Towards a Buruli Ulcer Rapid Diagnostic Test that Targets Mycolactone 2

Abstract


Introduction
Buruli ulcer (BU) is a progressive necrotizing skin infection caused by Mycobacterium ulcerans.The disease starts as a painless papule which can progress to a nodule and eventually degenerate into a large ulcer where the skin no longer covers the underlying tissue [1,2].Buruli ulcer has been reported in 34 countries [3], and is mostly found in rural areas of West Africa, though a recrudescence has been noted in Australia [4] where its transmission has been ascribed to possums and other animal reservoirs [5,6].In 2019, before the COVID-19 pandemic which affected active case finding campaigns, 2,271 new cases were recorded worldwide, and the incidence rate remained at the same level with 2,121 cases reported in 2022 [7].The disease is believed to be vastly under-reported partly due to its presence in very remote communities hindering access to care, and also because of the lack of awareness, of reporting system, and of fieldfriendly diagnostic tools [8].
Buruli ulcer remains one of the neglected tropical diseases (NTDs) in need of improved diagnostic tests.
With oral antibiotic treatment available [9], decentralizing diagnosis would facilitate strategies for early detection, timely treatment, and prevention of permanent disability.However, there are no diagnostics suitable for point-of-care applications.The well-established diagnostics are based on direct M. ulcerans pathogen confirmation by in vitro culture (definitive positive diagnosis), PCR, smear examination for acidfast bacilli, or histopathology.These diagnostics require either a sophisticated laboratory or have low sensitivity.An isothermal amplification technique (LAMP assay) has been proposed as alternative to PCR and, while not commercial, is promising [10,11].A method based on thin layer chromatography has also been evaluated and works better with fine needle aspirates from nodules than with swabs from open wounds [12,13].
Given the importance of securing new diagnostic tools for Buruli ulcer and to help guiding the development of new diagnostic tools, the World Health Organization (WHO) in consultation with different .partners developed a preliminary Target Product Profile (TPP) in 2018 [14] which was revisited in 2022 [15].The TPP served as the starting point for the efforts described herein.Of particular relevance to this work was the desire stated in the TPP to see a diagnostic test capable of detecting mycolactone.
Mycolactone is a unique metabolite of M. ulcerans composed of a macrolide core substituted with a fattyacid sidechain (MW = 743.02).Mycolactone has cytotoxic, immunosuppressive, and analgesic effects that induce the large debilitating skin lesions that develop when the infection is not treated [16].It is further believed that because mycolactone production is related to the viability of the mycobacterium, a mycolactone test has the potential to be used not only for diagnosis, but also to assess response to treatment and cure.
Here, we report the development of a prototype rapid test for Buruli ulcer, formatted as a lateral flow assay capable of identifying mycolactone as the disease biomarker.The test was specifically designed to isolate trace amounts of mycolactone from wound exudates, and then analyze the concentrated mycolactone on a lateral flow assay with minimal sample processing.

Materials and Methods
Assay Principle.We devised a competitive lateral flow assay featuring two innovations.The first was the use of a mouse monoclonal antibody specific for mycolactone [17], which has previously shown value in an ELISA [18,19].The second innovation was the use of custom-made nanoparticles specifically designed to enable immunomagnetic concentration of the analyte mycolactone while being darkly colored and providing a visual signal when run on a lateral flow test.These nanoparticles, dubbed magnetic gold nanoshells (Mag-GNS) consist of a ferrite center, a silica gel middle layer, and a gold outer shell (Figure 2).
The ferrite center allows for magnetic concentration, the silica gel layer encapsulates the ferrite and lowers the overall density of the nanoparticles allowing for their resuspension in aqueous media, and the gold shell surface absorbs light extremely efficiently due to plasmon resonance effects.This quantum physics phenomenon comes into play when metal nanoparticles reach the nm scale, in this case creating a dark brown-gray color, nearly black, a surprising hue for a gold particle.The gold shell is further functionalized with carboxylic acid groups for covalent conjugation to a detector antibody, and was custom made by nanoComposix, a Fortis LifeScience Company (San Diego, CA, USA).microscopy views of the particles being built starting from a ferrite core, which is then coated with silica gel (greyish), which is in turn studded with gold crystals (little black dots), used as nucleation points to grow the gold shell.
. In parallel, a test strip was developed as shown in Figure 3, where a sample pad (glass fiber), nitrocellulose, and an absorbent pad were laminated onto a backing card.The key property of the strip was a test line made of a mixture of biotinylated mycolactone analog PG-204 (Figure 1) and polystreptavidin, which effectively immobilized the PG-204 probe at the test line.The strip also included a control line of antimouse antibody to capture magnetic gold nanoshells that migrated past the test line.The strip was placed in a plastic cassette for ease of use and transportation.
Combined with the magnetic nanoshells covalently conjugated to an anti-mycolactone antibody, the strip allowed for a competitive assay.If the sample to be analyzed is devoid of mycolactone, then the conjugate binds to PG-204, itself immobilized on the test line, and produces a black test line.Conversely, if the test sample contains significant amounts of mycolactone, the latter saturates the binding sites on the antimycolactone antibody, and the conjugate is no longer captured by PG-204 at the test line.In summary, the resulting test line intensity is inversely proportional to the concentration of mycolactone in the sample.Assay workflow.We envisioned that the workflow of the finished assay would be as shown in Figure 4, with its components described in the next paragraphs of the Materials and Methods section.The magnetic nanoshells, after being conjugated to anti-mycolactone antibody (the "conjugate"), would be lyophilized for storage and transportation, and reconstituted in the laboratory or at the point of care.Next, a swab .containing a wound exudate would be transferred to a tube containing extraction buffer (step 1).The reconstituted magnetic nanoparticles would be added (step 2), incubated with the extracted sample to capture the mycolactone (step 3), and concentrated by means of a magnetic rack (step 4).The supernatant would be removed (step 5) and the magnetic nanoshells resuspended in chase buffer (step 6) before being added to the lateral flow assay (step 7).Monoclonal antibodies.The generation of mouse monoclonal antibodies to mycolactone has been previously described [17].The clone chosen for this application was JD5.1, produced by Swiss TPH (Basel, Switzerland) purified from hybridoma culture supernatants by affinity chromatography using a HiTrap Protein A HP column (Cytiva).Two other anti-mycolactone clones were deprioritized: JD5.11 which gave problems with reproducibility, and clone LW2.2 which appeared to have slower binding kinetics.
. Conjugate.Magnetic gold nanoshells were conjugated to mouse monoclonal antibody JD5.1 using EDC/sulfo-NHS chemistry.The magnetic nanoshells conjugated to JD5.1 will be referred to below as "the conjugate".
Lyophilization of the conjugate (optional).The conjugate was stored at 4 °C, and either used as such in solution, or later in the development cycle, in order to match the workflow of Figure 4, was freeze-dried as 3% solution in lyophilization medium (5% sucrose, 5% trehalose, 0.1% surfactant 10G in deionized water).
Mycolactone and mycolactone analogs.Synthetic mycolactone A/B, biotinylated analog PG-204, and acetyl analog PG-119 were prepared by the group of Prof. Altmann, ETH-Z, and stored frozen at -20 °C in DMSO [20,21].Mycolactone sourced from WHO in ethyl acetate gave inferior results for reasons that were not investigated and was not further used (data not shown).
Test strips.The test strips were built using a glass fiber pad (Ahlstrom grade 8951), nitrocellulose type CNPF-SN12, 10 µM (MDI, Advanced Microdevices PVT.LTD) and an absorbent pad (Ahlstrom, grade 222) affixed to a 300 x 60 mm backing card (Lohmann LC-58803) with 2 mm overlaps.The strips were 4 x 60 mm, placed in an off-the-shelf cassette.
Extraction buffer.Mycolactone was extracted from simulated or real samples using an extraction buffer made of 100 mM triethanolamine (TEA).When testing contrived samples, the pH was adjusted to 7.8.
Upon re-optimization with clinical samples, the optimal pH was found to be 10.4.For evaluation of clinical samples, addition of a Heterophilic Blocking Reagent (HBR) was found to be necessary to eliminate matrix effects.For these samples, a heterophilic blocking tube (Scantibodies, cat# PN 3IX762) was reconstituted with 0.5 mL of extraction buffer and used in optimized amounts for each lot of conjugate, at a final concentration of 2-5%.The TEA buffer was stored in glass bottles at room temperature, and HBR was prepared and stored in 1.5 mL Axygen vials (Axygen, cat# MCT-150-L-C) at 4°C.
. Chase buffer.The chase buffer was prepared with 2% proprietary detergent, 0.1% proprietary surfactant, 0.1% casein (Sigma, cat# G7078-500G) prepared by incubation in sodium tetraborate decahydrate pH 8.5 at 37 °C for 24 hours, and 0.05% Proclin (Sigma, cat# 48912-U) in PBS.The pH was adjusted with sodium hydroxide to pH 8.9 for experiments with clinical samples.Simulated wound exudate and other test materials.Based on a published biochemical analysis of wound fluids from leg ulcers [22], we concluded that wound fluid could be approximated by mixing equal volumes of negative human serum (NHS, ConeBio, cat# 4090) and PBS (FisherScientific, cat# 28372).We called this
Procedure to run the assay using a simulated sample and no swabs (Figure 4 Steps, 2-7).For the sake of simplicity, most of the assay development work was carried out without a swab, skipping steps 0 and 1 of was added to the extract, vortexed briefly, and incubated for 30 min, then placed on a magnetic rack to concentrate the nanoparticles for 30 min.During incubations the vials were protected from the light to avoid mycolactone degradation.The conjugate was concentrated with a neodymium magnet (Poichekailov magnetic racks, US, [24]), the supernatant was removed and the particles resuspended in 2 drops (approximately 60 µL) of chase buffer and transferred to the test sample port.The test developed for 30 .min prior to imaging and quantitation with a benchtop reader (Leelu v3.0, Lumos Diagnostics, Carlsbad,

CA).
Procedure to run the assay using a simulated sample absorbed onto a swab (Figure 4, Steps 1-7).To simulate a clinical specimen absorbed on a swab, a swab (Copan, 25-3306-H BT) was placed in an Axygen vial containing 50 µL of positive or negative SWE until all the liquid was absorbed.Next, the swab was placed in a fresh 1.5 mL Axygen vial and extracted with 0.5 mL of extraction buffer.The swab was rotated 3-5 times in the extraction buffer while pressing against the sides of the vial, left in the vial for 5 min, and pressed against the wall while rotating and lifting to recover as much liquid as possible.The swab was discarded, and the next steps (steps 2-7) were conducted with the extract as described in the previous paragraph for SWE.

Procedure to run the assay with clinical wound exudates (with HBR, pH 10.4 buffer).
The retrospective PCR negative clinical wound swabs from Ghana had been extracted in water prior to receipt.These samples were tested with two methods (1) as received, following the protocol for simulated extracts and (2) by first adding 10x concentrated TEA extraction buffer to final x1 concentration to improve conjugate collection on a magnet.A heterophilic blocking reagent (HBR) was added to clinical wound extracts at 2-5% concentration to prevent possible interference of the conjugate with heterophilic antibodies or other matrix effects, then samples were tested the same as described for the simulated extract.

Quantification.
Test line intensities were quantified with a benchtop reader (Leelu v3.0, Lumos Diagnostics, Carlsbad).We investigated both the test line intensity alone, as well as the ratio between test and control line (T/C) with the latter giving the most reproducible results.
Statistical Analysis.P-values were calculated using GraphPad Prism, version 10, assuming Gaussian distributions, and not assuming equal standard deviations (Welch's correction).
. developed over 30 minutes.The fully optimized conditions can be found in the Instructions for Use included in the Supplementary Information.
Limit of detection with simulated wound exudate.Samples of mycolactone A/B serially diluted in ethanol were added to the simulated wound exudates and processed according to the optimized conditions described in the previous two paragraphs.At this stage of development, the extraction buffer had a pH = 7.8 and the chase buffer had a pH = 7.4.The intensity of the test and control lines were quantified with a benchtop reader.The more mycolactone in the sample, the more it competes with PG-204 immobilized at the test for binding the conjugate.Hence, the higher the mycolactone concentration, the lower the test line intensity.When looking at the test with the unaided eye (Figure 5), the test line could be easily seen if no mycolactone was present, and the test line became barely visible when 7 mg of mycolactone were added to the simulated wound exudate.Similarly, when quantifying the intensity of the test lines with a benchtop reader (Figure 6), the intensity decreased from approximately 0.08 reader units in the absence of mycolactone to 0.02 units in the presence of 7 ng mycolactone, and the difference was statistically significant (p=0.006).All samples contained the same amount of DMSO, indicating that the dose response was due to mycolactone and was not an artifact due to DMSO.Test results with swabs and simulated wound exudates.Having in hand a test that was analytically sensitive and specific, we next investigated the use of a swab.The concern was that a swab would be .mandatory for sample collection but that mycolactone, being so lipophilic with a calculated logP of 9.0 [25], may stick to the fibers of which the swab is made, and not be easily extracted into an aqueous solution.To investigate this matter, an aliquot of 50 µL SWE was deposited on a swab and then extracted with the extraction buffer for 5-10 min.As reference, the same SWE was directly diluted with extraction buffer, without using a swab.The tests showed a nearly identical behavior when directly diluting the SWE in extraction buffer (not using a swab) versus first absorbing the SWE onto a swab and then extracting it (Figure 8A).Based on the quantification of the test line intensities, we calculated that 80% of the mycolactone was recovered from the swab, and the remaining 20% were left within the extract trapped in the swab fibers.Additionally, the swabs containing simulated samples could be stored for at least 48 hours at 4 °C before extraction without noticeable degradation of mycolactone (Figure 8B).This feature is useful for shipping samples from the field to a local laboratory.To our surprise, both the test and control line frequently disappeared, even though the samples had tested negative for Buruli Ulcer by PCR.Some clinical samples contained debris and particulates, sometimes in large amounts and possibly deriving from herbal medications and necrotic tissues.These debris interfered with the test, although could not explain alone all the cases where test and control lines disappeared.
Several parameters were optimized to achieve 100% validity (i.e., control line always visible) with clinical swabs.First, we reasoned that the test employs mouse monoclonal antibodies, and that heterophilic antibodies present in the wound exudates may have a deleterious impact, especially when using endemic samples from populations often exposed to rodents.This risk was mitigated by adding Heterophilic-Blocking Reagents (HBR) which led to a significant improvement of the line intensities (Figure 10).
. Second, we removed the particulate material from the swab extracts by centrifugation.This can be done easily in the laboratory but not in the field.Alternatively, doubling the amount of buffer with which the mycolactone is extracted from the swab (1.0 mL instead of 0.5 mL) lowered the impact of particulate on magnetic nanoparticle recovery (data not shown).Third, increasing the pH of both the extraction buffer (from 7.8 to 10.4) and chase buffer (from 7.4 to 8.9) increased the intensity of both the control and test lines (Figure 11).Taken together, these modifications provided 100% validity, as judged by the presence of a visible control line in all of the 207 clinical swabs analyzed.
. from LFA-negative lesions were pooled, and then spiked with synthetic mycolactone.Rather than reporting the absolute intensity of the test line as we did in the original LOD testing (Figures 5, 6), the test to control (T/C) line ratios were calculated, leading to more consistent results (Figure 12).In the absence of mycolactone, the T/C ratio was around 0.9.The presence of 3.7 ng of mycolactone/sample gave a T/C ratio in the 0.42-0.44range that was easily distinguished from the T/C ratio obtained without mycolactone.
Hence, under fully optimized conditions, the LOD was around 3.7 ng of mycolactone using extracts from clinical samples as the sample matrix.mycolactone in wound exudate.The test also met most of the WHO TPP parameters in that it was stable for transportation at ambient temperature, and that a stable positive control was developed.Detailed instructions for use were compiled and are available in the Supplementary Information.Based on this data, the prototype was deemed worthy of advancing to field evaluations, with fresh clinical samples.These results will be reported elsewhere.

Figure 3 .
Figure 3. Competitive-format test strip schematic.Biotinylated mycolactone is added to poly-streptavidin and the mixture is immobilized at the test line position.A secondary goat anti-mouse antibody is immobilized at the control line position.The magnetic-GNS anti-mycolactone conjugate will bind to the biotinylated mycolactone at the test line.In the presence of mycolactone in the sample, the mycolactone binding sites on the conjugate will become saturated, resulting in a test line intensity that is inversely proportional to the amount of mycolactone in the sample.

1: 1
mixture of NHS:PBS Simulated Wound Exudate (SWE).Positive simulated samples were prepared by spiking SWE with synthetic mycolactone A/B prepared from a mycolactone stock solution (1 mg/mL in DMSO) diluted in ethanol by a factor of 135 to reach a 10 µM concentration.Positive controls were prepared each containing 7 ng of mycolactone per sample, by diluting 1 µL of 10 µM mycolactone in ethanol to a final volume of 75 µL with SWE (for testing without a swab) or by diluting 1 µL of 10 µM mycolactone in ethanol to a final volume of 50 µL with SWE (for testing with a swab).Sterile swabs with a flock-tipped applicator and polystyrene handle (Copan, Cat #25-3306-H BT) were used for testing with swabs.Clinical wound exudates.Wound swab extracts from cases that were initially suspected to be Buruli ulcer but later found to be PCR-negative (Figure8) were kindly provided by Dr. Anthony Ablordey from the Noguchi Memorial Institute for Medical Research, University of Ghana.Those extracts were received already extracted in water instead of TEA.Additional samples were collected from suspected Buruli ulcer lesions but that proved negative for mycolactone with our rapid test were provided by The Pasteur Center in Cameroon, Yaoundé, Cameroon, Pasteur Institute Côte d'Ivoire, Abidjan, Côte d'Ivoire, and Hope Commission International, Abidjan, Côte d'Ivoire. .Clinical samples from Ghana were taken at the hospital or health centres as part of the routine diagnostic procedure and case confirmation recommended by WHO and sent to the Noguchi Memorial Institute for Medical Research (NMIMR) for analysis.Ethical approval for collection and use of patients' samples was obtained from the Institutional Review Board of the Noguchi Memorial Institute for Medical Research (NMIMR; Study number: NMIMR-IRB CPN 024/18-19).Patients were required to sign an informed consent.To ensure anonymity, personal information such as names, and other personal identifiers associated with patient's samples were replaced with codes.Similarly, Clinical Samples from Ivory Coast and Cameroon were collected under IRB approved protocols for the sake of this study and with informed consent.The protocols were cleared by the Comité National d'Ethique des Sciences de la Vie et de la Santé, Ministère de la Santé, Côte d'Ivoire, 2021, IRB 000111917 and the Comité National d'Ethique de La Recherche pour la Santé Humaine, Ministère de la Santé, Cameroun, IRB 2021/06/1367.

Figure 4 .
Figure 4. Instead, 75 µL of negative SWE or positive (mycolactone spiked) SWE were placed in a 1.5 mL

Figure 5 .
Figure 5. Limit of detection of the rapid test using simulated samples spiked with synthetic mycolactone (n=3).Visually, a difference can be seen between no mycolactone and 7 ng of mycolactone, where the latter causes a nearly complete inhibition of the test line.

Figure 6 .Figure 7 .
Figure 6.Limit of detection of the rapid test using simulated samples spiked with synthetic mycolactone.When measured with a reader, the difference between no mycolactone and 7 ng of mycolactone is statistically significant (p < 0.05).

Figure 8 .
Figure 8. A. Effect of swab on test results and sample stability when absorbed on a swab.Panel A. The test results are not impacted if simulated wound exudate is pre-absorbed on a swab.Panel B. Once absorbed on a swab, simulated wound exudates are stable for at least 48 hours at 4 °C.

Figure 9 .
Figure 9. Aspect of 6 different wound exudate extracts as received from Ghana.

Figure 10 .
Figure 10.HBR rescues the control line with negative wound swabs and improves the aspect of all tests.The test line intensity determined with a visual score card is indicated on the cassettes next to the test line.

Figure 11 .
Figure 11.Increasing the pH of the extraction buffer from 7.9 to 10 leads to stronger

which was not certified by peer review)
. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(