ABSTRACT
Aspergillus fumigatus has widely evolved resistance to the most commonly used class of antifungal chemicals, the azoles. Current methods for identifying azole resistance are time-consuming and depend on specialized laboratories. There is an urgent need for rapid detection of these emerging pathogens at point-of-care to provide the appropriate treatment in the clinic and to improve management of environmental reservoirs to mitigate the spread of antifungal resistance. Our study demonstrates the rapid and portable detection of the two most relevant genetic markers linked to azole resistance, the mutations TR34 and TR46, found in the promoter region of the gene encoding the azole target cyp51A. We developed a lab-on-a-chip platform consisting of: (i) tandem-repeat loop-mediated isothermal amplification; (ii) state-of-the-art complementary metal-oxide-semiconductor microchip technology for nucleic acid amplification detection; and (iii) a smartphone application for data acquisition, visualization, and cloud connectivity. Specific and sensitive detection was validated with isolates from clinical and environmental samples from 6 countries across 5 continents, showing a lower limit of detection of 10 genomic copies per reaction in less than 30 min. When fully integrated with a sample preparation module, this diagnostic system will enable the detection of this ubiquitous fungus at the point-of-care, and could help to improve clinical decision making, infection control, and epidemiological surveillance.
INTRODUCTION
The rapid emergence of multidrug resistance in pathogenic fungi poses a widespread emerging threat to human health and food security (1). Annually, the estimated global mortality associated with fungal diseases is more than 1.6 million (2), which is 3-fold more than that of malaria and is comparable to that for tuberculosis and HIV (3, 4). Aspergillus fumigatus, a prevalent airborne organism, is one of the principle fungal pathogens responsible for serious fungal diseases. Recent global estimates suggest 3,000,000 cases of chronic pulmonary aspergillosis and over 250,000 cases of invasive aspergillosis (IA) (4). In order to combat A. fumigatus infections, the azole antifungal drugs are widely used as first line therapies. However, increases in azole-resistant A. fumigatus (ARAf) infections worldwide are threatening our ability to combat aspergillosis, as evidenced by increased rates of mortality in intensive-care patients infected by ARAf compared to those infected by azole-susceptible genotypes (5). Two main pathways contribute to the emergence of ARAf. First, patients undergoing long-term treatment with azoles may evolve resistant genotypes by de novo mutation and in vivo selection (6). Second, the widespread use of azoles has led to the de novo evolution of the resistance in the environment (7, 8). The most ubiquitous mechanism of ARAf is associated with the cyp51A gene and its promoter region through TR34/L98H and TR46/Y121F/T289A mutations, with recent studies showing 40 to 90% of clinical azole-resistant isolates carrying the TR34/L98H allele (9–12). Discovered more recently and conferring higher-grade resistance to azoles, TR46/Y121F/T289A is defined by a tandem repeat of 46 base pairs (bp) (TR46) in conjunction with the Y121F and T289A mutations (13, 14). So far, both of these ARAf genotypes have been repeatedly reported in clinical and environmental isolates from numerous countries over five continents (15).
The worldwide emergence of ARAf has reinforced the need to accelerate the development of rapid and specific diagnostics. Diagnostic techniques, including culture and phenotypic evaluation, struggle to detect azole resistance owing to difficulty in growing the fungus (16, 17) and time-to-result (5 to 7 days). Lateral-flow devices (LFD) which detect circulating antigens in a patient’s serum have been developed (18); however, these LFDs do not detect the presence of the canonical ARAf-determining alleles. To overcome these limitations, molecular-based detection methods are needed. Several commercial products based on DNA detection are available for the detection of A. fumigatus and ARAf. Three examples include PathoNostics AsperGenius assay (PathoNostics, Maastricht, the Netherlands), MycoGENIE A. fumigatus kit (Ademtech, Pessac, France), and Fungiplex Aspergillus Azole-R in vitro diagnostics (IVD) PCR (Bruker Daltonik GmbH, Bremen, Germany), all of which use gold-standard fluorescent-based qPCR instruments and therefore are limited to specialized laboratories, preventing their deployment at the point-of-care and environmental settings.
Recently, we have introduced a new smartphone-based lab-on-a-chip (LoC) platform, offering an affordable and portable alternative to gold-standard laboratory-based instruments for detection of infectious diseases (19). This platform combines ion-sensitive field-effect transistors (ISFETs), fabricated in unmodified complementary metal-oxide semiconductor (CMOS) technology, with loop-mediated isothermal amplification (LAMP) and an AndroidOS application for data acquisition, visualization, and cloud connectivity. Electrochemical sensing using ISFETs removes the need for bulky and complex optical equipment, while LAMP eliminates the requirement for thermocycling—reducing the hardware complexity, power consumption, and size of the diagnostic platform (20–24).
In this study, we have developed a new LAMP assay for the detection of TR46 based on a recent method we developed, TR-LAMP (25). We have implemented TR-LAMP for TR46 and TR34 detection into our LoC platform and validated its performance in detecting ARAf within 30 min with high sensitivity (10 genome copies per reaction). The workflow for detecting TR34 and TR46 alleles using this LoC platform is depicted in Fig. 1. We have analyzed 30 isolates from 6 countries, including both clinical and environmental samples, and compared the results with a commercially available kit. This LoC diagnostic system, when integrated with a sample preparation module, represents an important milestone toward rapid, specific, and sensitive detection of azole-resistant alleles in A. fumigatus at the point-of-care, thereby improving patient outcomes, clinical decision making, and environmental surveillance.
Diagnostic workflow for detection of TR34 and TR46 using lab-on-a-chip platform. Clinical and environmental samples are cultured, and nucleic acids are extracted off-device. Subsequently, the extracted nucleic acids are mixed with TR-LAMP reagents and loaded onto a disposable cartridge mounted with a CMOS microchip (78 × 56 ISFET sensors). This cartridge is plugged into the LoC diagnostic platform and connected with a smartphone via Bluetooth for data acquisition, visualization, and cloud connectivity.
MATERIALS AND METHODS
Study design.The objective of this study was to develop a rapid diagnostic test for azole-resistant alleles in A. fumigatus at the point-of-care using CMOS technology and TR-LAMP. The two key hypotheses of this work were as follows: (i) the method developed for detecting TR34 using LAMP can be extended to other clinically relevant tandem repeats such as TR46; (ii) The TR-LAMP chemistry can be successfully translated onto a CMOS based lab-on-a-chip platform for detection of ARAf within 30 min in clinical and environmental isolates. To test the first hypothesis, 30 A. fumigatus isolates (6 clinical and 24 environmental) carrying the TR34 allele (n = 15), TR46 allele (n = 4), and wild-type allele (n = 11) were used (validated with whole-genome sequencing [WGS]). Detection of TR34 and TR46 by TR-LAMP was performed in a qPCR platform and compared with the commercially available AsperGenius (PathoNostics, Maastricht, The Netherlands). To test the second hypothesis, 20 isolates were tested against the TR34 and TR46 LAMP assays on a CMOS-based lab-on-a-chip platform.
To estimate the number of required isolates we used the following formula (26):
TR-LAMP assay.Primer sequences for TR34-LAMP were taken from the work reported in reference 25. TR46-LAMP primers were designed based on 100 cyp51A nucleotide acid sequences of A. fumigatus, including TR34, TR46, and wild-type strains, using Primer Explorer V5 (Eiken Chemical Co. Ltd., Tokyo, Japan; https://primerexplorer.jp/e) and optimized manually to ensure specificity. The list of all the accession numbers of the cyp51A gene nucleotide sequences used in this study are provided in Table S4 in the supplemental material. Sequences were downloaded from the NCBI database (27) and a multiple sequence alignment was constructed using the MUSCLE algorithm (28) in Geneious 10.0.5 software (29).
Each TR-LAMP reaction contained the following: 1.5 μl of 10× isothermal buffer (New England BioLabs), 0.9 μl of MgSO4 (100 mM stock), 2.1 μl of dNTPs (10 mM stock), 0.375 μl of bovine serum albumin (BSA) (20 mg/ml), 2.4 μl of betaine (5 M stock), 0.375 μl of SYTO 9 Green (20 μM stock), 0.6 μl of Bst 2.0 DNA polymerase (8,000 U/ml), 3 μl of different concentrations of synthetic DNA or genomic DNA (gDNA), 1.5 μl of 10× TR-LAMP primer mixture (20 μM of BIP/FIP, 10 μM of LF/LB, and 2.5 μM B3/F3) and enough nuclease-free water (Thermo Fisher Scientific) to bring the volume to 15 μl. Each solution was split into 5-μl reactions (triplicates) on a 96-well PCR plate, heated to 63°C for 30 min using a Light-Cycler 96 real-time PCR system (Roche Diagnostics). Subsequently, melting curve analysis was performed. All primers used in this study were purchased from Integrated DNA Technologies (IDT, California, US).
Transfer of the TR-LAMP assay on the lab-on-a-chip platform required modification of the reaction mix to pH-LAMP, as the detection mechanisms rely on proton detection. In this case, the buffering conditions were adjusted such that pH changes could be measured. Each reaction contained the following: 1 μl of 10× isothermal customized buffer (pH 8.5 to 9.0), 0.6 μl of MgSO4 (100 mM stock), 1.4 μl of dNTPs (10 mM stock), 0.6 μl of BSA (20 mg/ml stock), 1.6 μl of betaine (5 M stock), 0.1 μl of Bst 2.0 DNA polymerase (120,000 U/ml stock), 1 μl of different concentrations of synthetic DNA or gDNA, 0.25 μl of NaOH (0.2 M stock), 1 μl of 10× LAMP primer mixture, 0.25 μl of SYTO 9 Green (20 μM stock) only for qPCR experiments, and enough nuclease-free water (Thermo Fisher Scientific) to bring the volume to 10 μl.
Clinical, environmental, and synthetic DNA.Synthetic double-stranded DNA (dsDNA) fragments (gBlocks gene fragments) containing TR46 sequence were purchased from Integrated DNA Technologies and were used to study the sensitivity and specificity of the TR46-LAMP assay. High molecular weight DNA from 6 clinical and 25 environmental A. fumigatus isolates was extracted using the MasterPure yeast DNA purification kit with an additional bead-beating step to improve yield (29). The isolates were collected for this study and have not been previously described. Conidia, harvested from 2- to 3-day A. fumigatus cultures grown on Sabouraud dextrose agar, were homogenized in a Qiagen TissueLyser II (Qiagen, Hilden, Germany) using 1.0 mm zirconia/silica beads. Purified genomic DNA (gDNA) was quantified using a Qubit 2.0 fluorometer and the dsDNA BR assay kit (Life Technologies, Carlsbad, CA). Prior to WGS library preparation, the integrity of gDNA was determined using the TapeStation 2200 system (Agilent, Santa Clara, CA) and gDNA ScreenTape. DNA was stored at –20°C until required.
Illumina whole-genome sequencing.Whole-genome single nucleotide polymorphism (SNP) data were generated according to an established workflow for A. fumigatus isolates (30). Genomic DNA libraries were constructed using the TruSeq Nano kit (350 bp) (Illumina, San Diego, CA) according to Illumina protocols. Prepared libraries were sequenced in a 150-base paired-end format using a HiSeq 4000 sequencer (Illumina) based at Edinburgh Genomics. Raw genomic reads were quality controlled, trimmed, and aligned against the Af293 reference genome (31) using the Burrows-Wheeler Aligner (BWA) v0.7.15 alignment tool (32). The detection of high-quality SNP variants was conducted using HaplotypeCaller function from the Genome Analysis Toolkit (GATK) v4.0 package (33), according to confidence filters described in Rhodes et al. (34). We also removed any base with a minimum genotype quality of 50, a minimum percent alternate allele of 80%, or a minimum depth of 10.
Phylogenetic analysis of WGS SNP data.Maximum-likelihood trees based on WGS SNP data were constructed using RAxML-HPC v8.2.9 under the GAMMA model of approximation (35) and visualized in R using the package ggtree (36). The rapid bootstrap algorithm was used to bootstrap all phylogenies using 500 replicates.
Lab-on-a-chip platform.The diagnostic platform includes a microchip fabricated in standard CMOS technology, coupled to the TR-LAMP assays and integrates 3,874 ISFETs and 494 temperature sensors arranged in a 78 × 56 array (4 × 2 mm). The ISFET is a solid-state potentiometric sensor with a structure similar to a metal-oxide-semiconductor field-effect transistor (MOSFET), biased using a reference electrode (typically Ag/AgCl) (37, 38). Fig. S1 illustrates the cross-section of the ISFET in unmodified CMOS technology. Binding of protons on the top silicon nitride (Si4N3) layer modifies the electrical sensor behavior, which allows the sensor to track DNA amplification through the release of protons when nucleotides are incorporated into DNA.
The microchip is mounted on a printed circuit board (PCB) test cartridge and an acrylic manifold is attached at the surface to form a reaction chamber. The cartridge is plugged on the main PCB device which: (i) performs temperature regulation via a Peltier module underneath the cartridge for isothermal amplification; (ii) powers the system through a rechargeable battery; and (iii) sends the sensing data via Bluetooth. The user interface is provided by a smartphone application which acquires the data. Hydration of the microchip surface layer in contact with the liquid causes slow temporal drift during the measurement (37, 38), which presents challenges to extract signal associated with pH change. In this work, we performed postprocessing compensation by assuming that the drift follows an exponential relationship obtained by fitting the curves and extracting sensor output in the form of a traditional amplification curve, as shown in Fig. S2.
The algorithm for processing the data from the ISFET array is as follows. The outputs from all exposed sensors are averaged to yield the raw output curve, which reflects both sensor drift and pH change during DNA amplification. To extract the signal, the algorithm finds the inflection point at a time which corresponds to the amplification, i.e., when the contribution from the pH dominates over the initial drift. Three reaction regions are then identified in the output curve using a thresholding method:
(i) 0 to T1: no amplification; the output is only drift.
(ii) T1 to T3: amplification is happening; the output is the sum of drift and pH signal.
(iii) T3 to end: amplification products have saturated; the output is only drift.
Using the exponential approximation, drift is canceled in regions 1 and 3. In region 2, we assume that the drift rate varies linearly from T1 to T3 and subtract the resulting curve from the output. Converting mV to pH based on the standard pH sensitivity of unmodified CMOS (21), we yield the pH readout curve. Mapping the pH readout curve to a linear scale by exponentiating then results in the classical amplification curve, which tracks the DNA copy numbers during the reaction.
Statistical analysis.Time-to-positive (TTP) data are presented as mean TTP ± standard deviation; correlation is determined using Pearson’s correlation coefficient.
Data availability.All data associated with this study are present in the paper and in the supplemental material.
RESULTS
Whole-genome sequencing and phylogenetic analysis.The WGS analysis was performed on 30 isolates from clinical and environmental samples positive for A. fumigatus (confirmed by a selective medium). Results showed the presence of TR34 (n = 15) and TR46 (n = 4) alleles, with the rest of isolates being wild type (n = 11) (Table S1). A maximum likelihood phylogeny was generated (Fig. 2) to ensure that the isolates selected for diagnostic assessment encompassed the two-clade population structure of A. fumigatus (30, 39, 40). Whole-genome SNP analysis using RAxML places the collection of isolates into two distinguishable clades. Distribution of TR34, TR46, and wild-type (WT) alleles across the breadth of the phylogeny broadly fits with the association of drug resistance to one of the two clades (40). There was no discernible relationship between geographic origin across the phylogeny, with isolates from India, the Republic of Ireland, South Africa, the United States, and the United Kingdom represented on both clades.
Phylogenetic analysis of clinical and environmental isolates. Maximum-likelihood phylogeny of A. fumigatus isolates representing azole-resistant and -susceptible genotypes from environmental and clinical sources. Bootstrap analysis (500 replicates) was performed on WGS SNP data from 30 A. fumigatus isolates, with all branches shown having 87% support or higher. Isolates are color coded according to the country of origin (see key). Tip labels are annotated according to the cyp51A genotype. Environmental isolates are indicated by a triangle symbol and clinical isolates are identified by a circle symbol. Branch lengths represent the number of nucleotide substitutions per bases of SNPs between taxa. Isolate labels refer to unpublished data.
Optimization of TR-LAMP assay for detection of the TR46 allele.The TR-LAMP assay for rapid detection of the TR46 allele was based on the TR34-LAMP assay (25), consisting of F3-TR34, B3-TR34, BIP-TR34, and two novel highly specific primers FIP-TR46 and LB-TR46. These two primers were carefully designed to detect the 12 bp that differ between the TR34 (34 bp) tandem repeat and the TR46 tandem repeat (46 bp), to avoid any cross-reactivity. The TR46-LAMP optimization process consisted of the following steps: (i) insertion of 5′ mismatches in F2 to enhance assay specificity; (ii) design and testing of loop primers (LF and LB) to speed up the reaction; and (iii) assay screening at low copy numbers. These steps were repeated until the TR46-LAMP assay performed as expected with high specificity and sensitivity (Fig. S3, Fig. S4, Table S2). Detailed location and composition of the optimized TR46-LAMP primer sets is shown in Fig. 3 and Table 1, respectively.
Schematic representation of the TR46-LAMP primers. (A) The location and direction of each primer used in the TR46-LAMP assay are indicated with arrows (blue, forward; orange, reverse). Gray lines indicate the location of 34-bp and 46-bp tandem repeats. Red asterisks identify the locations where the mismatches are introduced. LF (shown with a dashed arrow) is not included because it is overlapping with the TR region. (B) Schematic construction of the TR46-LAMP primers used in this study.
TR46-LAMP primer sequences
The optimized TR46-LAMP assay achieved a lower limit of detection (LOD) of 10 copies per reaction within 25 min. Figure 4 shows the normalized amplification curves of the TR46-LAMP assay tested with 10-fold serial dilutions of synthetic DNA ranging from 106 to 101 copies per reaction. The average time-to-positive (TTP) for synthetic TR46 dsDNA from 106 to 101 copies per reaction were 13.15 ± 0.12 min, 15.18 ± 0.32 min, 16.27 ± 0.43 min, 18.40 ± 0.71 min, 19.46 ± 1.16 min, and 21.94 ± 2.91 min, respectively. Obtained TTP results versus the log of the serial DNA dilutions showed a strong linear relationship (R2 = 0.99).
TR46 LAMP assay design and performance. Real-time amplification plots (A) for a 10-fold dilution series of TR46 LAMP assay using synthetic DNA and the corresponding standard curve (B).
Performance of the TR-LAMP assay for detection of TR34 and TR46.Performance of the TR-LAMP assays targeting TR34 and TR46 was validated on extracted nucleic acids from 30 cultured isolates from clinical and environmental samples. Specific amplification with no cross-reactivity or false positives was demonstrated with both assays (Fig. S5, Table S3). All obtained TTPs were under 20 min and estimated concentrations ranged from 5.6 × 103 to 2.9 × 105 copies per reaction.
Translating TR-LAMP assays onto the lab-on-a-chip platform.We tested all extracted nucleic acids from the A. fumigatus isolates on our CMOS-based LoC platform and a commercial qPCR instrument for performance comparison. All isolates were analyzed using both TR34-LAMP and TR46-LAMP assays, including a nontemplate control (NTC). The output from all the exposed sensors was averaged, preprocessed as shown in Fig. S2, and a time-to-positive was extracted using the gold-standard cycle-threshold method. The obtained TTPs were under 20 min for both TR34-LAMP and TR46-LAMP assays with no cross-reactivity between assays and negative nontemplate controls (Fig. 5A).
On-chip LAMP amplification curves of 20 A. fumigatus isolates. (A) Normalized TR-LAMP amplification curves for the testing isolates. (B) Scatterplot showing the correlation between TTP from the LoC platform (eLAMP) compared to a real-time instrument (qLAMP). The black cross indicates an outlier removed from the statistical analysis.
The amplification curves obtained with our LoC platform (voltage measurements), eLAMP, were highly comparable to the fluorescence signals acquired by the qPCR instrument. More specifically, the correlation between the TTP from qLAMP and eLAMP was shown to be 0.91 with R2 equal to 0.83 (Fig. 5B).
DISCUSSION
In this study, we developed a rapid, quantitative, specific and sensitive isothermal nucleic-acid amplification method for the diagnosis of azole-resistant A. fumigatus (ARAf) based on TR34 and TR46 allele detection in under 30 min with an LOD of 10 copies per reaction. Furthermore, we transferred the new TR-LAMP assays onto our LoC platform to show its potential to be used as a point-of-care (PoC) diagnostic device for on-site detection of azole resistance.
Current diagnostic methods include culture-based approaches and nucleic acid detection. The culture-based approaches are limited by low sensitivity, in particular during the early phase of infection, as well as time to results (41). Nucleic acid detection emerged as a highly sensitive method targeting point mutations in certain genes, such as the cyp51A gene, which encodes the azole target enzyme required for the biosynthesis of ergosterol. In particular, several methods based on PCR have been developed (42) and made commercially available, including AsperGenius (PathoNostics, Maastricht, The Netherlands), MycoGenie (Adamtech, Pessac, France), and Fungiplex Aspergillus Azole-R in vitro diagnostics (IVD) PCR (Bruker Daltonik GmbH, Bremen, Germany). On the other hand, isothermal amplification methods such as LAMP have become an alternative method to PCR due to not requiring temperature cycling, and showing high speed, specificity, and tolerance to inhibitors (43). Compared to PCR, LAMP primer design is more complicated as it requires six primers targeting 8 different regions; however, several LAMP assays for IA diagnoses (44, 45) and ARAf detection (25) have been developed, which have demonstrated the benefits of this approach.
In our previous study, we developed a novel method based on LAMP for the detection of tandem repeat regions, called TR-LAMP. We also demonstrated its ability to detect the most prevalent 34 bp tandem repeat mutations in A. fumigatus, TR34, within 30 min (25). Following the principle of TR-LAMP primer design, we developed a novel TR-LAMP assay for the detection of TR46, the second most common mutation related to the development of azole resistance. The TR46-LAMP assay was designed to detect the 12 bp difference between the TR34 and TR46 alleles. Furthermore, three mismatches in F2 were introduced to increase allele-specific discrimination and obtain a yes/no TR46-LAMP assay. The developed TR46-LAMP assay detects the TR46 allele in under 30 min, showing high specificity and sensitivity with an LOD of 10 copies per reaction. This is a sensitive and rapid approach to IA detection compared to common diagnostic techniques (44–51).
To evaluate the analytical performance of TR34-LAMP and TR46-LAMP assays, a total of 30 isolates were tested from the United Kingdom, United States, Republic of Ireland, South Africa, India, and Australia. A whole-genome SNP phylogeny confirmed that the two major clades of A. fumigatus were also represented in the sample panel (30, 39, 40). Experimental results provided quantitative data with high sensitivity and specificity (100%) without cross-reactivity. TR-LAMP assays can be combined to develop a multiplex reaction azole-resistance detection, which will significantly speed up clinical diagnoses (20, 52).
We performed a comparison between our TR-LAMP assays using our LoC platform and one of the most commonly used commercial kits, AsperGenius (Table S3). From the obtained results we can conclude the several advantages. (i) The detection time can be greatly reduced due to the high efficiency of the TR-LAMP assay. All the samples were detected within 30 min by TR-LAMP assays, whereas the AsperGenius assay needed more than 1 h and a lab-based instrument (due to the required melting curve analysis). Sample preparation times were excluded from this analysis as all diagnostic solutions mentioned here require the same sample processing procedure. (ii) The TR-LAMP assay uses only two resistance markers (TR34 and TR46 mutant markers), whereas the AsperGenius assay required four resistance markers, L98H, T289A, TR34, and Y121F (41). The main limitation of the proposed TR-LAMP assay is that it does not consider resistant phenotypes in patients on antifungal therapy, which may not be related to tandem repeats, such as other SNPs, efflux-mediated mechanisms, or other less common and complex mechanisms (53). Additionally, the tandem repeat TR34 is commonly present with the L98H SNP to confer azole resistance, although it has been demonstrated in patient samples and a very small number of environmental isolates that TR34 may be present without L98H SNP (42). Therefore, the proposed assay may overestimate phenotypic resistance in those uncommon events. It is likely that the same situation may occur for TR46. To circumvent this issue, the relevant SNP resistant markers should be included in the assay and this work will follow. Last, the majority of cases of human aspergillosis have negative cultures. The commercially available assays used in this study circumvent this in part by amplifying directly in samples. Although this work has focused on positive culture, in order to characterize the performance of the assay and the POC platform, a follow-on study directly on human samples should be conducted.
In this study, we have established and demonstrated the smartphone-based LoC system for azole-resistant screening by implementing our TR-LAMP assays with a lab-on-a-chip platform based on ISFET sensors (21–24, 54–56). The amplification curves obtained from our label-free LoC platform are comparable to those obtained with a fluorescence-based qPCR instrument. Real-time amplification on our LoC platform was monitored through an android smartphone application which allows synchronization to a cloud server and electronic health records in hospitals, as well as real-time geo-tagging. There are many applications where our proposed system could be suitable. For example, given the simplicity of operation, the system would be highly valuable for soil sampling and bedside screening, where skills and resources are limited. Our next steps will be to explore the feasibility of direct detection of azole resistance from samples on our system and also develop an integrated self-contained, sample-to-answer portable device for point-of-care deployment.
ACKNOWLEDGMENTS
The authors are affiliated with the National Institute for Health Research Health Protection Research Unit (NIHR HPRU) in Healthcare Associated Infections and Antimicrobial Resistance at Imperial College London in partnership with Public Health England (PHE) in collaboration with Imperial Healthcare Partners, University of Cambridge and University of Warwick. The views expressed in this publication are those of the author(s) and not necessarily those of the NHS, the National Institute for Health Research, the Department of Health and Social Care, or Public Health England. The authors also acknowledge: (i) the NIHR Imperial Biomedical Research Centre (BRC) (P80763); (ii) the Department for Health and Social Care-funded Centre for Antimicrobial Optimisation (CAMO) at Imperial College London; and (iii) the MRC Centre for Global Infectious Disease Analysis (MR/R015600/1) and the NERC Environmental Bioinformatics Centre (NE/P001165/1). Matthew Fisher is a fellow in the CIFAR “Fungal Kingdoms” program. Alison Holmes is a National Institute for Health Research (NIHR) Senior Investigator.
Authors contributed as follows. L.-S.Y., J.R.-M., A.H.H., M.C.F., and P.G. conceived the idea. L.-S.Y., J.R.-M., M.C.F., and P.G. planned the experiments. L.-S.Y., N. Moser, N. Miscourides, T.S., T.K., A.M., and K.M.-C. conducted the experiments. L.-S.Y., J.R.-M., A.M., N. Moser, K.M.-C., N. Miscourides, T.S., and J.R. curated the data. L.-S.Y., J.R.-M., A.M., N. Moser, K.M.-C., N. Miscourides, T.S., and J.R. analyzed the results. M.C.F., T.S., A.B., and J.R. provided isolates. L.-S.Y., J.R.-M., A.M., and N. Moser wrote the initial manuscript. J.R.-M., M.C.F., and P.G. acquired the funding. All authors reviewed and edited the manuscript.
We report no conflicts of interest.
FOOTNOTES
- Received 22 April 2020.
- Returned for modification 16 June 2020.
- Accepted 27 August 2020.
- Accepted manuscript posted online 9 September 2020.
Supplemental material is available online only.
- Copyright © 2020 American Society for Microbiology.
