Multiplexed CRISPR-based Methods for Pathogen Nucleic Acid Detection

Abstract

Bacterial and viral pathogens are devastating to human health and well-being. In many regions, dozens of pathogen species and variants co-circulate. Thus, it is important to detect many different species and variants of pathogens in a given sample through multiplexed detection methods. CRISPR-based nucleic acid detection has shown to be a promising step towards an easy-to-use sensitive, specific, and high-throughput method to detect nucleic acids from DNA and RNA viruses and bacteria. Here, we review the current state of multiplexed nucleic acid detection methods with a focus on CRISPR-based methods. We also look toward the future of multiplexed point-of-care diagnostics.

1. Introduction:

Infectious diseases caused by viral and bacterial pathogens have continued to threaten human health and welfare for millennia. Viral pathogens, including influenza A virus, Ebola virus, and Coronavirus, have caused and will cause epidemics and pandemics [1]. Additionally, bacterial pathogens such as M. tuberculosis are a leading cause of death worldwide [2]. Early detection is essential to combat deadly viral and bacterial pathogens. Early detection within an individual will ensure better treatment outcomes, and early detection within a population will allow for informed policies that can minimize spread. A popular approach for detecting either a single pathogen or multiple pathogens at a time is nucleic acid-based detection, which includes Nucleic Acid Amplification Tests (NAATs) and sequencing-based methods [3]. In many circumstances, multiplexed methods are preferable since many pathogens are cocirculating, and there is a possibility that a person may be coinfected with multiple pathogens. Furthermore, some pathogens may mutate rapidly, leading to several subtypes of the pathogen to circulate in a given region at the same time. Therefore, multiplexed methods that can detect dozens or hundreds pathogens and subtypes within a given sample would be beneficial for viral and bacterial pathogen diagnostics. The discovery of clustered regularly interspaced short palindromic repeats (CRISPR)-CRISPR-associated (Cas) systems has enabled the development of powerful single- and multiplexed diagnostics tools. This review aims to cover the current state of multiplexed pathogen nucleic acid detection methods and discuss areas of improvement.

2. Non-CRISPR-based Nucleic Acid Detection:

The current gold standard for nucleic acid diagnostics is quantitative polymerase chain reaction (qPCR) or qPCR with reverse transcription (RT-qPCR), as required for RNA viruses [4]. qPCR methods can be divided into two categories: dye-based methods and probe-based methods. The former uses a dye such as SYBR Green that binds to double-stranded DNA, whereas the latter is more specific to the target DNA amplicon. Probe-based qPCR uses Taqman probes containing a fluorescent dye and quencher molecule that are designed to bind to a part of the target amplicon. During the PCR reaction, the polymerase moves from 5′ to 3′, and the exonuclease activity of the polymerase cleaves the quencher molecule from the fluorescent probe when it is bound to the target [5]. The cleavage of the Taqman probe leads to the accumulation of fluorescence over time proportional to the target amount. Each Taqman probe can be designed with a different fluorescent molecule, allowing for limited multiplexing of up to five targets in a single reaction (Table 1) [5-7].

Table 1.

Comparison of Multiplexed Methods

	Turnaround Time (Hours)1	Through put (Samples /Day2	Cost per Sample (USD)3	Multiplexing	Limit of Detection (copies/microliter)	Clinical Sensitivity7	Clinical Specificity7	Input Sample Type	Input Volume (microliter)	References
SHERLOCKv2	0.75-2	70-150	$5-6	4	1-10	Not reported	Not reported	Direct sampling (extraction free, RNA)	1	[29]
MiCaR	1	100-200	$5-10	9	0.16	45/46	53/54	Direct sampling (extraction free, DNA)	5	[30]
CARMENv1	7	20-100	$8-12	20-200	0.5	Not reported	Not reported	RNA extract	10	[31]
mCARMEN	5	576	$4-5	24-96	0.1	127/133	30/33	RNA extract	2	[31],[33]
LEOPARD	6	11-44	$7-10	5	1.2	4/4	5/5	RNA extract	1	[22]
RT-qPCR4	1-2	256	$8-12	5	1	27/34	48/48	RNA extract	2	[7], [46]
NovaSeq	24-96	96-3845	$50-250	50-1000+6	30-500	94/94	282/286	DNA or RNA extract	2	[40],[41],[47]
Sequel II	24-96	96	$75-100	50-1000+6	50-500	Not reported	Not reported	DNA or RNA extract	2	[42]
MinION	6-8	50-100	$20-100	500+	10	84/86	24/27	DNA or RNA extract	2	[43], [48]

^1:Turnaround time considers the duration of the target extraction, amplification (if any), and reaction steps. In the event that samples must be sent to an external facility for processing, the wait time until results are received is also considered in the turnaround time estimates.

^2:Sample throughput assumes an 8-hour workday and the availability of a skilled practitioner to complete the detection assays.

^3:The price assumes the multiplexing described in the multiplexing section of the table. Further, if plates/ chips were involved, the price assumed that the full chip or plate will be used for the given run.

^4:Automated RT-qPCR exists that may have higher throughputs than what we calculated. These technologies are reviewed and summarized in [44].

^5:These values are limited by library construction and, as mentioned in (3), could be higher with automation or pooling.

^6:Sequencing methods can demonstrate a wide range in multiplexing capabilities depending on the use of amplicon-based or metagenomic next-generation sequencing (mNGS). Amplicon sequencing can be less expensive and more sensitive but have lower multiplexing, while mNGS permits higher multiplexing at a higher limit of detection.

^7:Clinical sensitivity and specificity values are based on reported values for SARS-CoV-2 except for MiCaR, whose clinical sensitivity and specificity values are based on detection of nine HPV subtypes.

In addition to qPCR, digital PCR (dPCR) is a sensitive method to detect and quantify nucleic acids from pathogens. For instance, droplet-based dPCR (ddPCR) uses microfluidics to dilute the sample into many nanoliter-sized droplets such that each droplet can contain few to no target nucleic acids[5]. PCR then amplifies individual droplets, and the fluorescence of each droplet is measured. Poisson statistics are used to quantify the number of templates present [5]. Compared to qPCR, dPCR has higher sensitivity, precision, and reproducibility [5]. Like qPCR, dPCR can implement probes with different fluorescent molecules to allow multiplexing. In addition, three different targets were detected in a single sample with the fluorescence channel by designing the assay such that there was a large difference in the concentration of the dPCR reaction mixture [8]. Though dPCR has great potential for single and multiplexed pathogen detection, it still trails behind because there is currently a lack of automation and throughput compared to qPCR.

Sequencing, and in particular next-generation sequencing (NGS), has become the standard technique for de novo pathogen and variant identification [9-11]. For instance, what was once done using Sanger sequencing methods, Tzou et al. used showed that NGS can be used to detect Human Immunodeficiency virus (HIV) drug resistance mutations (DRMs) with greater sensitivity than the standard method, Sanger sequencing [12,13]. Additionally, NGS technologies have allowed us to discover new viral strains through metagenomic sequencing and phylogenetic analysis and provide a quick way to detect them, along with other viruses that may be present [14]. NGS is limited by the high cost of equipment and reagents and the training required for complex data analysis (Table 1). Third-generation sequencing methods, such as Nanopore sequencing, are promising for future viral and bacterial diagnostics. Third-generation sequencing allows for the sequencing of longer reads in a shorter time and the ability to directly sequence RNA without the need to convert RNA to cDNA before analysis [9]. However, third-generation sequencing is limited by its high error rate, the complex workflow of nucleic acid extraction and library preparation, and difficult data analysis [15,16]. As the cost of sequencing continues to decrease and data analysis software becomes more user-friendly, we may see an increase in the use of NGS and third-generation sequencing as diagnostic tools.

3. CRISPR-based Nucleic Acid Detection:

CRISPR-based diagnostics have emerged as powerful tools to detect a wide range of bacterial and viral nucleic acids with ease of use, high sensitivity, and exquisite selectivity [17-19]. CRISPR-based diagnostics use Cas enzymes, such as Cas12 and Cas13, to detect target DNA or single-stranded RNA (ssRNA) targets, respectively. Both Cas 12 and Cas13 have a crRNA that is complementary to the target nucleic acid sequence [18]. In a detection reaction, a crRNA-Cas complex binds to the target nucleic acid through complementary base-pairing, after which the Cas enzyme activates and cleaves the target nucleic acid through cis cleavage. Then the Cas enzyme remains activated and cleaves nearby ssDNA or ssRNA through a process called collateral or trans cleavage (Figure 1A). CRISPR-based detection systems couple this collateral cleavage activity with a reporter, a fluorescent molecule attached to a quencher molecule by single-stranded DNA or RNA, whereby collateral cleavage leads to the fluorescent molecule being released to generate a signal (Figure 1B) [20]. It is important to note that Cas12 and Cas13 remain activated, and so one target-binding event leads to a collateral cleavage rate on the order of 1-10 cleavage events per second [21]. In addition to the signal amplification that comes with collateral cleavage, coupling CRISPR-based methods with an amplification step greatly increases its sensitivity and allows for the detection of clinically relevant levels of nucleic acids. In addition to collateral cleavage offered through Cas12 and Cas13-based detection methods, there are other approaches to nucleic acid detection. Cas9’s programmable DNA binding can also be harnessed for nucleic acid detection using a gel-based or electrochemical readout (Figure 1A) [22,23]. Though there are many ways to detect nucleic acids, the basic workflow for CRISPR-based diagnostics remain the same. After sample collection, the RNA or DNA must be extracted, then there is an optional preamplification step that allows for greater sensitivity, and finally, nucleic acids can be detected using the different Cas proteins (Figure 1D).

Figure 1:

(A) Schematic showing Cas9, Cas12, and Cas13 binding to its target nucleic acid. Cas12 and Cas13 have collateral cleavage activity that allows for highly sensitive and specific nucleic acid detection. (B) Cas13 binds to target RNA through complementary-base pairing with its crRNA. After binding, Cas13 is able to cleave non-target RNA including a reporter to generate a fluorescent signal. (C) Workflow of CRISPR-based diagnostics. Briefly, the nucleic acid must be extracted. Then there is an optional pre-amplification of the target, followed by a CRISPR-based detection reaction.

Specific High-sensitivity Enzymatic Reporter unLOCKing (SHERLOCK), first described in 2017 by Gootenberg et al., was one of the earliest works that showed the potential of CRISPR-Cas systems for viral and bacterial pathogen diagnostics [24]. SHERLOCK used recombinase polymerase amplification (RPA) coupled with T7 transcription to amplify the nucleic acid before detection by Cas13 [24,25]. With this amplification step, SHERLOCK achieved attomolar sensitivity in a single reaction system with low variability across replicates [24]. Furthermore, SHERLOCK detected nucleic acids from clinical samples from individuals infected with various viruses, including Zika Virus (ZIKV), Dengue virus (DENV), and bacterial pathogens by targeting 16s rRNA gene from E. Coli and P. aeruginosa [24,25].

Soon after, one-HOur Low-cost Multipurpose highly Efficient Systems (HOLMES), which uses Cas12, also showed an attomolar sensitivity following a PCR amplification step [26]. Furthermore, HOLMES detected DNA viruses, such as pseudorabies virus (PRV), and RNA viruses, such as Japanese encephalitis virus (JEV), following a reverse transcription step, from infected cells [26]. The researchers recently improved this method by combining the amplification and detection steps in a one-pot reaction called HOLMESv2 [27].

Like HOLMES, DETECTR uses Cas12 to detect its nucleic acid target. This method. After coupling this CRISPR-based method with RPA amplification, the authors were able to detect HPV from cultured cells and were able to detect HPV from patient samples at attomolar sensitivity successfully [28].

Though the mentioned methods are sensitive and selective, the majority of the methods mentioned are singleplex methods, so they can only test for one target or pathogen within a given reaction. Thus, it is important to discuss current CRISPR-based multiplexed nucleic acid detection methods.

3.1. Multiplexed methods of detection

The programmability of CRISPR-Cas systems has recently been leveraged to create multiplexed pathogen detection methods. Different means of multiplexing, including orthogonal reporters, microfluidics, and color coding, have been used to achieve high-throughput identification of pathogens in patient samples. These different multiplexing strategies are demonstrated by SHERLOCKv2, microfluidic device with CRISPR-Cas12a and multiplex RPA (MiCaR), and Combinatorial Arrayed Reactions for Multiplexed Evaluations of Nucleic acids (CARMEN), respectively (Figure 2A) [29-31].

Figure 2:

(A, left) Orthogonal enzyme approach to CRISPR-based diagnostics. Multiple Cas12/13 orthologs are introduced within a given reaction mixture. Each ortholog has specific cleavage preferences for dinucleotide motifs, thus, when these motifs are paired with different fluorescent reporters, one can achieve multiplexed detection. (A, center) MiCaR and mCARMEN use custom or commercially available microfluidics chips in order to achieve multiplexed detection. (A, right) CARMEN uses color-coded emulsions whereby assay droplets containing crRNA, Cas13, and reporters and sample droplets are mixed together to initiate detection. If the target is present, a fluorescent signal is generated. (B and C) Graph depicting different multiplexed nucleic acid detection methods and how they compare with each other. (B) CRISPR-based diagnostics offer short turnaround time and relatively high multiplexing capabilities compared to sequencing and PCR-based methods (C) CRISPR-based diagnostics offer relatively high sample throughput and short turnaround time compared with PCR and most sequencing methods. The sample throughput was estimated with the assumption of detecting at least 20 pathogens and the multiplexing capabilities described in Table 1.

SHERLOCKv2 built upon the work of the first iteration of SHERLOCK through the use of orthogonal CRISPR-Cas enzymes for the simultaneous detection of four targets [29]. Through dinucleotide cleavage screens, Gootenberg et al. found that LwaCas13a, CcaCas13b, LbaCas13a, and PsmCas13b exhibited specific cleavage preferences for particular dinucleotide motifs and that these motifs could be paired with multiple reporters to achieve multiplexed detection [29].

The use of orthogonal enzymes and reporters can, however, be limited due to overlapping reporter emission spectra and the diversity of reporter motif preferences across Cas orthologs. In the SHERLOCKv2 study, the authors were able to identify four motifs that were preferentially cleaved through the collateral activity of four different Cas13 orthologs, but found that Cas13 orthologs shared many of the other motifs in the screen. As a result, other groups have taken a microfluidics-based approach to multiplexed detection (reviewed in [32]). For example, MiCaR uses a microfluidics chip to spatially distribute a pre-amplified sample mixture across 30 different outlets on the chip containing unique Cas12a-crRNA combinations [30]. The authors validated MiCaR by detecting the nine HPV subtypes targeted by 9-valent HPV vaccines [30]. All nine HPV subtypes could be detected in triplicate with high specificity and low background using the 30-well microfluidic approach [30].

Recent efforts have sought to increase the multiplexing capabilities of Cas-based diagnostics to enable new modalities of pathogen diagnostics such as SNP detection through the use of droplet color-codes and commercial microfluidic instruments. By emulsifying pathogen-specific detection mixes and amplified sample mixtures, CARMEN-Cas13 could detect up to 169 unique pathogens on a microwell plate via fluorescence microscopy [31]. This level of massive multiplexing has been further simplified through commercially available microfluidics chips from companies like Standard Biotools. The mCARMEN platform works similarly, but instead of relying on color codes to identify combinations of guide RNAs and samples, the Standard Biotools microfluidics platform distributes combinations of assay and sample mixes across thousands of observable nanoliter-scale chambers [33]. The commercial availability of the integrated fluidic circuits abstracts away the complexity of the required instrumentation and thereby makes the mCARMEN method more accessible for end users. Additionally, the fluid mixing facilitated through the chip instrumentation and the higher initial reaction temperature for mCARMEN allows it to be more sensitive and exhibit faster overall reaction kinetics. However, CARMENv1 can provide more multiplexed pathogen detection because it is not limited to the number of assay wells available on the Standard Biotools chip.

As we discover additional properties of different Cas enzymes, we will be able to create new CRISPR-based diagnostics. For instance, Jiao et al. discovered that they could engineer transactivating CRISPR RNAs (tracrRNAs) to detect RNA sequences by hybridizing to the RNA of interest and forming a non-canonical crRNA (ncrRNA) to direct Cas9 to the matching DNA leading to DNA cleavage [22]. Since the DNA sequence is of a specific length, the authors were able to detect five different RNA target sequences in a single reaction at attomolar sensitivity. They did this by cutting different DNA sequences for a given RNA target and resolving cleavage on a gel to multiplex their RNA detection platform called leveraging engineered tracrRNAs and on-target DNAs for parallel RNA detection (LEOPARD) [22]. Some limitations are that LEOPARD has relatively low sample throughput, and can only be used to detect RNA.

4. Current Challenges and Future Outlook

Nucleic acid detection workflows can be broken down into 3 major steps: sample preparation, amplification, and detection. Each step comes with its unique challenges and can vary based on the equipment or condition chosen (Figure 1C). The ideal method would involve improvements to all three steps, allowing for the rapid, multiplexed detection of nucleic acids from any source.

4.1. Sample Preparation:

The methods mentioned typically rely on a nucleic acid extraction step, which may be hard to implement in a low-resource or point-of-care setting. Additionally, bacterial and viral pathogens may require different extraction steps, which may complicate downstream multiplexed detection of viral and bacterial targets. Therefore, to allow for faster detection and screening in low-resource settings, extraction-free, direct detection of nucleic acids should be developed. Myhrvold et al. successfully combined SHERLOCK methods with Heating Unextracted Diagnostic Samples to Obliterate Nucleases (HUDSON), allowing for extraction-free detection of viral targets, such as ZIKV and DENV, from bodily fluid samples [34]. Qui et al. have developed Extraction-free one-step CRISPR-assistant detection (ExCad) to obtain nucleic acids from unextracted colonies and sputum samples [35]. They were able to successfully obtain and detect nucleic acids from S. pneumoniae [35]. In the future, it would be vital to create a single method that can be used for both viral and bacterial pathogens. Furthermore, though these methods are currently used for singleplex experiments, it would be interesting to test whether they can be adapted and optimized for detecting viral and/or bacterial pathogens in a multiplexed system.

4.2. Nucleic Acid Amplification

Cas-based detection assays often use target amplification methods to increase sensitivity. Depending on the type of pathogen, the simplest amplification approach often involves reverse transcription of the target followed by PCR. Conventional implementations of PCR-based pathogen detection typically involve the use of thermocycler equipment that can be difficult to deploy in field settings. However, smaller, more portable thermocyclers that enable RT-PCR amplification of viral targets have been developed [45]. These hand-held PCR assays can generate a readout within 1-2 hours and have been shown to exhibit high agreement with current gold-standard qPCR methods. While hand-held PCR assays have made sensitive point-of-care testing more accessible, the cost and required training can be prohibitive for use in resource-depleted regions.

As a result, many CRISPR-based detection methods use isothermal amplification. The DETECTR, Streamlined Highlighting of Infections to Navigate Epidemics (SHINE), and SHERLOCK platforms use loop-mediated isothermal amplification (LAMP) or RPA, isothermal amplification procedures which has the potential to be adapted for equipment-independent amplification of target material [25,28,36]. Isothermal amplification is inherently challenging to multiplex due to competition between amplicons, so future engineering efforts will be needed to address this issue.

Another way to amplify and accelerate detection signals can be through the use of signal-amplifying enzymes like Csm6, which is a CRISPR-associated enzyme activated by adenylates containing 2’,3’-cyclic phosphate ends [29]. Through collateral cleavage activity, Cas13 can create RNA products that have 2’,3’-cyclic phosphate ends, which can then be used as a mechanism for activating Csm6 and amplifying the reporter signal [29]. Csm6 has been shown to be activated efficiently by hexadenylated molecules [29]. Gootenberg et al. use Csm6 in conjunction with (A)₆-(U)₅ activator molecules in their SHERLOCKv2 protocol to enhance the rate of reporter cleavage and enhance the intensity of positive detection bands in the lateral flow assay format of SHERLOCK [29].

Amplification-free strategies have allowed for easier direct detection and quantification of nucleic acids using CRISPR-Cas-based systems. These amplification-free strategies have been reviewed by others [37]. Briefly, amplification-free strategies have the advantage of ease of use as there is no need to optimize the conditions to allow for amplification and detection and speed as one can directly detect the target and not wait for the amplification of the nucleic acid of interest. However, there are a few drawbacks as they will have lower sensitivity, so less abundant nucleic acids may not be detected [38]. In addition, with lowered sensitivity, amplification-free methods require more input volume. Therefore, multiplexing may become challenging when there is a limited amount of sample. Different amplification or amplification-free strategies can be used depending on the need for the experiment. If simplicity is needed, amplification methods may favor RPA for amplification as it can be coupled with detection for one-pot reaction systems. If sensitivity is not an issue, amplification-free methods may be favored.

4.3. Potential for multiplexed point-of-care diagnostics

For effective point-of-care pathogen detection, assays must be inexpensive, fast, and largely equipment-independent. Existing point-of-care assays include lateral flow formats, mobile-readable colorimetric outputs, and paper-based microfluidics with lyophilized assay reagents [39].

SHERLOCKv2 has been adapted for use in a lateral flow assay format by using anti-fluorophore antibodies to capture cleaved reporter molecules for visual readout [29]. In the sample loading region, reporter molecules are immobilized on the surface of the flow strip. If a pathogen of interest is present in the extracted sample, Cas13 will become active and cleave the reporter molecules, allowing free fluorophores to flow down the strip, where they will eventually be captured by immobilized anti-fluorophore antibodies. A multiplexed version of this assay could be constructed through the use of multiple distinct fluorophore-based reporters that are preferentially cleaved by orthogonal Cas13 orthologs. Fluorophore-specific antibodies can be placed on separate lines for clear visual detection of multiple pathogens, as illustrated in Figure 3. While the lateral flow assay format is unlikely to support the detection of dozens of unique pathogens like the CARMEN or MiCaR panels, it provides an inexpensive alternative for rapid pathogen testing in resource-limited settings.

Figure 3:

(A) Potential lateral flow-based workflow of point-of-care diagnostics. (Step 1) After sample collection, there is an extraction-free chemical nuclease inactivation and (Step 2) direct room-temperature isothermal amplification of the target nucleic acid and simultaneous reporter molecule cleavage via Cas13 collateral cleavage reactions. Different Cas13 orthologs can be paired with reporter molecules that have unique fluorophores. (Step 3) Finally, the sample would be loaded onto the lateral flow strip to allow for the detection of various pathogen nucleic acids. Fluorophore-specific antibodies are immobilized on visual detection lines to bind and immobilize fluorophores if cleaved in the Cas13 detection reaction. A biotin control band can be used to immobilize uncleaved reporter molecules to ensure minimal background. (B) Potential application of PDMS microfluidic chip technology for highly multiplexed Cas13-based pathogen detection. Samples are loaded into input wells across the chip and distributed spatially into reaction chambers containing lyophilized detection reagents. Inexpensive LEDs can be used to excite fluorescent reporter molecules and cell phone cameras can be used to obtain diagnostic readouts from reaction chambers.

Acronyms:

NAATs

Nucleic Acid Amplification Tests

CRISPR

Clustered Regularly Interspaced Short Palindromic Repeats

Cas

CRISPR-associated

qPCR

quantitative polymerase chain reaction

RT-qPCR

quantitative polymerase chain reaction with reverse transcription

dPCR

digital polymerase chain reaction

ddPCR

droplet-based digital polymerase chain reaction

crRNAs

CRISPR RNAs

HIV

Human immunodeficiency virus

NGS

Next-generation sequencing

ssRNA

single-stranded RNA

ssDNA

single-stranded DNA

SHERLOCK

Specific High-sensitivity Enzymatic Reporter unLOCKing

HOLMES

one-HOur Low-cost Multipurpose highly Efficient Systems

DETECTR

DNA endonuclease targeted CRISPR trans-reporter

RPA

Recombinase Polymerase Amplification

ZIKV

Zika Virus

DENV

Dengue Virus

PRV

Pseudorabies Virus

JEV

Japanese Encephalitis Virus

MiCAR

microfluidic device with CRISPR-Cas12a and multiplex RPA (MiCaR)

CARMEN

Combinatorial Arrayed Reactions for Multiplexed Evaluations of Nucleic acids

HUDSON

Heating Unextracted Diagnostic Samples to Obliterate Nucleases

ExCad

Extraction-free one-step CRISPR-assistant detection

LAMP

Loop-mediated Isothermal Amplification

LEOPARD

leveraging engineered tracrRNAs and on-target DNAs for parallel RNA detection

mNGS

metagenomic Next-generation Sequencing

SHINE

Streamlined Highlighting of Infections to Navigate Epidemics