Results of a phase 1, randomized, placebo-controlled first-in-human trial of griffithsin formulated in a carrageenan vaginal gel

HIV pre-exposure prophylaxis (PrEP) is dominated by clinical therapeutic antiretroviral (ARV) drugs. Griffithsin (GRFT) is a non-ARV lectin with potent anti-HIV activity. GRFT’s preclinical safety, lack of systemic absorption after vaginal administration in animal studies, and lack of cross-resistance with existing ARV drugs prompted its development for topical HIV PrEP. We investigated safety, pharmacokinetics (PK), pharmacodynamics (PD), and immunogenicity of PC-6500 (0.1% GRFT in a carrageenan (CG) gel) in healthy women after vaginal administration. This randomized, placebo-controlled, parallel group, double-blind first-in-human phase 1 study enrolled healthy, HIV-negative, non-pregnant women aged 24–45 years. In the open label period, all participants (n = 7) received single dose of PC-6500. In the randomized period, participants (n = 13) were instructed to self-administer 14 doses of PC-6500 or its matching CG placebo (PC-535) once daily for 14 days. The primary outcomes were safety and PK after single dose, and then after 14 days of dosing. Exploratory outcomes were GRFT concentrations in cervicovaginal fluids, PD, inflammatory mediators and gene expression in ectocervical biopsies. This trial is registered with ClinicalTrials.gov, number NCT02875119. No significant adverse events were recorded in clinical or laboratory results or histopathological evaluations in cervicovaginal mucosa, and no anti-drug (GRFT) antibodies were detected in serum. No cervicovaginal proinflammatory responses and no changes in the ectocervical transcriptome were evident. Decreased levels of proinflammatory chemokines (CXCL8, CCL5 and CCL20) were observed. GRFT was not detected in plasma. GRFT and GRFT/CG in cervicovaginal lavage samples inhibited HIV and HPV, respectively, in vitro in a dose-dependent fashion. These data suggest GRFT formulated in a CG gel is a safe and promising on-demand multipurpose prevention technology product that warrants further investigation.

Currently, once daily oral Truvada (tenofovir disoproxil fumarate-emtricitabine [TDF/ FTC]) is the only product for HIV pre-exposure prophylaxis (PrEP) approved by the Food and Drug Administration (FDA) for cisgender women. Approval was based on the results of clinical trials demonstrating significantly reduced risk of HIV acquisition in cisgender men and transgender women who have sex with men (iPrEX) and in serodiscordant couples (Partners PrEP Study) [2,3]. Truvada was shown to be efficacious in cisgender men and transgender women who have sex with men even if taken intermittently (STRAND) [4] or on-demand (IPERGAY) [5]. In contrast, daily Truvada did not demonstrate efficacy in two studies in cisgender women (FEM-PrEP and VOICE) [6,7]. Low adherence rates [6,7] potentially explain these results. Recent clinical trials (HPTN083 and HPTN 084) have demonstrated high efficacy of long-acting (LA) injectable Cabotegravir against HIV acquisition which was superior to Truvada [8,9]. LA parenteral products (injectables and implants) could offer simpler administration regimens and may be preferred over daily oral regimens for some women. However, these products require health care provider delivery, have injection site and potential systemic side effects, and, for the case of implants, may not be easily removable. Alternatively, user-controlled vaginally administered PrEP (gel, fast dissolving vaginal insert (FDI), intravaginal rings (IVR), vaginal film) offers the distinct advantage of increasing local exposure to the active pharmaceutical ingredients (APIs) at the HIV transmission site and low to no systemic side effects.
Studies in the US and Africa demonstrated variability in preferences for HIV prevention products across groups and geographies and emphasized that women need options of different routes of PrEP administration and formulations suitable for their lives [10][11][12]. Expanding available PrEP options is likely to improve uptake of HIV prevention products just as contraceptive choice increased uptake, retention and contraceptive effectiveness (reviewed in [13,14]). Therefore, it is critical to develop diverse PrEP products that align with preferences of people at risk of HIV infection.
In addition to adherence, other factors such as drug and dosage form-related characteristics (tenofovir distribution in colorectal tissues and female genital tissues [15]) and genital inflammation promoting susceptibility to HIV [16] may impact PrEP efficacy. Therefore, a thorough analysis of relationships between PrEP and mucosal environment (including transcriptome and proteome) may provide important data early in the product development process to identify PrEP candidates with favorable pharmacokinetics (PK) profiles across the conditions that do not induce perturbations in the mucosa that could lead to enhanced susceptibility to HIV and decreased PrEP efficacy.
The PrEP development pipeline has been dominated by repurposing approved therapeutic antiretroviral (ARV) drugs [13], which are undoubtedly an important addition to the toolbox of HIV prevention methods for women. However, a product that does not contain ARVs (particularly those approved for HIV treatment) offers several potential advantages. First, a product without an ARV is less likely to lead to drug-resistant HIV that may compromise subsequent ARV treatment options. Second, because the risk of resistance is low, non-ARV-based products are unlikely to require labeling indicating the need for HIV testing before and during use (once safety has been demonstrated in individuals with HIV). Removing the need for frequent HIV testing would overcome one of the barriers cited by individuals who are reluctant to use PrEP or who initiate PrEP and then discontinue [17][18][19]. Third, a non-ARVbased PrEP is likely to be approved faster by regulatory authorities as an over-the-counter (OTC) product than one with ARVs.
The Population Council is developing griffithsin (GRFT) as topical PrEP for on-demand use or sustained delivery. GRFT is a non-ARV lectin that has an outstanding anti-HIV activity (50% effective concentration (EC 50 ) of 0.13 nM, and 90% effective concentration (EC 90 ) of 0.58 nM) [20]. Unlike the "first generation" or broad-spectrum microbicides that did not demonstrate efficacy [21][22][23][24][25][26][27][28][29], GRFT acts by binding to the viral gp120 glycoprotein and blocking virus entry [30]. GRFT prevents both cell-free and cell-associated HIV transmission and blocks virus-cell fusion [31]. GRFT also has activity against other highly prevalent and morbid sexually transmitted infections (STIs), such as herpes simplex virus type-2 (HSV-2) and human papillomavirus (HPV), with overlapping HIV epidemiology that increase the risk of HIV acquisition and represent a significant public health burden in their own right [32]. GRFT inhibits HSV-2 by targeting viral entry and cell-to-cell mediated transmission [33,34]. GRFT's moderate activity against HPV is mediated by promoting HPV receptor internalization [34]. A GRFT multipurpose prevention technology (MPT) product that protects against HIV, HSV and HPV would be a promising addition to products in development based on anti-HIV and anti-HSV monoclonal antibodies [35,36].
This report describes the results of a Phase 1 trial of PC-6500 (0.1% GRFT in a CG gel) in healthy female volunteers conducted in the Bronx, NY. The aim of this trial (Population Council Protocol No. 728) was to gather first-in-human safety data and support further development of GRFT in various topical formulations. The primary objectives were (1) to evaluate the safety of PC-6500 gel used vaginally for a single dose, and then once daily for 14 consecutive days of dosing, and (2) to assess PK of GRFT in blood after a single dose, and then after 14 days of dosing. Exploratory safety and pharmacodynamics (PD) objectives included analysis of soluble immune mediators, cervical transcriptome and anti-HIV/HPV activity of cervicovaginal secretions.

Trial design
A 14-day randomized, placebo-controlled, parallel group, double-blind Phase 1 trial of PC-6500, sponsored by the Population Council (New York, NY), was conducted at the Albert Einstein College of Medicine (Einstein) in the Bronx, NY. The main (randomized) period of the trial was preceded by a single-dose, open label (OL) period among a different group of participants. The protocol was approved by the Institutional Review Boards of the Population Council and Einstein prior to initiation of data collection. The protocol was submitted to the US Food and Drug Administration as IND # 123512 and registered at Clinical Trials.gov., number NCT02875119. All participants provided written informed consent before undergoing any procedures. Participants and staff at the Council and Einstein were blinded to the product assignment for the randomized portion of the study, but not time point, through the end of data collection.

Study population
Healthy, HIV-negative, nonpregnant women aged 18-49 who were using an effective contraceptive method other than condoms or a vaginal ring, and who agreed to remain sexually abstinent during the study were invited to participate. Potential participants were assigned a unique identification number and screened for eligibility based on medical history, physical examination, pelvic examination, Pap smear if indicated, blood (CBC and chemistry) and STI testing, including HIV, chlamydia, gonorrhea, and trichomonas. Women with a laboratory abnormality, STI or abnormal Pap test were ineligible. Women with symptomatic BV, vulvovaginal candidiasis, or urinary tract infection at screening could be rescreened following completion of treatment.

Study products, randomization, and blinding
All participants (n = 7) in the OL period were to receive one single 4 g dose of PC-6500 administered vaginally by a study clinician.
In the randomized period, 20 participants were instructed to self-administer 14 doses of PC-6500 or its matching CG placebo (PC-535) once daily for 14 days; OL participants were not eligible for the randomized period to prevent participants using both PC-6500 and placebo, which could confound data interpretation. Day 1/Dose 1 was scheduled as soon as possible after menstruation ended to maximize the likelihood that all 14 days of dosing would occur during the non-bleeding days of the menstrual cycle. Participants administered five of the 14 doses in the clinic with a study staff member in the room, but with a curtain drawn for privacy. The other nine doses were to be self-administered at home at approximately the same time each day. Both gels (Table 1) were manufactured and packaged in identical individually wrapped, prefilled, single-use, metered dose applicators (HTI Plastics; Lincoln, NE) at the Population Council's Good Manufacturing Practices facility (Center for Biomedical Research, New York, NY). PC-6500 was manufactured using recombinant plant-produced GRFT [20]. The two study gels (PC-6500 and PC-535) differed imperceptibly in appearance; both were translucent with a faint beige color.
A Population Council statistician not associated with the trial created the randomization scheme that pre-assigned participants to either PC-6500 or placebo gel in a ratio of 7:3. Half of the participants were assigned blood and cervicovaginal lavage (CVLs) sample collection at 4h post dose; half were assigned to the 8h time point. Having different two time points for CVL collection allowed us to explore the length of time product remains present and active in the vagina. Randomization occurred in blocks of ten. Study staff received a list of kit numbers and specimen time point randomization assignments for 20 participants. All investigators, study staff and participants were blinded to product assignments. The post-dose specimen time points (4h, 8h and 24h) were not blinded. Due to inclement weather, specimen collection for one participant in the PC-6500 group was done at 6h instead of 8h post first dose.
Due to slower than anticipated accrual, it was not feasible to enroll all 20 women within the time and budget available. Therefore, the protocol was amended so that starting with the tenth enrollee, participants were assigned a kit number and time point, on a case-by-case basis, to maximize the number of women assigned to product vs. placebo, and to maintain the balance in post-dose specimen collection time points.

Outcome measures and analytical methods
The schedule of visits and procedures performed in the OL and randomized periods of the study are presented in Tables 2 and 3, respectively. Outcome measures and analytical methods are described in detail below. The schedule was designed to collect BL samples and 24h post last gel samples (randomized phase) in late luteal phase to minimize effects of the menstrual cycle on exploratory outcomes.

Clinical safety and exposure
Safety was evaluated at each study visit (Tables 2 and 3). Safety endpoints included treatment emergent adverse events (TEAEs) and medical significance, in the investigator's judgment, of abnormalities in physical exams, pelvic exams, cervical and vaginal biopsies, and clinical laboratory parameters once product had been administered. AEs were coded using the Medical Dictionary for Regulatory Activities (MedDRA, Version 21.0), and graded according to the Division of AIDS Safety data were summarized by treatment group (OL, randomized PC-6500, placebo). Exposure was calculated by participant based on the number of applicators inserted vaginally, as measured by dye stain assay (DSA) test (sensitivity 93%, specificity 100%) [38,39]. As needed, HCl and NaOH were used to bring the in-process pH into the 6.8 to 7.5 range. https://doi.org/10.1371/journal.pone.0261775.t001

Rapid Stain Identification (RSID) test
Swabs and CVLs were tested for presence of semenogelin using the RSID test (Independent Forensics, Lombard, IL).

Histopathology
The histopathological assessment was performed by board-certified gynecologic pathologist using inflammation scoring as detailed in S1 Table.

PK and concentrations of GRFT in CVLs
Concentrations of GRFT in plasma and in CVLs collected using 10 ml of sterile normal saline were measured by a validated indirect sandwich ELISA. 96-well Immulon 2HB microplates (Thermo Scientific, Rockford, IL) were pre-coated with HIV-1 BaL gp-120 (NIH Reagent Program Cat#49610, Germantown, MD). After blocking, the samples were added to the plate in duplicate. A goat anti-GRFT detection antibody (Pacific Immunology, Ramona, CA) was added and the mixture, followed by a rabbit anti-goat-HRP secondary antibody (Southern Biotech, Birmingham, AL). Ultra-TMB substrate (Thermo Scientific) was added followed by sulfuric acid (Thermo Scientific). Plates were washed and read on the Emax microplate reader (Molecular Devices, Sunnyvale, CA) at 450 nm (570 nm for reference) using the SoftMax Pro GxP 5.4.6 software (Molecular Devices). The lower limits of quantification (LLOQs) were 10 ng/mL and 1.25 ng/mL for plasma and CVL, respectively.

Cytokine and chemokine (CC/CK) concentrations in CVL supernatants before and after gel use
Interleukin (IL)-1β (lower limit of detection (LLD) 0.3pg/mL), IL1 RA (LLD 13.1 pg/mL), IL-6 (LLD 0.5 pg/mL), CCL5 (LLD 0.19 pg/mL), CXCL8 (LLD 0.4 pg/mL), CCL20 (LLD 70.4 pg/ mL) were measured using an electrochemiluminescence (ECL) platform (Meso Scale Discovery (MSD), Gaithersburg, MD) [40]. Each ECL immunoassay was optimized to allow detection of each biomarker within the linearity concentration range in the clarified CVL samples. IL1RA was tested in duplicates at 50-and 100-fold dilutions and when values exceeded the precision range of the assay, at an additional 1000-fold dilution. Other mediators were tested at a 2-fold or a 4-fold dilution, and where applicable undiluted, to generate duplicate values within linearity. The serial dilution of CVL samples resulted in reproducible values within the linearity range of each assay proving by accepted immunoassay methodology [41, 42] no matrix interference due to potential presence of study gel or other CVL matrix components. In addition, to rule out interference with the presence of gels, serial dilutions of a CG-containing gel were prepared in saline and mixed with equal volumes of biomarker solutions of known concentrations. Full CC/CK recovery was achieved in the concentration range of CG detected in CVL [43].

ADAs in serum
ELISA plates were coated with 20 μg/mL of GRFT. Following blocking and washes, serum samples and NM479 standards were added to the plate. For the samples, collected human serum was diluted 1:50. For the standard curve, 2X concentrated NM479 (1000 ng/mL-0.98 ng/mL) was diluted 1:1 in 25X diluted serum (Innovative Research), for a final serum dilution of 50X and final NM479 concentrations ranging from 500 ng/mL to 0.49 ng/mL. Then biotinylated KQ-GRFT was added to the plates. KQ-GRFT is a modified version of Q-GRFT that contains a lysine at the N-terminus of the protein. Then plates were washed followed by incubation with horseradish peroxidase conjugated with streptavidin (SA-HRP) (Pierce). ADAs were detected using TMB (SeraCare), and the reaction was stopped with addition of sulfuric acid. OD 450 nm was measured on a Biotek Synergy HT plate reader.

Anti-HIV-1 ADA-M activity of CVLs in TZM-bl MAGI assay
The multinuclear activation of a galactosidase indicator (MAGI) assay in TZM-bl cells (NIH AIDS Reagent Program) [45] was modified for evaluation of activity of GRFT in CVLs. Briefly, TMZ-bl cells cultured in 96-well white-opaque flat bottom microplates (Thermo Scientific) received different dilutions of non-clarified CVLs. GRFT solutions (between 100 and 0.015 ng/mL) prepared by diluting CMB-BDS-0900-003 GRFT stock (17 mg/mL) with cell culture medium were used as a control in the antiviral assay. All wells (including virus controls but excluding cell controls) were immediately challenged with 100-200 HIV-1 ADA-M infectious particles and stained using the MAGI assay following incubation [45]. All CVL dilutions and controls were tested in triplicate.

Anti-HPV activity in CVLs
The anti-HPV activity of selected CVL samples was tested in HeLa cells using the luciferase assay [46,47]. Briefly, HeLa cells cultured overnight at 10 5 cells/mL in 96-well clear flat bottom microplates (Thermo Scientific) were incubated with different dilutions of non-clarified CVLs in triplicate. A 3% carrageenan gel (Population Council, lot#160613B6500 placebo TZ) diluted to between 1000 and 0.15 ng/mL with cell culture medium was used as a control in the antiviral assay. All wells (including virus controls but excluding cell controls) were immediately challenged with HPV16 PsV (5x10 5 copies per well) and stained using the Pierce Firefly Luciferase Glow Assay (Thermo Scientific) [46] following incubation. All CVL dilutions and controls were tested in triplicate.

Anti-HIV-1 BaL activity in CVLs in ectocervical explants
Ectocervical explants (5x5mm) were prepared as we previously described [48], stimulated with PHA (5 μg/mL) and IL-2 (100 U/mL) and then challenged with 500 TCID 50 HIV-1 BaL /explant (in duplicate) in the presence of non-clarified CVLs. HIV-1 BaL was generated and titered as previously described [43]. Ten μL of viral stock diluted in CVL (500 TCID 50 of HIV BaL ; 1.79μl of virus and 8.21μl of CVL) was applied on the top of the epithelium for 2h at 37˚C in 5% CO 2 followed by a wash out. Tissues were then cultured in complete DMEM for 14 days (d). Where feasible, untreated control (culture medium) was included. Also, to discriminate between the viral inoculum and de novo viral replication lamivudine (3TC: 4μL of 500μM stock applied on the epithelium) control was included. Tissue culture supernatants were collected right after washes (d0) and during culture (d3, 7,11,14).

Epithelial integrity and ectocervical tissue viability in vitro
We explored if exposure to PC-6500 induces tissue toxicity, which may interfere with testing of anti-HIV BaL activity of CVLs. Effects of PC-6500 on epithelial integrity and ectocervical tissue viability were analyzed by Hematoxylin and Eosin (H&E) staining and the MTT assay [3-(4,5-dimethyl-2-thiazolyl)-2,5-diphenyl-2H-tetrazolium bromide] assay [50], respectively. Human ectocervical tissues without gross pathological changes were obtained from routine hysterectomies through the National Disease Research Interchange (NDRI; Philadelphia, P) and transported overnight (ON) in RPMI medium. 5x5mm and 3x3mm explants were prepared as we previously described for macaque and human cervicovaginal tissues [48,50]. Undiluted gels, PC-6500 (batches 131004B6500TR and 141027A6500TZ) and PC-535 (batch 130918A525TR), were applied on the epithelial surface of 5x5m explants and left ON (polarized cultures). Alternatively, 3x3mm explants were immersed in 1:30 and 1:100 diluted gels ON. Following incubation, tissues were processed for H&E staining (polarized cultures) or the MTT assay (immersion cultures).

Statistical analysis
As is typical for first-in-human studies, no formal sample size calculation was carried out for this study. A sample size of seven participants in the single dose OL portion of the study, and 20 participants in the randomized portion (14 active; 6 placebo) was deemed sufficient to evaluate the safety of PC-6500 and to assess the PK of GRFT. Number and percent of participants with TEAEs are summarized by treatment arm and relationship and toxicity. Descriptive statistics are presented by treatment arm for demographic and baseline characteristics and for clinical endpoints. Analysis was done with SAS Version 9.4.
Log-normal generalized linear mixed models were used to analyze changes in CC/CK concentrations in CVL adjusting for treatment type. Treatment (PC-6500 or placebo) was a fixed effect in the model and a random effect for each participant was included. Values below LLD were assumed LLD/10. Pairwise differences were calculated using Tukey-adjusted 95% confidence intervals (CIs).
For transcriptome analysis, edgeR � v 3.16.5 [51] was used to normalize the samples and Voom from limma# v 3.30.11 was applied to estimate the differential log fold change in the expression of genes. An FDR adjusted p value <0.05 was used to determine if there was a change in cervical gene expression post gel application.
For analysis of anti-HIV-1 ADA-M activity of CVLs in TZM-bl MAGI assay, the EC 50 and 95% CIs were calculated using a curve-fitting analysis (GraphPad Prism 5.0 Software, Inc., La Jolla, CA). GRFT concentrations and anti-HIV activity in CVLs were correlated using the Spearman correlation analysis (GraphPad Prism). Similarly, for analysis of anti-HPV activity in CVLs, the EC 50 and 95% CIs were calculated using a curve-fitting analysis (GraphPad Prism).
For analysis of anti-HIV-1 BaL activity of CVLs in ectocervical explants, log-normal generalized linear mixed models were used. The models examined the relationship between SOFT and CUM as the independent variables and treatment type, CVL collection time and their interaction as the dependent variables (fixed effects). A random intercept for each participant was included. Pairwise comparisons were calculated and stepdown simulation-adjusted 95% CIs reported. GRFT concentrations and anti-HIV activity in CVLs were correlated using the Spearman correlation analysis (GraphPad Prism).
For analysis of tissue viability after exposure to gels in vitro, log-normal generalized linear mixed models with random intercepts were used. Optical density (OD) 570/tissue weight (g) was the response variable. Predictors were gel treatments. Tissue donor was included as a random effect (SAS).

Disposition and demographics
Seven participants were enrolled in the single dose OL period, all of whom completed the study (October-November 2017). As summarized in CONSORT Fig 1, 24 women were screened for the randomized portion of the study and 15 participants enrolled: 12 in the PC-6500 group and 3 in the placebo group. Thirteen participants (10/12 PC-6500; 3/3 placebo) completed the study (March-August 2018). Of the nine women who did not enroll, one chose to withdraw after starting the screening process and the other eight were ineligible due to underlying conditions (n = 6), including anemia, irregular bleeding, and trichomonas; inability to tolerate the pelvic exam (n = 1) and the investigator's discretion (n = 2). Data collection ended based on the timeframe and resources available from the donor. Two participants in the PC-6500 arm withdrew prior to dosing, one due to a trichomonas infection detected at enrollment, and the other for personal reasons.
Participants in both periods of the study were approximately 32 years of age, on average, representing a mix of races, ethnicities and other background characteristics (see Table 4). Fig 2 illustrates the study design, including in-clinic and at-home dosing. All 7 participants in the OL period received one PC-6500 dose administered by the study clinician. The median dose exposure was 13 (range [3][4][5][6][7][8][9][10][11][12][13][14] in the randomized PC-6500 group and 14 (range 8-14) in the placebo group. Two of the three participants in the placebo group inserted all 14 doses per dye stain applicator (DSA) testing [38,39]. In the randomized PC-6500 group, five of the ten participants inserted all 14 doses per DSA testing. Of the five participants in the randomized PC-6500 group who did not insert all applicators per the DSA, two inserted 13/14 doses, one inserted 12/14 doses, one inserted 8/14 doses and one appeared to have inserted only 1 out of 14 doses, including 4 of the 5 doses self-inserted in the clinic.

Clinical safety
A total of 34 treatment emergent adverse events (TEAEs) were reported (Table 5) in five out of seven participants in the OL period and in 11 out of 13 participants (8 PC-6500, 3 placebo) in the randomized period. Only one event (vaginal discharge reported by a woman in the randomized PC-6500 group) was considered possibly related to study drug. The event was moderate in severity and resolved spontaneously without any intervention.
Most adverse events (AEs) were transient and mild. Three TEAEs were judged to be of moderate severity: hyperkalemia, thrombocytopenia and hypophosphatemia, none of which were related to study drug or required additional interventions; these TEAEs occurred in 2

PLOS ONE
Phase 1 trial of griffithsin formulated in a carrageenan vaginal gel participants in the PC-6500 arm. TEAEs that required additional interventions were seasonal allergies (medication), and musculoskeletal pain (medication).
No serious AEs occurred. There were no clinically significant abnormalities found in physical or pelvic exams, and no clinically relevant shifts in any laboratory parameters.

Histopathology
Ectocervical and vaginal biopsies were collected from 15 participants. All 15 participants had baseline (BL) biopsies and 13 had biopsies collected both at BL and one day after last gel application (10 PC-6500, 3 placebo) (n = 1 ectocervical and n = 1 vaginal). Evaluation of histologic data disclosed no findings of dysplasia or hyperplasia at BL or post gel administration. Most of the biopsies had mild to moderate inflammation (grades 1-2) in the epithelium/stroma at BL and post PC-6500 and placebo gel administration (S1 Table). In paired BL and post gel samples within PC-6500 group, the semiquantitative scores in cervical epithelium and stroma ranged from 0 to 2.5+ at BL and from 0 to 2 post gel. In vaginal epithelium and stroma the scores ranged from 0 to 1+ at BL and from 0 to 2 post gel. In placebo group, the scores in cervical epithelium and stroma ranged from 0 to 1 at BL and from 0.5 to 1+ post gel. In vaginal epithelium and stroma the scores ranged from 0 to 2 at BL and from 0 to 2+ post gel. No worsening of inflammation post gel administration was evident (S1 Table).

GRFT PK in plasma
A total of 77 plasma samples from the OL period were collected, of which 23 were unevaluable due to coagulation resulting in a non-specific background. A total of 54 (71%) samples were analyzed. All samples from the randomized period were analyzed. GRFT levels in all samples

PLOS ONE
Phase 1 trial of griffithsin formulated in a carrageenan vaginal gel were below LLOQ (10 ng/mL). As GRFT was undetectable after a single dose, or after multiple doses, no PK parameters were calculated.

GRFT concentrations in CVLs
Among participants randomized to PC-6500, CVLs collected at BL, 4h (or 6h) and 8h after single gel administration and 24h after dose 14 were evaluated. GRFT concentrations post gel administration ranged from between 0.0 and 82.1 μg/mL (Fig 3).

CC/CK concentrations in CVLs
Biomarkers of inflammation were chosen for their established role in vaginal inflammation and HIV risk, reliable detection in cervicovaginal secretions, and clinically validated acceptance for microbicide safety evaluation [41, 52, 53]. CC/CK were measured in CVLs collected at BL and 24h post last gel administration in the randomized study (Fig 4)

Cervical tissue transcriptome
Gene expression was analyzed in ectocervical biopsies collected at BL and 24h post last gel administration in the randomized period. An MDS plot revealed that most of the analyzed samples clustered together (Fig 5A), suggesting similar tissue gene expression before and after gel exposure. No specific patterns of gene expression before or after gel exposure were evident on the heatmap presentation ( Fig 5B). Gene expression after repeated gel administration was similar to BL, as shown on the volcano plots (Fig 5C and 5D).

ADAs
ADAs were evaluated in the randomized period of the study. No ADAs were detected at the tested time points (BL; d14, d21 and d28 post first gel application). Only one participant had a low positive optical density (OD) value of approximately 0.1 at BL. However, because we did not observe any increase in the later time points, we attributed the positive signal at BL to be an artifact or a pre-existing immune response to something that cross-reacts with GRFT.

Pharmacodynamics
Anti-HIV-1 ADA-M and anti-HPV16 PsV activities in CVLs. The antiviral activity of CVLs was tested in well-established TZM-bl and luciferase assays, as we previously published [45,46]. EC 50 values in the TZM-bl assay (0.7 ng/mL ± 0.6) based on GRFT concentrations suggest that CVL components did not affect the antiviral properties of GRFT. We were unable to estimate EC 50 values in two CVLs where GRFT was quantified due to a sub-optimal curve fitting analysis (R square <0.8 and 95% CIs higher than 1log). As shown in Fig 6A, samples with high GRFT concentrations resulted in lower EC 50 values based on CVL dilution; the more GRFT in the sample, the more dilution required to reach the EC 50 value. Anti-HPV activity was evaluated in all CVLs from participants in the placebo and PC-6500 groups. CVLs  (Fig 6B). Lack of anti-HPV activity in CVL samples was noted for one participant in PC-6500 gel group, who appeared to have inserted only 1 of 14 doses according to the DSA.
Anti-HIV-1 BaL activity of CVLs in human ectocervical explants. Our previously published methodology for testing CVL activity in explants using an immersion tissue culture model [43] was adapted for use with polarized tissue cultures. PHA/IL2 activated explants  (Fig 6C). GRFT concentrations in CVLs directly correlated with HIV-1 BaL inhibition (Fig 6C). To rule out potential contribution of toxicity in these experiments, epithelial integrity and tissue viability post gel application were evaluated in vitro. No changes in epithelial integrity (e.g., epithelial fractures) or decreased viability were observed (S2A and S2B Fig).  and (C, D) volcano plots. MDS plot was prepared through plot MDS function within edgeR (doi: 10.1093/bioinformatics/btp616). The heatmap was constructed by edgeR using top 500 genes based on LogCPM. Induction (red) and inhibition (blue) of expression (z-scores) are shown. The volcano plots of gene expression at BL and post gel administration according to the fold change and adjusted p values (q < 0.05) were prepared through in-house R code. The y-axis corresponds to the mean expression value of log10 (FDR; q), and the x-axis displays the log2 fold change (FC) value. Samples from 9 subjects in PC-6500 group and from n = 3 subjects in placebo group were analyzed. Due to poor RNA quality, sample from one subject in PC-6500 group was excluded from the analysis. https://doi.org/10.1371/journal.pone.0261775.g005

Discussion
In this first-in-human trial of 0.1% GRFT in a CG gel (PC-6500) there was no measurable systemic absorption, and no safety signals were detected in healthy, HIV-negative women using the gel for up to 14 consecutive daily doses. PC-6500 remained active against HIV and HPV in the presence of vaginal fluids in vitro. These results concur with preclinical assessments of GRFT and the GRFT/CG combination by the Population Council and others, demonstrating lack of systemic absorption, safety and activity against HIV and other STIs (HPV and HSV) [33,34,54].
No significant histopathological changes in cervical and vaginal mucosa and no ADA were detected. Consistent with this data, toxicology studies in small animal models following repeated vaginal administration of a GRFT/CG gel or intravenous (IV) administration of GRFT demonstrated no adverse events and showed that GRFT/CG gel is non-irritating. IV administration of GRFT in rats resulted in no detectable anti-drug antibodies (ADAs) [54].
Furthermore, rather than an increase in inflammatory CC/CK post gel application, a decrease in concentrations of CXCL8, CCL5 and CCL20 was detected post PC-6500 administration. Notably, CXCL8 and CCL20 are induced early during vaginal SIV infection in macaques followed by secondary inflammatory process likely driven by CCL5 and other CKproducing cells, leading to recruitment of immune cells and fueling infection [55]. Increased gene expression of CXCL8 and CCL20 in the cervix following exposure to Nonoxynol-9 (N9) gel is considered to be a "harm signal" [56]. CCL5 is involved in both the blocking of HIV binding to CCR5 on target cells and the recruitment of these target cells to the female genital tract [57][58][59]. The decreased concentrations of CXCL8, CCL5 and CCL20 post gel administration suggest a possible anti-inflammatory effect of PC-6500 gel.
Analysis of changes in gene expression may provide additional safety information in PrEP trials. PC-6500 or placebo gel did not induce changes in the ectocervical tissue transcriptome.
CVLs from women who used PC-6500 showed potent anti-HIV activity in cell-based and explant assays. In the polarized cervical explant model, which allows application of CVL and HIV on the epithelial surface mimicking the in vivo scenario, infection inhibition was detected after exposure to CVLs collected 4h or 6h post gel administration, but not 8h post gel. These data are consistent with in vivo data demonstrating significant protection of GRFT/CG FDI against vaginal simian human immunodeficiency virus (SHIV SF162P3) challenge [54]. DMPA-treated macaques were challenged with high dose vaginal SHIV SF162P3 4h after vaginal administration of either GRFT/CG FDIs (2 of 10 infected) or control CG FDIs containing all the same components except GRFT (10 of 10 infected). The activity of GRFT against SHIV SF162P3 infection in mucosal targets (in explants) was predictive of in vivo efficacy [54]. GRFT/CG FDIs were also effective against HSV-2 in mice [54]. Consistent with our published data on anti-HPV activity of CG and GRFT [34,54], CVLs from participants in the PC-6500 and placebo groups inhibited HPV in vitro.
The study had several limitations. Although no sample size calculation was carried out, which is typical for first-in-human studies, the small sample size resulted in imbalance in baseline characteristics and may have been underpowered to detect some PD effects and may have limited the safety, PK and PD assessments. The potential limitations of the small sample size were compounded by imperfect adherence in half of the women in the PC-6500 group, which may have further limited the chances of detecting safety signals. In particular, the DSA results that indicated one participant inserted only a single dose of PC-6500 highlights the importance of emphasizing and assessing adherence even in early-stage clinical trials [60]. The study was not designed to assess intra-person heterogeneity or to formally monitor ADA response, which involves testing of samples over an extended period of time. Measurement of cervicovaginal fluid dilution in CVL was not feasible. Variation in the cervicovaginal fluid volume could have contributed to the variation in detected GRFT concentrations, CC/CK concentrations and could have affected PD results. Although we did not detect CG-mediated anti-HIV activity in CVLs derived from subjects administered placebo gel, we cannot exclude contribution of CG to the observed infection inhibition [43,61]. Given modest anti-HPV activity of GRFT, it is likely that observed anti-HPV activity in the CVLs was mediated by CG [34].
In conclusion, PC-6500 represents a promising on-demand MPT product. PC-6500 containing potent non-ARV lectin GRFT may be preferred for individuals at high risk of HIV acquisition and who do not know their HIV status, decreasing the risk of developing drug resistance. Due to the anti-HPV and anti-HSV-2 activity of GRFT and CG, PC-6500 may also decrease HPV infection and HSV-2 infection/shedding, bolstering the anti-HIV activity of this prevention product. Lack of significant adverse events in clinical, laboratory or histopathological evaluations points to the safety of this MPT product.
Overall, a GRFT-containing on-demand or sustained-release product would fill an important gap in the current ARV-dominated prevention product pipeline. Data from this trial support the further clinical testing of GRFT/CG formulations for informing doses of GRFT for future trials and other GRFT delivery systems.