Fast, inexpensive, and multiplexed detection of pathogens is of great value to human health and global security. Several diagnostic applications require the rapid detection of diverse nucleic acids. These include the identification of different types of pathogens (e.g., the quick differentiation of SAR-CoV-2 from the influenza virus or other respiratory viruses during the COVID-19 pandemic) and the discrimination of viral variants (e.g., subtyping of human papillomavirus [HPV], which has >100 variants). Next-generation sequencing (NGS) can provide detailed information and allow the identification of a wide range of pathogens. However, the turnaround time and cost are both high. Although various nucleic acid amplification methods can be used to detect multiple targets at a lower cost, some limitations and challenges remain. An ideal multiplexed detection platform would have the capability to address these above-mentioned problems. Such a platform should be able to: (i) correctly and efficiently amplify multiple (e.g., ≥4) nucleic acid targets; (ii) amplify and identify the targets at a suitable temperature; (iii) precisely recognize the targets without any interference; and (iv) accurately distinguish between signals and correlate them with each individual target.

Polymerase chain reaction (PCR) has long been utilized for multiplexed amplification. Real-time PCR is commonly used as the gold standard method for pathogen identification. To examine multiple targets, real-time PCR often relies on the detection of different fluorophores during amplification through nucleic acid hybridization. However, this detection approach has two primary limitations. First, the emission spectra of fluorescent probes often overlap, and therefore, the number of targets that can be tested is limited. Additionally, off-target amplification can confound target identification owing to the unavailability of rectification mechanisms7, and this problem worsens during multiplexed detection. Thus, although real-time PCR can address the first two challenges, it does not solve the last two.

Recently, CRISPR-based diagnostic (CRISPR-Dx) methods have been widely implemented for diagnosing infections due to their high specificity and sensitivity. These platforms harness the collateral cleavage activity of various Cas proteins (e.g., Cas12a, Cas13a, Cas14) to cut the designed reporters after activation. To further increase detection sensitivity, CRISPR-powered detection strategies are typically combined with isothermal amplification methods such as recombinase polymerase amplification (RPA) and loop-mediated isothermal amplification (LAMP). RPA, a fast and high-fidelity amplification method, is used more commonly because the reaction requires only a single temperature (37–42 °C) that can be easily achieved with simple heaters and is also compatible with the optimum temperature of CRISPR-Cas systems. Moreover, the detection specificity is guaranteed due to the sequence specificity between CRISPR RNA (crRNA) and the amplicons.

The first CRISPR-Dx platform called SHERLOCK was developed using Cas13a and reverse-transcription RPA (RT-RPA) for the detection of Zika and Dengue viruses. This platform had attomole (aM) sensitivity and single-base mismatch specificity. Subsequently, the multiplexed platform SHERLOCKv2 was developed by combining four CRISPR enzymes (LwaCas13a, PsmCas13b, CcaCas13b, and AsCas12a) with specific cleavage preferences and four types of fluorescent probes (FAM, TEX, Cy5, and HEX) to detect four targets. This platform had high potential as a multiplexable and sensitive platform for nucleic acid testing (NAT). However, it was challenging to increase the number of targets because of the difficulties in choosing additional appropriate CRISPR enzymes and probes. Therefore, in order to identify more pathogen targets, another Cas13-based platform called CARMEN was developed recently7,30. This system allows the simultaneous detection of up to 169 human-associated viruses using microwell array technology to pair CRISPR reagents and amplified samples. The readouts are obtained by identifying a pool of color codes that result from different ratios of four Alexa Fluor dyes. Though CARMEN represents a milestone for scalable and multiplexed pathogen detection, its widespread use may be compromised due to the multi-step color coding/decoding and the complicated assay protocol30. Additionally, owing to the four-fluorophore-based color-coding strategy, concerns regarding signal overlap due to the wide emission spectra of the fluorophores remain. Therefore, despite advancements in CRISPR-Dx technologies for multiplexed pathogen detection, the fourth challenge (generating distinguishable signal codes for multiple targets) remains unmet. Indeed, a recent work developed an improved CARMEN system (named microfluidic CARMEN, mCARMEN), which relies on commercially available Fluidigm microfluidics and instrumentation. It could be a promising strategy for the detection of multiple viruses, but the instrumentation setup and the device are relatively expensive and these might compromise its wide application in resource-restricted areas. Therefore, an “ideal” suitable and versatile platform that fulfills all the four requirements remains elusive.

Herein, we report the development of a unique platform that couples a microfluidic device with CRISPR Cas12a and multiplex RPA (MiCaR) to address all the challenges associated with multiplexed NAT. In addition to deploying the promising features of RPA and the CRISPR system, our strategy utilizes the unique properties of microfluidics to allow precise fluid control and spatial reagent arrangement. Instead of using multiple fluorophores or color coding to indicate different targets, MiCaR distinguishes between target signals based on spatial positions or space coding using only a single fluorophore. By preloading various crRNAs into the 30 designated wells, the system simultaneously initiates up to 30 CRISPR-based cleavage assays after the sample is pipetted onto the center of the Starburst-Shaped microchip (SS-Chip).

In this study, we evaluate the performance of the SS-Chip to ensure homogenous liquid division and mixing. We select the nine HPV subtypes targeted by 9-valent HPV vaccine (9vHPV). Subsequently, we carefully optimize and demonstrate the efficiency and specificity of the multiplex RPA assay. Additionally, we conduct a test containing 9 × 9 crRNAs against HPV subtypes to determine the optimal crRNA combination. Moreover, extensive testing demonstrates that MiCaR can achieve a detection sensitivity of 0.1 nM and 0.26 aM for unamplified and amplified targets, respectively. Furthermore, we apply MiCaR to test samples from patients at risk of HPV infection. In total, we test 100 samples, and 3000 assays were performed. MiCaR shows great stability, and the on-chip HPV subtyping results are highly consistent with clinical results, achieving a sensitivity of 97.8% and specificity of 98.1%. Finally, we apply MiCaR to successfully test eight common respiratory viruses to confirm the generalizability of our approach. Therefore, the findings show that MiCaR allows rapid, low-cost, and multiplexed NAT with high sensitivity, specificity, and reliability. This approach could be widely applied for the diagnosis of specific infections and provide significant health-related and clinical benefits.

Results

Principle of the MiCaR platform

To address all the four challenges mentioned earlier, we developed the MiCaR system (Fig. 1) as a universal platform for multiplexed NAT. Herein, we demonstrate its use in HPV subtyping. After sampling, cervical cells were heated to induce the release of HPV DNA. Then, direct multiplex RPA was performed without DNA extraction (Fig. 1a). The amplified products were loaded onto the microfluidic device and tested using the CRISPR-based cleavage assay. The readout was obtained via an automatic fluorescence imaging system. As shown in Fig. 1b, the detection by MiCaR relies on a Starburst-Shaped Chip (SS-Chip), which is designed as a hub-and-spoke network. The sample is loaded into the hub well at the center, and the spoke microchannels homogeneously distribute the sample into 30 labeled wells pre-loaded with a specific CRISPR Cas12a-detection mix containing Cas12a, crRNA, and a fluorescence reporter. The reporter used is a fluorophore quencher (FQ)-labeled oligonucleotide named TBA11 (GGTTGGTGTGG), a truncated form of thrombin binding aptamer (TBA), that has been proven to have higher sensitivity than a normal ssDNA reporter31,32. If the HPV DNA matches the crRNA, the relevant wells (three wells used as a group for testing the same HPV subtype) will show a bright fluorescence signal. However, a low background signal will be obtained when there is a mismatch between the HPV DNA and the crRNA. A blank control (with reporter and Cas12a in the absence of crRNA) was tested on the device to obtain the basic background. This space-coding based strategy enabled the efficient identification of multiple targets by employing the “amplify together and detect individually” principle. The use of only one fluorescent probe eases system setup and data analysis, and the triplicate on-chip test (for the same target) also ensures detection accuracy and reliability.

Fig. 1: Scheme of the MiCaR-based HPV subtyping strategy.

Primer and crRNA design for the HPV panel

There are >100 HPV subtypes, including high-risk and low-risk types, which can cause cervical cancer and condyloma acuminatum, respectively. As the risk associated with different HPV subtypes varies, it is necessary to screen and identify infections and prevent subsequent complications, such as cervical cancer. Vaccines can prevent cancer and diseases caused by some HPV infections. Currently, 9-valent (HPV−6, −11, −16, −18, −31, −33, −45, −52, −58) HPV vaccines are considered the most powerful for preventing HPV infection. In this study, we selected the nine HPV subtypes targeted by the 9vHPV to develop the MiCaR-based detection system. We first designed the relevant RPA primers for the nine targets. RPA is a highly sensitive isothermal amplification method that requires minimal sample preparation (Fig. 2a, left). Since the HPV subtype is determined by the L1 gene sequence, we designed a series of primers against this region in the nine targets (Fig. 2a, right).

Fig. 2: Designing of RPA primers and crRNAs.

The L1 gene sequences of these nine HPV subtypes show high similarity (Supplementary Data 1, Supplementary Table 1). To select the applicable primers, two rounds of designing and testing were performed. In round one, 10 pairs of primers were designed (Supplementary Table 2, one for each HPV subtype, with an additional one for HPV-11 owing to its high similarity with HPV-6; Supplementary Table 1). The performance of these primers was determined using agarose gel electrophoresis (Supplementary Fig. S1a). Most of the primers worked well and displayed clear and dense bands. However, primers for HPV-6 and HPV-45 did not generate distinct bands. Therefore, three additional pairs of primers were designed for these two subtypes based on the general principles of primer design (Supplementary Table 3). It was observed that the Primer HPV-6a, -6b, -45a, -45b, and -45c amplified the relevant targets successfully (Supplementary Fig. 1b).

To choose the preferable primers for HPV-6, -11 and 45, we comprehensively checked the sequence similarity between the amplification regions of the primer candidates (HPV-6a, -6b, -11a, -11b, -45a, -45b and -45c) and the L1 gene of the other eight subtypes. As shown in Supplementary Fig. 2a and Supplementary Data 2, the amplification region of Primer 6b showed a relatively greater difference than Primer 6a, which should contribute to reducing potential cross-reactivity in the subsequent CRISPR-based assay. Similarly, Primer 11b and 45a were theoretically superior to their counterparts (Supplementary Data 2, Supplementary Table 4). Therefore, Primer HPV-6b, -11b, -16, -18, -31, -33, -45a, -52, -58 were selected for the final panel (Supplementary Table 5). The gel results demonstrated the successful amplification of all nine targets (Fig. 2b). The L1 gene sequences and amplification regions of the nine HPV subtypes are listed in Supplementary Data 3.

In addition to an appropriate amplification region, optimal primer constitutions and concentrations are important for successful multiplex RPA assays. The Basic Kit Quick Guide for single RPA recommends a primer concentration of 1 μM. Thus, we initially tested RPA performance with the nine targets and nine pairs of primers (1 μM for each pair, and 9 μM in total). As shown in Supplementary Fig. 3a, severe primer dimer formation was observed, and no amplicons were obtained. Hence, RPA assays were conducted using only one target (HPV-16) with lower primer concentrations (0.1, 1, and 5 μM in total) to identify the best primer pool. As indicated in Supplementary Fig. 3b, the pool with a total concentration of 1 μM (~0.11 μM for each pair) appeared optimal for efficiently amplifying the target HPV-16 without causing significant primer dimer formation. Hence, this primer constitution was used for subsequent assays.

Next, we evaluated the performance of the primer pool in multiplex RPA. The products of the 9-plexed PRA were identified through next-generation sequencing (NGS) (Supplementary Fig. 4). The results indicated that a variety of amplicons were produced for each of the nine targets (Fig. 2c). This was expected, as amplicons of different lengths can be generated within a reaction for a single target in RPA. The top 5 sequences of the products showed high-fidelity to the original sequences (Supplementary Data 4). The relatively lower amplicon fidelity of HPV-11 and HPV-18 was resulting from one or two amplicons with shorter lengths comparing to the original templates. Anyway, all these amplicons showed a 100% fidelity in the crRNA binding regions. These results demonstrated the effectiveness of the primer pool.

Subsequently, we tried to optimize the crRNA pool. Initially, the crRNA pool was generated by only comparing the crRNA of a certain target to the amplification regions of non-targets (Supplementary Table 6). However, this led to cross-activity between the initial crRNA of HPV-16 (HPV-6 crRNA-I) and HPV-11 (Supplementary Fig. 5). To efficiently avoid cross-reactivity among the targets, a refined approach was used to design the optimal crRNA pool. HPV-16 was selected to generate proof of concept. First, all four potential crRNAs were designed within the RPA region according to the basic principles of crRNA design (Fig. 2d). Then, each crRNA was compared with the full L1 gene (and not just the amplification region) of the other eight HPV subtypes (Supplementary Fig. 6). Given that HPV is a double-stranded DNA virus, the sequence comparison could be performed directly using the crRNA and gene sequence. Comparison results showed that crRNA-2, -3, -4, and -5 were highly similar (difference of only two nucleotides) to certain regions of the full L1 gene in the other eight HPV subtypes (Fig. 2d, Supplementary Fig. 6b–d). Accordingly, crRNA-1 that showed the most significant differences was selected for HPV-16 (Supplementary Fig. 6a). The crRNAs for the other subtypes were designed in a similar manner. All the crRNA sequences are listed in Supplementary Table 7.

Validation of crRNAs and RPA primers for the nine HPV subtypes

Using the comprehensive theoretical analysis described above, optimal crRNAs were designed for all nine HPV targets. Then, a 9 × 9 matrix test (9 crRNAs × 9 HPV L1 gene plasmids; Fig. 3a, left panel) was designed to verify crRNA performance (i.e., recognition specificity). Additionally, a control group without any target gene was set up. After testing, widespread cross-activity was not observed (Fig. 3a, right panel). Similar trends were observed when the crRNAs were analyzed separately (Fig. 3b). Of note, the crRNA for HPV-58 showed a relatively high background for HPV-33. However, this was negligible compared to the signal for the actual matched sample (crRNA [HPV-58] against HPV-58). These crRNAs with high specificity were then used for subsequent multiplexed NAT.

Fig. 3: Comprehensive characterization of the performance of the crRNA pool and multiplex RPA.

A TwistAmp Basic kit containing RPA freeze-dried pellets was purchased from TwistDx (Cambridge, UK). Lachnospiraceae bacterium ND2006 Cas12a (Magigen LbCas12a) was purchased from Magigen company (Guangzhou, China). Plasmids containing nine types of the HPV L1 gene, relevant gene sequences of the eight respiratory viruses and TBA11-FQ were obtained from Tsingke Company (Beijing, China). The primers and crRNAs used were synthesized by Sangon Biotech Company (Shanghai, China). DNA agarose and nucleic acid gel stains were purchased from Yeasen Company (Shanghai, China). Fluorescein and sulforhodamine B were purchased from Shanghai Aladdin Biochemical Technology Co. Ltd. (Shanghai, China). The steam eye mask was purchased from Kao Commercial (Shanghai, China).

More detail please visit:

Microfluidic space coding for multiplexed nucleic acid detection via CRISPR-Cas12a and recombinase polymerase amplification

Zhichen Xu, Dongjuan Chen, Tao Li, Jiayu Yan, Jiang Zhu, Ting He, Rui Hu, Ying Li, Yunhuang Yang & Maili Liu