Introduction

The accurate identification of the pathogen responsible for an infection is essential for effective treatment and control of infections. Currently, most molecular diagnostics for pathogen identification rely on antigen detection tests or nucleic acid-based tests (NATs). NATs are widely used in clinical and field settings to diagnose viral, bacterial, fungal, protozoan, and helminth infections in humans, plants, and animals by detecting pathogen-specific genetic material, including DNA, RNA, or both1,2,3,4. For example, PCR tests are widely used for the detection of SARS-CoV-2 RNA. Various molecular techniques that identify single or multiple genes from pathogens, including those from SARS-CoV-2, include quantitative real-time PCR (qPCR), droplet digital PCR (ddPCR), reverse-transcription loop-mediated isothermal amplification (RT-LAMP), CRISPR-Cas13a5, next-generation sequencing (NGS), and micro-NMR (µNMR). Although NATs have high sensitivity and specificity, they are limited by the requirement for amplification of target sequences from the pathogen and specialized laboratory infrastructure, which restricts their use in routine clinical or field settings6,7. CRISPR-Cas based NATs do not require sequence amplification and can be used with minimal instrumentation, thereby overcoming the limitations of other NATs. However, CRISPR-Cas based diagnostics may be limited by sensitivity to RNA secondary structures and the need for a protospacer adjacent motif (PAM) sequence in the target genes5,8,9,10,11.

Low-cost and rapid pathogen tests, such as enzyme-linked immunosorbent assays (ELISAs) or lateral flow assays (LFAs), have been developed to detect pathogen-specific antigens using capture antibodies12. While antigen tests are generally less sensitive than NATs, their affordability and quick results make them a suitable option when resources, cost, and speed are important. However, antigen tests are not ideal as public health tools since they may not detect infections early enough to prevent disease spread. In contrast, NATs can detect organism-specific DNA or RNA sequences in the early stages of an infection before the pathogen load is sufficient for a positive antigen test. In clinical settings, a negative rapid test in a symptomatic individual usually requires a confirmatory NAT, which increases the time to and cost of diagnosis. Thus, there is a need for low-cost, rapid pathogen tests that retain the sensitivity advantages of NATs.

Various low-cost techniques, including ELISA and LFA, have been developed to detect disease-specific nucleic acids5,10,13,14,15. One example is the HC2-HPV diagnostic assay for human papillomavirus (HPV) infection marketed by Qiagen (Digene), which employs polyclonal antibodies in a "hybrid capture" sandwich ELISA format16,17. The HC2-HPV assay uses a complementary RNA probe to hybridize with HPV DNA, which is then detected by a polyclonal DNA-RNA duplex-specific antibody. Recently, several methods were developed using DNA-RNA specific antibody for pathogen detection13,14,15. S9.6, a DNA-RNA specific monoclonal antibody, was isolated from mice immunized with a DNA-RNA heteropolymer antigen of bacteriophage ɸX17418. Previous studies have shown that S9.6 binds to a 6-bp epitope on DNA-RNA hybrids with high affinity and specificity in a largely sequence-independent manner18,19,20,21,22. Furthermore, S9.6 can bind with low affinity to double-stranded RNA (dsRNA)22,23,24.

In this study, we developed a pathogen-specific nucleic acid-based test in both ELISA and LFA formats by utilizing the unique properties of S9.6. Recently, we validated the use of S9.6 Fab in a sandwich ELISA25 and used it as a platform to develop our assays in this study. This assay, called DNA-RNA Hybrid Capture Immunoassay (HC), detects pathogen-specific RNA, and can be adjusted to detect any pathogen. The HC assay does not require molecular manipulations such as reverse transcription or pre-amplification of the pathogen's RNA or DNA, nor does it require advanced instrumentation or specialized training to perform. With the HC assay, we could specifically detect RNA for the pagA, rpoB, and gyrA genes of Bacillus anthracis, the Spike E gene of SARS-CoV-2, and the B1 gene of Toxoplasma gondii. This assay can be used for the detection of pathogens in low-resource settings, as well as for qualitative gene expression analysis.

Results

S9.6 specifically binds pre-existing hybrids in total cellular RNA extracts

To develop a specific assay for DNA-RNA hybrids, first, we investigated the nucleic acid species to which S9.6 binds in total cellular RNA extracts (henceforth referred to as “cellular RNA”) since most clinical samples contain a mixture of nucleic acids. Cellular RNA from B. anthracis was treated with nucleases of varying specificities, and a dot blot was performed using S9.6 (Fig. 1A). Treatment with RNase H, which selectively degrades RNA in DNA-RNA hybrids, caused a roughly 75% signal loss relative to the untreated control (Fig. 1A,B), confirming that S9.6 binds to pre-existing hybrids present in total nucleic acids. Treatment with the dsRNA-specific nuclease RNase III similarly caused roughly a 75% reduction in signal, suggesting that much of the signal comes from dsRNA despite S9.6's low affinity for dsRNA19,22,23,24.

Figure 1
figure 1

Analysis of S9.6’s specificity for DNA-RNA hybrids. (A) Representative dot blot analysis of S9.6 binding to a crude extract of cellular RNA from B. anthracis. One microgram of cellular RNA was spotted directly on a nitrocellulose membrane or first treated with respective nucleases. UT untreated, MBN mung bean nuclease. Nuclease specificities are as follows: RNase H, DNA-RNA hybrids; MBN, ssDNA; RNase T1, ssRNA; RNase III, dsRNA; RNase A, all RNA. (B) Quantification of signal intensity relative to untreated (UT) control for nuclease treated samples. Data represents mean ± SD of duplicate dot blots (A, Fig. S1A). (C) Dot blot analysis using S9.6 to detect in vitro synthesized pagA mRNA, ssDNA, and DNA-RNA hybrids in specified amounts.

Full size image

To determine whether S9.6 binds to ssRNA or ssDNA, we treated cellular RNA with mung bean nuclease and RNAse T1, which specifically digest ssDNA and ssRNA, respectively, and performed a dot blot analysis. We observed no difference in signal, indicating that S9.6 does not bind to ssDNA or ssRNA (Figs. 1A,B, S1A). We further validated this finding by hybridizing in vitro synthesized mRNA from the B. anthracis gene pagA, a gene encoding protective antigen of anthrax toxin, to complementary ssDNA produced using Lambda Exonuclease (Fig. S1B). We performed a dot blot using S9.6 and observed a signal from the hybrid dot, but not from the single-stranded DNA or RNA dots (Fig. 1C). Our results confirm that S9.6 does not bind to the secondary structures of ssDNA or ssRNA, which often form double-stranded structures (Fig. S2A). The limit of detection (LOD) of dot blot analysis for in vitro synthesized pagA hybrid was found to be in the sub-nanomolar range (0.82 nM or 1.25 ng/µL), corresponding to 5 × 108 molecules of pagA DNA-RNA hybrid.

Development of HC-ELISA

Our dot blot results confirmed that S9.6 can detect pre-existing hybrids in cellular RNA and may non-specifically bind to highly abundant dsRNA. Therefore, we aimed to develop an assay that excludes dsRNA and pre-existing hybrids by selectively enriching transcripts from cellular RNA. The result was the Hybrid Capture Immunoassay by ELISA (HC-ELISA), which captures biotinylated complementary ssDNA probes hybridized to RNA on a streptavidin-coated surface. The captured hybrids are then detected using S9.6 with chemiluminescence as the signal output (Fig. 2A). To determine the optimal probe type for the assay, HC-ELISA was performed using 5′-mono-biotinylated (hereafter referred to as PM) and poly-biotinylated (using 1 µM and 10 µM biotin-14-dATP, hereafter referred to as P1 and P10) probes hybridized to 2.21-kilobase-long pagA RNA synthesized via in vitro transcription. Assuming full hybridization from equal amounts RNA and ssDNA probes (Fig. S2B), the LOD of HC-ELISA for pagA RNA hybridized to the PM probe was 108 molecules, while for RNA hybridized to P1 and P10 probes, it was 106 and 104 molecules, respectively (Fig. 2B). Based on these results, we selected P1 and P10 probes to maximize the sensitivity of HC-ELISA in further experiments.

Figure 2
figure 2

Sensitivity and specificity of HC-ELISA to detect B. anthracis gene transcripts. (A) Schematic of the HC-ELISA format. Gene specific DNA-RNA hybrids are generated using 5’-biotinylated ssDNA probes (PM) or 5′-biotinylated ssDNA probes synthesized in the presence of 1 µM (P1) or 10 µM (P10) biotin-14-dATP. The hybrids are captured on a poly-streptavidin coated ELISA plate and bound to S9.6, which is detected using anti-mouse IgG conjugated with HRP (horseradish peroxidase) and a chemiluminescent substrate. Illustration created with BioRender.com. (B) HC-ELISA sensitivity to pagA hybrids incorporating PM, P1, and P10 probes. Data represents mean arbitrary units (AU) of chemiluminescence normalized to average of ssDNA and buffer-alone control ± SD (n = 3 wells). (C) Signal from HC-ELISAs detecting rpoB transcripts in B. anthracis cellular RNA. Data represent mean chemiluminescence normalized to average of ssDNA and buffer-alone control ± SD (n = 2 wells). (D) Assay specificity assessment using ribonucleases. Data represent mean chemiluminescence normalized to average of ssDNA and buffer-alone control ± SD (n = 2 wells). UT = untreated. (E) rpoB and pagA expression over reference gene gyrA between ambient and CO2-rich conditions as measured by chemiluminescence from HC-ELISA and (F) relative abundance of rpoB and pagA to gyrA as measured by RT-qPCR. Error bars represent mean ± SD of duplicates; statistical significance was determined with Student’s t-test. *p < 0.05, **p < 0.01, ***p < 0.001; ns, not significant. Panels (B–D) show a representative experiment of three repeated experiments; panels (E,F) present the data in full.

Full size image

Sensitivity and specificity of HC-ELISA

Observing that HC-ELISA was able to specifically identify DNA-RNA hybrids generated from a pure source of target gene transcripts (in vitro synthesized RNA; Fig. 2B), we next sought to validate the assay’s utility in the detection of target transcripts from a crude source of many gene transcripts since this condition more closely mimics a clinical or field test sample. We also observed that longer hybrids resulted in a higher signal intensity (Fig. S2C); thus, the ~ 3.5 kb mRNA of B. anthracis RNA polymerase beta subunit (rpoB) was selected as the target transcript in tests of assay sensitivity to hybrids generated in the cellular RNA of B. anthracis. PM, P1, and P10 probes (equivalent to ~ 3.5 kb full length gene) were hybridized to rpoB transcripts in cellular RNA and assayed by HC-ELISA (Fig. 2C). The P1 and P10 probes could identify the presence of rpoB mRNA in as little as 5 µg and 500 ng of B. anthracis cellular RNA, respectively, whereas the PM probe could not enable detection of rpoB mRNA in HC-ELISA. The signal to noise ratio (S/N) of the P10 probe was found to be fourfold greater than that of the P1 probe (Fig. S2D).

To test the specificity of HC-ELISA for detection of longer DNA-RNA hybrids, an array of ribonucleases including RNase H, RNase T1, RNase A, and RNase III was used (Fig. 2D). Samples treated with RNase H, which digests RNA in DNA-RNA hybrids, showed a ~ 85% signal loss in HC-ELISA. In contrast, samples treated with RNase T1, which degrades ssRNA, showed no decrease in signal. Samples treated with RNase III also showed a 65–70% signal loss when compared to untreated controls.

To test the quantitative power of HC-ELISA, the assay was used to evaluate a well-established conditional transcription pattern specific to B. anthracis. RT-qPCR and HC-ELISA were performed in parallel using the same extracts of cellular RNA from bacteria grown in either air or CO2 (Fig. 2E,F). Gene transcripts rpoB and pagA were selected as targets and the gyrA transcript (encoding gyrase A) was selected as a reference. Under enriched CO2 conditions, pagA is expressed at levels 130-fold greater than when B. anthracis is grown in air, while rpoB and gyrA are constitutively expressed independent of CO226. Hybrid capture signals from rpoB and pagA were normalized to the signal from gyrA to facilitate comparison with qPCR results. HC-ELISA was able to significantly (p > 0.01) identify ~ sixfold induction in the expression of pagA mRNA while showing no change in rpoB expression in CO2 (Fig. 2E) and RT-qPCR indicated a 40-fold induction of pagA transcription and no change in rpoB transcription under the CO2 condition (Fig. 2F).

HC-ELISA can detect simulated B. anthracis infections in mice

We observed that the HC-ELISA method is robust in detecting highly expressed gene transcripts from pathogens like B. anthracis. To evaluate the assay's performance in detecting simulated infections, B. anthracis A35 strain vegetative bacteria were spiked into mouse spleen suspensions at varying bacterial doses ranging from 107 to 109 CFU/100 mg of spleen tissue. The presence of B. anthracis rpoB and pagA mRNA in cellular RNA extracted from the spleens was then analyzed using HC-ELISA with P10 hybridization probes (Fig. 3A,B). As expected, the fold change in chemiluminescence over non-spiked controls was highest in the 109 CFU group and lowest in the 107 CFU group for both genes but rpoB showed a higher fold change relative to non-spiked spleen samples. The assay’s efficiency for detecting simulated infection was assessed in comparison to the standard curve generated from assays performed using cellular RNA from bacteria (Fig. S3). For the highest bacterial load (109 CFU/100 mg of spleen), rpoB and pagA RNA detection in the simulated infection was equivalent to detection from bacteria alone, but efficiency dropped as bacterial load decreased.

Figure 3
figure 3

HC-ELISA can detect simulated B. anthracis infection in mice. (A,B) Median fold change in chemiluminescence as a function of bacterial CFU/100 mg of spleen suspension as measured by HC-ELISAs detecting rpoB (A) and pagA (B). Fold change was calculated relative to uninfected control spleens. Error bars represent range of triplicate wells in a representative experiment of n = 3.

Full size image

HC-ELISA can be paralleled by a lateral flow assay (HC-LFA)

Next, we developed a hybrid capture lateral flow assay (HC-LFA) based on the HC-ELISA method to detect pathogen-specific RNA via DNA-RNA hybrids. The HC-LFA assay involves capturing biotinylated hybrids on a poly-streptavidin test line (T) and detecting them with gold nanoparticle-conjugated S9.6 (S9.6-GNP) (Fig. 4A). The control line (C) in HC-LFA contains anti-mouse IgG, which captures unused S9.6-GNP. We tested PM, P1, and P10 probes hybridized to rpoB mRNA in total nucleic acids from B. anthracis to determine the best probe for use in HC-LFA. The P1 probe provided an optimal balance of test line intensity for true versus false positives and was selected for use in all subsequent lateral flow experiments. False positives of low intensity were observed in assays performed with the rpoB P1 and P10 ssDNA probes alone, but not in assays performed using only cellular RNA (Fig. 4B).

Figure 4
figure 4

HC-LFA can detect simulated B. anthracis infections in mice. (A) Schematic of HC-LFA. S9.6-GNP = gold nanoparticle-conjugated S9.6. The arrow indicates the direction of mobile phase. Illustration created with BioRender.com. (B) Images of lateral flow assays showing HC-LFA’s specificity for hybrids by probe type (B. anthracis RNA, 2 µg). (C) Images of lateral flow assays showing HC-LFA to detect rpoB transcripts in cellular RNA from B. anthracis or B. anthracis-spiked murine spleen tissue. The spiked bacterial load ranged from 107 to 109 CFU/100 mg of tissue. For bacteria alone, 2 µg cellular RNA, equivalent to a bacterial load of 109 CFU was diluted. 5 ng of P1 probe was used in hybridization and ssDNA controls. Assays were imaged after 30 min. Individual lateral flow strips are placed separately and are grouped within the dashed line in (B) and (C).

Full size image

To validate the ability of HC-LFA for B. anthracis infection detection, cellular RNA was extracted from spleen suspensions spiked with vegetative B. anthracis and used as a source of rpoB transcripts for hybridization. Parallel assays were also performed using cellular RNA from bacteria alone to form a standard curve against which to evaluate the assay's efficiency in simulated infections. The assay was positive for hybrids starting at 107 CFU/100 mg of tissue for both spiked spleen and bacteria alone with comparable results at all tested bacterial loads (Fig. 4C, quantifications in Fig. S4A). However, a false positive of low intensity was observed in the assay performed using the rpoB P1 ssDNA probe alone but not in assays using cellular RNA alone. A 30-min incubation time provided only marginally greater sensitivity than a 20-min incubation, but improved readability with the naked eye when near or at the detection limit of 107 CFU (Figs. 4C, S4A). We next tested the ability to detect rpoB RNA in the spleens of mice infected with B. anthracis spores and euthanized at various stages of malaise. We found clear detection in 3 of 5 tested samples and less clear detection in one sample (Fig. S4B, Table S1). However, the ability to detect infection was not correlated to malaise grade.

HC assays can detect RNA from parasite and viral pathogens

After demonstrating the effectiveness of HC assays in detecting gene-specific transcripts from bacteria, we aimed to extend the assay's applicability to diverse pathogens, such as Toxoplasma gondii (strain Prugniaud (PRU)) and SARS-CoV-2. To assess HC-ELISA’s ability to detect the presence of T. gondii, P10 probe (~ 2.2 kb) was hybridized to transcripts of the highly repeated B1 gene in cellular RNA extracted from infected human foreskin fibroblasts (HFF)27. Figure 5A shows that HC-ELISA could specifically identify B1 gene RNA from T. gondii-infected HFF cells when approximately 40% of the monolayer was lysed. The fold change in signal increased at higher parasite loads (monitored visually) indicated by the increase in percent cell lysis; however, an increase in cell lysis from 50 to 90% did not lead to further increase in signal over uninfected controls. We also assayed cellular RNA from overly confluent uninfected cells (C) to assess the potential for non-specific probe binding. No non-specific detection of transcript was seen in over-confluent HFF cells.

Figure 5
figure 5

HC assays can detect RNA transcripts from T. gondii and SARS-CoV-2. (A) Average fold change in chemiluminescence from HC-ELISA to detect B1 gene mRNA of T. gondii-infected HFF cells over uninfected cells seeded at the same density at start of study and harvested at the same time as infected cells. The P10 probe was used in this experiment and C (control) is the probe hybridized with cellular RNA from overly confluent uninfected cells. Bars represent mean ± SD of duplicate wells in a representative experiment of n = 3. (B) Median chemiluminescence from HC-ELISA to detect in vitro synthesized Spike E mRNA of SARS-CoV-2. Data obtained using 2 × 109 molecules/well of specified nucleic acids (assuming full hybridization from equal amounts RNA and probe for P10 hybrid). Bars represent range of triplicate wells in a representative experiment of n = 3. (C) Lateral flow assays show the sensitivity of HC-LFA to Spike E gene hybrids generated from 5 ng (2.6 × 109 molecules) P1 probe and in vitro-synthesized mRNA in amounts specified. Assays were imaged after 30 min. Individual lateral flow strips are placed separately and are grouped within the dashed line.

Full size image

Next, we assessed HC-ELISA and HC-LFA for their ability to detect SARS-CoV-2 Spike E (also known as S) gene RNA. Spike E mRNA was synthesized by in vitro transcription and hybridized to an equal number of P10 probe (~ 3.7 kb) molecules for detection in HC-ELISA. HC-ELISA could clearly detect Spike E transcripts at 2 × 109 molecules of P10 hybrid (Fig. 5B). For HC-LFA, hybrids were produced using 5 ng (2.6 × 109 molecules) P1 probe and transcript abundance ranged from 0.02 to 20 ng (107 to 1010 molecules); the assay was able to detect down to 108 molecules of Spike E RNA with no false positives visible on the RNA strip and a minimal false positive visible on the ssDNA probe strip (Fig. 5C).

Discussion

The Hybrid Capture immunoassay (HC) developed in this study utilizes the DNA-RNA binding properties of S9.6 to specifically capture and detect pathogen-specific RNA transcripts without requiring sequence amplification. This assay was successful in identifying mRNA from known pathogens such as B. anthracis, T. gondii, and SARS-CoV-2, demonstrating its pan-pathogen adaptability.

S9.6 has been extensively utilized in recent years for investigating R-loop metabolism (single-stranded genomic DNA annealed to nascent mRNA) through techniques like DRIP-Seq (DNA-RNA immunoprecipitation followed by high-throughput DNA sequencing) and DRIPc-Seq (DNA-RNA immunoprecipitation followed by cDNA conversion coupled to high-throughput sequencing)28,29,30,31. Prior research has employed S9.6 for whole cell-based R-loop identification; however, since S9.6 displays some affinity for dsRNA24, the quality of results varies across studies, whether using immunofluorescence32,33,34,35,36 or dot blots37. In our dot blots, dsRNA-specific nuclease RNase III treatment showed that some signal from S9.6 bound to B. anthracis cellular RNA originated from dsRNA. Furthermore, treatment with RNase H—specific for RNA in DNA-RNA hybrids—revealed that approximately half of the signal stemmed from pre-existing hybrids in the form of R-loops. The dsRNA signal may be due to S9.6 detection of bacterial ribosomal RNA, which constitutes over 60% of a cell's total nucleic acid content24. Interestingly, samples containing only in vitro synthesized RNA showed no signal even with extensive double-stranded secondary structures predicted by mFold23. Any assay which uses S9.6 to detect pathogen transcripts hybridized to DNA probes in a cellular RNA sample must be designed to overcome the challenges posed by both pre-existing hybrids and competition from abundant dsRNA.

Recognizing the potential utility of this antibody beyond R-loop detection, we utilized the biotin-streptavidin interaction to enrich specific mRNA transcripts. Biotinylated ssDNA probes anneal to their RNA counterparts, which are then selectively captured on the streptavidin-coated plate surface. We assessed three types of biotinylated ssDNA probes for their impact on the sensitivity of HC assays. Mono-biotinylation in PM probes was accomplished by directly using 5′-biotinylated primers. However, these probes displayed suboptimal performance in both HC-ELISA and HC-LFA assays likely because of the low amount of biotin per probe. Previous studies suggested that Taq polymerase's 5′ flap endonuclease activity enables the incorporation of modified deoxyribonucleotides, such as biotin-14-dATP, into DNA strands via nick-translation38,39. We hypothesized that the incorporation of biotin-14-dATP would sufficiently biotinylate probes to enable sensitive detection of probe-target hybrids in HC assays. Indeed, the use of P1 and P10 pagA probes proved successful, facilitating the detection of pagA transcripts from 106 and 104 molecules, respectively.

Our primary selection criterion for target transcripts from each pathogen was specificity; secondary considerations included abundance and length. Transcripts encoding the protective antigen (PA) toxin (pagA, 2.2 kb), whose expression is induced by CO226, and AtxA (atxA, 0.8 kb), a transcription factor that regulates virulence genes including pagA, are unique to B. anthracis and provide a high degree of pathogen specificity on this basis. Transcripts encoding RNA polymerase beta subunit (rpoB, 3.5 kb) and gyrase A (gyrA, 2.5 kb) are constitutively expressed and are found in all bacteria, with rpoB displaying sequence variation which enables distinction even between members of Bacillus40. Similarly, the B1 gene of T. gondii is established as specific for the parasite and its multicopy locus is transcribed in tandem to produce 1.3 kb-long spliced transcripts, which were of sufficient length for our assay27. Lastly, transcripts of the Spike E (also known as S) gene (3.7 kb) encoding the spike protein from SARS-CoV-2 provided assay specificity to this virus.

After establishing that HC-ELISA could specifically detect hybrids from in vitro-synthesized pagA mRNA, we assessed its capability to identify hybrids generated from total bacterial RNA. Given S9.6’s largely sequence-independent 6-bp epitope, we reasoned that longer hybrids would enhance assay sensitivity and thus chose to detect rpoB in cellular RNA; rpoB is longer (3.5 kb) than pagA (2.2 kb), providing a larger number of S9.6 epitopes per hybrid. As observed with pagA, HC-ELISA using P1 and P10 rpoB probes generated signal above the cellular RNA background.

With the limit of detection (LOD) for HC-ELISA ranging from 104 to 106 molecules, we inferred that the assay could be employed for semi-quantitative estimation of differentially expressed genes. Several studies have shown that CO2 induces pagA expression in B. anthracis without affecting the expression of constitutive genes like rpoB and gyrA26. Our HC-ELISA results were consistent with these findings, successfully measuring the differential expression of pagA between bacteria cultured with and without supplemental CO2. To evaluate our assay's performance relative to an established quantitative method, we measured the abundance of pagA, rpoB, and gyrA transcripts in the same samples using RT-qPCR. The results demonstrated a 40-fold increase in pagA expression under CO2-enriched conditions. In comparison, the sixfold increase detected by HC-ELISA suggests that our assay can effectively identify significant changes in the expression of highly expressed genes.

After establishing the sensitivity and specificity of our HC assay, we then aimed to use the assay to diagnose anthrax infection. Organ infections where vegetative B. anthracis was spiked into spleen suspensions at 107 CFU/100 mg were used as a model for anthrax infection. HC-ELISA could detect B. anthracis rpoB RNA with a 24-fold greater signal and pagA RNA with a fivefold greater signal relative to uninfected controls. Two factors can account for the observed difference in signal intensity between genes: firstly, rpoB is a constitutive gene expressed at a higher level than pagA, a gene with transcription regulated by various environmental cues. Further, the rpoB transcript is also ~ 60% longer, presenting a higher number of epitopes for S9.6 binding.

To demonstrate that the HC format can be employed in low-cost diagnostics, we adapted the HC-ELISA into a lateral flow assay (LFA). Our findings indicate that HC-LFA is sufficiently sensitive to identify pathogen-specific DNA-RNA hybrids in a spiked organ. The assay could also detect rpoB transcripts in some but not all spleens of mice infected with B. anthracis A35 strain spores. This was not unexpected, as the murine model of anthrax A35 infection of C57BL/6 J mice with 2 × 107 spores rarely yields CFU values exceeding 105 bacteria in pools of whole spleen, liver and kidneys harvested at onset of malaise41. Thus, a portion of the spleen from one or two infected animals as used in our studies is unlikely to reach sufficient bacterial counts for detection until terminal stages of disease.

The ability of S9.6 to indiscriminately identify all DNA-RNA hybrids offers pan-pathogen adaptability, which is a significant advantage of HC assays. We successfully detected the presence of T. gondii and B. anthracis in simulated infection models, as well as SARS-CoV-2 in vitro. However, we observed substantial variability in the limit of detection (LOD) among pathogens and HC formats. The differing LOD between HC-ELISA and HC-LFA can be attributed to the inherent tradeoffs between time, sensitivity, and semi-quantitative detection in HC-ELISA, compared to speed, simplicity, and qualitative detection in HC-LFA. Variability between pathogens, on the other hand, is multifactorial. To achieve maximum sensitivity for a specific pathogen, careful analysis of potential target gene expression characteristics is necessary. This analysis relies on baseline characterization of the pathogen in existing literature, which highlights the assay's primary limitation: it is unlikely to be useful for detecting uncharacterized pathogens. An appropriate target should either be pathogen-specific (like pagA and Spike E) or exhibit enough sequence variation between species (like rpoB) to avoid falsely detecting closely related species. Target transcript abundance is also an important consideration, as highly expressed genes are superior targets for HC assays due to a higher target-to-cellular RNA ratio in samples, resulting in a lower LOD. Transcript length also plays a crucial role in determining the assay’s sensitivity, as demonstrated by the differences in maximum signal and LOD among atxA, gyrA, pagA and rpoB hybrids; longer hybrids with more S9.6 epitopes per molecule have a lower LOD than shorter hybrids.

Multiplexed probes targeting multiple genes specific to the pathogen have the potential for enhancing the assay's sensitivity. To assay for multiple pathogen-specific target transcripts using full-length probes would generate signals from a greater proportion of the cellular RNA, while using multiple small detection probes combined with a capture probe could overcome reduced hybridization efficiency caused by secondary structures in long RNA transcripts. The HC-ELISA assay is simple and can be adopted into sentinel labs, reference labs and national laboratories for pathogen detection to detect active infection at low cost. As this manuscript was prepared for submission, new reports on the use of S9.6 in diagnostics emerged, supporting our findings that it can be adopted for detecting any gene or transcript specific to a disease or pathogen13,14,15. Considering the usage of non-amplification-based nucleic acid detection for pathogen-specific targets, we find our approach to be 10 to 100-fold more sensitive in detecting in-vitro synthesized transcripts than the assay developed by Dey et al.13. However, the introduction of transcription-mediated-isothermal-RNA-amplification by Dey et. al. does improve the assay sensitivity to LOD = 10 copies/µl. Additionally, a different approach using primer exchange reaction-amplified protein-nucleic acid interactions could identify small DNA-RNA hybrids in micro-RNAs at a range of 100 aM, corresponding to 103 copies/sample14.

Conclusion

We developed a DNA-RNA Hybrid Capture immunoassay (HC) capable of detecting pathogen RNA transcripts using the S9.6 antibody, which specifically recognizes DNA-RNA hybrids in a sequence-independent manner. We developed the HC method in an ELISA format (HC-ELISA) and adapted it to a lateral flow format (HC-LFA). HC-ELISA has demonstrated the sensitivity to detect as few as 104 molecules of 2.2 kb-long RNA transcripts, while HC-LFA has shown the ability to identify the presence of B. anthracis equivalent to a bacterial load of 107 CFU. We successfully detected B. anthracis rpoB, pagA, and gyrA RNA from in vitro cultures' cellular RNA extracts and rpoB from simulated and real infections in mouse spleen with a total turnaround time of ~ 5 h. Moreover, we showcased HC's pan-pathogen adaptability by detecting B1 gene RNA from a simulated T. gondii infection and Spike E gene RNA of SARS-CoV-2 from in vitro transcription. Consequently, HC-ELISA and HC-LFA can detect the presence of long RNA transcripts (> 1.5 kb) from any organism without molecular manipulations such as reverse transcription or nucleic acid amplification. Future work could explore utilizing multiple pathogen-specific probes to further enhance detection limits.

Materials and methods

Plasmids and bacteria

Plasmids and bacterial strains used for this study are listed in Table 1. The SARS-CoV-2 gene Spike E was obtained from plasmid pET30a-Spike-E, which was a gift from Rick Tarleton (Addgene plasmid # 165096)42. B. anthracis pagA gene plasmid pET15b-His-PA was created in this laboratory. The B1 gene of T. gondii is a 1384 bp-long gene with 35 tandem copies from head to tail, harboring an intron of 388 bp. To clone the B1 gene without introns for the purpose of probe preparation, the fragments upstream (F1) and downstream (F2) of the intron were PCR amplified using the primers listed in Table 2. The amplified F1 and F2 fragments of the B1 gene were cloned into the pJET1.2 blunt end vector from the CloneJET PCR Cloning Kit (Thermo Fisher Scientific) following the NEBuilder HiFi DNA Assembly Kit protocol (New England Biolabs [NEB]). The non-encapsulated Sterne-like B. anthracis A35 strain43 was grown in NBY media at 37 °C to A650 of ~ 1.0 prior to harvest. Growth conditions for some studies included 0.8% NaHCO3 and 5% atmospheric CO2. Spores used in mouse infections were prepared as previously described41.

Table 1 List of plasmids and strains used in this study.
Full size table
Table 2 List of primers and oligonucleotides.
Full size table

Toxoplasma

Tachyzoites from type II Prugniaud (luciferase-expressing) T. gondii parasites44 were grown in Human foreskin fibroblast (HFF) cells following standard protocols45. Briefly, cells were grown in DMEM media supplemented with 10% fetal bovine serum, 2 mM GlutaMAX and 1 mM sodium pyruvate at 37 °C with 5% CO2 to 70–80% confluence in T-25 cells prior to infection with various doses of parasite.

For harvest, cells were scraped from flasks showing different levels of parasite-induced lysis as assessed by monolayer destruction (40–90%). This material was passed through a syringe (27-gauge needle) once and centrifuged at 300 rpm for 5 min to remove bulk of cell debris. From each T-25 flask, half of the post-spin supernatant containing tachyzoites was centrifuged at a higher speed (1300 rpm, 10 min) and pellets resuspended in 1 mL of TRIzol (Invitrogen) followed by processing as described for B. anthracis samples.

Mice

C57BL/6 J mice (in-house, male, 8 weeks old) were used as a source of uninfected spleens or in spore challenge studies. Mice were subcutaneously (SC) infected with 2 × 107 B. anthracis A35 spores prepared in PBS (1 mL) in the scruff of neck and monitored daily for signs of malaise prior to euthanasia and spleen harvest. All animal experiments were performed in strict accordance with guidelines from the NIH and the Animal Welfare Act, under protocols approved by the Animal Care and Use Committee of the National Institute of Allergy and Infectious Diseases, National Institutes of Health (protocol LPD9E).

Cellular RNA isolation

Cellular RNA from bacteria, protozoan, and mouse spleen was isolated using TRIzol (Invitrogen). 50–100 mg wet-weight B. anthracis culture was resuspended in 1 mL TRIzol and 250 µL of 0.1 mm zirconia beads was added (BioSpec Products). The cells were homogenized at 6.0 m/s speed for 60 s using FastPrep-24 (MP Biomedicals) then incubated on ice for 5 min. Following three rounds of homogenization, RNA was isolated according to the manufacturer’s protocol and used directly as cellular RNA for dot blot and hybrid capture assays. For C57BL/6J mouse spleen, a single cell suspension was made from 100 mg of organ material filtered through a 70 µm cell strainer (BD Falcon) and combined with TRIzol (1 mL) prior to RNA extraction. B. anthracis vegetative bacteria were spiked in select samples at specific doses in some studies.

In vitro transcription

Plasmids carrying the B. anthracis pagA and SARS-CoV-2 Spike E genes were used as templates for transcription following linearization (which improved the reaction’s quality and yield) with FastDigest BamHI for pET15b-His-PA or FastDigest XhoI for pET30a-Spike-E (Thermo Scientific). In vitro transcription was performed using T7 RNA Polymerase according to the manufacturer’s protocol (NEB). The reaction was then treated with TURBO DNase I (Invitrogen) and purified using the RNeasy Mini Kit (Qiagen). Generally, 500 ng of plasmid yielded ~ 50 ng of in vitro-transcribed RNA.

Single-stranded DNA (ssDNA) probe synthesis

Full length individual genes were amplified by PCR using LongAmp Taq DNA polymerase (NEB), equimolar concentrations of 5′-phosphorylated forward primer and 5′-biotinylated reverse primer (Table 2), and 0, 1, or 10 µM biotin-14-dATP (Invitrogen) (Fig. S1C–G). Probes synthesized without added biotin-14-dATP are termed “PM” for “mono-biotinylated probe”, while those synthesized with added biotin-14-dATP are termed “P1” and “P10” for “1/10 µM poly-biotinylated probe”, respectively. Lambda Exonuclease (NEB) was used to digest the 5′-phosphorylated sense strand leaving behind a biotinylated ssDNA probe. In general, 4 µL enzyme could digest 750 ng PCR product in a 50 µL reaction over 3 h at 37 °C. The ssDNA band was excised from 1% agarose gel and purified using Zymoclean Gel DNA Recovery Kit (Zymo Research). Probe purity was confirmed with Thermolabile Exonuclease I treatment (NEB).

Expression and purification of S9.6 monoclonal IgG

S9.6 IgG expression and purification protocols were followed as previously reported20. S9.6 IgG-producing hybridoma cells (HB-8730) were grown in DMEM supplemented with 10% fetal bovine serum, 2 mM GlutaMAX, 1 mM sodium pyruvate, and 50 μg/mL gentamycin (Life Technologies, Carlsbad, CA). The S9.6 IgG was purified from culture supernatants using HiTrap Protein G columns (GE Healthcare Life Sciences, Piscataway, NJ).

Nucleic acid hybrid formation

DNA-RNA hybrids were generated from respective ssDNA probes and in vitro transcribed mRNA or cellular RNA. For in vitro transcribed mRNA, 1.25 ng of respective DNA probe was mixed with an equal amount of mRNA in a hybridization buffer (25 mM Tris–Cl, pH 8.0, 150 mM NaCl, 1 mM MgCl2). For cellular RNA, 5 ng of probe (unless mentioned; based on calculations from Ct values in qPCR) was mixed with 2 µg of cellular RNA. To achieve hybridization, the reaction was heated at 95 °C for 3 min and transferred to a boiling water bath covered with aluminum foil, which was allowed to gradually cool to room temperature. Alternatively, hybridization was achieved using a thermocycler programmed to hold sample temperature at 95 °C for 3 min, then cool 0.1 °C every 6 s to the equivalent of room temperature.

Nuclease treatment

For the dot blot assay, cellular RNA from B. anthracis was treated with nucleases of various specificities including RNase H, ShortCut RNase III, and Mung Bean Nuclease (NEB); RNase T1 and TURBO DNase I (Invitrogen); and RNase A (Sigma-Aldrich). The treatments were carried out following the manufacturer’s protocols, with incubation at 30 °C for Mung Bean Nuclease and 37 °C for all other nucleases, prior to spotting on the membrane. For HC-ELISA, nucleic acids were treated after hybridization with the nucleases listed above, except Mung Bean Nuclease. Enzyme treatment was performed according to each manufacturer’s protocol with the following quantities of enzyme used: RNase H, 0.2 uL (1 U); RNase III, 1 uL (2 U); Mung Bean Nuclease, 0.5 uL (5 U); RNase T1, 1 uL (1000 U); DNase I, 0.5 uL (1 U); RNase A, 1 uL (1.5 mU).

Dot blot assay

A 200 µl of DNA-RNA hybrids and related controls were treated with nucleases (described above) and transferred in 20× SSC by vacuum manifold (Merck Millipore) onto a Nylon H + Hybond membrane (Amersham, Merck). Nucleic acids were crosslinked to the membrane by baking (2 h, 80 °C) and the membrane was then rehydrated in 20X SSC buffer prior to blocking with Intercept TBS Blocking Buffer (LiCOR; 1 h) and adding S9.6 (1 µg/mL, 16 h, 4 °C). The membrane was washed 3 times with 1X TBS + 0.05% Tween-20 prior to incubation with DyLight 800 dye conjugated anti-mouse IgG (Rockland; 1:10,000, 1 h). The membrane was again washed 3 times and imaged using Odyssey CLx (LiCOR). Pixel intensity for each sample on the dot blot was calculated using Image Studio software (LiCOR).

Quantitative real-time polymerase chain reaction (qPCR)

Cellular RNA was treated with TURBO DNase I and purified using RNeasy Mini Kit; 1 µg purified RNA was used to synthesize cDNA with iScript™ cDNA Synthesis Kit (Bio-Rad). qPCR was performed with 5 ng of cDNA template per reaction in the QuantStudio™ 7 Flex Real-Time PCR System (Thermo Fisher Scientific).

Hybrid capture immunoassay by ELISA (HC-ELISA)

For DNA-RNA detection by a hybrid capture ELISA format (HC-ELISA), DNA-RNA hybrids containing biotinylated DNA were captured onto streptavidin-coated clear 96-well plates (Pierce, Invitrogen) in binding buffer (50 mM TrisCl, pH 7.5, 150 mM NaCl, 0.005% Tween 20, 0.1% BSA) at room temperature for 2 h. Post-binding, the plate was washed 3 times with wash buffer (25 mM TrisCl, pH 7.5, 150 mM NaCl, 0.1% Tween 20, 0.1% BSA). S9.6 in Intercept TBS Blocking Buffer was added to the plate (1 µg/mL, 16 h, 4 °C). Three washes were repeated prior to incubation with HRP-conjugated chicken anti-mouse antibody (Aves Labs; 1:10,000, 1 h) and development with SuperSignal ELISA Femto Substrate (Thermo Fisher Scientific), reading signal on a Victor3V microplate reader (PerkinElmer).

DNA-RNA hybrid-capture immunoassay by lateral flow assay (HC-LFA)

Lateral flow strips, chase buffer and S9.6 gold nanoparticle conjugates (S9.6-GNP) were used in HC-LFA using the proprietary knowledge of DDTD, CA. BioReady 150 nm gold nanoshells functionalized with carboxylic acid groups (nanoComposix) were used to conjugate S9.6 IgG, using manufacturer’s protocol. The control (C) and test (T) lines on the nitrocellulose lateral flow strips were prepared using solutions of 1 mg/mL of goat anti-mouse IgG (Lampire Biological Laboratories, Pipersville, PA) and 0.5 mg/mL of streptavidin (Sigma-Aldrich), respectively. Nucleic acids were added to the sample port in volumes up to 10 µL followed by sequential addition of 2–4 µL S9.6-GNP and ~ 100 µL of chase buffer (2–3 drops) to the conjugate port. The strip developed for 30 min prior to imaging and analysis on a Leelu Reader (Lumos Diagnostics, Carlsbad, CA).