Generation of a Dual-Target, Safe, Inexpensive Microbicide that Protects Against HIV-1 and HSV-2 Disease

HSV-2 infection is a significant health problem and a major co-morbidity factor for HIV-1 acquisition, increasing risk of infection 2–4 fold. Condom based prevention strategies for HSV-2 and HIV-1 have not been effective at stopping the HIV-1 pandemic, indicating that alternative prevention strategies need to be investigated. We have previously developed an inexpensive HIV-1 specific microbicide that utilizes the S-layer mediated display capabilities of Caulobacter crescentus, and have shown that recombinant C. crescentus displaying HIV entry blocking proteins are able to provide significant protection from HIV-1 infection in vitro. Here we demonstrate that recombinant C. crescentus are safe for topical application and describe 5 new recombinant C. crescentus that provide protection from HIV-1 infection in vitro. Further, we demonstrate protection from disease following intravaginal infection with HSV-2 in a murine model using C. crescentus expressing the anti-viral lectins Cyanovirin-N and Griffithsin, as well as α-1-antitrypsin and indolicidin. Interestingly, C. crescentus alone significantly reduced HSV-2 replication in vaginal lavage fluid. Protection from HSV-2 disease was strongly associated with early cytokine production in the vaginal tract. Our data support the potential for a dual-target microbicide that can protect against both HIV-1 and HSV-2, which could have an enormous impact on public health.


Results
Expression of Recombinant Proteins on the Surface of C. crescentus. Four proteins, elafin, BmKn2, α-1-antitrypsin and indolicidin, that could prevent HIV-1 attachment or entry were identified from the literature and expressed in the S-layer protein of C. crescentus as previously described [20][21][22]28 (Table 1). The sequence of each expression vector was confirmed before transformation into C. crescentus. A low pH extraction method, which has been a reliable method to monitor export, assembly and surface attachment of S-layer protein, and SDS-PAGE with Coomassie Brilliant Blue R stain or Western blot was used to assess the expression levels for the chimeric S-layer proteins. All constructs resulted in expression of substantial quantities of protein, and demonstrated small size shifts by protein gel analysis, which, when combined with the sequencing data, provides evidence that the recombinant protein is expressed within the S-layer (Fig. 1a). An indolicidin multimer, containing 2 tandem copies of indolicidin for each one RsaA protein was also constructed. Cc-Indo2 was successfully expressed but produced approximately half the amount of S-layer protein as Cc-Indo. Western blot analysis for C. crescentus RsaA protein indicated that each recombinant C. crescentus produced S-layer protein, providing further indication that the bacteria are producing the recombinant proteins (Fig. 1b). While there is variation in the amount of RsaA that is present for each recombinant compared to the Cc-Control, this is unlikely to have a great impact on the ability of each recombinant to prevent HIV-1 infection as a unmodified C. crescentus SCIentIfIC RePoRTS | (2018) 8:2786 | DOI: 10.1038/s41598-018-21134-1 expresses 40,000-60,000 copies of the RsaA protein on its' surface 26,31 , suggesting that even a 50% reduction in RsaA proteins will still lead to substantial expression of the recombinant protein, which is supported by our previous HIV-1 studies 20,22 . Protection from in vitro HIV-1 infection using S-layer display of Elafin, BmKn2, α-1-Antitrypsin and Indolicidin. We have previously demonstrated that 10 8 recombinant C. crescentus were sufficient to prevent HIV-1 infection using HIV-1 pseudotyped viruses 20,22 . To confirm that this number of bacteria were sufficient for protection from infection when replication competent HIV-1 was used, we performed a titration of C. crescentus using 10 6 -2 × 10 8 C. crescentus (Cc-Control, Cc-Elafin, Cc-BmKn2, Cc-A1AT and Cc-Indo) and HIV-1 89. 6 (Fig. 2). There was no protection from HIV-1 infection with Cc-Control or any of the recombinant C. crescentus when 10 6 , 10 7 or 5 × 10 7 C. crescentus were added. When 10 8 C. crescentus were added the 4 recombinant C. crescentus provided protection from HIV-1 89.6 infection whereas the Cc-Control did not. Although protection from HIV-1 89. 6 infection with all 4 recombinants did increase when 1.5 × 10 8 recombinant C. crescentus were added the Cc-Control also lowered HIV-1 infection at this dose, which suggests that at high doses C. crescentus may have some ability to prevent HIV-1 infection, and this was clearly evident when 2 × 10 8 recombinant C. crescentus was added, with no difference observed in HIV infection levels between Cc-Control and Cc-Elafin. As the Cc-Control is not predicted to interact with either HIV-1 or the TZM-bl cells, we hypothesize that the large number of C. crescentus present in such a small cell culture volume (200 μL) is creating a physical barrier, preventing HIV-1 and the TZM-bl cells from interacting. To avoid any impact of the C. crescentus bacterium itself on HIV-1 prevention 10 8 recombinant C. crescentus were used for subsequent experiments.
Recombinant C. crescentus displaying elafin, BmKn2, α-1-antitrypsin, indolicidin and an indolicidin multimer individually within the S-layer were incubated with HIV-1 89.6 for 1 hour before adding TZM-bl cells or  Cc-Indo2; 7) Cc-Control. The image has been brightened and cropped to minimize background. The original image has been provided to Scientific Reports and is available from the authors upon request. (b) Representative Western blot. RsaA was extracted from C. crescentus using a low pH method, normalized, run on 7.5% SDS-PAGE using a prestained protein ladder, and RsaA was detected by western blot using an anti-RsaA antibody. 1) Cc-Control (no insert); 2) Cc-A1AT; 3) Cc-BmKn2; 4) Cc-Elafin; 5) Cc-Indo; 6) Cc-Control.  (Table 2). There was no difference observed between Cc-Indo and Cc-Indo2 in TZM-bl cells. Although Cc-Indo2 contained two tandem copies of indolicidin per each RsaA monomer, the overall expression level of recombinant RsaA on Cc-Indo2 was less than the level of Cc-Indo, which may account for the similar levels of protection. In preliminary experiments, Cc-Indo and Cc-Indo2 provided complete protection of PBMCs from HIV-1 89.6 infection. Taken together these results suggest that each of these recombinant C. crescentus are good candidates for further microbicide development.
Immune response following vaginal application of C. crescentus. Previous work by our laboratory 45 and collaborators 24 suggested that C. crescentus is safe for inoculation. However, the mucosal response to C. crescentus has not been studied. We applied 10 8 Cc-Control, PBS or LPS isolated from Salmonella enterica serotype Minnesota to the vaginal tract of C57Bl/6 mice and collected vaginal lavage fluid from the mice 5 days prior to application, then 4, 24 and 48 hours post-inoculation. Vaginal lavage fluid was analyzed using a cytometric bead array kit for mouse inflammatory cytokines. There was no significant difference between PBS, Salmonella LPS and 10 8 Cc-Control for IL-10, IFNγ and IL-12p70 (Fig. 4a). Salmonella LPS induced an increase in MCP-1 and IL-6 production at 4 hours post-application compared to both PBS and Cc-Control, but there was no significant difference observed between PBS and Cc-Control (Fig. 4a). Although Cc-Control did cause a small increase in TNF at 4 hours post-application, this difference was not statistically significant compared to PBS and about a third of the amount of TNF that was produced after application of Salmonella LPS, indicating that C. crescentus does not induce inflammatory cytokines in a manner similar to pathogenic gram negative bacteria (Fig. 4a). We next investigated whether C. crescentus application initiated recruitment of immune cells to the vaginal tract. Female C57Bl/6 mice were inoculated intravaginally with PBS, Salmonella LPS or 10 8 Cc-Control. Four hours later vaginal tissue was excised, fixed and stained with hematoxylin and eosin. There was no evidence of immune cell infiltration to the vaginal tract at 4 hours post-application (Fig. 4b).
Finally, we investigated whether antibodies against C. crescentus would be produced in the vaginal tract after vaginal application of Cc-Control. As we anticipate that a C. crescentus microbicide would be used on a semi-regular basis, the mice were inoculated intravaginally bi-weekly for 9 weeks. Vaginal lavage fluid was collected and analyzed for the presence of antibodies against whole C. crescentus by ELISA, using an anti-RsaA antibody as a positive assay control. No IgG or IgM antibodies against C. crescentus could be detected in the vaginal fluids (Fig. 4c). Taken together, these results suggest that C. crescentus may be safe for vaginal application.
Cyanovirin-N and Griffithsin protect from HSV-2 disease in a murine model. C. crescentus expressing Cyanovirin-N, microvirin and Griffithsin were previously developed and shown to be able to provide significant protection from HIV-1 infection in vitro 22 . Based on published work with HSV-1 and HSV-2 we hypothesized that these recombinant C. crescentus would also provide protection from HSV-2 infection. To test this we used a murine model of HSV-2 infection. Progesterone treated C57Bl/6 mice were infected intravaginally with HSV-2 in the presence of Cc-Control, Cc-Cyanovirin, Cc-Microvirin or Cc-Griffithsin. All mice that received HSV-2 alone reached the humane endpoint by day 11 post-infection (Fig. 5a). Ten of twelve mice (83%) that were infected with HSV-2 in the presence of Cc-Control reached the humane endpoint by day 9 post-infection (Fig. 5a). Cc-Control application at the time of HSV-2 infection did not significantly enhance survival of the mice compared to HSV-2 alone. The remaining two mice developed mild symptoms of HSV-2 infection between days 6-14 post-infection, indicating that all mice that received Cc-Control were infected, with one of these mice reaching the humane endpoint at day 18. Cc-Griffithsin provided significant protection from HSV-2 mortality compared to both HSV-2 alone and HSV-2 + Cc-Control with 4 of 7 mice (57%) that did not reach the humane endpoint (Fig. 5a). Cc-Cyanovirin protected 3 of 7 mice (43%) from HSV-2 mortality although this was not statistically significant when compared to HSV-2 + Cc-Control (Fig. 5a). Cc-Microvirin provided no protection from HSV-2 infection, with 6 of 7 mice reaching the humane endpoint at the same time as mice that received HSV-2 alone (Fig. 5a). Cc-Griffithsin application at the time of HSV-2 infection led to a significant decrease in severity of HSV-2 symptoms when compared to both HSV-2 + Cc-Control and HSV-2 alone (Fig. 5b). Mice that received Cc-Cyanovirin had a statistically significant delay in developing symptoms of HSV-2 infection up to day 12 post-infection when compared to both HSV-2 alone and HSV-2 + Cc-Control (Fig. 5b). Cc-Microvirin did not have an impact on severity of HSV-2 symptoms (Fig. 5b). There was no statistically significant difference in the time to development or severity of HSV-2 symptoms between mice that received HSV-2 alone and Cc-Control (Fig. 5b).
Interestingly, when compared to HSV-2 alone, all recombinant C. crescentus tested, including Cc-Control and Cc-Microvirin, which did not prevent development of HSV-2 symptoms, provided a significant reduction of approximately 1 log in viral load in the vaginal lavage fluid at day 2 post-infection (Fig. 5c). Despite this significant decrease in HSV-2 replication, Cc-Control and Cc-Microvirin did not significantly increase survival or decrease herpetic scores in the mice when compared to those that received HSV-2 alone, suggesting that there may be some non-specific anti-viral effects of recombinant C. crescentus. This could potentially be due to the bacteria providing a physical barrier that makes it more difficult for HSV-2 to interact with epithelial cells to cause infection. While the same bacterial barrier would be present in mice that received Cc-Griffithsin or Cc-Cyanovirin, the additional ability of these recombinant proteins to prevent cell-to-cell spread of HSV-2 (Cc-Griffithsin) or HSV-1 entry and membrane fusion (Cc-Cyanovirin) 34,38 could account for the improved survival and lower herpetic scores observed in the mice that received these recombinant C. crescentus.

Prevention of in vivo HSV-2 disease by Cc-A1AT and Cc-Indo. All of the new inhibitors developed
for this paper were tested in vivo for protection against HSV-2 infection. Indolicidin has been demonstrated to reduce HSV-2 infectivity in vitro 43 and elafin has been demonstrated to have in vitro and in vivo anti-viral activity against HSV-2 45 . α-1-antitrypsin is a serpin antiprotease which are key players in the regulation of inflammatory responses 46 , so α-1-antitrypsin may be able to protect from HSV-2 infection by regulating inflammation in the vaginal tract. The exact anti-viral mechanism of BmKn2 has yet to be elucidated, and it was possible it may have broad anti-viral activity.
Cc-A1AT provided significantly improved survival following HSV-2 infection with 6 of 7 mice (86%) surviving challenge compared to both HSV-2 alone and to HSV-2 + Cc-Control (Fig. 6a). Cc-Indo protected 4 of 7 mice (57%) from HSV-2 mortality but this was not statistically significant (Fig. 6a). Cc-Elafin protected 3 of 7 mice (43%) from HSV-2, however this was not statistically significant (Fig. 6a). In preliminary experiments mice that received HSV-2 + Cc-BmKn2 rapidly reached the humane endpoint, at an earlier day post-infection than HSV-2 alone, indicating that Cc-BmKn2 was not a good candidate for HSV-2 prevention and additional experiments were not undertaken (Fig. 6a). There was a slight delay in the development of symptoms and a significant reduction in symptom severity with the mice that received HSV-2 + Cc-A1AT when compared to both HSV-2 alone and HSV-2 + Cc-Control (Fig. 6b). There was a significant decrease in severity of HSV-2 scores for mice that received Cc-Indo compared to HSV-2 alone, and on days 12-15 post-infection when compared to HSV-2 + Cc-Control (Fig. 6b). Although the disease score for mice receiving Cc-Elafin was lower, this was not statistically significant. When vaginal lavage fluid was analyzed for viral load, there was a significant reduction of approximately 1 log in viral load between HSV-2 alone and HSV-2 plus Cc-Control; Cc-Elafin; Cc-A1AT; and Cc-Indo (Fig. 6c). As all recombinant C. crescentus lowered viral load but not all prevented HSV-2 infection this suggests that there are non-specific anti-viral effects, and that those recombinants that protect from infection use an additional mechanism to provide protection. Based on previous studies, Cc-Indo is likely disrupting the viral membrane of HSV-2 to prevent infection 43,44 . However, Cc-A1AT was not anticipated to have direct anti-HSV-2 activity so further studies were undertaken to for further investigation.

Cc-A1AT produces early cytokine responses in the vaginal tract following HSV-2 infection.
Given that HSV-2 viral loads in the vaginal lavages were similar between the Cc-Control and each recombinant C. crescentus, but that not all C. crescentus improved survival and lowered HSV-2 symptom severity, we hypothesized that the C. crescentus themselves were providing non-specific protection by interfering with the ability of HSV-2 to interact with epithelial cells, similar to the effects observed with high amounts of bacteria added to in vitro HIV-1 assays (Fig. 2a). While proposed mechanisms of protection have been elucidated for Cc-Cyanovirin, Cc-Griffithsin and Cc-Indo, Cc-A1AT was not anticipated to have anti-HSV-2 ability as it interacts specifically with gp41 present on HIV-1. However, α-1-antitrypsin is a serpin antiprotease that can mediate inflammatory responses, so it is possible that the presence of α-1-antitrypsin at the time of HSV-2 infection could mediate a protective immune response. Preliminary in vitro viral blocking assays with Vero cells suggested that direct interaction with HSV-2 or target cells was an unlikely mechanism of action for Cc-A1AT as there was no impact on development of HSV-2 plaques with Cc-A1AT. We investigated the production of inflammatory cytokines following Cc-A1AT application in the presence or absence of HSV-2 to determine if the protection from HSV-2 infection was potentially immume mediated. Mice were inoculated intravaginally with HSV-2, Cc-Control, recombinant C. crescentus, HSV-2 + Cc-Control, or HSV-2 plus recombinant C. crescentus. Vaginal lavage fluid was collected 4 days prior to inoculation then at 4, 24 and 48 hours post-inoculation and analyzed by cytometric bead array for mouse inflammatory cytokines. Cc-A1AT was analyzed because it was the most successful at preventing HSV-2 disease, Cc-Griffithsin because it protected approximately half of the mice from HSV-2 disease, Cc-Control because it did not protect from HSV-2 infection and Cc-BmKn2 because it may enhance HSV-2 infection. At 24 hours post-infection, mice that received HSV-2 + Cc-A1AT had significantly increased (11 fold more) IFNγ in the vaginal lavage compared to mice that were infected with HSV-2 alone (Fig. 7a). There was 2.7 fold more IFNγ when mice received HSV-2 + Cc-A1AT compared to HSV-2 + Cc-Control, although this was not statistically significant. In comparison, although mice that received HSV-2 + Cc-Griffithsin had elevated IFNγ in the vaginal tract at 24 hours post-infection, this was not significantly greater than HSV-2 alone or HSV-2 + Cc-Control, and about a third the amount of IFNγ that the HSV-2 + Cc-A1AT mice produced (Fig. 7a). Furthermore, mice that received HSV-2 + Cc-BmKn2 or HSV-2 + Cc-Control had no increase in IFNγ in the vaginal tract at 24 hours post-infection (Fig. 7a). At 4 hours post-infection, all recombinant C. crescentus tested, including Cc-Control, caused a significant increase in the amount of TNF present in the vaginal lavage when given at the same time as HSV-2 (Fig. 7b). Cc-A1AT and Cc-Griffithsin produced similar levels of TNF. Although the levels of TNF produced by either HSV-2 + Cc-A1AT or HSV-2 + Cc-Griffithsin were not statistically significant compared to HSV-2 + Cc-Control, both of these recombinant C. crescentus led to production of a third more TNF than either HSV-2 + Cc-Control and HSV-2 + Cc-BmKn2, and ten times more TNF than HSV-2 alone. Furthermore, HSV-2 + Cc-A1AT produced significantly more IL-6 at 4 hours post-infection than HSV-2 alone (Fig. 7c). Although the comparison between HSV-2 + Cc-C-A1AT and HSV-2 + Cc-Control was not statistically significant, HSV-2 + Cc-A1AT produced 32% more IL-6 than HSV-2 + Cc-Control at this time point. HSV-2 + Cc-Griffithsin also produced significantly more IL-6 at 4 hours post-infection compared to virus alone, although this was about 20% less IL-6 than what was produced by HSV-2 + Cc-A1AT, and not statistically significant when compared to HSV-2 + Cc-Control (Fig. 7c). Although HSV-2 + Cc-BmKn2 did cause a slight elevation in IL-6 at 4 hours post-infection, this was not statistically significant (Fig. 7c). Cc-A1AT alone did not produce significantly more IFNγ, TNF or IL-6 in the vaginal tract compared to Cc-Control alone (Fig. 7). There was no significant difference between levels of IL-12p70, IL-10 and MCP-1 for HSV-2 compared to HSV-2 + Cc-A1AT, or HSV-2 + Cc-Griffithsin at any of the time points (Supplementary Figure 1). Taken together, these results suggested that Cc-A1AT could protect mice from HSV-2 infection by inducing an earlier cytokine response in the vaginal tract during HSV-2 infection, with IFNγ production 24 hours earlier, TNF about 20 hours earlier, and IL-6 produced 2 days earlier than HSV-2 alone. Although the comparisons between HSV-2 + Cc-A1AT and HSV-2 + Cc-Control for each of these time points were not statistically significant, we hypothesize that the increases in IFNγ (2.7 fold), TNF (1.4 fold) and IL-6 (1.4 fold) compared to HSV-2 + Cc-Control are biologically significant and contribute to the significant reduction in HSV-2 symptom severity and significant increase in survival observed in mice that received HSV-2 + Cc-A1AT compared to HSV-2 alone and HSV-2 + Cc-Control.

Discussion
In this paper we have continued our development of a C. crescentus based microbicide. We describe the creation of five new recombinant C. crescentus and demonstrate that these recombinant bacteria are able to provide 36-58% protection from HIV-1 infection in vitro. Furthermore, three of the anti-HIV C. crescentus that we previously published and the five new candidates presented here were tested for their ability to provide protection from HSV-2 disease. We found that C. crescentus expressing α-1-antitrypsin, Griffithsin, Cyanovirin-N and indolicidin are able to provide some protection of mice from vaginal HSV-2 disease. This seemed to be independent of viral load as C. crescentus alone was able to cause a significant reduction in HSV-2 in vaginal fluid. When we measured cytokine production in the vaginal fluid after inoculation with HSV-2 in the presence or absence of recombinant C. crescentus, we found that Cc-A1AT induced an earlier production of IFNγ, TNF and IL-6 compared to virus alone. In particular, the amount of IFNγ produced at 24 hours post-infection seemed to correlate with protection, with Cc-A1AT producing 15 times more IFNγ than virus alone and protecting 86% of mice, Cc-Griffithsin producing 10 times more IFNγ and protecting 57% of mice, Cc-Control providing no protection and producing 4 times more IFNγ and Cc-BmKn2 providing no protection and producing double the amount of IFNγ. Although the comparison between HSV-2 + Cc-Control and HSV-2 + Cc-A1AT was not statistically significant, there was 2.7 times more IFNγ, 1.4 times more TNF and 1.4 times more IL-6 produced with HSV-2 + Cc-A1AT. We propose that this increase in cytokine production is biologically significant and that this earlier and greater cytokine response could initiate an earlier recruitment of immune cells to the vaginal tract, which may be able to contain the viral infection before the virus has the chance to cause pathology, although further studies are warranted to investigate this. Although both HSV-2 + Cc-Control and HSV-2 + Cc-BmKn2 do lead to an increase in IFNγ, perhaps it is not enough to reach the threshold necessary to protect the mice from HSV-2 infection.
Our data has suggested that Cc-A1AT provides protection from HSV-2 disease by inducing an earlier immune response in the vaginal tract, possibly resulting in earlier immune cell recruitment to the vaginal tissue to control HSV-2 infection. This cytokine production could be problematic for HIV-1 prevention, as genital inflammation has been shown to increase HIV-1 acquisition [47][48][49][50][51][52][53] . In particular, in CAPRISA 002 increased levels of proinflammatory cytokines were associated with increased HIV-1 infection, and this finding was validated in the CAPRISA 004 clinical trial, with women having an elevation of three or more proinflammatory cytokines in the cervicovaginal lavage at a 3-fold increased risk of HIV-1 acquistion 54,55 . However, we do not anticipate this to be problematic with a C. crescentus based microbicide based on our data. First, when Cc-A1AT alone was applied to the vaginal tract of mice there was no production of inflammatory cytokines, indicating the recombinant bacterium alone does not induce inflammation. In addition, preliminary data investigating in vivo protection from vaginal HIV-1 infection in humanized bone marrow-liver-thymus (BLT) mice with Cc-A1AT indicated that protection from HIV-1 infection occurred (Farr Zuend et al., submitted). This suggests that it is the combination of Cc-A1AT plus HSV-2 that is responsible for the earlier cytokine response, and that in the absence of HSV-2 there is no concern for an increased immune response in the vaginal tract.
In these studies the recombinant C. crescentus was applied at the time of HSV-2 infection, which may not be feasible in a "real world" situation. We have undertaken preliminary studies in which Cc-A1AT or Cc-Control were applied to the vaginal tract 2 hours prior to HSV-2 infection and found that similar levels of protection from HSV-2 disease were maintained, with 100% of mice that received HSV-2 + Cc-Control reaching the humane endpoint and only 33% of mice that received HSV-2 + Cc-A1AT reaching the humane endpoint. Similarly, scores of HSV-2 infection were reduced by more than two-fold in mice that received HSV-2 + Cc-A1AT, indicating that recombinant C. crescentus do not need to be applied at the time of exposure for protection from HSV-2 disease.
Interestingly, C. crescentus alone had some ability to lower HIV-1 infection when used at high doses, and was able to provide a 1 log reduction in HSV-2 viral load in vaginal lavage. While we have hypothesized that this may be an indirect method of prevention, with the presence of large amounts of C. crescentus providing a non-specific barrier that can lower the ability of HIV-1 or HSV-2 to infect cells further studies are necessary to determine if C. crescentus itself exerts some sort of non-specific anti-viral effect. If C. crescentus has some inherent anti-viral activity this would further its suitability and desirability as a microbicide candidate as both the recombinant protein, which binds directly to the virus or target cells, as well as the bacteria itself could work together to enhance protection from infection.
HSV-2 infection is a major risk factor for HIV-1 acquisition, leading to a 2-4 fold increase in HIV-1 infection 56 . The use of condoms, disclosure of serostatus and anti-viral therapy are only 50% effective at preventing HSV-2 infection 7 . Several attempts have been made to develop a vaccine against HSV-2, with no reports of success in people 7,56-60 . Furthermore, HIV-1 positive women have impaired mucosal immunity to HSV-2 infection and diminished anti-HSV activity of cervico-vaginal secretions 6 . There is currently no vaccine for HIV-1, and prevention efforts focused on condom use have not been successful in protecting women from HIV-1 infection. A dual-target microbicide could have a major impact on women's sexual health.
Both HIV-1 and HSV-2 cause significant morbidity each year. By preventing HSV-2 infection we may be able to prevent numerous HIV-1 infections. A microbicide that is targeted to two sexually transmitted viruses could have a huge impact on the HIV-1 epidemic, and could prevent HSV-2 infections, having an enormous impact on public health. We have previously demonstrated that C. crescentus is easily modifiable to express a wide variety of anti-HIV-1 proteins, and that these recombinant C. crescentus can provide significant protection from HIV-1 infection in vitro. We have built on our previous findings to show preliminary safety data for a C. crescentus based microbicide, creation of new recombinant C. crescentus, protection from HIV-1 infection, and that we have created a dual-target microbicide with several candidates protecting against both HIV-1 and HSV-2 infection.

Materials and Methods
Cell Lines and Viruses. Vero  was added at a 1:1 ratio. Plates were incubated at 37 °C for 6 days before fixing with 25% paraformaldehyde and staining with 1% Crystal violet.
Primary cells. Whole blood was collected and processed using Lymphoprep (Stemcell Technologies Inc). PBMCs were grown in RPMI 1640 supplemented with 20% FBS. All procedures were approved by the University of British Columbia Clinical Research Ethics Board certificate H12-02480.

Preparation of C. crescentus Displaying Chimeric S-layer Proteins. C. crescentus strain JS4038 was
used for all experiments. Gene segments were synthesized by Integrated DNA Technologies, Inc. (Coralville, Iowa) with codon usage adapted for C. crescentus. See Table 1 for the amino acid sequences. The synthesized DNA segments specified BglII and SpeI restriction sites on the 5′ side and an NheI site on the 3′ end to facilitate directional cloning into p4BRsaA(723)/GSCC digested with BglII and NheI 29 . p4BRsaA(723)/GSCC is identical to p4ARsaA(723)/GSCC except an extra, irrelevant BamH1 site has been removed by digesting with the enzyme, filling in ends with Klenow DNA polymerase and blunt self ligation 26 . The 2X indolicidin was made in a fashion similar to other tandem repeat multimer clones 29 . Inserted sequences were confirmed by DNA sequencing before transfer to C. crescentus by electroporation.
Protein Analysis. C. crescentus were grown to an optical density at 600 nm of approximately 1 (3 × 10 9 cells/ mL) and protein was prepared from equal numbers of bacterium. S-layer proteins were prepared by a low-pH extraction method 76 . S-layer proteins were visualized using sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) using 7.5% separating gels. For Coomassie Brilliant Blue R stained gels the Coomassie Bio-Rad SDS-PAGE standards, low range (Bio-Rad) was used. The image has been brightened and cropped to minimize background. The original image has been provided to Scientific Reports and is available from the authors upon request. For immuno-blotting, proteins were transferred to BioTrace NT nitrocellulose membranes (Pall Life Sciences). Membranes were blocked with Tris-buffered saline (TBS)-milk, (20 mM Tris base, 150 mM NaCl, pH 7.4, containing 3% [wt/vol] nonfat milk) and incubated with appropriate primary antibodies in TBS-milk. Membranes were then washed (3X) in TBS-milk, then incubated with appropriate Horse Radish peroxidase (HRP) conjugated secondary antibodies in TBS-milk. Blots were washed with TBS and visualized with 4-chloro-1-naphthol as previously described 27,77 . Immunoblots to identify chimeric RsaA proteins were done using a rabbit anti-RsaA antibody (1:10,000 dilution) and goat anti-rabbit HRP secondary antibody (1:10,000 dilution). A lane containing Fermentas Page ruler prestained protein ladder (Thermo-Fisher) 26,78 was used as a size standard.
HIV-1 89.6 Viral Blocking Assays. C. crescentus were grown and prepared as described above immediately prior to use. All experiments were performed in quadruplicate wells and repeated at least three times. 200TCID 50 was determined as previously described 22 and used for all virus blocking assays. For titrations 10 6 , 10 7 , 5 × 10 7 , 10 8 , 1.5 × 10 8 or 2 × 10 8 C. crescentus were added to each well and 1 × 10 8 C. crescentus were added for viral blocking experiments. The virus and C. crescentus constructs were incubated for 1 hour at 37 °C before adding 10,000 TZM-bl cells or PBMCs and 75 μg/mL DEAE-dextran. After an overnight incubation at 37 °C and 5% CO 2 the plates were centrifuged at 800 rpm for 5 minutes and medium was changed in all wells (PBMCs only). After an additional 24-48 hours the level of infection was determined by β-galactosidase assay kit (TZM-bl cells) or p24 ELISA (ZeptoMetrix Corporation and ProSci Incorporated) according to the manufacturer's instructions (PMBCs). For β-galactosidase assays data are presented and determined as a percentage of infection of the TZM-bl + HIV-1 wells with the background from uninfected TZM-bl cells subtracted. p24 ELISA data is normalized to infection of the untreated control wells with PBMCs + HIV-1 set as 100%. Histology. Mice were inoculated intravaginally with 10 8 Cc-Control, PBS or 20 μg LPS from Salmonella enterica serotype Minnesota as described above. Four hours later vaginal tissue was harvested and placed in buffered formalin. Fixed vaginal tissue was embedded in paraffin, cut into 7 μM slices, and mounted on slides. Tissue was deparafinized in xylene and re-hydrated in ethanol before staining with Harris hematoxylin and eosin Y solution. Images were acquired on EVOS cell imaging system (ThermoFisher).
Caulobacter ELISA. Mice were inoculated biweekly intravaginally with 10 8 C. crescentus Cc-Control for 2 months. Vaginal lavage fluid was collected and prepared as described above before inoculation and 67 days post-inoculation. Vaginal lavage fluid supernatant was diluted in blocking buffer containing 1% bovine serum albumin, 10% FBS and 0.05% Tween-20. An anti-RsaA antibody raised in rabbits (concentration: 14 mg/mL) was used in a limiting dilution to create a standard curve for quantification of antibody levels, with limit of detection of 14 pg/mL. Ninety-six (96) well Nunc plates were coated with 2 × 10 8 Cc-Control diluted in carbonate buffer and incubated overnight at 4 °C. Plates were washed and blocked with 1% BSA. Diluted vaginal lavage fluid or anti-RsaA antibody was added to each well. After incubation at room temperature for 2 hours the plates were washed and 1:1000 anti-mouse IgG-HRP (Sigma) or 1:1000 anti-mouse IgM-HRP (Sigma) were added. A goat anti-rabbit HRP antibody (1:1000) was used for the standard curve. The plates were incubated at room temperature for 2 hours then 1:10 O-Phenylenediamine dihydrochloride (Sigma, 2 mg/mL) and 1:200 Urea hydrogen peroxide (Sigma, 40 mg/mL) were diluted in citrate phosphate buffer and added to each well. Twelve minutes later 25% H 2 SO 4 was added and the OD 490 was determined using a plate reader. The standard curve was used to interpolate the unknown values for antibody levels in vaginal lavage fluid. HSV-2 in vivo studies. Mice were given subcutaneous injection of 2 mg Medroxyprogesterone 17-acetate 4-5 days before infection with HSV-2. 10 8 recombinant C. crescentus cells prepared as described above were mixed with 10 5 pfu HSV-2 to a final volume of 20 μL and immediately inoculated intravaginally as described above. Vaginal lavage fluid was collected 24, 48 and 72 hours post-infection and prepared as described above. Mice were scored daily for disease progression using a 5 point scale: 1 -slight redness of external vagina; 2 -swelling and redness of external vagina; 3 -severe swelling and redness of both vagina and surrounding tissue and hair loss in genital area; 4 -genital ulcerations with severe redness, swelling and hair loss of genital and surrounding tissue; 5 -severe genital ulceration and hind limb paralysis. Mice were euthanized when they reached a score of 4.
Plaque assays. Vaginal lavage fluid was serially diluted in MEM and a limiting dilution was prepared. Plaque assays were performed on confluent Vero cells as described above.
Statistics. Statistical analysis was performed with Prism GraphPad software. As indicated, Student's t test or ANOVA with correction for multiple comparisons were used as appropriate to evaluate the significance of differences between groups. Kaplan-Meier survival curves were compared by log rank tests. Unless otherwise indicated, presented statistics represent the comparison between HIV/HSV2 + Cc-Control and HIV/HSV2 + recombinant C. crescentus. A p value of <0.05 was considered significant. Data availability. Data is available from the corresponding author upon request.