inhibitors, including but not limited to receptor analogues and antibodies to both host and viral antigens, have been demonstrated to inhibit transmission of infection in vitro. However, a system for sustained, stable delivery of these protein inhibitors will contribute to the successful use of these inhibitors in vivo.

Lactobacilli have been used as probiotics and, more recently, to express foreign proteins. When administered orally or vaginally, lactobacilli have been shown to restore the normal vaginal flora and alleviate bacterial vaginosis (BV) (14, 15, 29). Relevant to HIV, lactobacilli transfected with a number of anti-HIV proteins, including CD4, MIP1β, and single chain (scFv) anti-ICAM, were effective at secreting protein that neutralized HIV in vitro (5, 20). Despite the potential positive features of engineered lactobacilli for a microbicide formulation, considerable uncertainty remains for use in vivo. While persistent colonization (16) in the vagina might occur, stable maintenance of these recombinant bacteria within the flora at a sufficiently high level to provide minimally adequate ongoing microbicide activity is not likely. Similarly, expression of scFv activities on the engineered bacterium's surface may well not be at sufficiently high levels to achieve blocking of virus infection. In addition, long-term colonization of a live, engineered bacterium may have negative effects on native microflora or may cause unanticipated side effects for some individuals. Elimination of a stably colonized strain from the vaginal tract may be difficult or impractical. We sought to develop a second bacterial display system that could address these concerns.

Caulobacters are generally regarded as completely nonpathogenic bacteria; however, they are Gram-negative bacteria and so have lipopolysaccharide (LPS) in their cell wall. Many types of LPS cause sepsis reactions when injected. This is due to the lipid A portion of LPS stimulating a sepsis cascade. Fortuitously, we have learned that C. crescentus has a significantly different LPS lipid A composition (35). The result is that it induces a much reduced endotoxin response, as judged by tumor necrosis factor alpha (TNF-α) induction. In addition, in the course of whole-cell immunization projects, we have often injected high levels of live caulobacters into mice with no detectable endotoxin effects (3). Thus, we expect them to be compatible with exposure to human mucosal tissues (considering that E. coli and other enteric Gram-negative bacteria with potent LPS endotoxin are frequent inhabitants of urogenital tissues).

As described here, additive effects can occur with display of combinations of HIV blocking agents. Caulobacters expressing CD4, in addition to protein G, and then loaded with blocking antibodies can perform better than the separate, soluble blocking entities. In other studies, caulobacters displaying CD4 alone and MIP1α (the ligand for the HIV coreceptor CCR5) alone have been developed and shown to reduce infectivity of HIV (27), and there was enhanced activity when combined. We can imagine that additional additive effects can be found in dense codisplay on caulobacters, using combinations of blocking antibodies and other HIV blocking agents. In the case of antibodies, the amplified effectiveness achieved with caulobacter display may assist the case for use of antibodies as microbicides, despite their high cost. To allow for the analysis of multiple antibodies, in this study, caulobacter-expressed protein G was used to capture antibodies of interest. However, for use in vivo, these results suggest that the conversion of selected antibodies to scFvs and genetic insertion into the S-layer of caulobacter cells for direct antibody display represent a novel means to inhibit HIV.

Additional challenges remain, such as concern about the stability of the S-layer and antibody/antigen binding characteristics in the human vagina, where a pH of 4 can be expected in healthy women. The S-layer requires a pH of 3 or less for removal from the cell surface (39), and we anticipate chemical cross-linking of the bacteria (and its S-layer) to be part of microbicide formulation, Nevertheless, the ability to stably display high levels of scFvs on the S-layer and still retain strong specific binding remains an important and significant challenge for microbicide development using the caulobacter S-layer display system.