Programmable CRISPR-responsive smart materials

Cas12a has both targeted and collateral deoxyribonuclease activities

The Cas12a from Lachnospiraceae bacterium ND2006 (1, 2) displays a specific cleavage activity toward double-stranded DNA (dsDNA) fragments matching the gRNA spacer sequence (targeted cleavage) and subsequent indiscriminate single-stranded DNA (ssDNA) hydrolysis activity (collateral cleavage, fig. S2). Once the Cas12a-gRNA complex binds and cleaves its dsDNA target, the collateral cleavage of nearby ssDNA by Cas12a is highly efficient (~1250 turnovers per second) (2). To demonstrate the target programmability of CRISPR gels, we tested gRNAs to detect fragments of the mecA antibiotic-resistance gene of methicillin-resistant Staphylococcus aureus (MRSA). We selected MRSA as a clinically relevant model to demonstrate the target programmability of CRISPR gels, given the high concentration of extracellular DNA in MRSA biofilms (13).

We first validated the collateral cleavage activity of Cas12a-gRNA in solution (14). Cleavage of an ssDNA reporter containing a quenched fluorophore by Cas12a in response to mecA dsDNA triggers confirmed the performance of the MRSA1 gRNA (fig. S3). Cas12a programmed with this gRNA detected its cognate trigger down to ~16 pM in solution (fig. S4) and was selected for subsequent experiments. Sequence mismatches between the gRNA and trigger reduced the collateral cleavage rate, consistent with previously reported patterns (2, 15). Nonspecific activation of Cas12a by a scrambled dsDNA sequence was over 100-fold lower than that by the mecA trigger (fig. S4). Cas12a’s high dsDNA target specificity and rapid nonspecific ssDNA collateral cleavage activity make it an ideal candidate for modulating a wide range of physical and mechanical hydrogel properties (fig. S1).

Programmable materials capable of the controlled release of soluble compounds, as well as encapsulated cells, have broad utility in various therapeutic and research applications (16–19). Because of the diversity of DNA-based hydrogels in the literature (table S1), we selected material formulations that span a wide range of physical properties to demonstrate that they could be actuated by Cas proteins (table S2): (i) poly(ethylene glycol) (PEG) hydrogels with covalently bound nonstructural DNA pendants that are released by Cas12a nuclease activity without degrading the overall hydrogel structure; (ii) acrylamide hydrogels with structural ssDNA cross-links that could be cleaved by Cas12a, resulting in bulk gel degradation as well as particle or cell release; and (iii) conductive hydrogels loaded with carbon black (CB) that were cleaved from the surface of electrodes by Cas12a, thus behaving like an electronic fuse.

Collateral Cas12a activity releases ssDNA-anchored cargos from hydrogel matrices

The targeted dsDNA cleavage activity of Cas12a can be used to preferentially release anchored cargos with near-single turnover (fig. S5); however, we focused on the collateral ssDNase activity of the enzyme, as it allows for the efficient transduction of external stimuli into changes in material properties through catalytic signal amplification. To illustrate the programmable actuation of materials using Cas12a, we covalently tethered a fluorophore (Cy3) into PEG hydrogels through an ssDNA linker (Fig. 1A) and monitored its release into solution (fig. S6) upon Cas12a-induced cleavage. The Cas12a-gRNA complex was insufficient to catalyze cargo release; however, introduction of the mecA dsDNA trigger initiated the hydrolysis of ssDNA anchors (Fig. 1B and fig. S7). By contrast, a randomly permutated version of the mecA dsDNA (scrambled control) failed to do so. We then used horseradish peroxidase (HRP) enzyme as a model for larger biomolecule payloads, demonstrating that biological function was preserved after immobilization in hydrogels using ssDNA anchors and subsequent release by Cas12a (Fig. 1C). Within 10 min of exposure to a low-concentration (10 nM) dsDNA stimulus, we detected sufficient HRP activity in the supernatant for a visual readout (Fig. 1C, inset). Further incubation allowed us to routinely discriminate trigger and scrambled dsDNA down to 100 pM (fig. S8). These experiments are consistent with the efficient ssDNase activity of activated Cas12a (2).

• Download high-res image
• Open in new tab
• Download Powerpoint

To demonstrate that changes to the gRNA were sufficient to entirely reprogram the target responsiveness of the material, we designed gRNAs to target a panel of genes involved in S. aureus antibiotic-resistance mechanisms (fig. S9). These include the antibiotic-resistance genes ermA and ermC (20, 21), the virulence factor gene spa (22), and the vancomycin-resistance gene vanA (23). Out of 25 combinations of gRNAs and dsDNA, those in which the sequence of the trigger matched the gRNA resulted in substantially higher fluorophore payload release from the hydrogel matrix (Fig. 1D). These results correlated to similar observations of the reactions performed in solution (fig. S10) and suggest that different gRNA-trigger pairs activate Cas12a to different extents (2).

For many applications, the rate at which a molecule is delivered from a carrier conveys important biological information (16). The speed of CRISPR-mediated hydrogel actuation corresponds to the amount of input dsDNA (fig. S11); conversely, for a given level of input, the response dynamics can be hard-coded into the system by altering the properties of the starting material. For example, pore size is expected to alter the mobility of macromolecules in polymer networks (24). On the basis of our macroscopic observations of programmed anchor hydrolysis (fig. S12), we hypothesized that this could be used to further tune the relationship between dsDNA input and Cas12a-mediated response. By modulating the cross-linking density of a PEG-DNA hydrogel and measuring the rate of fluorophore release by Cas12a-gRNA, we established another strategy by which the behavior of the CRISPR-responsive material could be controlled (Fig. 1E and figs. S13 and S14).

In addition to controlling global dynamics of ssDNA cleavage through bulk material properties, we capitalized on the sequence-defined, addressable nature of the ssDNA linkers and the selectivity of the collateral cleavage activity of Cas12a for ssDNA over dsDNA (1, 2). We attached two different fluorophores (Cy3 and 6-FAM) into PEG hydrogels with distinct ssDNA linkers and preprogrammed the differential sensitivity of one linker over the other to Cas12a collateral degradation by hybridizing it with a complementary blocking strand in situ. Whereas the release of the unprotected fluorophore was unaffected, the speed of release of the hybridized reporter was markedly reduced (Fig. 1F and fig. S15).

Collateral Cas12a activity alters the large-scale mechanical properties of DNA hydrogels

The high catalytic efficiency of dsDNA-activated Cas12a-gRNA (k_cat/K_m ~ 1.7 × 10⁹ s⁻¹ M⁻¹) (2) makes it well suited for converting dsDNA signals into bulk material changes. To demonstrate this, we designed DNA cross-linked polyacrylamide (PA) hydrogels (9, 25) by separately incorporating two noncomplementary oligonucleotides into PA chains (fig. S16). We then cross-linked the PA-DNA precursors using an oligonucleotide strand that forms bridges between the PA-DNA chains. These cross-links contained single-stranded, AT-rich Cas12a collateral cleavage sites (Fig. 2A). In these hydrogels, degradation of DNA cross-links physically disrupts the polymer networks (fig. S1B) (24, 26).

The Cas12a-induced degradation of PA-based CRISPR gels was initially evaluated with a DNA-intercalating dye to label bridge sequences in PA-DNA gels and track gel integrity. The bridges were degraded upon exposure to gRNA-Cas12a and trigger dsDNA, as revealed by the dissipation of gel fluorescence at rates dependent on trigger concentration (Fig. 2B). Compared with experiments performed in solution (fig. S4), gel degradation appeared more robust to the introduction of sequence mismatches between the gRNA and dsDNA trigger (fig. S17). Using fluorescein isothiocyanate–dextran particles physically entrapped in the hydrogel, we also visualized the degradation of millimeter-scale PA-DNA hydrogels (fig. S18).

Programmable degradation of PA-DNA hydrogels was assessed by testing 25 combinations of different gRNAs and dsDNA triggers. Consistent with the nondestructive cargo-release experiments (Fig. 1E), PA-DNA hydrogel degradation occurred only when the gRNA and dsDNA sequences were complementary (Fig. 2C and fig. S19), demonstrating Cas12a-gRNA’s ability to discriminate between inputs.

Though biomolecules can be tethered to materials through well-defined, single linkers, physical entrapment in a polymer matrix represents a more general strategy to control the release of larger payloads. We tested the release of NPs by encapsulating PEG-coated gold NPs (AuNPs) (figs. S16C and S20) in PA-DNA hydrogels. Loading gels with both Cas12a-gRNA and a dsDNA trigger led to total NP release through Cas12a activation and gel degradation, whereas gels loaded with a scrambled dsDNA trigger showed no significant release of AuNPs relative to a buffer-only background (Fig. 2D and figs. S21 and S22). This was consistent with the disruption of the percolated network upon cross-link cleavage (27, 28).

The complex interactions between cells and surrounding materials have implications for tissue engineering and other therapeutic applications. We reasoned that Cas12a-gRNA would have the capacity to modify the extracellular matrix of cells encapsulated in DNA materials in response to predefined cues. We tested the reversible encapsulation of human primary peripheral blood mononuclear cells (PBMCs) in PA-DNA hydrogels by exposing the gels to activated gRNA-Cas12a. Complete gel degradation and cell release were observed within 2 hours in the presence of 1 μM dsDNA trigger (Fig. 2E and fig. S23), without compromising cell viability (figs. S24 and S25). Conversely, gels exposed to a scrambled dsDNA control remained intact in the same time frame. These results demonstrate that NPs and live-cell payloads can be immobilized in biocompatible hydrogels and released upon addition of trigger dsDNA sequences without the need for hydrogel redesign to accommodate different input signals.

Cas12a-controlled hydrogel formation in a paper fluidic device enables diagnostic readouts

We used a tunable PA-DNA hydrogel to control the permeability and electrical readout of a paper-based microfluidic device (Fig. 4). Paper-based technologies have shown promise for point-of-care diagnostics, as they are low cost, equipment-free, and easy to use (38, 39). Our device (Fig. 4A and fig. S24) expands on the concept of microfluidic paper-based analytical devices (μPADs) that rely on the capacity of hydrogels to obstruct flow through porous channels (40).

The layers of the device were folded to create a multilayered structure in which the hydrophilic regions are topologically aligned. Capillary-driven flow through the device terminated in a fifth layer where the output was measured (Fig. 4A and fig. S27). In this system, an intermediary layer contains PA-DNA gel precursors (Ps-X and Ps-Y) that, when mixed with ssDNA cross-linker, form a hydrogel in the paper channels (41, 42). The extent of gel formation, and therefore the rate of buffer flow, is dependent on the extent of degradation of the ssDNA gel cross-linker during a preincubation step. The activation of Cas12a can be confirmed by adding a fluorescent ssDNA reporter during this step (fig. S28). By degrading the cross-linker by using Cas12a, we were able to couple the level of buffer flow to the concentration of dsDNA trigger added to a Cas12a reaction incubated for 4 hours.

When nonspecific dsDNA trigger is present during preincubation, ssDNA cross-linkers are not cleaved, allowing for hydrogel assembly in the microchannel (Fig. 4A). Conversely, in the presence of a specific dsDNA trigger, unimpeded flow can be visually detected by adding dyes to the μPAD channel. We found the rate of buffer flow through a μPAD to be inversely related to the concentration of an MRSA dsDNA trigger. Using this visual output, we were able to detect dsDNA concentrations down to 400 pM (fig. S29).

To optimize our CRISPR-μPAD for field diagnostic applications, we used reverse transcription (RT) to expand the range of detectable biomarkers to RNA and coupled the RT to an isothermal amplification [recombinase polymerase amplification (RPA)] step to improve the limit of detection. We used RT-RPA followed by a μPAD readout to detect synthetic Ebola genomic RNA (43) down to 11 aM (Fig. 4B), a sensitivity matching other state-of-the-art CRISPR-based diagnostics (fig. S30) (2, 4, 5). This approach is promising for point-of-care diagnostics and has overall better performances in terms of sensitivity, portability, and cost than other molecular diagnostics (table S3).

Visual readouts of buffer flow are commonly used, yet they are difficult to couple to downstream hardware for data processing. To overcome this limitation, we modified the CRISPR-actuated fluidic system to read buffer flow as an electric signal: the microfluidic channel in the final layer was sandwiched between two electrodes and connected to an ohmmeter (Fig. 4A and fig. S27B). Electrical conductivity between the electrodes relied on electrolytes provided by the flowing buffer and was directly correlated to the buffer penetration length in the μPAD channel (fig. S31A) (44). Using this approach, sub-nanomolar concentrations of dsDNA trigger were successfully detected at a 5-min end point, without DNA amplification, demonstrating the potential of the CRISPR μPAD for embedded sensor applications (Fig. 4C). Nonspecific dsDNA trigger did not activate Cas12a, thus leaving the electrical circuit open (Fig. 4C and fig. S31). We were able to reduce the preincubation time required to observe a signal to 1 hour by tuning the properties of the acrylamide precursors (fig. S31B).

The wireless, decentralized logging of individual clinical tests during infectious disease outbreaks could address challenges with record keeping and logistics. To integrate CRISPR-Cas reactions with electronic monitoring systems through hydrogel actuation, we incorporated a wireless radio-frequency identification (RFID) module into the μPAD. The original design was modified such that buffer flow would short-circuit an interdigitated silver electrode, thereby modulating the efficiency of signal transmission by a flexible RFID tag (Fig. 4D and fig. S31). We then conducted an experimenter-blinded trial consisting of 12 samples (containing either 11 or 0 aM Ebola ssRNA amplified by RT-RPA) divided across three geographic locations (Fig. 4E and fig. S32). The experimenter preincubated the samples with Cas12a and Ebola-specific gRNA for 4 hours and then recorded the RFID-μPAD signals over the course of 2 min. Buffer flow through the μPAD in Ebola-positive samples caused short-circuiting of the RFID tag antenna, which was detected in real time as a change in the signal strength compared with an unmodified reference RFID tag (Fig. 4E). All positive and negative samples were correctly assigned using the RFID-μPAD (fig. S32).

We have demonstrated several strategies to interface biological signals with materials that combine the inherent programmability of Cas enzymes with hydrogel systems. These strategies offer control over a variety of complex behaviors and properties, including the release of molecules, NPs, and live cells, as well as bulk hydrogel degradation, electronic signal transduction, and microfluidic valve actuation. By exploiting the enzymatic properties of Cas12a, we have designed a platform that improves on hydrogel programmability and versatility, as only the gRNA molecule needs to be changed to allow hydrogel response to a user-defined DNA sequence. The catalytic activity of Cas12a improves sensitivity compared with DNA-responsive hydrogels requiring stochiometric amounts of DNA triggers for material activation. Finally, we demonstrate various forms of output that expand the capabilities of CRISPR-responsive materials and enhance existing biomaterial-based approaches for tissue engineering, molecular diagnostics, and bioelectronic interfaces with programmable readouts.