CRISPR-Cas-amplified urinary biomarkers for multiplexed and portable cancer diagnostics

The study addresses the challenge of multiplexed, non-invasive detection and classification of complex diseases, particularly cancer, using synthetic biomarkers. Endogenous biomarkers (e.g., shed proteins, nucleic acids) often suffer from low abundance or instability, limiting sensitivity and specificity. DNA barcoding offers vast multiplexing potential but unmodified nucleic acids are rapidly degraded in vivo, hindering use as circulating barcodes. Therapeutic oligonucleotide chemistries confer nuclease resistance but are incompatible with standard polymerase amplification and low-cost detection. The authors hypothesized that chemically stabilized DNA barcodes could be released in vivo in response to disease-associated protease activity, survive circulation, be concentrated in urine, and be decoded sensitively and specifically using CRISPR-Cas nucleases without upstream amplification. They engineered DNA-encoded synthetic urine biomarkers (DNA-SUBs) that target diseased sites, sense dysregulated proteolysis, emit nuclease-resistant DNA barcodes filtered into urine, and trigger Cas12a collateral activity for fluorescent or paper-based detection, enabling classification of disease states and potential point-of-care use.

- Liquid biopsy biomarkers are important in oncology but limited by abundance/stability of endogenous molecules (cfDNA fragmentation patterns, tissue-of-origin inference). Prior synthetic biomarker strategies use mass-encoded reporters or engineered cells to amplify signals but often require sophisticated instrumentation. - DNA barcoding has transformed multiplexing across applications but is hindered in vivo by nuclease susceptibility; stabilized therapeutic oligonucleotides exist (phosphorothioates) and reduce immunostimulation but cannot be amplified by conventional polymerases. - CRISPR-Cas systems (Cas12a/Cas13/Cas14) enable sequence-specific nucleic acid detection with collateral cleavage and portable readouts, often paired with isothermal amplification in infectious disease diagnostics. - Tumor microenvironment proteases (MMPs, serine, aspartic proteases such as cathepsins, PLAU) are elevated in cancers and have been profiled previously for activity-based diagnostics; prior nanosensors classified prostate cancers and profiled pulmonary proteases, and mass-encoded urine biomarkers enabled multiplex monitoring but required MS readouts. These works motivate a platform that merges in vivo stable DNA barcoding, protease-activated release, renal concentration, and CRISPR-mediated, multiplexed, low-cost detection.

- Chemistry and CRISPR interface: Designed phosphorothioate-modified single-stranded DNA (ssDNA) barcodes (20–24-mers) that retain the ability to activate LbaCas12a (Lachnospiraceae bacterium Cas12a) through complementary crRNA binding, triggering collateral ssDNA cleavage. Assayed activation in vitro and in unprocessed mouse urine using a fluorophore-quencher reporter and a lateral-flow FAM-biotin reporter. Length optimization identified 20-mer PAM-free barcodes as optimal in vivo due to renal filtration and concentration. - Singleplex targeted DNA-SUB: Constructed nanobody-based protease-activatable sensors targeting cMET-expressing prostate tumors. Engineered a peptide substrate responsive to PLAU (uPA) with an unpaired cysteine for site-specific conjugation; added a rigid linker to maintain nanobody folding. Conjugated a 20-mer phosphorothioated DNA barcode to the cMET nanobody (cMET-Nb-DNA) via click chemistry. Control: non-targeting GFP nanobody with PLAU-activated linker (GFP-Nb-DNA). Validated targeting and tumor accumulation via IVIS and immunofluorescence in PC-3 xenografts. - Multiplex untargeted DNA-SUBs: Built a panel of 5 protease-activated peptides (PAP7, PAP9, PAP11, PAP13, PAP15) covering metallo-, serine-, and aspartic proteases, each barcoded with a 20-mer phosphorothioated DNA and conjugated to an eight-arm 40 kDa PEG scaffold to prolong circulation. Selected PAPs by screening 16 candidates against recombinant proteases and ex vivo tissue homogenates from CRC-bearing lungs versus healthy lungs; integrated top five into 5-plex DNA-SUBs. - Animal models: (1) PC-3 human prostate cancer xenografts in NCr nude mice for singleplex cMET-Nb-DNA testing; (2) Syngeneic metastatic colorectal cancer (CRC) lung nodules established by intravenous MC26-LucF cells in BALB/c mice for multiplex monitoring at days 11 and 21; (3) Autochthonous KP (KrasLSL-G12D/+; Trp53fl/fl) lung adenocarcinoma model initiated by intratracheal Adeno-SPC-Cre, assessed at 7.5 and 12 weeks. - Dosing and collection: Intravenous injections of ssDNA activator (1 nmol), nanobody-DNA sensor (1 nmol by DNA content), or pooled 5-plex PEG sensors (0.2 nmol each; 1 nmol total DNA). Urine collected 1 h post-injection and assayed directly without processing. - Readouts: Cas12a trans-cleavage kinetics measured fluorometrically to determine initial reaction velocities (Vo); paper-based lateral-flow assays (LFA) for end-point visualization, quantifying sample and control bands (ImageJ). ROC analyses for classification performance. PCA of normalized urine barcode levels across models. - High-throughput multiplexing: Implemented a Fluidigm 96.96 microfluidic platform to test 44 distinct 20-mer ssDNA barcodes with corresponding crRNAs (pairwise matching including 2 non-targeting) in human urine and solution (2,116 reactions), assessing orthogonality and AUCs. - Additional characterizations: Hydrodynamic sizing, cryo-TEM of PEG sensors; biodistribution and pharmacokinetics of labeled DNA and nanobodies; safety histology and cytokine/chemokine profiling; qPCR and immunohistochemistry for target validation; statistical analyses with appropriate tests (t-test, Welch’s correction, Mann–Whitney, ANOVA).

- Chemically stabilized DNA enables in vivo CRISPR readout: Phosphorothioate-modified ssDNAs retained the ability to activate LbaCas12a; native ssDNA failed to activate after in vivo administration due to nuclease degradation. In vivo, 20-mer PAM-free barcodes provided optimal urinary activation consistent with size-dependent renal filtration. Paper LFA provided visible sample bands upon Cas12a activation with subnanomolar sensitivity; Michaelis–Menten analysis on paper yielded Km ≈ 0.2104 nM. - Singleplex targeted PCa detection: cMET-Nb-DNA accumulated in PC-3 tumors and yielded significantly increased Cas12a trans-cleavage rates in urine from tumor-bearing mice compared to healthy controls (normalized Vo; P=0.028 vs normal; P=0.0221 vs GFP-Nb-DNA; GFP-Nb-DNA vs normal NS P=0.9902). Paper LFA showed stronger sample bands for tumors. ROC AUC for cMET-Nb-DNA: 0.89 ± 0.10 (P=0.03); GFP-Nb-DNA AUC ~0.52 ± 0.21. - Multiplex CRC lung metastasis monitoring: In syngeneic CRC lung nodules, DNA-PAP7-SUB distinguished tumors at day 11 (small 1–2 mm nodules; Mann–Whitney P=0.0343) and remained significant at day 21 (P=0.0379). Additional sensors became significant by day 21 (PAP9: P=0.0054; PAP15: P=0.0171). Combining metallo- (PAP7, PAP15) and serine (PAP9) sensors improved classification (fluorescent readout AUC for 5-plex: 0.99 ± 0.01; selected combo PAP7+9+15 AUC: 0.94 ± 0.03). Paper LFA fingerprints mirrored fluorescence; ratio-based ROC for PAP7+9+15 AUC: 0.88 ± 0.07. - Autochthonous KP lung adenocarcinoma: At 7.5 weeks, PAP9 (serine-responsive) significantly distinguished tumors from controls (P=0.044). At 12 weeks, PAP9 (P=0.004) and PAP15 (MMP-responsive; P=0.045) were significant in both fluorescent and paper readouts. Combined PAP9+15: fluorescent AUC 0.94 ± 0.05; paper AUC 0.89 ± 0.09. - Disease signatures and specificity: PCA of urinary barcode patterns separated primary KP lung tumors from CRC lung metastases. Differential reporter enrichment across PCa, KP lung, and CRC metastases revealed tumor-type- and state-specific protease activity signatures (e.g., PAP7 and PAP15 enriched in CRC metastasis; PAP15 also enriched in KP; PAP9 decreased in lung nodules vs controls but not in PCa). - Massively multiplexed decoding: On Fluidigm microfluidics, 44 orthogonal 20-mer ssDNA barcodes with matched crRNAs showed selective detection in human urine and solution with median AUC 0.99 across base-paired sets and no cross-reactivity above a stringent threshold (6× s.d. over background).

The platform addresses the key barrier to DNA-barcoded synthetic biomarkers—instability of nucleic acids in vivo—by employing phosphorothioate-modified ssDNA that survive circulation and concentrate in urine for detection. Coupling disease-localized protease activation with CRISPR-Cas12a collateral cleavage provides a sensitive, programmable readout without upstream polymerase amplification and enables portable paper-based diagnostics. In targeted format (nanobody scaffold), the approach achieves tumor-specific activation and accurate detection in PSA-negative prostate tumors. In an untargeted, multiplexed PEG format, it captures heterogeneous protease activity in vivo to detect early-stage CRC lung nodules and track disease progression, and it discriminates primary lung tumors from metastatic CRC within the same organ via activity fingerprints. The findings indicate that small panels (2–3 sensors) can classify disease states in controlled murine models, while larger, orthogonal barcode sets—validated here up to 44—could scale to the complexity of human comorbid conditions. The size-optimized 20-mer barcodes ensure efficient renal filtration and robust urinary detection, enabling rapid, non-invasive monitoring. Overall, the work demonstrates a modular synthetic biomarker system with high signal-to-noise, specificity, and scalability, positioning it as a complementary alternative to endogenous liquid biopsy assays and as a foundation for point-of-care diagnostics.

This work introduces DNA-encoded synthetic urine biomarkers that combine protease-activated release of nuclease-resistant DNA barcodes with CRISPR-Cas12a-mediated detection, enabling multiplexed, non-invasive cancer diagnostics with fluorescent and paper-based readouts. The platform detects and classifies disease across multiple murine cancer models, distinguishes tumor types within the same tissue microenvironment, and scales to dozens of orthogonal barcodes using microfluidics. These advances establish a programmable, high-throughput diagnostic modality suitable for point-of-care deployment. Future directions include expanding sensor panels and enzymatic targets beyond proteases, optimizing Cas12a–crRNA–DNA interactions to further improve sensitivity, validating performance in comorbidity-bearing animal models and human studies, exploring additional delivery routes (oral, inhaled, topical), and integrating computational tools and multiplexed reader hardware to enhance diagnostic accuracy and usability.

- Current validation is in murine models; human disease heterogeneity and comorbidities will likely require larger, more diverse sensor panels and clinical studies. - Protease substrate specificity in vivo is imperfect; background cleavage in healthy tissues necessitates multiplexing and careful panel optimization. - The multiplex panel optimized for CRC metastasis performed suboptimally in the PC-3 model, reflecting model- and tissue-specific protease profiles. - While phosphorothioate DNA enables stability, modified duplexes have lower melting temperatures; further optimization of crRNA–DNA pairing may be needed to maximize Cas12a kinetics and sensitivity. - Translation requires thorough safety, dose, and off-target assessments across diverse physiological conditions; performance under renal impairment or urinary variability was not assessed.