Charles Kelly, Department of Oral Immunology, Dental Institute, King’s College London, Floor 28 Tower Wing, Guy’s Hospital, London SE1 9RT, UK. (fax: +44 20 7188 4375; e-mail: firstname.lastname@example.org).
Abstract. Kelly CG, Shattock RJ (Dental Institute, King’s College London, London; and Imperial College, London; UK). Specific microbicides in the prevention of HIV infection (Symposium). J Intern Med 2011; doi: 10.1111/j.1365-2796.2011.02454.x.
Microbicides are products that are designed for application at vaginal or rectal mucosae to inhibit or block early events in HIV infection and thereby prevent transmission of HIV. Currently, the most advanced microbicides in the development pipeline are based on highly active anti-retroviral drugs (ARVs). Significant protection of women by vaginally applied tenofovir gel, demonstrated in the CAPRISA 004 trial, has provided proof-of-concept that microbicides can be effective. The rationale for investigating ARVs and other compounds as vaginal or rectal microbicides is discussed together with approaches to improve efficacy by the development of combination microbicides and by new formulations that may increase user acceptance.
Microbicides are products that are designed for application at vaginal or rectal mucosae to inhibit or block early events in HIV infection and thereby prevent transmission of HIV. Despite promising data from in vitro studies and, in some cases, nonhuman primate (NHP) models of infection, early generation microbicides failed to demonstrate efficacy in clinical trials [1–5]. More recently, the CAPRISA 004 trial of tenofovir gel has provided clinical data to support the efficacy of vaginally applied microbicides . While providing a significant boost to the field of microbicide development, the CAPRISA trial also reflects the increased focus on development of more potent and specific inhibitors including antiretroviral drugs (ARVs).
First-generation microbicides were aimed at disrupting HIV virions or at preventing fusion and cell entry. Recent studies of early events in infection using both tissue explant and NHP models [7, 8] suggest that rectal and vaginal mucosal surfaces present a significant barrier to HIV such that the majority of infections are initiated by a single virus. There is a period of a few days before the infection becomes self-sustaining. Thus, while fusion and entry inhibitors continue to be developed, drugs that target intracellular events in the HIV life cycle during this early phase may also prevent systemic infection. Antiretroviral drugs that are currently used for therapy in individuals infected with HIV are highly potent inhibitors of critical stages in HIV replication and are increasingly being investigated as microbicides. The long tissue half-lives of some of these drugs are likely to be beneficial for microbicide-dosing regimens. A further advantage of using ARVs for prophylaxis is that some are also active against cell-associated virus . Although the weight of evidence suggests transmission is predominantly mediated by cell-free virus, the possible contribution of cell-associated HIV to infection cannot be discounted; cell-associated virus is present in semen , and NHP models have shown that cell-associated virus can initiate infection [11, 12].
In this review, we discuss findings from studies of early events in HIV transmission which provide a rationale for some of the microbicide strategies under investigation. We also describe some of the classes of inhibitors being developed as microbicides and consider approaches to formulation that address the challenges of ensuring transport of microbicides across epithelial surfaces to target cells.
Early events in sexual transmission provide a rationale for specific microbicide interventions
Single-genome amplification and sequence analyses of env genes  or of whole viral genomes  from HIV-infected individuals indicate that HIV infection through genital or rectal mucosae may be established by a single isolate in most cases. Consistent with this, studies of SIV infection in NHPs indicate that only a small number of cells in cervicovaginal tissues are infected at 3 days after vaginal exposure to virus [7, 8]. Although some spread of infection to lymphatic tissue is detectable within this time period, it may be at too low a level to establish productive infection at these sites . Expansion of the original foci of infection results in further seeding of lymphatic tissue leading to self-sustaining propagation and systemic infection from day 6 on.
Data from the in vivo model of SIV infection and from ex vivo cervicovaginal tissue explant models of HIV infection [15, 16] indicate that CD4+ T cells form the founder populations of infected cells at the portal of entry as shown in Fig.1. An innate inflammatory response to virus results in the recruitment of CD4+ T cells that provide target cells for the expansion of infection. Explant studies indicate that Langerhans cells also acquire virus at an early stage, but their role as primary targets for infection remains controversial. Nevertheless, Langerhans cells and dendritic cells may contribute to viral dissemination to lymphoid tissues and enhance amplification of viral replication in CD4 T cells at mucosal sites  (Fig.1).
The mucosal barriers appear to be effective in limiting entry of virus and further ‘modest improvements in barrier integrity’ could be sufficient to prevent infection . Such improvements could be provided by microbicides. In addition, the delay before a sustainable infection is established provides a potential opportunity for interventions, such as microbicides, that could prevent the expansion of the initial founder population and thereby prevent the establishment of systemic infection.
Early generation microbicides [1–5] were aimed at preventing HIV fusion and entry because complete blockade at these stages would prevent any possibility of infection becoming established. Ultimately, they lacked sufficient potency to prevent HIV infection. However, in view of the delay before an infection becomes self-sustaining as outlined previously, inhibitors that act postentry have become an additional focus for investigation as microbicides. Ideal properties of a microbicide have been discussed extensively elsewhere , but in addition to efficacy and safety, potential for large-scale production at an affordable cost is an essential property. Candidate microbicides and their mechanisms of action are reviewed below. Targets for potential microbicide intervention so as to inhibit stages in the HIV life cycle are shown in Fig. 2.
Microbicides that target specific stages in HIV infection and replication
Binding of HIV-1 envelope protein to CD4 triggers a series of events leading to fusion of viral and cell surface membranes with entry of the viral core into the cytosol. Binding of CD4 induces conformational changes in gp120 that expose the co-receptor binding site and allows formation of a complex of virus bound through gp120 to CD4 and co-receptor. This complex induces further conformational changes in gp120 that allow the formation of the six-stranded helical bundle structure of gp41 necessary for fusion .
Inhibitors that could potentially block each step in this process have been investigated as microbicides. The demonstration that vaginal application of the monoclonal antibody b12 (directed against the CD4 binding site of gp120) conferred protection in a NHP challenge model provided important proof-of-principle of the potential of a highly specific microbicide . Other antibodies have also been investigated including the combination of anti-gp120 antibodies 2F5, 2G12 and 4E10 that showed efficacy when applied systemically [21, 22]. The antibodies were tested as a microbicide in a phase I clinical trial . It is not clear that antibodies can currently be produced on the scale required for a microbicide at reasonable cost. Single domain antibodies with potent in vitro neutralizing activity have been isolated from immune libraries [24, 25] and, because they can be produced by microbial fermentation, may become a more viable alternative. A peptide mimic of CD4 based on the highly stable scyllatoxin  has been developed which also shows potent in vitro activity. Both the anti-gp120 antibodies and the CD4 mini-protein act by binding to gp120 and blocking the binding of CD4.
A different mechanism of inhibition has been demonstrated for a series of small molecule inhibitors that also bind to gp120 and prevent fusion. Investigation of one member of this series, BMS-806, demonstrated that it does not inhibit binding to CD4 but prevents the conformational changes in gp120 that lead to the formation of the six-stranded helical structure that mediates membrane fusion . BMS-806 effectively prevented infection in a NHP model of SHIV vaginal challenge , and a derivative of BMS-806 (BMS-793) is in preclinical development as a microbicide [http://www.ipmglobal.org/products-development].
The six-stranded helical structure adopted by gp41 following receptor binding comprises a three-stranded parallel coiled coil structure formed by a sequence (HR1) towards the N-terminal from each gp41 in the envelope trimer. A second sequence (HR2) towards the C-terminal region of each gp41 ectodomain also adopts a helical conformation and each of these helices pack in an anti-parallel manner in the grooves of the HR1 coiled coil to form the six helical hairpin bundle . A peptide (Fuzeon/T20) corresponding to the HR2 sequence inhibits bundle formation and fusion  and was developed as an ARV for systemic administration. More recently, a modified form of T20, CL52 (designed for production by microbial fermentation) was shown to be effective as a microbicide when applied vaginally in the NHP model . The cost of production of peptides has been considered too expensive for further development of Fuzeon-based peptides as a microbicide; however, the technology developed for large-scale synthesis (on the scale of >100 kg per batch) together with economies of scale in production of the raw materials (e.g. derivatized amino acids) has considerably reduced the cost of production . The same considerations may apply to the development of the mini-CD4 protein, above.
Transmitted strains of HIV-1 use almost exclusively the CCR5 coreceptor [31, 32], and compounds that prevent binding to CCR5 have, therefore, been investigated as microbicides. Targeting a cellular rather than viral target has the potential advantage that any drug will have broad spectrum activity but also has the disadvantage of a higher likelihood of toxicity . Compounds that bind to CCR5 are of pharmaceutical interest not only because they may inhibit HIV fusion but also because they may have immunomodulatory functions of potential benefit as anti-inflammatory agents. Arguably, a combination of these properties would be beneficial in a microbicide to confer the dual activity of blocking viral fusion and reducing the anti-viral inflammatory response which could otherwise increase target cells . Several small molecule CCR5 inhibitors are in clinical development as drugs for therapy  although to date only maraviroc has been approved for treatment. Maraviroc is also under development as a microbicide, and a vaginally applied gel formulation of maraviroc effectively protected against infection with SHIV in a macaque model . Site-directed mutagenesis and binding studies together with modelling of the CCR5 structure (based on experimental data for other members of the G-protein coupled receptor family) suggest that the small molecule inhibitors bind to a predominantly hydrophobic pocket formed by the transmembrane helices near the extracellular surface [37, 38]. Binding induces conformational change in CCR5 that prevents attachment of HIV.
In contrast, another class of CCR5 inhibitors, protein-based RANTES analogues, binds to the extracellular domain of CCR5 and inhibits fusion either by receptor internalization or by receptor blockade [39, 40]. Modification of the N-terminus of RANTES results in loss of chemotactic activity and leucocyte activation but retains receptor binding properties . A chemically modified analogue, PSC-RANTES, with increased receptor binding affinity relative to wild type was effective in vivo as a vaginal microbicide . The requirement for chemical modification, together with the strong CCR5 agonist activity of PSC-RANTES potentially leading to mucosal inflammation, presents problems for further development as a microbicide. For these reasons, further RANTES analogues were developed with the N-terminus modified solely by incorporation of naturally occurring amino acids that would allow production by microbial fermentation . Subsequently, analogues that retained agonist and CCR5 internalization activities (6P4-RANTES) or that retained CCR5 binding activity only, with no detectable signalling activity (5P12-RANTES), were shown to prevent vaginal transmission of HIV in NHP models . These analogues show promising stability , and formulation studies are in progress to produce clinical grade material.
Reverse transcriptase inhibitors
Following fusion of viral and host cell membranes and virus decapsidation, RNA is released in the host cytosol and is transcribed to produce viral double-stranded DNA by the HIV-1 enzyme reverse transcriptase. Reverse transcriptase has been successfully targeted by nucleoside and nucleotide reverse transcriptase inhibitors (NRTIs/NtRTIs) for therapy in HIV infection. Activation of NRTIs requires three consecutive phosphorylation reactions by intracellular kinases, whereas NtRTIs carry a phosphonate group and require only two consecutive phosphorylation reactions for activation. The active forms of these drugs compete with intracellular nucleotide triphosphates for incorporation into the growing DNA strand. Incorporation of the drug results in termination of synthesis because analogues lack the 3′ hydroxyl group necessary for the formation of a phosphodiester linkage with the next incoming nucleoside .
Tenofovir (9-[(R)-2-(phosphonomethoxy) propyl] adenine monohydrate], an NtRTI licensed for therapeutic use, has shown considerable potential as a microbicide. When applied as a microbicide, tenofovir gel prevented infection in rectal  and vaginal  NHP models of SIV and SHIV transmission, respectively. Even more significantly, tenofovir is the first microbicide to demonstrate efficacy in a clinical trial . The CAPRISA 004 trial, carried out in South Africa, was a double-blind, randomized, placebo controlled trial of vaginally applied 1% tenofovir gel. Tenofovir gel reduced HIV infection by 39% over a 30-month period, and for high adherers (women who used the drug with >80% compliance), the reduction was 53%. Overall protection was higher at 50% after 12 months use of tenofovir gel.
Non-nucleoside reverse transcriptase inhibitors (NNRTIs) are a second class of drug being investigated as microbicides. Non-nucleoside reverse transcriptase inhibitors are noncompetitive inhibitors that bind to an induced allosteric hydrophobic pocket approximately 15Å from the reverse transcriptase catalytic site . As well as inhibiting reverse transcriptase, some NNRTIs have an inhibitory effect on downstream virion formation . A number of NNRTIs are currently being used for treatment. More recent NNRTIs have been designed to maximize interactions with protein main chain atoms and conserved amino acid residues lining the hydrophobic pocket resulting in both increased binding affinity and presenting a higher barrier to resistance than first-generation compounds . Four NNRTIs, UC781, dapivirine, MIV-150 and MC1220 that are not currently used for therapy are being investigated as potential microbicides. Two of these, MIV-150 and MC1220, prevent infection in NHP models of vaginal transmission [49, 50]. Dapivirine has been tested in human tissue explant models and shows very high activity . Studies in rabbits and macaques of vaginal administration of dapivirine showed accumulation of drug in the keratinized epithelial layer of the vagina with some penetration of the superficial cell layers . Dapivirine was detectable in tissue at 48 h after administration. Phase I trials of dapivirine have been performed [53, 54], and clinical efficacy trials are planned.
HIV-1 integrase integrates proviral reverse-transcribed DNA into the host genome in a two-step reaction. First a conserved dinucleotide is removed from the 3′ of each strand of viral DNA. Next, in a single-strand transfer reaction, the host genomic DNA is simultaneously cleaved, and the 3′end of the viral DNA is attached to the 5′end of the genomic DNA . Raltegravir, an inhibitor of the strand transfer reaction, is the only integrase inhibitor currently in use for therapy and has recently shown efficacy as a vaginally applied microbicide in macaques .
Several cellular co-factors have been identified that contribute to integrase activity [reviewed in 57]. Some are required for catalytic activity, while others are involved in cytoplasmic trafficking and nuclear import of the preintegration complex. Potentially, these present several targets for inhibitory drugs that block interaction with the integrase complex, but currently candidates under development are at early preclinical stages .
The final stage in HIV replication is conversion of noninfectious immature virus particles (that bud off from the cell) to mature infectious virus. Maturation is initiated by the HIV protease cleavage of gag and gag-pol polyproteins . Ten protease inhibitors are licensed for therapeutic use . Protease inhibitors present a higher barrier to resistance than other classes of ARV, and boosted protease inhibitors have been tested for use as monotherapy [60, 61]. The findings, described previously, that establishment of systemic infection requires seeding from the original small foci in subepithelial mucosae provide a rationale for investigating protease inhibitors as microbicides. Protease inhibitors may block the spread of infection from the initial focus to lymphatic tissue while simultaneously limiting local spread such that the infection cannot be sustained.
Formulation of microbicides
Gel formulations of microbicides have been used in all late stage efficacy clinical trials to date. The advantage of a gel formulation is that a well-defined dose of microbicide can be delivered at a defined time. Several studies have documented relatively high rates of acceptability of gel formulations both for vaginal [6, 62, 63] and rectal  administration of microbicides. Acceptability of tenofovir gel was reported to be as high as 97.4% of study participants in the CAPRISA 004 trial . However, assessment of acceptability does not necessarily reflect compliance. In the CAPRISA 004 trial, 41% of study participants used gel for only 50% or less of self-reported sex acts. Different dosing regimens may affect compliance with microbicide administration, and the problems associated with coitally dependent dosing have been well documented [65, 66]. Based on data from NHP studies and timing of drug dosages in preventing mother-to-child transmission of HIV, the CAPRISA 004 trial  adopted a coitally related dosing procedure. Participants were required to follow a ‘before and after sex’ dosing regimen [BAT-24] in which 1% tenofovir gel was applied any time in the 12 h preceding intercourse and again in the 12 h after intercourse with no more than two applications in a 24-h period. In contrast, the Vaginal and Oral Interventions to Control the Epidemic (VOICE) trial of tenofovir gel [http://www.mtnstopshiv.org/news/studies/mtn003/backgrounder] has adopted noncoitally dependent daily dosing in anticipation that because of the long tissue half-life of tenofovir, this regimen will provide an effective concentration of drug. This approach is anticipated to lead to greater compliance because it can be incorporated as a routine act in daily life.
Noncoitally dependent sustained release formulations for microbicides have been developed based largely on silicone elastomer intravaginal rings . These intravaginal rings (designed for self-insertion and removal) are intended to be worn continuously so as to deliver and maintain effective vaginal microbicide concentrations over weeks or months. High user acceptability has been reported for the rings . Different forms of ring have been developed to allow loading with a variety of potential microbicides ranging from small lipophilic drugs to relatively large polar proteins. Silicone elastomer vaginal rings have been tested extensively as a formulation for dapivirine, the small molecule lipophilic NNRTI, in both preclinical [68, 69] and clinical [53, 54] studies. In separate clinical trials, dapivirine loaded rings released drug over 7 days  or over 28 days  resulting in concentrations of dapivirine in vaginal fluid that were more than 3–4 logs higher than the in vitro EC50. There were no serious adverse effects in these trials. Efficacy trial of dapivirine rings is planned.
Other formulations are being investigated including films, tablets, thermoplastic rings and diaphragms [67, 70]. Different formulations will provide more choice and should encourage microbicide use.
Formulations of the first-generation microbicides that were aimed at preventing HIV fusion or entry were required to deliver effective concentrations of microbicide at mucosal surfaces. With the recognition that postfusion events can be effective targets for microbicides, formulations now need to allow for penetration of drugs into mucosae and entry into target cells. Drug uptake across mucosal epithelium is complex. Uncharged lipophilic compounds can diffuse across mucosal cell membranes, whereas charged compounds may be transported by a range of solute carrier uptake transporters including organic ion transporters, nucleoside transporters and peptide transporters [71, 72]. Net uptake of drug is influenced by the balance between uptake and efflux, with the latter being mediated by transporters of the ATP-binding cassette family, as well as intracellular drug metabolism . In addition, small polar drugs may cross the epithelium by paracellular routes where the presence of tight junctions and the small pore size of the paracellular canal present a selective barrier . Drug transporters (both uptake and efflux) have been well characterized in the human intestinal tract including the colon  but less well in cervicovaginal epithelium where only the P-glycoprotein efflux transporter  has been identified.
Problems associated with the complexity of drug uptake are largely mitigated in the case of ARV-based microbicides because mucosal and cellular uptake of these compounds is well established. Factors such as pH of the formulation may be significant in promoting drug uptake either by maintaining lipophilic compounds in the uncharged state or by providing optimal conditions for uptake transporters at least some of which are stimulated by low extracellular pH . Drug interactions may also need to be considered in optimizing uptake although these may also be complex. Some protease inhibitors including ritonavir and saquinavir inhibit P-glycoprotein function  and, therefore, in combinations could potentially enhance drug uptake. However, other reports indicate that saquinavir and darunavir  as well as ritonavir  may upregulate P-glycoprotein and hence drug efflux.
Resistance to antiretroviral microbicides
The potential for development of drug-resistant virus strains must be considered when ARV-based microbicides are to be used in settings where the same ARV or ARVs of the same class are in use for treatment. Resistance develops in response to sustained selective pressure when ARVs are used to treat systemic infection. In contrast, use of an ARV-based microbicide that effectively prevents infection would not provide sustained selective pressure. If an individual using the ARV-based microbicide does not become infected, there is no possibility of resistance developing.
Within populations, two situations which could result in selection of resistant HIV strains have been considered . If a particular ARV is being used both for treatment and as a microbicide, the microbicide may not prevent transmission of resistant strains that develop in patients undergoing treatment. Resistance, however, is not absolute, i.e, at very high doses of drug, resistant strains are susceptible. Microbicides are typically applied at doses at least three logs higher than the EC50 of the drug and local application ensures a high dose at the site of infection. Thus, even resistant strains may be inhibited. The second possibility of driving resistance is when an ARV-based microbicide is used by an individual infected with HIV. Depending on the systemic levels of the ARV provided by microbicide application, resistant virus could be selected.
The likelihood of either of these occurrences is unknown. In current clinical trials of microbicides, only HIV-negative participants are enrolled and are frequently monitored for HIV infection. In the CAPRISA 004 trial , women who became infected during the trial provided no evidence for transmission of tenofovir-resistant isolates. The infected women were estimated to have used tenofovir gel over 3–4-week period after infection. Similar monitoring in further clinical trials of ARV-based microbicides will inform the extent to which screening and monitoring of infection would be required for product development and use.
Development of combination microbicides
Combinations of microbicides are being investigated as next-generation products in part to address the possibility of resistance. Combinations present a higher barrier to resistant viruses and would prevent transmission of a virus resistant to any single component of the microbicide. In addition, targeting more than one event in virus transmission and replication is likely to provide increased protection. Finally, use of combination microbicides may allow exploitation of useful drug–drug interactions either as additive or synergistic inhibitors or as a potential means of increasing drug uptake and retention in local tissues by effects on drug transporters or drug metabolism.
Formulation of microbicide combinations may present challenges particularly when the drugs differ in chemical properties. However, a number of different approaches to co-formulation of microbicides have been described that take advantage of the variety of different formulation options available for microbicides. Three lipophilic protease inhibitors (ritonavir, lopinavir and saquinavir) were successfully co-formulated in a silicone elastomer ring and displayed favourable in vitro release kinetics , while a combination of maraviroc and dapivirine was formulated in thermoplastic (ethylene-vinyl acetate copolymer) rings . Polyvinyl alcohol-based films have been used to co-formulate maraviroc and tenofovir as well as mariviroc and dapivirine . Dapivirine has also been co-formulated with the fusion inhibitor BMS-793 in a tablet form for vaginal delivery .
Testing whether combination microbicides show increased efficacy in clinical trials will present considerable problems for future trial design particularly with regard to numbers of participants and expense. Pharmacokinetic and pharmacodynamic studies, used routinely in drug development, may become more important in microbicide development as a means of optimizing microbicides, combinations and formulations prior to large-scale efficacy trials as discussed elsewhere.
The CAPRISA 004 trial has provided critical proof-of-concept that a microbicide can be effective and, together with studies of early events in HIV infection, provides a rationale for further development of ARV-based microbicides. If the VOICE trial provides a similar level of protection, it is likely that tenofovir will be advanced as a microbicide product. The challenge will be to increase efficacy. Compliance with microbicide protocols is clearly a crucial aspect of clinical trials but equally clearly is considered by several researchers [85, 86] to be one that is difficult to control and prone to overestimation. New compounds under development, perhaps in combinations, together with improved, more user-acceptable formulations leading to increased compliance may improve efficacy.
Future trials will carefully monitor the potential for ARV-based microbicides to drive development of resistant strains and provide further data to assess the extent to which this could be a problem. In any event, it may be prudent to continue development of the most promising new classes of microbicides such as the RANTES analogues that act by mechanisms distinctly different to those of the ARVs that are used for treatment and are being investigated as microbicides. Future trial protocols may also have to solve the issue of how to demonstrate improved efficacy of new microbicides or combinations over an established relatively effective product. Such a dilemma could be regarded as a marker of real progress in microbicide development.
Conflict of interest statement
The authors have no conflict of interest.
The authors are supported by grants from the European Commission: LSHP-CT-2006-037611 [EUROPRISE] and Health-F3-2009-242135 [CHAARM].