Rapid Point-of-Care Test Kit for Bacterial Vaginosis: Detection of Vaginolysin and Clue Cells Using Paper Strips and a Smartphone.

Despite the ongoing investigation into understanding the mechanisms behind BV, there is still a pressing need for a simple, reliable, and cost-effective diagnostic method for BV. BV is inherently challenging to diagnose as BV arises from the overgrowth of various polymicrobial flora accompanied by the decreased presence of the usual Lactobacillus -dominant vaginal microbiota. A pore-forming toxin, VLY, produced by one of the microbes commonly overgrown in BV, G. vaginalis , plays an important role in the pathogenesis of BV. 26 , 31 , 32 The susceptibility of human erythrocytes to VLY-mediated hemolysis, specifically human cervical epithelial cells and vaginal epithelial cells, has been previously reported, and an ELISA for VLY was developed accordingly. 29 , 30 Building upon this knowledge, we developed an inexpensive POC detection method, which includes identification of the toxin VLY, the increased presence of which can indirectly indicate an unhealthy vaginal microenvironment, via a LFA, coupled with a cutting-edge, easy-to-use, and portable visualization method for the identification of clue cells, the pillars of BV. A sandwich-format colorimetric microtiter plate-based immunoassay was then developed for the detection of VLY to create a platform for translation to a lateral flow assay. For this purpose, anti-PLY antibodies were coated on the surface of the microtiter plate wells. In the presence of VLY, anti-PLY antibodies capture the antigen. Anti-VLY antibodies are then added to the wells, which bind to a different epitope of the antigen. Finally, an HRP conjugated anti-human secondary antibody then binds to the VLY-specific primary antibody, and a signal is produced as a result of the HRP oxidizing the TMB substrate. Each step of the immunoassay, as well as the reagent concentrations, were optimized to achieve the optimum sensitivity and the selectivity of the assay while keeping the assay time minimum (see Supporting Information ). Following immunoassay optimization, the assay was evaluated for precision and accuracy ( Figure 1 ). Spike recovery tests were performed over five consecutive days to determine the accuracy of the assay by spiking known concentrations of VLY and calculating the interpolated results back to the measured concentrations. For these studies, we selected three different concentrations of VLY that fall within the linear range of the immunoassay. Specifically, a low concentration was selected that is close to the detection limit (37.0 ng/mL), a mid-range level (147.0 ng/mL), as well as a high concentration that is at the tail end of the linear portion of the assay (440.0 ng/mL). All percent recoveries are within the ±20% clinically accepted values indicating very high accuracy for our VLY assay and are found in Table 1 as Recovery Min, Recovery Med, and Recovery Max. 33 The precision of the assay was calculated by measuring the same three concentrations that were described above and calculating the CV for each CV Min, CV Med, and CV Max that correspond to concentrations 37.0, 147.0, and 440.0 ng/mL, respectively. Data are an average of six measurements on five separate days ± 1 standard deviation. Our results show that the CV is less than 10% for each value, indicating high precision for our assay. The assay validation data are summarized in Table 1 . The precision of the data was only calculated for the linear portion of the calibration curve as described by He et al. 33 Following immunoassay validation, patient samples were tested in duplicates using the developed ELISA. Swabs were collected from female patients ( n = 16) and processed as described in the Materials and Methods section. The negative control was a negative sample confirmed by both the Amsel criteria and Nugent scoring, and the positive control was a negative sample confirmed by both the Amsel criteria and Nugent scoring spiked with 728 ng/mL of the purified VLY. Purchased patient samples from Discovery Life Sciences confirmed positive for BV were also tested. All patient samples tested showed elevated levels of VLY, indicating a 100% agreement of the positive BV diagnosis obtained via the Amsel criteria and Nugent scoring ( Figure 2 ). The ELISA was then translated to a lateral flow assay for future use at the point of care. In the field, the sample will be added to the paper strip and the solution will migrate, via capillary action, through the various regions of the strip. If the sample contains VLY, the VLY will bind to colored conjugate particles. Two colored lines will be indicative of a positive test, a single line in the control region is indicative of a negative test, and the absence of a control line is an inconclusive test. Each component of the LFA was optimized to yield the lowest LoD, including the type of membrane, the pore size of the membrane, the concentrations of the primary antibody, secondary antibody, and cellulose nanobeads, as well as the composition of the lateral flow running buffer (LFRB), which included 10% sucrose and 1% Tween 20 for the facilitation of flow through the pores of the nitrocellulose membrane. A CN140 nitrocellulose membrane (Sartorius, Göttingen, Germany) was selected as the microfluidic platform as this membrane demonstrated minimal to no nonspecific binding (DCN Diagnostics, Carlsbad, CA). CNBs were used due to their increased sensitivity, faster detection time, and improved reproducibility when compared to colloidal gold and because they do not require external equipment to visualize. 34 Reproducible membrane strips were manufactured using a BioDot XYZ3210 Dispense Platform (Biodot, Irvine, CA) capable of printing precise antibody test and control lines mirroring commercially available LFAs. Patient samples were processed as described in the Materials and Methods section. Analytical sensitivity, the assay’s ability to detect a low concentration of a given substance, is usually reported as the limit of detection. 35 The limit of detection (LoD) for the LFA was calculated using the optimized lateral flow assay described above. The LoD for both spiked VLY in LFRB and spiked VLY in a patient sample was determined by titrating the level of toxin from 38.8 ng/mL to 1.165 μ g/mL with a positive control strip of 365 μ g/mL in duplicates. The pictures of the developed strips were then recorded, and the color density of the test lines was measured using ImageJ software. ImageJ results showed that even at the lowest concentration of 38.8 ng/mL, the signal intensity was statistically different than the blank ( Figure S8 ). This band, however, was not easy to see with the naked eye. Therefore, qualitative determination of the results using the naked eye was also performed. For this purpose, the results were interpreted by five independent individuals, all of whom were blinded to the study. All five individuals selected the same strips, which indicate the current LoD of both the LFA in buffer and the LFA in spiked patient samples as 194 ng/mL demonstrating the sensitive nature of the assay developed thus far and the functionality of the assay in the vaginal matrix ( Figure 3 ). Following the optimization of the LFA, the assay was run using purchased BV positive patient samples ( n = 6), Amsel-confirmed positive BV patient samples ( n = 16), Amsel-confirmed negative BV patient samples ( n = 5), a negative control using only LFRB, and a positive control using VLY (194 ng/mL) spiked in LFRB. After 30 min, all positive patient samples presented two bands, all negative patient samples presented one band, and the negative and positive control samples presented one and two bands, respectively ( Figure 4 ). These findings are 100% in agreement with the Amsel score-based diagnosis for these samples, which is the currently accepted clinical criterion. Our assay showed 100% selectivity and 100% sensitivity and correspondingly demonstrated 100% negative predictive value and 100% positive predictive value, respectively. The calculation of the assay selectivity and sensitivity as well as their descriptions are given in the Supporting Material . As some bands were fainter than others, indicating a lower level of VLY, which was confirmed via the ELISA, samples were then investigated for the presence of clue cells. Traditional diagnosis of BV using the Amsel criteria requires the use of a microscope, which is a significant expense that not only requires electricity but also trained technicians for proper use. A simple to use, inexpensive, battery-operated DIPLE phone microscopy kit (SmartMicroOptics, Genova, Italy) was employed to visualize clue cells at the PoC as described in the Materials and Methods section. Briefly, collected samples both positive (n = 22) and negative (n = 3) for BV were Gram stained and observed using the DIPLE kit via the DIPLE black lens and an iPhone 11 Pro version iOS 14.4.2. The DIPLE black lens provides magnification of 150× without using the digital zoom function of the cell phone. The battery-operated light source was powered on, the prepared slide was inserted into the slide holder, the DIPLE black lens was lowered onto the viewing area and screwed into place, and the phone was rested directly on top of the DIPLE black lens ( Figure 5 ). Four photos were taken of the slide at random, one photo per quadrant of the smear. The presence of 20% or more of clue cells, which appear fuzzy and purple in the image due to the presence of bacteria covering the epithelial cell, in all of the four images combined were indicative of a positive parameter and the lack thereof indicated a negative parameter ( Figure 5E , F ). All samples confirmed positive for BV via the Amsel criteria and Nugent scoring presented more than 20% presence of clue cells, and all samples confirmed negative via the Amsel criteria and Nugent scoring either presented less than 20% clue cells or no clue cells at all demonstrating the efficacy of this visualization method.

The 516 amino acid sequence for VLY was obtained from the NCBI GenBank (accession number EU522488 , “vaginolysin [ G. vaginalis ]”) for expression in E. coli . The VLY sequence was synthesized commercially (GenScript Biotech, Piscataway, NJ) and inserted into the pET-30 a (+) plasmid (New England Biolabs Inc., Ipswitch, MA) vector using the Bam HI and HindiIII restriction sites of the multiple cloning site. The plasmid was then transformed into the BL21(DE3) expression strain. Small cultures were prepared for overnight growth at 37 °C by inoculating 5 mL of Miller LB supplemented with 30 mg/mL kanamycin from glycerol stocks. Bacterial genomic DNA from culture was extracted using a QIAprep spin miniprep kit (Qiagen, Hilden, Germany) according to the manufacturer’s instructions. For specific molecular identification of the VLY gene, the primers R (5′ ATT ATT CGT GGA TCC GTC GTT CTT CAC GGT 3′) and F (5′ TAA TAA GCA CAT ATG ATG AAG AGC ACC AAG) were used. The reactions were performed in a final volume of 50 μ L containing 0.5 μ M of each primer, 1.0 μ L of DNA template, and 25 μ L of Q5 Hot Start High-Fidelity 2X Master Mix (New England Biolabs, Ipswich, MA). The conditions for amplification of the VLY gene were initial denaturation at 98 °C for 30 s (sec) followed by 30 cycles of denaturation at 98 °C for 10 s, annealing at 67.5 °C for 30 s, extension at 72 °C for 1 min followed by a final extension at 72 °C for 2 min. PCR reactions were performed in triplicate in an Eppendorf Mastercycler X50s (Eppendorf, Hamburg, Germany). The amplicons were visualized on a 1.0% agarose gel in 1.0X TAE after electrophoresis at 110 V by means of a transilluminator ultraviolet light. Quick-Load 100 bp DNA ladder (New England Biolabs, Ipswich, MA) and Gold Bio 1kb DNA ladder (Gold Biotechnology, St. Louis, MO) were used as molecular weight standards. As a quality control, Sanger sequencing was performed by Genewiz (South Plainfield, NJ). Small cultures were prepared for overnight growth at 37 °C by inoculating 5 mL of Lysogeny Broth (Miller) supplemented with 30 mg/mL kanamycin from glycerol stocks. Following overnight growth, small cultures were used to inoculate large 200 mL cultures, which were grown to an OD 600 of 0.6 in Lysogeny Broth (Miller) supplemented with 30 mg/mL of kanamycin. The cultures were then centrifuged at 8000 g for 10 min at 4 °C and resuspended in fresh LB Broth (Miller) (Thermo Fisher Scientific, Waltham, MA) supplemented with 30 mg/mL kanamycin. Cultures were then induced with a final concentration of 0.1 mM isopropyl β - d -1-thiogalactopyrano-side (IPTG) and allowed to grow overnight at room temperature (RT). The cells were collected by centrifugation at 8000 g at 4 °C for 10 min and resuspended in a denaturing lysis buffer of 100 mM sodium phosphate,10 mM Tris base, and 8 M urea, pH 8.0. The cell suspension was supplemented with 1× ProBlock Gold Bacterial Protease Inhibitor Cocktail (Gold Biotechnology Inc. St. Louis, MI) to prevent protein degradation following cell lysis, which was subsequently carried out via sonication. The insoluble material was removed via centrifugation at 17,000 g at 4 °C for 20 min, and the crude protein was then allowed to incubate with 1 mL of Ni-NTA agarose rotating for 1 h at 4 °C. The solution was collected on a disposable column by gravity flow and washed with 20 column volumes of denaturing lysis buffer followed by 20 column volumes of denaturing lysis buffer supplemented with 20 mM imidazole. Then, 20 column volumes of denaturing lysis buffer were passed through the column in decreasing increments of 1 M urea until there was no urea in the buffer. The protein was then eluted in 1 mL increments with an elution buffer of 50 mM Tris base, 150 mM sodium chloride, 250 mM imidazole, 1% (vol) nonyl phenoxypolyethoxylethanol (NP-40), 0.2% (vol) Tween 20, and 10 mM 2-mercaptoethanol ( β -ME). Protein elution fractions were analyzed via SDS-PAGE using 4–20% gradient Mini-PROTEAN TGX Precast Gels (Bio-Rad Laboratories Inc., Hercules, CA) under denaturing conditions with running buffers containing sodium dodecyl sulfate (SDS). The elution fractions with a purified VLY band present at 57 kDa were collected and concentrated via an Ultra Centrifugal Filter Device 30 kDa filter, and the purified VLY was then dialyzed using Micro Tube-O-DIALYZER Dialysis devices (G-Biosciences, St. Louis, MO) with a molecular weight cutoff (MWCO) of 15 kDa into phosphate buffered saline (PBS) of 10 mM sodium phosphate, pH 7.2, and 150 mM sodium chloride. The concentration of purified VLY was measured using a nanodrop device (Thermo Fisher Scientific, Waltham, MA) as well as using a Pierce BCA Protein Assay Kit (Thermo Fisher Scientific, Waltham, MA). Typically, 50 and 500 μ g/mL of the purified VLY protein were separated on a 4–20% polyacrylamide gel (Bio-Rad Laboratories, Inc., Hercules, CA), which was then transferred to a 0.2 μ m polyvinylidene difluoride (PVDF) membrane (Bio-Rad Laboratories, Inc., Hercules, CA) and blocked with Odyssey Blocking Buffer (PBS) (LI-COR Biosciences, Lincoln, NE) for 2 h at RT. After washing the membrane with PBST (PBS with 0.05% Tween 20), the membrane was probed overnight at 4 °C using recombinant human anti-vaginolysin (anti-VLY) monoclonal IgG1 antibody from the hybridoma 9B4 (Absolute Antibody, Oxford, United Kingdom) and mouse anti-pneumolysin (anti-PLY) (MyBioSource, Inc., San Diego, CA) monoclonal IgG1 antibodies diluted 1:2500 in blocking solution, both of which have been previously shown to bind to the VLY toxin. 30 The following day, the membranes were again washed three times with PBST, and detection was carried out using IRDye 800CW goat anti-mouse IgG secondary antibody (LI-COR Biosciences, Lincoln, NE, 1:3000 dilution in blocking buffer). The membranes were then scanned using a LICOR Odyssey CLx (Lincoln, NE) imaging system. The wells of a 96-well, clear bottom, high-binding polystyrene microtiter plate (Thermo Fisher Scientific, Waltham, MA) were coated with 100 μ L of 1.0 μ g/mL anti-PLY antibody in coating buffer (100 mM NaHCO 3 , pH 9.60) overnight at 4 °C. The next day, the immunoassay procedure was carried out via a sandwich-format colorimetric immunoassay in BV incubation buffer (10 mM sodium phosphate dibasic, 2 mM potassium phosphate monobasic, 137 mM sodium chloride, 2.7 mM potassium chloride, 0.05% Tween 20, 1.0% BSA, and 2% poly(ethylene glycol), pH 7.40) at RT with a washing step performed between each incubation step using wash buffer (10 mM sodium phosphate dibasic, 2 mM potassium phosphate monobasic, 137 mM sodium chloride, 2.7 mM potassium chloride, 0.1% Tween 20, pH 7.40). The washing steps were carried out using Molecular Devices (Sunnyvale, CA) MultiWash+ Plate washer using five cycles of 250 μ L/well of wash buffer, employing a 10 sec shaking step at the end of each cycle. Following the overnight coating step, 100 μ L of purified VLY spiked in BV incubation buffer was added at various concentrations between 5.0 ng/mL and 0.7 μ g/mL into the wells and incubated for 90 min. BV incubation buffer was used as blank. Aliquots of 100 μ L of 0.25 μ g/mL anti-VLY antibody were then added to the BV incubation buffer and incubated for 30 min. Then, 100 μ L of biotinylated goat anti-human IgG Fc preadsorbed secondary Ab (Southern Biotech, Birmingham, AL) was added at a concentration of 62.5 ng/mL for 60 min into each well, followed by a 30 min incubation period of 100 μ L of Pierce Streptavidin Poly-Horseradish Peroxidase (SA-poly-HRP) (Thermo Fisher Scientific, Waltham, MA) at 50 ng/mL. The wells were then incubated with 100 μ L of 1-Step Ultra TMB-ELISA substrate solution (Thermo Fisher Scientific, Waltham, MA) for 2 min and followed by immediate addition of 100 μ L of KPL TMB Stop Solution (SeraCare Life Sciences Inc., Milford, MA) to stop the color development. The absorbance measurement for each well was recorded using a Clariostar Optima UV/Vis Spectrophotometer (BMB Labtech, Ortenberg, Germany). The wells of a 96-well microtiter plate were coated with 100 μ L of 1.0 μ g/mL of anti-PLY overnight at 4 °C. The wells of the plate were washed, and an aliquot of 100 μ L of VLY was added at varying concentrations from 5.0 ng/mL to 8.5 μ g/mL in BV incubation buffer. BV incubation buffer was used as a negative control. The data was fitted to a variable slope nonlinear sigmoidal dose–response curve using GraphPad Prism (GraphPad Software, La Jolla, CA) to generate the calibration curve for the assay validation. On the same plates but in different wells, 100 μ L of VLY was added at concentrations that correspond to low, medium, and high levels of the antigen (37.0, 147.0, and 440.0 ng/mL) in the BV incubation buffer, and the plate was incubated for 90 min. The remaining steps of the immunoassay were carried out as previously optimized. The generated signals corresponding to low, medium, and high concentrations of VLY were interpolated using the generated calibration curve to calculate the projected concentrations, and the corresponding percent recovery was calculated as an immunoassay parameter for accuracy. Finally, the limit of detection ( LoD ) and limit of quantitation ( LoQ ) values were interpolated from the previously generated calibration curves using LoD = S B + ( 3 × SD B ) and LoQ = S B + ( 10 × SD B ) , where S B and SD B are the mean and standard deviation of the blank measurements, respectively. The primary anti-PLY antibody was conjugated to commercially available colored cellulose nanobeads (CNBs) following the protocol provided (DCN Dx, Carlsbad, CA). A 1.0 mg/mL solution of primary anti-VLY antibody, which serves as the test line, and a 0.5 mg/mL solution of secondary goat anti-human antibody, which serves as the control line, were printed at 22 mm/s using a BioDot XYZ3210 Dispense Platform (Biodot, Irvine, CA) on a CN140 nitrocellulose membrane (Sartorius, Göttingen, Germany) and dried in a 40 °C incubator for 2 h. Preparation of the half-stick assays was achieved by attaching the freshly printed and dried nitrocellulose membrane to a backing card (DCN Dx, Carlsbad, CA) and overlapping a 2 mm section of the membrane with an adsorbent pad (MilliporeSigma, Burlington, MA) to facilitate capillary action. A Biodot CM5000 Guillotine Cutter (Biodot, Irvine, CA) was used for high-quality precision cuts of the LFA strips at 4 mm in diameter, and the strips were stored in a low humidity environment at RT. A portable microscope adaptor designed for use with smartphones and tablets, DIPLE (SmartMicroOptics, Genova, Italy), was employed for investigation into the presence of clue cells. A slide smear was prepared by adding one drop of sample to a microscope slide, and gram staining was carried out following the Gram staining protocol for bacterial differentiation (Gold Biotechnology, St. Louis, MO). Briefly, the slide was fixed, stained with gram crystal violet solution, stained with gram iodine solution, decolorized with ethanol and acetone, counterstained with gram safranin solution, and observed using the instructions included with the kit via the DIPLE black lens and the camera application of an iPhone 11 Pro version iOS 14.4.2. One photo was taken in each quadrant of the slide for a total of four photos, and the total number of clue cells and healthy vaginal epithelial cells were counted. Samples in which clue cells exceeded 20% were considered positive for this parameter of BV as defined by the Amsel criteria (23). Healthy reproductive-aged women in Miami, FL, were screened for BV first using the Amsel criteria and then confirmed positive for BV via Nugent scoring. The vaginal swabs were collected by swabbing the inside of the vaginal canal with a FLOQswab (Copan Diagnostics Inc., Murietta, CA) as approved by the Institutional Review Board committee (IRB protocol #20180758 titled, “Women, HIV, Immunology, Microbiome and Sexual Health (WHIMS)”). As a positive control, six patient samples confirmed positive for BV via sequencing for G. vaginalis were purchased from Discovery Life Sciences (Discovery Life Sciences, Los Osos, CA). Collected swabs, which were negative for BV using both the Amsel criteria and Nugent scoring, were vigorously vortexed in 500 μ L of the BV incubation buffer and spiked with VLY. Then, 100 μ L of the sample was added directly to a 96-well plate for ELISA investigation, while 50 μ L of the sample was mixed with 50 μ L of a separate lateral flow running buffer (LFRB) (12 mM phosphate-buffered saline containing 10% sucrose, 1.0% bovine serum albumin (BSA), 0.5% polyvinyl alcohol (PVA), 0.75% Triton X, pH 7.4). For LFA investigation, 0.5 μ L of conjugated CNBs were added to the 100 μ L of sample, and the strip was added directly to the microcentrifuge tube. Results were recorded after a 30 min incubation period. For clue cell investigation, 20 μ L of the sample in the PBS was added directly to a microscope slide, and the slide was Gram stained following the protocol provided by Gold Biotechnology (St. Louis, MO), and the sample was assessed following the instructions provided by the DIPLE manufacturer. Following ELISA, LFA, and microscopic optimizations, experiments were run directly using patient samples confirmed positive for BV via the Amsel criteria and Nugent scoring with the developed ELISA and LFA platforms as well as the DIPLE technology.

Herein this manuscript, we described a rapid, accurate, and easy method for the diagnosis of BV at the POC. Both an ELISA and LFA for the accurate and sensitive detection of the VLY toxin secreted by G. vaginalis was designed, developed, and validated in patient samples. The accuracy and precision of the methods developed are high, demonstrating the robustness of the assay. Human patient samples were also investigated for the presence of clue cells employing an easy-to-use portable, battery-operated microscopic device, DIPLE. These technologies, paired together, were able to accurately diagnose BV in 100% of the human vaginal samples tested that were confirmed positive via the Amsel criteria and Nugent scoring. The development of this unique pairing of diagnostic methods represents an important milestone, as the diagnosis of BV at the POC has been an ever-present challenge to address, especially in resource-limited settings in which routine testing is not performed due to lack of resources such as reagents and trained personnel. Overall, it has been extremely challenging to break down the specific epidemiology of BV for a plethora of reasons, the main ones being the lack of accurate and available testing for all populations. Thus, the development of this methodology described provides a platform to investigate the prevalence of BV in any setting including previously unreachable populations of underdeveloped countries where BV persists. Our approach and paper-based detection platform provide the foundation for expansion to develop tests for Lactobacillus, Mobiluncus, Prevotella, Bacteroides, Peptostreptococcus , and the other bacteria that leave the vagina in an unbalanced state. Additionally, shifting from the traditional microscopic diagnosis methodology to a smartphone-based approach allows for future incorporation of machine learning and artificial intelligence-based screening for objective identification of clue cells. Similarly, advanced technology is of great value for the analysis of LFAs. Applications are currently under development that blank subtract a test line from a control sample, minimizing background artifacts and eliminating the subjective quality that arises between differences in room lighting, cell phone models, and inherent human differences such as eyesight and judgement. 36 Synergistically, all of these components will ultimately provide for a fully comprehensive POC test in which any individual can diagnose bacterial vaginosis in any setting.

Bacterial vaginosis (BV) is the most common vaginal infection among reproductive-aged females and is characterized as a microbial imbalance of the vagina. 1 Various species of lactobacillus flourish in a healthy vaginal microbiome, while anaerobic and facultative bacteria are only present in low concentrations, if at all. This specific balance of flora produces hydrogen peroxide and lactic acid to maintain a low vaginal pH and create a natural barrier against foreign pathogens. 2 However, various activities such as sexual intercourse, douching, smoking, and the presence of an intrauterine device (IUD) can disrupt the balance of this delicate microenvironment, resulting in BV. 3 - 6 Approximately one-third of North America’s female population has BV, the high prevalence of which is partially attributed to the fact that only approximately half of affected females are symptomatic. 7 When present, symptoms include increased vaginal discharge and/or malodor, both of which can contribute to social isolation, decreased self-esteem, and an overall decreased quality of life. 8 Whether symptoms are present or not, the risks of untreated BV comprise a two-fold increase in the acquisition of sexually transmitted infections (STIs) due to the decreased host defense system of the imbalanced vaginal microbiome as a result of BV. 9 - 12 Apart from STIs and their associated diseases, research has investigated BV’s correlation with pelvic inflammatory disease (PID), low birth weight, spontaneous abortion, preterm delivery, and postpartum endometriosis. 13 , 14 It is important to note that there is an extensive list of possible variables that influence the epidemiology of BV, including but not limited to, sample collection method, ethnicity, socioeconomic status, contraceptive use, lack of circumcision with male partners, vitamin D levels, dietary influence, chronic stress, and genetic variation of host genes. 3 , 15 - 20 These variables are hypothesized to have an influence in the fact that there is a higher prevalence of BV found in developing countries than in developed countries (35 vs 24.8%). 21 However, the explanation for this discrepancy is difficult to unveil without widely available, easy-to-use methods of diagnostics for BV, an area which has been largely unsatisfied thus far. Although there is not one widely accepted gold standard for the diagnosis of BV, the most common methodology currently used to diagnose BV in the clinic is the Amsel criteria. 22 The Amsel criteria use four parameters to diagnose BV, three of which must be present to confirm a diagnosis, although the presence of two suffices albeit at the tradeoff of sensitivity and specificity. 23 , 24 The four parameters include a thin, white homogeneous discharge, a vaginal fluid of over pH 4.5, the release of a fish odor upon the addition of 10% potassium hydroxide solution to a wet mount (a “whiff” test), and the presence of clue cells, which are visualized as dark fuzzy epithelial cells, through wet mount microscopy via Gram staining. Interestingly, the presence of clue cells on their own has been shown to be 98% specific and 89.9% sensitive for BV. 24 Another method used for the diagnosis of BV, the Nugent score, is carried out solely through the assessment of vaginal smears under a microscope. This method uses a complex scoring system dependent on the presence of large Gram-positive rods ( lactobacillus morphotypes), small Gram-negative to Gram-variable rods ( Gardnerella vaginalis and Bacteroides spp. morphotypes), and curved Gram-variable rods ( Mobiluncus spp. morphotypes). The presence, or absence, of these bacteria is then counted and given a corresponding score. Patients scoring between 4 and 6 points are classified as having intermediate flora with future testing required, while patients scoring 7 or above are considered to have BV. Although the Amsel criteria are the most widely used due to their relative ease of execution in comparison to other methods available such as the Nugent scoring, both methods still require a microscope, electricity, and a trained technician to accurately identify clue cells through wet mount microscopy. To minimize the misdiagnosis of BV and the corresponding negative health complications, ultimately improving the overall wellbeing of the female population, a method of diagnosis that does not require electricity, expensive or large laboratory equipment, or extensively trained personnel is required. A POC diagnostic assay that a relatively inexperienced diagnostician can perform as efficiently as the most trained individual would revolutionize the diagnosis of BV, both in the clinics as well as in many remote regions of the world in which this condition persists. Accordingly, a proof-of-concept POC diagnostic was developed to identify key factors associated with BV. Gardnerella vaginalis ( G. vaginalis ), the bacteria with the highest frequency of overgrowth in patients diagnosed with BV, has recently been split into 13 different infectious strains. 25 G. vaginalis produces the toxin vaginolysin (VLY), quantification of which can be correlated with the current concentration of G. vaginalis . 26 VLY is a member of the cholesterol-dependent cytolysin family of toxins and recognizes the complement regulatory molecule CD59 on the surface of human cells. 27 Higher concentrations of G. vaginalis have been shown to lead to a harmful vaginal environment, facilitating BV. 28 Interestingly, there has been contradicting data concerning VLY’s role and concentration in vitro and in vivo, with little minimal data for the latter. 25 Because of this, two assays were designed building on previous VLY assay development and subsequently tested using clinical patient samples to investigate BV in a quick and easy fashion as the diagnosis of this condition significantly decreases the negative effects presented as a result of untreated BV. 27 , 29 To achieve this, a method was developed for the expression, purification, and detection of VLY via an enzyme-linked immunosorbent assay (ELISA), which was then translated into a point-of-care device by adapting it into a lateral flow assay (LFA) that delivers visual results in under 30 min (min). This paper-based technology was paired with a commercially available kit that harnesses smartphone technology to visualize clue cells, the identification of which can be used to diagnose BV in any setting without the need for a conventional microscope. The coupling of these two technologies minimizes the risk of overtreatment of BV that might result from the detection of vaginolysin alone. The combination of these two diagnostic methods was used to accurately characterize BV in comparison with the Amsel criteria and the Nugent score in a fast and easy-to-follow manner, using portable components that are significantly less expensive than the reagents and machinery currently required to accurately diagnose BV.

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acs.analchem.2c02094 . Methods for assay optimization as well as the demographic information of the patient samples are given in the supplementary material (Word Document) (PDF)