|Home | About | Journals | Submit | Contact Us | Français|
Vaginal mucosal microfloras are typically dominated by Gram-positive Lactobacillus species, and colonization of vaginal mucosa by exogenous microbicide-secreting Lactobacillus strains has been proposed as a means of enhancing this natural mucosal barrier against human immunodeficiency virus (HIV) infection. We asked whether an alternative strategy could be utilized whereby anti-HIV molecules are expressed within the cervicovaginal milieu by endogenous vaginal Lactobacillus populations which have been engineered in situ via transduction. In this study, we therefore investigated the feasibility of utilizing transduction for the expression of two HIV coreceptor antagonists, the CC chemokines CCL5 and CCL3, in a predominant vaginal Lactobacillus species, Lactobacillus gasseri. Modifying a previously established transduction model, which utilizes L. gasseri ADH and its prophage Φadh, we show that mitomycin C induction of L. gasseri ADH transformants containing pGK12-based plasmids with CCL5 and CCL3 expression and secretion cassettes (under the control of promoters P6 and P59, respectively) and a 232-bp Φadh cos site fragment results in the production of transducing particles which contain 8 to 9 copies of concatemeric plasmid DNA. High-frequency transduction for these particles (almost 6 orders of magnitude greater than that for pGK12 alone) was observed, and transductants were found to contain recircularized expression plasmids upon subsequent culture. Importantly, transductants produced CC chemokines at levels comparable to those produced by electroporation-derived transformants. Our findings therefore lend support to the potential use of transduction in vaginal Lactobacillus species as a novel strategy for the prevention of HIV infection across mucosal membranes.
In sub-Saharan Africa, HIV infections are acquired predominantly via heterosexual contact and women are at greatest risk of being infected, accounting for 60% of HIV infections (27). Currently, there is no effective vaccine against HIV, and therefore, the development of topical microbicides for the prevention of viral entry at the cervicovaginal and rectal mucosal surfaces could serve as a potentially important alternative means of preventing HIV infection via vaginal and possibly rectal intercourse. To eliminate the requirement for precoital application of microbicides, an alternative live-microbicide strategy whereby nonpathogenic bacteria capable of colonizing the genital or gut mucosa are engineered to secrete HIV inhibitors has been investigated (13, 3, 20). Vaginal mucosal microfloras are typically dominated by Gram-positive Lactobacillus species, usually L. crispatus, L. jensenii, L. gasseri, and L. iners, which serve as an important natural barrier to HIV infection (1, 29, 17, 8). The colonization of the vaginal mucosa by engineered microbicide-secreting Lactobacillus strains would therefore potentially provide an economical and long-lasting method of enhancing this natural mucosal barrier. However, the persistence of such engineered strains within the vaginal mucosal milieu may require that these strains exhibit a selective advantage over endogenous bacterial populations, which could lead to potentially undesirable perturbations of the host's existing microflora.
We asked whether an alternative and potentially less disruptive strategy than the introduction of exogenous strains into the cervicovaginal mucosa could be utilized, whereby anti-HIV molecules are expressed within the cervicovaginal milieu by endogenous vaginal Lactobacillus populations which have been engineered in situ via bacteriophage-mediated transfer of plasmid DNA (transduction). Transduction is a well-established phenomenon for a variety of bacterial species and has been used as a convenient and often more reliable method of DNA transfer than conventional methods (e.g., electroporation) for Gram-positive bacteria such as Staphylococcus species (4). Thus, the introduction of transducing phage particles specific for resident Lactobacillus species into the cervicovaginal milieu, subsequent transduction, and the secretion of antiviral molecules would negate the need for the introduction of exogenous engineered bacteria, which would have to compete with resident microflora.
Few studies to date, however, have investigated transduction in Lactobacillus (26, 23, 21), and to our knowledge, there has been no attempt as yet to utilize transduction as a means of transferring expression plasmids into Lactobacillus species for the expression and secretion of anti-HIV molecules. Therefore, in this study, we undertook a proof-of-concept investigation to determine the feasibility of using transduction for this purpose. We show that high-frequency transduction into L. gasseri ADH by transducing particles derived from the Φadh phage can be achieved, where transducing particles contain pGK12-based expression plasmids (in concatemeric form) and transduction results in the production of transformants that express and secrete the CC chemokines and HIV coreceptor antagonists CCL5 (RANTES) and CCL3 (30).
L. gasseri ADH (NCK99) and L. gasseri NCK102, a strain cured of Φadh, were routinely propagated in de Man-Rogosa-Sharpe (MRS) broth (Oxoid, United Kingdom) at 37°C. NCK374, an L. gasseri ADH strain harboring the plasmid pTRK170 (23), was propagated as described above in MRS broth containing 7 μg/ml chloramphenicol. The Escherichia coli strain NCK240 containing the plasmid pGK12 was propagated in Luria-Bertani (LB) broth (BD) with 7 μg/ml chloramphenicol at 37°C. All of the above-mentioned strains were from the culture collection of the Department of Food, Bioprocessing and Nutrition Sciences, North Carolina State University. The plasmid pBR322 was purchased from Fermentas, Canada, while transformations were carried out using supercompetent E. coli XL1-Blue cells per the instructions of the supplier (Stratagene). Transformants containing pBR322-based constructs were selected on and propagated in LB plates and LB broth, respectively, supplemented with 0.1 mg/ml ampicillin, while transformants containing pGK12-based constructs were selected on and propagated in LB plates and LB broth containing 2 μg/ml chloramphenicol and 50 μg/ml erythromycin or 7 μg/ml chloramphenicol alone at 37°C. All L. gasseri ADH electroporation-derived transformants were selected on and propagated in MRS plates and MRS broth containing 9 μg/ml chloramphenicol at 37°C. The plasmid pSPS9 was purchased from the American Type Culture Collection (ATCC).
The strong Lactobacillus promoter P6 (5) was amplified from the DNA of a commercially available, over-the-counter L. acidophilus probiotic strain (Biomox Pharmaceuticals, South Africa), from which DNA was extracted using a ZR fungal/bacterial DNA kit (Zymo Research), by high-fidelity PCR with primers based on the P6 sequence (5) and previously designed by Russell and Klaenhammer (24), with the following modifications: an internal HindIII site was removed and a ClaI restriction site was included in the forward primer, and a HindIII restriction site was included in the reverse primer (Table (Table1).1). The resultant Pfu (Promega)-amplified P6 was restricted with ClaI and HindIII (Fermentas, Canada) and inserted into ClaI- and HindIII-restricted pBR322, and ampicillin-resistant transformants were selected. After being checked by sequencing using a pBR322-specific primer (designated pBR322 Seq) (Table (Table1),1), P6-containing pBR322 constructs were restricted with HindIII and EcoRV (Fermentas, Lithuania) and ligated to a synthetically synthesized, HindIII-restricted, 105-bp Usp45 secretion signal and propeptide leader sequence-containing oligonucleotide (designated USP45L) (Table (Table1)1) (19). USP45L contained a 5′ HindIII site for sequential ligation into P6-pBR322 constructs. Sense and antisense strands of USP45L were purchased (from Integrated DNA Technologies) and annealed prior to restriction. P6- and secretion signal-containing pBR322 constructs were checked by sequencing. To obtain ccl5 and ccl3 gene inserts, RNA from phytohemagglutinin (PHA)-stimulated peripheral blood mononuclear cells (isolated and stimulated as described previously ) was extracted using a QIAamp RNA blood kit (Qiagen) and converted to cDNA by using the SuperScript III first-strand synthesis for reverse transcriptase-PCR (RT-PCR) system per the instructions of the manufacturer (Invitrogen). Genes (ccl5 and ccl3) were then amplified from the resultant cDNA by high-fidelity PCR using Pfu and gene-specific primers (Table (Table1).1). Reverse primers for each gene contained terminal (3′-end) ClaI restriction sites. Pfu-amplified CC chemokine genes were blunt-end ligated into EcoRV-restricted, P6- and secretion signal-containing pBR322 constructs. Directional PCR and sequencing were used to confirm the correct orientation and sequences of chemokine gene inserts. P6-secretion signal-chemokine gene cassettes were then excised from pBR322 by ClaI restriction and inserted into pGK12, which was restricted with HpaII (Promega).
CCL3 expression in L. gasseri ADH was achieved by replacement of P6 with the lactococcal promoter P59 (28). To this end, P6-secretion signal-CCL3 gene-containing pBR322 constructs were digested with EcoRI and HindIII (Fermentas, Lithuania) to excise the P6 promoter and the resultant P6-lacking constructs were religated with P59. P59 was obtained from the shuttle vector pHPS9 (7) by high-fidelity PCR using an EcoRI and ClaI restriction site-containing forward primer and a HindIII restriction site-containing reverse primer (Table (Table1).1). P59-secretion signal-CCL3 gene constructs were restricted with ClaI and inserted into HpaII-restricted pGK12 as described above for P6 expression cassettes.
pTRK170 was amplified by high-fidelity PCR using Pfu with pGK12-specific primers (Table (Table1).1). The resultant 6.6-kb product was recircularized by blunt-end ligation with P6-secretion signal-CCL5 gene or P59-secretion signal-CCL3 gene expression cassettes, which were amplified from pBR322 constructs using pBR322-specific primers flanking the EcoRI and EcoRV sites (Table (Table1).1). Amplified expression cassettes were phosphorylated using T4 polynucleotide kinase (Promega) prior to ligation, and the resultant ligation mixture was cleaned, concentrated, and used directly for electroporation as pTRK170-based constructs did not yield stable E. coli transformants.
A 232-bp PCR product derived from Φadh DNA and encompassing the phage cos site was obtained using primers with BclI restriction sites (Table (Table1).1). After digestion with BclI (Promega), this cos fragment was inserted into the BclI site of pGK12. Directional PCR was used to assess the orientation of the cos fragment relative to the chloramphenicol resistance (Cmr) gene. Cos fragment-containing pGK12 was then restricted with HpaII and ligated to ClaI-restricted expression cassettes, as described above. Transformants were selected as described above, and plasmid minipreps (prepared using a Zyppy plasmid miniprep kit [Zymo]) were used for electroporation experiments.
Introduction of 0.5 to 1 μg of plasmid DNA into L. gasseri ADH by electroporation was performed using a Gene Pulser system (Bio-Rad) as described previously (2). After electroporation, 0.2-μl samples were plated onto MRS plates supplemented with 9 μg/ml chloramphenicol and cultured anaerobically at 37°C for 48 h.
Expression cassette-containing L. gasseri ADH transformants were cultured in MRS broth with 9 μg/ml chloramphenicol for 18 h, and CCL5 and CCL3 levels in conditioned broth passed through a 0.2-μm-pore-size filter were quantified by using an enzyme-linked immunosorbent assay (ELISA) kit (DuoSet) per the instructions of the manufacturer (R&D Systems).
L. gasseri ADH transformants containing pGK12 alone, pTRK170 (NCK374), pTRK170-expression cassette constructs, pGK12 with the Φadh cos site fragment, and pGK12-Φadh cos site-expression cassette constructs were cultured in MRS broth (with 9 μg/ml chloramphenicol) to an optical density at 590 nm of 0.3 to 0.4, and cultures were then diluted to an optical density of 0.1 (final volume, 30 ml). Mitomycin C was added to obtain a final concentration of 0.3 μg/ml. Cultures were maintained at 37°C for up to 18 h, after which 0.9 g of NaCl (~0.5 M final concentration) was added and samples were centrifuged for 30 min at 3,500 × g. The resultant supernatants were filtered through 0.45-μm filters, and polyethylene glycol 8000 was added to obtain a final concentration of 10% (wt/vol). After incubation on ice for 1.5 h, samples were centrifuged at 25,000 × g for 1 h and crude phage pellets were resuspended in 0.5 ml MT buffer (10 mM MgSO4, 50 mM Tris-HCl [pH 7.5]). Numbers of PFU per milliliter for individual samples were obtained from plaque counts by using the L. gasseri strain NCK102, cured of Φadh, as described previously (22).
Phage suspensions produced as described above were treated with DNase I (3 U) at 37°C for 40 min, diluted 1/4,000 to 1/80,000, and used for transduction in CaCl2-treated L. gasseri ADH cultures as described previously (23). After 48 h of anaerobic culture on MRS plates with 9 μg/ml chloramphenicol at 37°C, colonies were counted to determine transduction frequency and some were selected for overnight broth culture to determine levels of chemokine production by transductants.
Plasmid extraction from L. gasseri ADH transductants was performed as described previously (16).
Transducing-particle suspensions (0.2 ml) prepared as described above were subjected to phenol-chloroform extraction and ethanol precipitation as described previously (25). The resultant DNA was HpaII restricted for 8 h, and fragments were separated on a 0.3% agarose gel. DNA was blotted onto a Hybond-N+ nylon membrane (GEC Healthcare) as described previously (25) and probed with digoxigenin (DIG)-labeled linearized pGK12 by using DIG high-prime DNA labeling and detection starter kit II per the instructions of the manufacturer (Roche, Germany).
Transducing-particle suspensions were DNase I treated as described above, and 10 μl of each sample was subjected to extraction with 100 μl Prepman Ultra solution (Applied Biosystems). Real-time quantitative PCR analyses of appropriately diluted samples were then performed with an Applied Biosystems 7500 real-time PCR system using pGK12-specific primers designed to amplify a 123-bp fragment of the pGK12 Cmr gene (Table (Table1)1) and Maxima SYBR green quantitative PCR master mix per the instructions of the manufacturer (Fermentas, Canada). The number of moles of pGK12 per microliter of each sample was determined using a pGK12 standard curve (the molecular weight of pGK12 was determined to be 2,904,000), and the value was multiplied by Avogadro's number to give the number of copies of plasmid per microliter. Based on the Southern blot results, the number of copies of plasmid per concatemer was estimated for each construct, and thus, the number of transducing particles per microliter was determined by dividing the former value by the latter estimate.
All experiments were performed at least in triplicate. Values are presented as averages ± standard deviations (SD; n = 3).
Plasmids based on the rolling circle replication shuttle vector pGK12 (9) were constructed for the expression and secretion of the CC chemokines CCL5 and CCL3 in L. gasseri ADH. CC chemokines were chosen for this study due to their small size and ability to prevent HIV coreceptor binding (30, 15). pGK12 contains a broad-host-range origin of replication (pWV01), functional in E. coli and Lactobacillus species, and a Cmr gene for antibiotic selection of transformed cells or cells with transduced material (9). Therefore, by standard molecular techniques, expression-secretion cassettes consisting of a constitutive promoter, a secretion signal, and a chemokine gene were constructed using the scaffold plasmid pBR322. Cassettes were then inserted into the unique HpaII site of pGK12, yielding plasmids with the following components: (i) a broad-host-range origin of replication (pWV01), (ii) a Cmr gene, (iii) the constitutive promoter P6 or P59 (5, 28), (iv) a Usp45 secretion signal from Lactococcus lactis followed by a sequence encoding a 5-amino-acid propeptide (DTNSD), shown previously to increase secretion efficiency (19), downstream of the promoter, and (v) a chemokine (CCL5 or CCL3) gene inserted downstream of the DTNSD sequence (Fig. (Fig.11).
We obtained stable L. gasseri ADH transformants for pGK12 constructs containing a P6-secretion signal-CCL5 expression cassette; however, no stable transformants were obtained for P6-secretion signal-CCL3 expression constructs, and we surmised that this result was likely to be due to chemokine toxicity because of overproduction under the control of the P6 promoter. Replacement of P6 with the significantly weaker promoter P59 allowed for the production of stable CCL3-expressing transformants with predictably lower levels of chemokine expression and secretion than transformants expressing P6 constructs. Chemokine levels in overnight broth cultures, as determined by ELISA, indicated that P6-CCL5 gene transformants produced 24,679 pg/ml of CCL5 (SD, ±5,412 pg/ml; n = 3) and that P59-CCL3 gene transformants yielded 181 pg/ml of CCL3 (SD, ±10.7 pg/ml; n = 3). CCL5 production was therefore almost 140 times greater than CCL3 production.
Having established the functionality of P6-secretion signal-CCL5 gene and P59-secretion signal-CCL3 gene expression cassettes in L. gasseri ADH, the synthesis of transducing particles from plasmid constructs and their transduction frequencies were investigated next. Since it has been established previously that transduction frequency is greatly enhanced by using plasmids that contain restriction fragments of the transducing phage genome (23), two strategies were adopted for the incorporation of Φadh restriction fragments into the pGK12-based, chemokine expression-secretion plasmids synthesized as described above. In strategy 1, pTRK170, a plasmid derived by the insertion of two Φadh BglII fragments (0.8 and 1.4 kb) into the unique BclI site of pGK12 and shown previously to yield high-frequency transduction (23), was amplified by high-fidelity PCR and the resultant 6.6-kb product was recircularized by blunt-end ligation with the P6-secretion signal-CCL5 gene or P59-secretion signal-CCL3 gene expression cassette. In strategy 2, a 232-bp PCR product derived from Φadh DNA and encompassing the phage cos site was inserted into the BclI restriction site of pGK12 (Fig. (Fig.1).1). ClaI-restricted CCL5 and CCL3 expression cassettes were then inserted into the unique HpaII restriction sites of the cos site-containing pGK12 constructs. In preliminary experiments, we observed that the orientation of the cos site fragment had little effect on transduction frequency (data not shown), and therefore, for convenience we utilized plasmids containing the 232-bp Φadh cos site fragment in the opposite orientation from the pGK12 Cmr gene. Having constructed the above-described plasmids, we used electroporation to create L. gasseri ADH transformants 1 to 5 with the following respective constructs: construct 1, pTRK170 containing the P6-secretion signal-CCL5 gene cassette; construct 2, pTRK170 with the P59-secretion signal-CCL3 gene cassette; construct 3, pGK12 containing the 232-bp Φadh cos site fragment and the P6-secretion signal-CCL5 gene cassette; construct 4, pGK12 containing the Φadh cos site fragment and the P59-secretion signal-CCL3 gene cassette; and construct 5, pGK12 as a negative control. Transducing phage particles were then prepared from broth cultures of transformants 1 to 5 by induction with 0.3 μg/ml mitomycin C, and phage particles were used for transduction into L. gasseri ADH.
We obtained stable transductants for all constructs used, with those derived from cos site fragment-containing pGK12 constructs (constructs 3 and 4) producing chemokine levels almost the same as or higher than levels of chemokine production by electroporation-derived transformants (with ca. 5% decrease for CCL5 and 23% increase for CCL3) (Table (Table2).2). Furthermore, transductants from constructs 3 and 4 subcultured daily over 7 days exhibited stable chemokine production at levels comparable to those in 18-h cultures (21,780 pg/ml [SD, ±2,169 pg/ml; n = 3] and 216.3 pg/ml [SD, ±11.4 pg/ml; n = 3] for constructs 3 and 4 in day 7 subcultures, respectively). However, transductants derived from pTRK170-based constructs yielded lower chemokine levels. Transductants from construct 1-derived particles exhibited an approximately 24% decrease in CCL5 levels and construct 2-derived transductants exhibited a 74% decrease in CCL3 levels in comparison to electroporation-derived transformant chemokine levels (Table (Table2).2). Due to its poor functionality, construct 2 was excluded from further study. Next, we investigated the transduction frequencies for transducing particles derived from transformants 1, 3, and 4. These results exhibited a trend similar to that described above, with pTRK170-based construct 1 exhibiting almost 100 and 80% lower transduction frequencies than cos site fragment-containing constructs 3 and 4, respectively, but only an approximately 3% difference in transduction frequency from pTRK170 alone (construct 7) (Table (Table2),2), indicating that PCR amplification of the expression cassette and insertion into pTRK170 had little effect on the transduction frequency.
As the DNA compositions of pTRK170-derived transducing particles have been characterized previously and shown to consist of plasmid-phage DNA cointegrates (23), we undertook an investigation of the DNA compositions of transducing particles derived from cos site fragment-containing pGK12 constructs. We speculated that as observed previously (14), the inclusion of a cos site within pGK12 would yield particles with linear multimers or concatemeric DNA. Southern blots of HpaII-restricted and unrestricted DNA samples from transducing particles derived from construct 4 were analyzed and probed with DIG-labeled pGK12. The promoter P59 has an internal HpaII site, and therefore, if concatemeric, construct 4-derived phage DNA would yield a single major band of approximately 5.1 kb upon restriction. We found that unrestricted, construct 4-derived phage DNA (from non-DNase-treated samples) yielded a high-molecular-weight pGK12-hybridizing band corresponding to the 43-kb Φadh wild-type DNA sequence (Fig. (Fig.2A,2A, lane 2, and B, lane 8). This was also true for construct 5 (data not shown). Upon restriction, a major hybridizing band corresponding to 5 kb on the DNA ladder was observed (Fig. (Fig.2A,2A, lane 1), indicating that this high-molecular-weight pGK12-hybridizing band was indeed a concatemer and therefore consisted of approximately 8 copies of construct DNA per transducing particle. The restriction of construct 1-derived phage DNA with HpaII yielded a major hybridizing band smaller than the pGK12 band, consistent with the finding that pTRK170-derived particles contain phage-plasmid DNA cointegrates (Fig. (Fig.2A,2A, lanes 3 and 4). Upon investigation, transductants from constructs 3 and 4 were found to contain plasmid DNA corresponding to or just above 5 kb, indicating the recircularization of concatemeric DNA after transduction (Fig. (Fig.33).
Transduction frequency, as traditionally determined by the number of transductants relative to PFU (for wild-type phage), does not take into account the number of transducing particles per sample. We wished to investigate whether the observed differences in transduction frequency between constructs 3 and 4 (Table (Table2)2) could be due to the various numbers of transducing particles per sample. Therefore, having determined that cos fragment construct-derived particles are concatemeric, we were able to use real-time quantitative PCR to obtain the numbers of plasmid copies per milliliter for DNase-treated transducing particles derived from constructs 3 and 4 and the construct consisting of pGK12 and the cos fragment alone (construct 6) and estimate the number of transducing particles per milliliter for each sample by dividing the number of plasmid copies per milliliter by the number of plasmid copies per concatemer. The latter value was estimated to be 8.2, 8.5, and 9.4 for constructs 3, 4, and 6, respectively, based on the size of the Φadh genome (6) and the size of each expression plasmid. As expected, the number of transduction particles per milliliter for construct 3 (and thus the multiplicity of infection [MOI]) was almost 30% higher than that for construct 4 (Table (Table3),3), partly explaining the observed 40% increase in the number of transductants per milliliter for construct 3 compared to that for construct 4 (Table (Table2).2). However, interestingly, we observed an almost 14% increase in the number of transductants per milliliter for construct 6 in comparison to that for construct 3 although construct 6 corresponded to 16% fewer transducing particles per milliliter than construct 3 (Table (Table3).3). In addition, an approximately 15% increase in the number of transducing particles per milliliter for construct 6 compared to that for construct 4 (Table (Table3)3) yielded a 50% increase in the number of transductants per milliliter for construct 6 compared to that for construct 4 (Table (Table2).2). These findings, although not statistically significant, did suggest that for cos site-derived constructs, transduction frequency was affected by both different numbers of transducing particles per sample and the presence/type of an expression cassette within the plasmid. Unfortunately, it was not possible to use the above-described approach for pTRK170-derived constructs, as the number of plasmid copies per transducing particle is difficult to determine due to the nonconcatemeric composition of the particles.
In this study, we investigated the feasibility of utilizing transduction with pGK12-based expression plasmids and the lysogenic phage Φadh for the expression of anti-HIV molecules (specifically the CC chemokines CCL5 and CCL3) by L. gasseri ADH. Although much higher transduction frequencies (0.1 to 1) for another Lactobacillus transduction model, which utilizes L. delbrueckii and the lytic phage LL-H, have been reported previously (21), we chose the former model due to the extensive characterization of L. gasseri ADH and its prophage (Φadh) (22, 6) and, importantly, because L. gasseri has been found to be a predominant Lactobacillus species in the vaginal mucosa (1, 29, 17). We have successfully shown that by using expression plasmids that contain a fragment which encompasses the phage cos site, transducing particles that contain 8 to 9 copies of plasmid DNA in concatemeric form can be synthesized. Transduction frequency for these particles is almost 6 orders of magnitude greater than that for pGK12 alone (Table (Table2),2), and importantly, transductants were found to stably produce chemokines at levels comparable to those produced by electroporation-derived transformants and contained recircularized expression plasmids upon subsequent culture.
We found that transduction frequencies for cos fragment constructs were decreased by the insertion of expression cassettes, but this effect does not appear to be related to the size of the insert since, with adjustment for differences in MOI, construct 6 (cos fragment-containing pGK12 alone) yielded 27 and 40% increases in transductants in comparison to constructs 3 and 4, respectively, although the sizes of these constructs (5.36 and 5.13 kb for constructs 3 and 4, respectively) differ by only 4%. Therefore, it is likely that the observed decrease in transductants for constructs 3 and 4 in comparison to those for construct 6 was due to chemokine toxicity, and given the observed differences in promoter strengths and thus chemokine expression (P6 [CCL5] P59 [CCL3]), CCL3 appears to have significantly greater bactericidal activity than CCL5. This conclusion is supported by our inability to obtain electroporation-derived P6-CCL3 gene transformants. Interestingly, we observed no bactericidal effect when L. gasseri ADH was exposed to recombinant CCL3 or CCL3-L1 (up to 50 μg/ml) according to the method of Yang et al. (31) (data not shown). Therefore, it is likely that the observed toxicity of CCL3 was due to endogenous effects rather than antibacterial defensin-like activity, which has been shown previously for a variety of other chemokines (31).
A previous transduction study by Raya and Klaenhammer utilizing L. gasseri ADH and Φadh established that a pGK12 plasmid with a 7.9-kb cos site-containing phage fragment (termed pTRK177) exhibits a lower transduction frequency (40%) than pTRK170 (23). We observed the opposite effect: a significant increase in transduction frequency for construct 6 over pTRK170. This apparently contradictory observation may be a consequence of increased structural instability of pTRK177 in comparison to construct 6 since rolling circle plasmids, like pGK12, exhibit structural instability with increasing size (usually exceeding 8 kb) (18) and pTRK177 is almost 3 times larger than construct 6 (12.3 versus 4.63 kb). It is therefore fortuitous that we undertook a parallel investigation with cos site-containing plasmids, as this added advantage of smaller size potentially allows for the expression of larger genes in the absence of increased plasmid instability.
Interestingly, we observed that transduction frequencies and protein expression levels for plasmids based on the pTRK170 plasmid were poor in comparison to those for cos fragment-containing plasmids. The reasons for this observation are likely multiple and may include potentially lower numbers of transducing particles relative to the total number of phage-forming units for the former and potentially lower numbers of functional plasmids introduced per cell for the latter. The observation that transductants obtained from pTRK170 expression cassettes with the weaker promoter, P59, yielded significantly lower protein levels than electroporation-derived transformants supports the latter hypothesis. A recombination event has been shown to occur for pTRK170, resulting in phage DNA-plasmid multimer cointegrates' being packaged into transducing particles (23). While it has been shown previously that pTRK170 transductants do contain recircularized pTRK170 (23), the efficiency of this process (excision from phage DNA and recircularization into functional plasmids) may be lower than that for concatemeric pWV01-based plasmids (e.g., pGK12), which ordinarily form long, single-stranded intermediates during replication (12) and may therefore rapidly recircularize into functional plasmids after transduction.
In conclusion, we believe that the results of this investigation lend credible support to the use of transduction as a means of genetically manipulating endogenous vaginal Lactobacillus populations for the expression of anti-HIV molecules within the vaginal mucosal milieu and potentially other mucosal environments. The feasibility of this approach as a future, viable therapeutic strategy is naturally dependent on a number of complex issues, such as the stability of transducing particles, their efficient delivery to lactobacilli within the mucosal environment, and importantly, the host range of the transducing particles used, which require extensive further investigation. Preliminary studies indicate that Φadh has a limited host range, producing plaques in only 1 of 13 clinical isolates tested thus far (T. R. Klaenhammer, unpublished data). Therefore, the discovery of phages with broader host ranges is critical for the success of this strategy.
Advantages of this approach, if it can be successfully implemented in more complex models, however, are numerous and include negation of the need for the introduction of exogenous bacteria, which would have to compete with resident microflora, as well as the potential for the addition of cocktails of phage particles containing multiple expression plasmids, which would allow for the inducible or constitutive simultaneous expression of multiple natural or artificial anti-HIV molecules. Bacteriophages are also stable, compatible with simple storage and formulation methods, inexpensive to produce, and suitable for large-scale production and have an established safety record for use in humans (11). Therefore, if successful, this expression system could potentially lead to development of an easily implementable and economical strategy for the prevention of HIV infection across mucosal membranes.
This work was funded by a grant (no. 51825) from the Bill & Melinda Gates Foundation (to L.H.D.) through the Grand Challenges Explorations Initiative.
We thank Maria Paximadis and Anabela Picton from the AIDS Unit, National Institute for Communicable Diseases, Johannesburg, South Africa, for valuable advice and technical assistance with sequencing.
Published ahead of print on 23 April 2010.