|Home | About | Journals | Submit | Contact Us | Français|
Women are at significant risk of heterosexually transmitted human immunodeficiency virus (HIV) infection, with the mucosal epithelium of the cervix and vagina serving as a major portal of entry. The cervicovaginal mucosa naturally harbors dynamic microflora composed predominantly of lactobacilli, which may be genetically modified to serve as a more efficient protective barrier against the heterosexual transmission of HIV. We selected a vaginal strain of Lactobacillus, L. jensenii 1153, for genetic modification to display surface-anchored anti-HIV proteins. Genomic sequencing analyses revealed that L. jensenii 1153 encodes several unique high-molecular-weight cell wall-anchored proteins with a C-terminal cell wall sorting LPQTG motif. In this report, we employed these proteins to express a surface-anchored two-domain CD4 (2D CD4) molecule in L. jensenii 1153. Our studies indicated that the C-terminal cell wall sorting signal LPQTG motif alone is insufficient to drive the surface expression of heterologous proteins, and the display of surface-anchored 2D CD4 molecules required native sequences of a defined length upstream of the unique C-terminal LPQTG cell wall sorting signal and the positively charged C terminus in a Lactobacillus-based expression system. The modified L. jensenii strain displayed 2D CD4 molecules that were uniformly distributed on bacterial surfaces. The surface-anchored 2D CD4 molecule was recognized by a conformation-dependent anti-CD4 antibody, suggesting that the expressed proteins adopted a native conformation. The establishment of this Lactobacillus-based surface expression system, with potential broad applicability, represents a major step toward developing an inexpensive yet durable approach to topical microbicides for the mitigation of heterosexual transmission of HIV and other mucosally transmitted viral pathogens.
The predominant mode of human immunodeficiency virus type 1 (HIV-1) transmission worldwide is via heterosexual contact (38). Women are particularly at risk for HIV infection, as the efficiency of HIV transmission from male to female is greater than that from female to male (1). There are few means by which women can actively protect themselves against HIV infection, particularly in the absence of a protective vaccine or the inability to negotiate condom use. The need to develop new methods of HIV prevention that are controlled by women is urgently recognized by health organizations worldwide.
The cervicovaginal mucosa is the main site of HIV entry in women. While the exact cell type and site of transmission are being actively investigated (7, 17, 42, 50), it is believed that a potential microbicide product will be most effective if it offers broad protection against the mucosal transmission of HIV. In healthy women of childbearing age, the protective mucosa in the vagina is populated with commensal bacteria typically dominated by H2O2-producing lactobacilli. The dominance of lactobacilli over pathogenic anaerobes is positively associated with vaginal health (35). The principal Lactobacillus species isolated from the vaginal mucosa of healthy women are L. jensenii, L. crispatus, L. gasseri, and L. iners (2, 48, 51). These species are efficient colonizers of the vaginal mucosa and likely exist as a natural “biofilm” composed of bacteria and extracellular matrix materials (15). The depletion or disturbance of vaginal Lactobacillus flora has been associated with the establishment of opportunistic infections in the urinary tract, bacterial vaginosis, and increased risk of acquiring HIV and herpes simplex virus type 2 in women (11, 41, 44). Thus, vaginal lactobacilli play a critical role in the maintenance of reproductive health in women.
Through genetic engineering, a member of the vaginal microflora may be enhanced to form an efficient protective shield against the transmission of sexually transmitted diseases such as HIV. Our novel approach involves genetically modifying a natural human isolate of H2O2-producing lactobacilli to express the first two domains of the high-affinity HIV-binding protein, human CD4 (39). CD4, a member of the immunoglobulin (Ig) superfamily, is the primary host receptor for HIV entry into susceptible cells. The extracellular portion of CD4 (residues 1 to 371) is a concatenation of four Ig-like domains, D1 to D4. The two N-terminal domains, two-domain (2D) CD4 (K1-S183) (39), encode and properly fold to form the gp120 binding epitope (4) and, when expressed in the absence of the remaining domains of CD4, retain the high-affinity binding to HIV-1 gp120 (40). Given that human CD4 is an endogenous protein in the human immune system, it has less potential to have immunogenic properties when expressed on the mucosal surface in vivo. Importantly, glycosylation of CD4 is not required for binding to gp120 (25). Soluble 2D CD4 that adopts a native disulfide-bonded conformation has been expressed in a number of well-established systems including L. jensenii (6, 10, 12).
A biologically active CD4 molecule surface expressed by L. jensenii could potentially trap viruses at the bacterial surface, thus impeding the access of viruses to underlying epithelial cells and lymphocyte targets. These trapped viruses may be rendered unstable (28), undergo an aborted infection process, and/or be inactivated locally by antiviral compounds, such as lactic acid and hydrogen peroxide, secreted by the lactobacilli (21), thereby significantly reducing the number of infectious viral particles.
The surface expression of heterologous proteins has been achieved via the sortase-catalyzed cell wall-anchoring mechanism in gram-positive bacteria (30) including Streptococcus gordonii, Lactobacillus paracasei, and Staphylococcus carnosus (5, 18, 23, 31, 43). While the genetic manipulation of two human vaginal isolates of L. fermentum and L. jensenii has been reported (10, 33), this is the first report on the surface expression of a heterologous mammalian protein in human vaginal isolates of L. jensenii. In this report, we describe a broadly applicable approach for the genetic modification of lactobacilli enabling the expression of 2D CD4 molecules covalently linked to peptidoglycan in the cell wall of L. jensenii 1153. The surface-anchored 2D CD4 molecule adopted a native conformation, recognizing two conformation-dependent anti-CD4 antibodies (Leu3a and Sim.4). We demonstrated that a native cell wall sorting signal alone was insufficient to drive the surface expression of 2D CD4 and required fusion with native upstream sequences of a defined length. The approach reported here also affords the surface expression of heterologous proteins in other Lactobacillus species.
Naturally occurring human vaginal isolates of Lactobacillus, including L. jensenii 1153, L. jensenii Xna (10, 27), L. gasseri 1151, and L. casei Q were routinely cultivated at 37°C and 5% CO2 in either MRS broth or Rogosa SL broth (Difco) as described previously (10). A nonvaginal isolate, L. paracasei 343 (from Peter H. Pouwels), was cultured similarly. Shuttle plasmids were constructed and maintained in Escherichia coli cells, purified, and electroporated into lactobacilli (10). Transformed lactobacilli were routinely propagated either on MRS agar plates or in liquid medium containing 20 μg/ml erythromycin (10).
The genome sequence of L. jensenii 1153 was determined as described previously (27). Cell wall-anchored proteins of gram-positive bacteria share several common features that enable them to be covalently anchored to the cell wall peptidoglycan (16). Among them, they have a conserved C-terminal LPXTG motif, followed by a hydrophobic stretch of amino acids and a short charged tail, which are collectively called the cell wall sorting signal (16). Other motifs, including LPXTA in L. paracasei (19), have been identified. Accordingly, a computer script was written to identify motifs similar to LPXTG and LPXTA in all reading frames of the assembled contigs of the partially sequenced L. jensenii 1153 genome. We used independent PCR amplification and manual sequencing to verify the contigs containing putative cell wall anchor motifs. For anchor sequence C370, a forward primer, 5′-ATGTTCTATCAAATTGACCCAGCTTTGG-3′, was used, in pair with the reverse primer 5′-CCTGCGCCTAATGCCATCAATCCAATA-3′, to PCR amplify the 5,712-bp coding region. The sequences were also subjected to a BLAST search for sequence homology to cell wall-anchored proteins in gram-positive bacteria.
To facilitate protein surface anchoring in L. jensenii, an expression cassette was constructed and subcloned into the SacI and XbaI sites of a shuttle vector, pOSEL175 (27). The expression cassette contains four components, including a Lactobacillus-compatible P23 promoter, a CbsA signal sequence (10), a DNA sequence encoding a heterologous protein, and cell wall-anchoring domains from known or putative cell surface proteins in gram-positive bacteria. 2D CD4 (K1-S183) was synthesized and subcloned as previously described (10). Unique restriction sites, including SacI, EcoRI, NheI, MfeI, and XbaI, were placed between each component from the 5′ to 3′ ends, respectively. Amplification of each component was performed using conventional PCR with Pfu DNA polymerase.
Various lengths of repetitive sequence upstream of the LPQTG motif in the C370 sequence were amplified with flanking MfeI and XbaI restriction sites from the genomic DNA of L. jensenii 1153. The same reverse primer (5′-CCGTCTAGATTATGCTTCATCATCTTTTCT-3′ [the restriction site is underlined]) was used in pair with the forward primers listed in Table Table11 for each PCR. In addition, the forward primer 5′-GCGCAATTGAAGAAGGCAGAAGAAGT-3′ (the restriction site is underlined) was paired with the reverse primer described above to amplify the nucleotide sequences corresponding to the region containing the C-terminal LPQTG domain and their upstream 200 amino acids (CWA200) in the C370 sequence that contains the last 2.5 repeats (Fig. (Fig.1A1A).
To drive the surface expression of 2D CD4 molecules, the protein coding sequence of 2D CD4 was C terminally fused to sequences of a defined length upstream of the LPQTG motif in the C370 sequence. The DNA fragments resulting from the above-described PCRs were digested with both MfeI and XbaI. The resulting fragments were ligated with MfeI/XbaI-double-digested pOSEL651 (10). The resulting plasmids were designated pOSEL262 (with no repeat), pOSEL268 (with one repeat), pOSEL278 (with two repeats), pOSEL280 (with four repeats), pOSEL281 (with seven repeats), and pOSEL276 (with eight repeats). Similarly, the MfeI/XbaI-digested fragment corresponding to the CWA200 sequence upstream of the LPQTG motif in C370 was cloned, yielding pOSEL249 (with 2.5 repeats).
Oligonucleotide primers corresponding to c-Myc epitope EQKLISEEDL were designed (forward primer 5′-GCGCTAGCGAACAGAAACTGATCTCCGAAGAGGACCTGTTGAAGAAGGCAGAAGAAGT-3′ and reverse primer 5′-CCGCAATTGTTATGCTTCATCATCTTTTCT-3′ [the restriction sites are underlined]), allowing the fusion of the c-Myc epitope to the N terminus of sequences containing the C-terminal cell wall sorting signal and upstream CWA200 in the C370 sequence. The amplified PCR fragments were digested with both MfeI and NheI and then subcloned into MfeI/NheI-double-digested pOSEL651, resulting in pOSEL241.
A series of deletion mutants was generated by PCR amplification. An oligonucleotide complementary to the 2D CD4 sequence in pOSEL249 (5′-GATCGTGCTGATTCACGTCGT-3′) was used as a forward primer. Reverse primers listed in Table Table22 were used to delete the positively charged amino acids from the C terminus of the C370 sequence. All reverse primers contained an XbaI restriction site. The amplified PCR fragments were digested with both MfeI and XbaI and then subcloned into MfeI/XbaI-double-digested pOSEL249. The clones were verified by nucleotide sequencing, and the constructs were electroporated into L. jensenii cells for protein analysis.
Point mutations were generated using a QuikChange XL site-directed mutagenesis kit (Stratagene, La Jolla, CA). Plasmid pOSEL249 (containing the protein coding sequence of 2D CD4 and the C-terminal LPQTG domain and their upstream CWA200 of the C370 sequence) was used as a template. The mutagenic primers were designed based on preferred codon usage in L. jensenii and the nucleotide sequences corresponding to LPQTG and its flanking sequences in C370 (Table (Table2).2). After the PCR, the DpnI enzyme was added to the amplification mixture to degrade the methylated parental plasmids. Newly synthesized plasmids were introduced into chemically competent E. coli Top10 cells (Invitrogen, Carlsbad, CA) in LB broth supplemented with 200 μg/ml erythromycin. Plasmids were subsequently isolated for DNA sequencing (Biotech Core, Mountain View, CA) to identify clones with the desired mutations. Plasmids with verified sequences were electroporated into L. jensenii 1153 cells.
Bacterial cultures, at late log phase or close to stationary phase, containing 109 cells were centrifuged at 12,000 × g for 5 min. The resulting cell pellets were washed once in 20 mM HEPES (pH 7.2) and suspended in 100 μl of a solution containing 10 mM Tris-HCl, (pH 8.0), 1 mM EDTA, and 25% sucrose (34). The bacterial cell wall was digested in the presence of a muramidase, mutanolysin (Sigma Chemical Co., St. Louis, MO), at a final concentration of 15 units/ml for 1 h at 37°C. Afterward, the cells were centrifuged at 2,500 × g for 10 min to isolate the cell wall-enriched fraction from the protoplast-enriched one (34). The resulting samples were heat denatured after the addition of 25 μl of 4× and 125 μl of 1× reducing sodium dodecyl sulfate (SDS)-polyacrylamide gel electrophoresis (PAGE) buffer to the cell wall- and protoplast-enriched fractions, respectively.
The modified lactobacilli were grown at 37°C and 5% CO2 in MRS or Rogosa SL broth buffered with 100 mM HEPES (pH 7.4) to late log phase or early stationary phase. Cell-free supernatants for evaluation of soluble proteins were collected by centrifugation (12,000 × g for 10 min), and proteins were heat denatured in the SDS-PAGE loading buffer (50 mM Tris-HCl [pH 6.8], 10 mM dithiothreitol, 0.4% SDS, 6% sucrose, 0.01% bromophenol blue). Afterwards, soluble proteins were resolved and detected, as described previously (10), with the rabbit polyclonal anti-CD4 antibody T4-4 (NIH AIDS Reagents Reference Program), the mouse monoclonal anti-c-Myc antibody (Invitrogen), or the rabbit polyclonal anti-C370 antibody OI9 (developed at AbboMax, Inc., San Jose, CA). The antigen-antibody reaction was then visualized by using horseradish peroxidase-conjugated secondary antibody and enhanced chemiluminescence reagents (GE Healthcare Life Sciences, Piscataway, NJ).
Cultures of L. jensenii 1153 harboring plasmids designed for heterologous protein expression or control plasmid pOSEL175 grown overnight were subcultured at 1:50 dilutions in erythromycin-containing MRS or Rogosa SL broth that was buffered with 100 mM HEPES (pH 7.4). Bacteria at early log phase were washed and suspended in phosphate-buffered saline (PBS) containing 2% fetal bovine serum (FBS). About 2 × 108 cells were probed with rabbit anti-CD4 antibody T4-4 or rabbit anti-C370 anchor protein antibody OI9 at 1:3,000 dilutions in PBS-2% FBS at 4°C for 30 min. After washes in PBS containing 2% FBS, the bacteria were then probed with Alexa Fluor 488 goat anti-rabbit IgG (Molecular Probes, Eugene, OR) at 1:200 dilutions at 4°C for 30 min. Alternatively, modified L. jensenii cultures were probed with anti-CD4 monoclonal antibody (MAb) Sim.4 or Leu3a at a final concentration of 6 μg/ml or with anti-c-Myc MAb (Invitrogen), followed by probing with 10 μl of neat fluorescein isothiocyanate (FITC)-conjugated goat anti-mouse IgG (Becton Dickinson, San Jose, CA). The anti-CD4 MAbs Sim.4 and Leu3a recognize conformational epitopes in the HIV-1 gp120-binding site (domain 1 of CD4) (39). The fluorescence of 20,000 labeled cells in triplicate samples was analyzed using a FACScan system (Becton Dickinson) running with CellQuest software. Density plot output (side scatter or forward scatter versus fluorescence) was obtained from modified L. jensenii cultures, with those harboring plasmid pOSEL175 as a background control. The shift in mean fluorescence intensities between the plots was taken as a measure of antibody binding to bacterial surface and calculated using FLOWJO software (Tree Star, Inc., Ashland, OR).
For the quantitation of CD4 molecules expressed on the modified L. jensenii surface, anti-CD4 MAb Leu3a directly conjugated with phycoerythrin (PE) (Pharmingen, San Diego, CA) was used to label bacteria and Quantum Simply Cellular (QSC) beads (Bangs Laboratories, Inc., Fishers, IN). The QSC beads contain a blank bead and four populations of microbeads with various capacities to bind mouse monoclonal IgG. Quantum R-PE medium reference beads (Bangs Laboratories, Inc.) were also used for the quantitation of the fluorescence intensity of a sample in terms of the number of molecules of equivalent soluble fluorochrome using QuickCal Sample Report software. The modified bacteria were grown to an optical density at 600 nm of about 0.4, harvested, and washed with PBS-1% FBS. The bacteria and the QSC beads were incubated with 10 μl of anti-CD4 MAb conjugated with PE for 1 h and then washed three times with PBS-1% FBS. Finally, the bacteria and both QSC and QR-PE beads were visualized using a FACScan system (Becton Dickinson).
The modified L. jensenii cultures harboring pOSEL175 or plasmids designed for surface expression were cultured to early log phase in Rogosa broth as described above. Approximately 1.4 × 108 bacteria were washed with PBS-1% FBS and resuspended in 50 μl of buffer. Bacteria were incubated directly with 20 μl of Leu3a-FITC (BD Biosciences, San Jose, CA) for 20 min, washed three times with PBS-1% FBS, and visualized. Alternatively, bacteria were incubated with rabbit anti-CD4 antibody T4-4 at 1:100 dilutions for 20 min. After three washes in PBS-1% FBS, the bacteria were incubated with goat anti-rabbit IgG conjugated with FITC (Jackson ImmunoResearch Laboratories, Inc., West Grove, PA) or anti-rabbit Rhodamine Red X (Jackson ImmunoResearch Laboratories) at a 1:100 dilution and incubated for 20 min. After three washes, the labeled cells were visualized. In dual antibody binding experiments, bacteria were first incubated with MAb Leu3A-FITC and then incubated with polyclonal antibody (PAb) T4-4 and finally incubated with anti-rabbit Rhodamine Red X prior to visualization. Images were collected using a Leica TCS-NT/SP confocal microscope (Leica Microsystems, Mannheim, Germany). Standard filters were used for FITC and Texas Red. Rhodamine Red X was detected in the same channel as Texas Red. The detector slits for the FITC channel were configured to collect between 500 and 547 nm to minimize cross talk between the green and yellow channels. All compared images were detected using the same gain. In dual antibody binding experiments, images were collected separately and then superimposed. For all negative controls, differential interference contrast images were collected simultaneously with the fluorescence images using the transmitted light detector. Composed images were organized using Adobe Photoshop (Adobe Systems, Inc., San Jose, CA).
The nucleotide sequence reported in this paper has been submitted to the GenBank/EMBL data bank under accession number EU332140.
In order to optimize the expression of surface proteins, we sought to identify native surface anchors from vaginal isolates of Lactobacillus in our collection, focusing on a H2O2-producing L. jensenii strain, 1153, that is a fast grower and amenable to genetic manipulation (28). Accordingly, genomic sequencing was conducted with 484 contigs assembled from the sequence reads that cover approximately 75% of the genome. Analyses of this partial genomic sequence database led to the identification of more than 30 contigs containing putative LPQTG sequences. A more detailed sequence homology search was performed with those in the databases of the National Center for Biotechnology. Six protein-coding open reading frames were selected, including a protein sequence tentatively assigned as C370 (Fig. (Fig.1)1) that contained structural features that are commonly present in known cell wall-anchored proteins (30). A BLAST search showed that the C370 protein sequence was most similar to the EF0109 protein in a vancomycin-resistant Enterococcus faecalis strain (32) and shared moderate homology to some other cell wall-associated proteins in gram-positive bacteria. C370 contains eight nearly identical tandem repeats (approximately 72 amino acids per repeat unit) and a cell wall sorting signal (17), including an LPQTG motif preceding a hydrophobic region and a positively charged C-terminal tail (Fig. (Fig.1),1), which are features commonly seen in cell wall-anchored proteins. Indeed, antiserum raised against CWA200 of C370 (contains 2.5 repeats) (Fig. (Fig.1)1) readily bound to L. jensenii 1153 based on flow cytometric analysis (data not shown), further suggesting that the C370 protein is displayed on the cell surface.
It was unknown whether the cell wall sorting signal alone in the C370 sequence would be sufficient to support the efficient surface expression of 2D CD4 molecules in L. jensenii or whether an additional upstream sequence is required to maximize surface protein display. Accordingly, several constructs were engineered, with the protein coding sequence of 2D CD4 C terminally fused to the C370 cell wall sorting signal sequence and the upstream sequences with various lengths. pOSEL262, pOSEL268, pOSEL278, pOSEL249, pOSEL280, pOSEL281, and pOSEL276 contain 0, 1, 2, 2.5, 4, 7, and 8 repeats of the C370 sequence, respectively. To determine whether 2D CD4 was expressed on the bacterial surface, flow cytometric analysis was used (Fig. (Fig.2).2). The bacterial cells harboring parental plasmid pOSEL175 had minimal binding of anti-CD4 PAb T4-4 (Fig. (Fig.2A).2A). The level of anti-CD4 PAb T4-4 bound to bacterial cells harboring pOSEL262 (zero repeats) was indistinguishable from levels of those harboring the negative control pOSEL175. Consistent with these observations, the surface expression of 2D CD4 molecules was not detected when the cell wall sorting signal of a similar length from Streptococcus pyogenes or L. paracasei was employed (data not shown). These data suggested that the 36-amino-acid cell wall sorting signal alone in the C370 sequence, although native, was not sufficient to display 2D CD4 molecules on the cell surface. In contrast, a significant increase in fluorescence intensity was detected when the number of repeats was increased to two copies in pOSEL278. The fluorescence intensity of each construct continued to increase up to seven repeats in the bacteria harboring pOSEL281. To determine the level of 2D CD4 molecules that adopt a correctly folded conformation, the transformed bacteria were probed with anti-CD4 MAb Sim.4 for flow cytometric analysis (Fig. (Fig.2B).2B). Again, the mean fluorescence intensity in bacteria harboring pOSEL262 (0 repeat) was indistinguishable from that in the negative control pOSEL175. There was a significant increase in fluorescence intensity in bacteria harboring pOSEL249 (2.5 repeats, ~200 amino acids in length). The employment of a similar length of nonrepetitive sequences upstream of the native cell wall sorting signal also enabled the surface display of 2D CD4 molecules in L. jensenii 1153 (data not shown). The insertion of additional repeats did not yield a significant increase in fluorescence intensity. Additional flow cytometric analyses were performed when the modified lactobacilli harboring above-mentioned plasmids were probed with anti-CD4 MAb Leu3a that also recognizes a conformation-dependent epitope in CD4. These studies confirmed the above-described findings (data not shown). These data suggested that the sequence of a minimally defined length upstream of LPQTG is required for the surface display of 2D CD4 molecules.
The cell wall-anchored proteins in gram-positive bacteria contain a characteristic stretch of positively charged amino acids at the extreme C terminus (30). These signature charged sequences serve as a critical cell surface retention signal (30). It was unknown whether the positively charged amino acids KKKRKDDEA1903 present in the C370 sequence (Fig. (Fig.1B)1B) serve the same function. Therefore, a series of deletions were constructed, resulting in pOSEL249-8, pOSEL249-9, and pOSEL249-10 (Fig. (Fig.3).3). The modified L. jensenii cultures harboring these C-terminal deletion constructs were probed with anti-CD4 PAb T4-4 or MAb Sim.4 for detection by flow cytometric analysis. A marked decrease in the mean fluorescence intensity in the bacterial cells harboring these mutant plasmids was observed relative to that of cells harboring pOSEL249 (Fig. 3B and C) as a result of the abolished surface display of 2D CD4 molecules. Furthermore, a 48-kDa protein of 2D CD4 in fusion with the CWA200 sequence was detected by Western analyses using anti-CD4 PAb T4-4 in cell-free conditioned medium of all of the modified L. jensenii cultures harboring C-terminal charged tail deletion constructs, suggesting that the tail-deleted fusion proteins were abundantly released instead of being anchored on cell surfaces (data not shown).
It was previously reported that replacing proline (P) with asparagine (N) in the LPETG cell wall sorting motif in protein A of Staphylococcus aureus decreased the efficiency of protein surface display (30). Conversely, replacing threonine (T) with serine (S) has little effect (30). To determine whether the LPQTG motif within C370 is indeed the critical sorting signal, the importance of P, T, and G within the LPQTG sequence was investigated. Point mutations were generated by PCR within the LPQTG motif in the C370 sequence. The P residue was mutated to alanine (A) or asparagine (N); T was mutated to A, S, or glycine (G); and G in the LPQTG motif was mutated to A. Plasmids with the altered LPQTG motif were designated pOSEL249(P→A), pOSEL249(P→N), pOSEL249(T→A), pOSEL249(T→S), pOSEL249(T→G), and pOSEL249(G→A). To further determine the effect of the mutagenesis of LPQTG on L. jensenii surface protein display, the L. jensenii strains harboring pOSEL175 (empty vector, negative control), pOSEL651 (no LPQTG motif, negative control), and pOSEL249 (wild-type LPQTG motif, positive control), along with the various mutants, were probed with anti-CD4 PAb T4-4 or MAb Sim.4, followed by the appropriate fluorochrome-conjugated secondary antibody. Flow cytometric analysis of antibody binding in modified bacteria detected a substantial decrease in the mean fluorescence intensity in bacterial cells harboring pOSEL249(P→A) (with the LAQTG motif) and pOSEL249(P→N) (with the LNQTG motif) compared to that of cells harboring pOSEL249 (Fig. (Fig.4),4), indicating that there was a significant reduction in the amount 2D CD4 protein displayed on the cell surface. However, the mean fluorescence intensities in the bacterial cells harboring pOSEL249(T→S) (with the LPQSG motif) and pOSEL249(T→A) (with the LPQAG motif) were comparable to those of cells harboring the wild type, demonstrating that replacing T with S or A has little effect on the efficiency of cell wall anchoring (Fig. (Fig.4).4). The mean fluorescence intensity in the bacterial cells harboring pOSEL249(G→A) (with the LPQTA motif) decreased to about 40% relative to those harboring the wild type. Taken together, the modulation of surface-displayed 2D CD4 molecules, as a result of either amino acid substitutions in the LPQTG motif or deletions in the C-terminally-charged tail of the C370 sequence, reflected the behavior of a native surface protein in L. jensenii 1153.
To determine if the native C370 anchor sequences of L. jensenii 1153 could function in protein surface display in other human lactobacilli or L. jensenii strains, pOSEL241 was constructed as a fusion of the c-Myc epitope tag to the CWA200 and the cell wall sorting signal of the C370 sequence (Fig. (Fig.5A).5A). pOSEL241 or negative control plasmid pOSEL175 was introduced into L. jensenii Xna, L. gasseri 1151, and L. casei Q by electroporation. The transformed bacteria were analyzed by Western and flow cytometric analyses and compared to positive control L. jensenii 1153 harboring pOSEL241. All of the transformed lactobacilli were digested with mutanolysin, an N-acetyl muramidase that cuts the β1-4-glycosidic bond between MurNAc and GlcNAc of the glycan strands in mature peptidoglycan. Cell wall-anchored proteins typically migrate as a large spectrum of fragments following muramidase treatment and SDS-PAGE separation (30). Western analyses of cell wall digests followed by probing with anti-c-Myc MAb detected laddering patterns, consistent with the pattern of anchored proteins when digested with mutanolysin (34). This pattern was found in transformed L. jensenii Xna cells and L. gasseri 1151 cells harboring pOSEL241 and was similar to those in L. jensenii 1153 and, to a lesser extent, in L. casei Q (Fig. (Fig.5B).5B). We also detected an increase in fluorescence intensity in L. jensenii Xna cells and L. gasseri 1151 cells harboring pOSEL241, indicating that the c-Myc epitope was displayed on the cell surface of these bacteria (Fig. (Fig.5C).5C). When analyzed by flow cytometry following immunostaining of the bacterial cells with anti-c-Myc MAb, a very low level of fluorescence was detected in all Lactobacillus species harboring pOSEL175. There was approximately a 19-fold increase in the fluorescence intensity of L. casei Q harboring pOSEL241 relative to that of L. casei Q harboring pOSEL175 (Fig. (Fig.5B),5B), indicating that c-Myc-tagged protein was being surface anchored albeit at a much lower level than that in other strains. In summary, the native anchor sequence of L. jensenii 1153 clearly exhibits a broad utility for displaying proteins in various Lactobacillus species of human origin, although there appears to be a species preference for sorting signals.
In order to properly engage trimeric HIV-1 gp120 in a virion, we believe that the 2D CD4 molecules expressed on the Lactobacillus surface needed to be available in sufficient numbers and evenly distributed in the correct conformation. Since the native anchor sequence C370 surface protein had never been described and had only limited homology to other surface proteins, we knew nothing of its surface topology. Therefore, we needed to determine the pattern of distribution of 2D CD4 molecules on the bacterial cell surface. When engineered L. jensenii cells harboring pOSEL249 (2D CD4-CWA200, cell wall sorting signal of the C370 sequence) were probed with anti-CD4 PAb T4-4, the fluorescence detected by flow cytometry increased dose dependently with the amount of available T4-4 antibody, indicating an even distribution of the 2D CD4 protein displayed on the cell surface (Fig. (Fig.6A).6A). In addition, both anti-CD4 PAb (which detects both native and linear epitopes of CD4) T4-4 and MAb Leu3a (detects a native conformational epitope in the gp120-binding site) were used to visualize the three-dimensional image reconstruction by confocal laser scanning microscopy. The modified L. jensenii cells, upon probing with antibody, were optically sectioned with images collected every 0.2 μm through the cell wall (z stack). Examination of composite images revealed a very uniform distribution of the 2D CD4 protein on the cell surface (Fig. (Fig.6B).6B). Based on the colocalization of anti-CD4 PAb T4-4 and MAb Leu3a in these composite images, the surface-expressed 2D CD4 molecules adopt a native conformation (Fig. (Fig.6B6B).
In addition, we used quantitative flow cytometric analysis to determine the numbers of anchored molecules on the bacterial surface in reference to two series of microbead standards with known numbers of surface molecules. In one set, the microbeads contained a mixture of bead populations, which have various capacities to bind mouse MAb, and reacted with MAb Leu3a under the same conditions as those for the bacterial cells. The second set of microbeads was directly conjugated to PE. The results showed that there were approximately 1,250 correctly folded 2D CD4 molecules per bacterium harboring pOSEL249 (data not shown).
The vaginal mucosal epithelium is a major natural route of HIV entry into underlying target cells. Numerous avenues to curtail the HIV epidemic are currently being investigated (9, 26, 27, 37, 42, 47, 49). While an effective HIV vaccine remains the most important goal, the ability of HIV to mutate rapidly and evade the immune response has made the development of a vaccine problematic, necessitating the pursuit of additional innovative approaches. Topical microbicides represent a possible approach to inhibit the transmission of sexually transmitted disease pathogens, including HIV, in women (42, 46). Here, we propose a unique mode of impeding the transmission of HIV on vaginal and other mucosal membranes. This approach involves the engineering of human vaginal lactobacilli to express proteins for intercepting infectious viral particles and reintroducing the lactobacilli to recolonize the vaginal mucosa. The success of this approach depends largely on whether human vaginal isolates of lactobacilli are amenable to genetic manipulation for the surface display of heterologous virus-binding proteins and whether the modified bacteria can recolonize.
The surface expression of proteins via covalent linkage with peptidoglycans in gram-positive bacteria involves unique sorting signals. One of the best-studied systems is the emm6 gene of S. pyogenes that encodes the M6 structural protein (20). Since the universal cell wall sorting motifs for a variety of gram-positive bacterial species have been identified (30), we initially attempted a plasmid-based modular approach to anchor heterologous proteins on the surface of L. jensenii cells by utilizing two well-characterized cell wall sorting signals from either the PrtP protease of L. paracasei, the M6 protein (emm6) of S. pyogenes, or the cell wall sorting signal from the M6 protein of S. pyogenes plus an upstream 100-amino-acid region (CWA100) (14). The inefficient levels of expression using these heterologous systems necessitated the development of a surface expression system employing cell wall anchor sequences native to vaginal lactobacilli. Interestingly, the LPQTG sorting motif, which accounts for only 7% of the LPXTG motifs found in gram-positive bacteria (30), is present in almost all putative cell wall-anchored proteins identified in L. jensenii 1153. Recently, the same LPQTG motif was also identified in L. reuteri (36), L. plantarum (22), and L. gasseri (NCBI Microbial Genomes Annotation Project). We previously attempted the surface expression of several proteins in Lactobacillus using cell wall sorting signals of other gram-positive bacteria but with only limited success. The presence of unique cell wall sorting signals that do not match those in M6 of S. pyogenes and PrtP of L. paracasei (Fig. (Fig.1B)1B) suggests that regions of native sequences should be exploited to covalently anchor heterologous peptides and proteins to the cell surface of L. jensenii.
It was apparent from our analysis that although there is flexibility in amino acids at the T or G position of the native LPQTG motif, there are additional sequence requirements for efficient surface display. First, we discovered that L. jensenii requires ~120 to 200 amino acids of native cell wall anchor sequences upstream of the C-terminal cell wall-anchoring motif to display 2D CD4 molecules. These upstream sequences may facilitate the retention or extension of substrate sequence and thus proper proteolytic cleavage by membrane-associated sortase. Second, we showed that the positively charged C terminus is essential for the retention of 2D CD4 molecules on the bacterial surface. By employment of these sequences, we were able to achieve the surface display of 2D CD4 molecules not only in L. jensenii 1153 but also in L. zeae (data not shown), L. jensenii Xna, and L. gasseri 1151, which indicates a high level of conservation in cell wall-anchoring motifs and machinery among these lactobacilli. In our analysis, the surface expression of 2D CD4 molecules by modified L. jensenii cells was sustained when bacteria were cultured in broth at broad pH ranges or at different growth phases (data not shown).
Despite the successful demonstration of the surface expression of protypical 2D CD4 molecules in this report, we believe that the surface expression of 2D CD4 can be further optimized to increase the density of 2D CD4 expression on the cell surface by the employment of strong L. jensenii promoters (27) and/or to increase the steric accessibility or surface extension of cell wall-anchored molecules with enhanced affinity to bind gp120. Furthermore, there are other promising protein candidates that could more efficiently bind and neutralize HIV within the mucosa. These include CD4-antibody fusion proteins (13) or variants of CD4 capable of forming oligomers that exhibit an enhanced avidity for binding primary isolates of HIV-1 (24). In a recent article by Arthos et al., a dodecameric CD4-Ig fusion was constructed using a polyvalent antibody backbone to display D1 and D2 of CD4. This multimeric form of CD4 showed potent anti-HIV activity in vitro (3). In a follow-up paper, the dodecameric CD4-Ig, now termed D1D2-IgP, was shown by cryoelectron tomography to induce the virion rupture of simian immunodeficiency virus upon binding to gp120 (28). Although the mechanism of membrane rupture was not elucidated, McKeating et al. suggested that the geometric constraints of the gp120 bound to CD4 could cause a destabilization of the virion. Lactobacillus surface-anchored viral binding proteins may trap, immobilize, and destabilize virions on the bacterial surface, thus decreasing the number of infectious viral particles. Since the cervicovaginal transmission of HIV is already an inefficient process, a reduction in the infectious viral load should further reduce transmission frequencies (8).
Our current proof-of-concept experiments involve the use of modified L. jensenii cells harboring plasmids that contain antibiotic resistance markers. Since these plasmids might be transferred to opportunistic human pathogens such as staphylococci and enterococci (45), the plasmid-based experimental approach is clearly undesirable for clinical analysis. As an important step toward the development of engineered L. jensenii strains with the desired genes stably maintained, we identified sequences for site-specific chromosome integration by homologous recombination (27) and succeeded at stably integrating the gene encoding 2D CD4 into the L. jensenii chromosome for the surface anchoring of 2D CD4 molecules. The resulting strain had a 40% to 50% reduction in the surface expression of 2D CD4 molecules, while the erythromycin resistance gene was excised (data not shown).
An important element in this concept is whether the surface display of heterologous proteins in L. jensenii adversely affects the persistence, colonization, and competitive advantages of the engineered lactobacilli on the vaginal mucosa. While this remains to be tested in vivo, evidence that modified gram-positive bacteria or exogenously applied lactobacilli can colonize and persist in relevant animal models already exists. Recombinant strains of S. gordonii and L. zeae designed for secreting or surface displaying single-chain antibody or immunogen stably colonized oral or vaginal mucosa in rat or mouse models of bacterial colonization (5, 23, 29).
The likely purpose of the native vaginal lactobacilli is to protect the reproductive tract from pathogens that are introduced during intercourse. By subtle genetic modifications, such as the surface expression of an antiviral protein on the lactobacillus, the floras themselves can add an additional layer of protection against HIV-1 and possibly reduce its ability to infect the host.
We thank Jan-Fang Cheng at the Lawrence Berkeley National Laboratory for directing genomic sequencing of L. jensenii 1153, Ken Frank for the associated bioinformatics support in this work, Ernie Tai for technical assistance, and Yang Liu and Rosa Yu for critical reading of the manuscript. We also thank John Lewicki, Jim Turpin, John Mascola, and Peter Kwong for helpful discussions and Owen Schwartz of the Biological Image Facility, NIAID, for help with the confocal microscopy.
This work was supported in part by NIH Integrated Preclinical/Clinical Program for Topical Microbicides grant 5U19AI060615.
Published ahead of print on 21 April 2008.