Multiplex logic processing isothermal diagnostic assays for an evolving virus

Sanchita Bhadra; Miguel A. Saldaña; Hannah Grace Han; Grant L. Hughes; Andrew D. Ellington

In contrast, isothermal nucleic acid amplification tests (iNAATs) have been developed that rival PCR in sensitivity, cost far less, and do not necessarily rely on complex instrumentation. 4 To further the adoption of these assays, we and others have begun to develop 'smart molecular diagnostics' that can integrate information at the molecular level. For example, to increase signal specificity, a variety of sequence-specific probes, including molecular beacons, nucleasedependent probes, and fluorescence resonance energy transfer (FRET) pairs have been adapted to isothermal amplification assays such as rolling circle amplification (RCA), recombinase polymerase amplification (RPA), or loop-mediated amplification (LAMP). 5, 6, 7 More recently, RNA-guided CRISPR enzymes, such as Cas13a and Cas12a, have been used to signal the presence of isothermally generated amplicons. 8, 9 In our own previous work, we developed oligonucleotide strand displacement (OSD) probes, 10 that were triggered by strand exchange reactions with transiently single-stranded stem-loop sequences. 11 This led to the exquisitely sensitive detection of LAMP amplicons without interference. 12 One-pot LAMP-OSD assays that can work with crude samples are especially appealing for POC use. 13, 14 These assays have been coupled to highly sensitive and reliable 'yes/no' output signals such as fluorescence, glucose, or hCG that can be readily read using offthe-shelf cellphones, glucometers, or pregnancy test strips, respectively. 12, 13, 15 16, 17, 18, 19 Toehold switch RNA sensors have also been used to link iNAATs with in vitro reporter protein translation, leading to colorimetric signals. 20 Even as these iNAAT tests move towards wider adoption, they all still face the same problem as other POC assays, in that the answers they give are relatively simplistic. However, given that the strand exchange reactions that underlie OSD probes were originally derived from far more complex DNA computations, 10, 21, 22, 23 it may be possible to go beyond mere improvements in specificity and to integrate additional desirable features directly into the molecular diagnostic itself, such that the reaction helps to 'compute' its own outcome. As examples, we have previously developed strand exchange computation modules that can quantitate inputs to isothermal amplification reactions, 24 or that can integrate multiple molecular signals via Boolean logic operations. 17 We now attempt to take on real-world problems with computations that are embedded in the smart molecular diagnostics themselves. It is unfortunately a fundamental fact that diagnostic targets often evolve faster than assays designed to detect them. The resulting target sequence variations can easily prevent recognition of assay primers and probes, leading to test failure and false negative readouts, a problem which to date can only be solved via regular (and impractical) assay updates. 25, 26 For instance, primers and probes for all eight published reverse transcription (RT) qPCR assays (qRT-PCR) for Zika virus (ZIKV) were found to have numerous mismatches with multiple ZIKV genomes sequenced from recent outbreaks. 27 Some 20%-80% of ZIKV-infected patients are believed to have remained undiagnosed due to this incompatibility. 27, 28 To overcome this problem and increase the overall accuracy of POC nucleic acid testing we have engineered additional computations into our smart molecular diagnostics that can compute the presence of almost