Cell-Free Recombinase Integrated Boolean Output System (CRIBOS)

Summary

Cell-free gene expression systems are increasingly important in fundamental research and biomanufacturing, offering a versatile platform for studying gene circuits and bio-computation. We present the Cell-Free Recombinase Integrated Boolean Output System (CRIBOS), a site-specific recombinase-based multiplex genetic circuit platform designed for cell-free environments. With CRIBOS, we build over 20 multi-input-multi-output circuits, including 2-input-2-output genetic circuits and a 2-input-4-output decoder. Combined with allosteric transcription factors (aTFs)-based sensors, the circuits demonstrate multiplex environmental sensing. Moreover, utilizing paper-based CRIBOS, which demonstrates remarkable portability and stability in a paper-based setting, we present a biological memory storage logic circuit device that can preserve DNA-based biological information for over four months with minimal resources, energy costs, and maintenance requirements. Implementing CRIBOS not only expands the application of multiplex Boolean logic gates from cellular systems to the cell-free environment but also augments their overall versatility, opening new avenues for designing and applying sophisticated genetic circuits.

A record of this paper’s Transparent Peer Review process is included in the Supplemental Information.

Introduction

Genetic circuits are foundational tools in synthetic biology with applications in medicine^1,2, diagnostics^3–5, agricultural production^6,7, and environmental surveillance^8,9. Synthetic gene circuits enable dynamic, precise, interactive, and logical control of cell function, which can facilitate drug discovery by enabling real-time monitoring of diverse molecular events and disease-relevant dynamics¹. Genetic circuits have also been applied to develop new classes of rapid, cost-effective, deployable diagnostic devices applicable to various diseases and chemicals⁴. Moreover, in chemical manufacturing, engineered cell factories equipped with sense-and-response systems enable dynamic control of expression pathways and optimal allocation of cellular resources¹⁰. In agriculture, crop yield can be boosted by smart plants incorporated with synthetic circuits capable of sensing and adapting to environmental challenges¹¹.

Increasingly, cell-free systems have been recognized as valuable platforms for fundamental interrogation and biomanufacturing^12–14.

A cell-free system is a membrane-free, simplified version of cells that operates without concern for cell viability. It provides many advantages for developing and characterizing circuits and bio-computation, such as a large dynamic range for chemical measurements¹⁵, reduced interference from irrelevant cellular pathways¹⁶, and a high level of control over the composition of the genetic circuits^16,17. Moreover, a cell-free environment can detect small molecules that are toxic or impermeable to cells^18,19. It can provide clearer signals and exclude most of the intrinsic noise caused by complex organelles and biochemical reactions inside the cell by simplifying the protein composition²⁰. With its portability and ease of detection, cell-free systems have been combined with diverse genetic circuits for point-of-care diagnostics²¹, environmental sensing²², and fundamental research on metabolic pathways^23,24 (e.g., allosteric transcription factors (aTFs)-based water contaminant detection system ROSALIND (RNA Output Sensors Activated by Ligand Induction)²², and CRISPR-based infectious disease diagnostics device SHERLOCK and DETECTR^25,26). These examples demonstrate the high sensitivity and robustness of genetic circuits created in cell-free systems and provide an ideal platform for research in synthetic biology and the construction of bio-computing devices.

Boolean algebra has been foundational in the design and analysis of digital circuits, forming the basis of computer logic and programming^27,28. Boolean logic gates, such as AND, OR, and NOT, can describe intricate computational tasks and can be transformed into equivalent functionalities efficiently. This makes it particularly suitable for constructing complex genetic circuits in synthetic biology, where precise control and predictability are paramount^29,30. Various types of bio-computing devices and Boolean logic gates have been applied to different bacterial³¹, yeast³², mammalian cell strain^33,34, and animal models³⁵ to enable complex bio-computation. However, their implementation in the cell-free field is not as developed, and circuits with multiple inputs and multiple outputs are still limited in the cell-free system, primarily due to limited tools for building cell-free biocomputing circuits. This limitation necessitates the integration of multiple layers of genetic logic gates to construct biocomputing devices and requires extensive construction and characterization for circuit design. The restricted exploration and construction of cell-free Boolean logic gate toolkits hinder the broader application of cell-free systems. This paper creates a cell-free Boolean biocomputing platform for complex multi-input-multi-output genetic circuit constructions. It makes a streamlined procedure to simplify the construction of cell-free genetic circuits to facilitate the study of complex metabolic pathways, controlled biologics manufacturing, and multiplex environmental sensing.

Site-specific DNA recombinases are powerful tools for developing multiplex Boolean logic gates in various organisms and models, such as bacteria⁴⁰, mammalian cells^41,42, and mouse brains⁴³. In addition, they perform different types of modification, such as excision and inversion, based on the direction of the recognition sites. Moreover, many recombinases and their recognition sites have been developed or discovered, each targeting their corresponding recognition sites with high efficiency and orthogonality, further expanding the pool of usable genetic tools⁴¹. The success of recombinase-based genetic circuits in various living organisms proves that recombinases can be easily adapted to different platforms. However, recombinase-based Boolean logic gates have yet to be applied to a cell-free system.

To expand the toolkit for genetic circuits, we developed a cell-free recombinase system (Fig. 1).

Figure 1.

A cell-free recombinase-based biocomputing platform in test tube/microfluidics allows for the construction of orthogonal, multiplex, and versatile devices. Such a platform has potential applications in multiplex environmental sensing, biological information storage, and disease diagnostics.

We demonstrated the ability to engineer complex circuits that can be easily implemented and function with less optimization. Although recombinase has been widely studied and well-characterized in a cellular system, transitioning from cellular hosts to cell-free systems is not simple. To apply the recombinase circuits to cell-free systems, we comprehensively characterized circuit designs and established design rules applicable to cell-free genetic circuit development.

Additionally, the irreversibility of recombinase excision can produce a DNA writer for biological memory recording. With its high fidelity, recombinase has been previously applied to cellular systems to develop DNA-based memory devices, such as BLADE and Recombinase State Machine (RSM)^40,41. These works illustrate that recombinase is an ideal candidate for implementing genetic circuits in a DNA-based memory storage device. Thus, we also establish the potential of cell-free recombinase to develop DNA-based memory storage and characterize the stability of memory and biological information stored in such circuits. Such cell-free recombinase-based biocomputing platform demonstrates the ability to engineer modular, multi-layered, and programmable platforms, with potential applications in environmental monitoring, biological information storage, and disease diagnostics.

Results

Design Characterization of Recombinases in a Cell-free Environment

We first characterized Cre recombinase excision circuits to establish the potential of recombinase in a cell-free reaction. We chose to use the Protein Synthesis Using Recombinant Elements (PURE) system because of its high purity, simple composition, easy manipulation, and robust performance in building genetic circuits. The excision reporter comprises a T7 promoter, a ribosome binding site (RBS), a terminator flanked by two Cre recognition loxP sites, and a mCherry reporter gene (Fig. 2A). For the starting reporter design, we used a combination of rrnB T1 and T7Te terminator (Term 1), a double terminator combination verified in a bacterial recombinase system in the previous study, to suppress reporter expression before recombination⁴⁶. In addition, the expected reporter plasmid post-excision, referred to as a LoxP-mCherry plasmid, was also generated. This serves as a positive control for the successful expression of the reporter plasmid after excision by Cre (Fig. 2A). With the expression and activity of Cre recombinase, we expect the expression of the mCherry reporter gene to go from OFF to ON, while the LoxP-mCherry should constitutively produce high reporter output. However, opposite results were observed. The reporter-alone condition showed a basal output signal that is a 27-fold change over the blank condition. In addition, an 82% and a 55% decrease in translational output were observed for reporter and LoxP-mCherry, respectively, when incubated with Cre-expressing plasmids. (Supplementary Fig. 1A). These results suggest multiple issues with the current circuit design. First, the high reporter basal expression indicates that the terminator between loxP sites was too weak and failed to terminate the T7 RNA polymerase activity. Second, the decreased reporter expression in the reporter+Cre condition indicates that Cre protein expression negatively impacts reporter expression, with the underlying mechanism remaining unclear.

Figure 2: Characterization of cell-free recombinase excision circuits.

(A) Excision reporter and LoxP-mCherry plasmids design. Before Cre excision, the terminator after the promoter suppresses transcription of the mCherry reporter gene. When Cre is expressed, the recombinase will excise the terminator and allow for transcription, thus triggering a mCherry fluorescent output. LoxP-mCherry plasmid, an expected reporter sequence post-excision, is cloned as a positive control for excision circuit testing. Blue triangle: loxP recognition site for Cre recombinase. T: transcription terminator.

(B) Reporter optimization part 1: terminator optimization to prevent leaky reporter expression (left). mCherry expression of the pre-excision reporter (right). Blank: cell-free reaction mix with water only.

(C) Reporter optimization part 2: LoxP to mutLoxP. comparison of loxP and mutLoxP sites on the reporter (left) and the corresponding mCherry output (right).

(D) Reporter optimization part 3: insert length between promoter and recognition sites. Normalized transcription output (NTO) as a function of insert length between promoter and recognition sites.

(E) NTO comparison of original excision reporter and modified reporter.

Data shown are the means of technical triplicate samples with error bars indicating +1 standard deviation. P-values were calculated as described in the methods.

Reporter Optimization I: optimizing terminator to suppress T7RNAP readthrough

We focused on reducing the basal mCherry expression by utilizing strong terminators to limit the T7RNAP readthrough. We chose the T7 terminator, an engineered terminator for the T7RNAP expected to prevent T7RNAP readthrough efficiently, as the first candidate (Term 2). However, the results show that the T7 terminator also failed to thoroughly terminate the reporter transcription pre-excision (Fig. 2B). Our result agrees with a previous study of synthetic terminator design, in which the T7 terminator only showed 80% termination efficiency⁴⁷. In addition, the previous research also screened an array of terminator designs and created a synthetic terminator (Term 3 - synTerm) that can prevent 99% T7RNAP readthrough. The synTerm is composed of a synthetic T7 terminator, rrnBT1 terminator, and T7 terminator (Fig. 2B)⁴⁷. Replacing the terminator in our original reporter design with the synTerm reduced reporter leaky expression by >78%. Adding one extra copy of synTerm in the reporter (Term 4 – 2XsynTerm) can further lower 94% of reporter basal expression to a level similar to the blank condition (Fig. 2B).

Reporter Optimization II: loxP sites vs. mutLoxP sites

We next investigated the mechanism of the decreased reporter expression in the reporter+Cre condition. We hypothesized that the Cre binds to the loxP on the reporter post-recombination, thus blocking the T7 transcription. One loxP site is left in the reporter plasmid after excision, thus serving as a repressive operator site for Cre to inhibit the promoter activity. To decrease the Cre binding to loxP sites post-excision, the loxP sites were replaced by the asymmetric, mutant loxP (mutLoxP) sites lox66/lox72. The lox66/72 pairs carry 5bp of mutation on the left or right end of the lox66 or lox72 sites, respectively, resulting in 10bp mutation on the recombined loxP sites post-excision (Fig. 2C)⁴⁸. We hypothesize that the 5bp sequence mutation isn’t strong enough to prevent Cre recombinase from binding and modifying the DNA but will result in a 10bp mutation post-excision, which can effectively prompt Cre recombinase to dissociate from the reporter plasmid. To test this hypothesis, we created a mutLoxP-mCherry plasmid by replacing the original loxP site with the mutLoxP site. Both LoxP-mCherry and mutLoxP-mCherry plasmids were then exposed to Cre recombinase. Implementing the mutated lox66/72 sites recovers over 39% of reporter expression in the LoxP-mCherry plasmids (Fig. 2C).

Reporter Optimization III: insert length between promoter and recombinase recognition site.

The success of recovering output signal with the mutLoxP site suggests potential interference between T7 RNA polymerase and Cre binding to DNA. We hypothesized that the insert length, or distance between the promoter and the loxP sites, could impact the excision circuit expression. Thus, we also varied insert lengths to elucidate their impact on circuit performance. Among all the insert lengths that were tested, 10bp, 20bp and 30bp all provide strong reporter expression, with 10bp provides the most significant enhancement of reporter expression under condition with Cre plasmid, and lowest background reporter expression under condition without Cre plasmid (Fig. 2D). We rationalize that it might be due to the competitive binding between the Cre recombinase and the T7RNAP – the two enzymes might interfere the other’s binding efficiency and activity if their binding sites are too close. However, ribosome binding sites should not be too far downstream of the T7 promoter since a longer sequence behind the transcription starting site will result in a longer non-coding mRNA sequence with a more complex secondary structure that can negatively affect translation efficiency. Thus, finding an optimal distance between promoter and recognition sites is critical, and 10bp is determined to be the insert length that can maximize the translation efficiency and T7RNAP activity in our system.

A cell-free system is supplemented with limited energy, enzymes, and substrates for transcription and translation. Thus, to better leverage the limited sources in the genetic circuits, we fine-tuned the ratio of reporter and recombinase plasmids for the excision circuits (Supplementary Fig. 1B). The dose-response curve of recombinase plasmids demonstrates that while a sufficient amount of recombinase plasmids is needed to trigger reporter signals efficiently, high-level of recombinase plasmids will also lead to decrease in reporter fluorescence. We hypothesize that this is due to the rapid depletion of energy and resources for recombinase expression, leaving insufficient resources for reporter fluorescence expression post recombinase excision.

With the characterization of the terminator sequence, recognition sites, insert length, and plasmid dosage, we present a cell-free Cre excision circuit that starts with a 0.2-fold induction, eventually optimized to 41-fold induction, resulting in more than 200X improvement of the excision reporter output signal (Fig. 2E and Supplementary Fig. 1C). This systematic optimization approach was also applied to excision circuits of five other recombinases in the cell-free environment. These recombinases can trigger > 20-fold change of reporter output signal with high efficiency and orthogonality in cell-free systems (Supplementary Fig. 2).

Environmental Inputs for Cell-free Recombinase Genetic Circuits

Next, we interfaced the cell-free recombinase genetic circuits with biochemically relevant inputs using allosteric transcription factors (aTFs) to regulate T7RNAP activity.

Allosteric transcription factors (aTFs) can respond to various stimuli and control transcription⁴⁹. Many aTFs have been discovered and engineered in bacteria to sense a wide range of small molecules and environmental signals, including antibiotics^50,51, heavy metals^52–56, metabolites^57–60, aromatic compounds^61,62, and changes in pH and osmolarity⁶³. Their extensive range of potential sensing chemicals has positioned them as invaluable assets in fundamental research, protein biomanufacturing, and environmental sensing across diverse cellular contexts. Building upon these advantageous features and leveraging insights from previous studies on aTFs, we integrated aTFs into the CRIBOS system, building a sensing platform compatible with recombinase-based Boolean circuits to broaden the potential applications of the CRIBOS system.

To control the transcription of recombinase, the cognate operator sequence is placed after the promoter. We reason that the binding of aTFs to the operator will prevent T7RNAP from accessing the promoter. In contrast, ligand binding will result in a conformation change of the aTFs, causing their release from the operator and thus triggering transcription of recombinase and the downstream excision circuit (Fig. 3A).

Figure 3. Characterization of aTFs-implemented recombinase genetic circuits in a cell-free condition.

(A) Mechanism of aTFs-controlled recombinase genetic circuits. aTFs can bind to transcription operators behind T7RNAP and suppress transcription of the recombinase gene. Corresponding inducers bind to aTFs to release them from the operators, thus triggering recombinase expression and genetic circuit activation.

(B) Characterization of tetR and smtB aTFs.

(C) Cre, PhiC, and Bxb1 excision circuits controlled by tetR (left) and smtB (right) aTFs are tested. To set up the experiment, aTFs were produced, purified from bacteria, and added to the PURE reaction with genetic circuit plasmids with and without chemicals.

(D) Chemical detection range is determined for each aTFs-implemented recombinase genetic circuit under optimal aTFs for each recombinase. To set up the experiment, 12.5 μM tetR protein is used for the Cre excision circuit, 1.25 μM tetR protein is used for PhiC and Bxb1 excision circuits, 10 μM smtB protein is used for Cre, PhiC, and Bxb1 excision circuits.

Data shown are the means of technical triplicate samples with error bars indicating +1 standard deviation. P-values were calculated as described in the methods.

We chose a group of aTFs that consists of tetR, SmtB, HucR, QacR, OtrR, Ctcs and TtgR. These aTFs were previously tested and characterized in an in vitro transcription reaction, whose reaction environment is similar to that of a cell-free system²². To verify that the aTFs-operator interaction can efficiently control the expression of a mCherry reporter, an operator was inserted after the T7 promoter, following the design characterized in the previous study⁶⁶. Within the seven pairs of aTF-operator that function in the cell-free environment, we selected well-studied tetR and smtB, and their corresponding ligands tetracycline and zinc ions, due to their highest yield and best performance (Fig. 3B and Supplementary Fig. 3). Their corresponding cognate operators were then inserted into the Cre, PhiC, and Bxb1 plasmids to control recombinases expression (Fig. 3C). Dose-response curves for aTFs were performed to determine optimal aTF concentrations (Supplementary Fig. 4A).

After the optimal aTFs concentration was determined, a dose-response curve for small molecule inducers was performed to determine the detection window of the recombinase-based sensing circuits (Fig. 3D). It is essential to consider that the detection window for inducers varies among recombinase genetic circuits due to differences in their activities. For instance, stronger recombinases such as Cre necessitate higher concentrations of aTFs to fully deactivate circuit activity. As a result, these circuits exhibit a detection window that is shifted to the right compared to weaker recombinases like PhiC and Bxb1, detecting higher dosages of inducers. The observed shift in the detection window can be attributed to variations in aTFs concentration requirements, as different recombinase circuits demand distinct concentrations of aTFs. This diversity in recombinase activities allows us to broaden the detection window for inducers, as illustrated in Fig. 3D.

Furthermore, we assessed the limit of detection (LoD) for both zinc and tetracycline-induced sensors to determine the input dose that achieves detection above negative controls (Supplementary Fig. 4B). The fold change was calculated to determine the maximum change in reporter output. We found the limit of detection (LoD) for zinc-induced recombinase sensing was 1.17 μM for Cre, 2.69 μM for PhiC, and 4.45 μM for Bxb1, with fold changes of 20.7x, 8.14x, and 12.8x, respectively. These values indicate that for zinc-induced sensing, Cre achieves the highest fold change in reporter response and detection at the lowest dose of zinc input. For tetracycline-induced sensing, the detection window was between 1.08 – 59.0 μM for Cre, 0.254 – 19.6 μM for PhiC, and 0.366 – 22.5 μM for Bxb1, with fold changes of 14.0x, 14.6x, and 5.76x, respectively. These results show that while PhiC has an initial lower limit of detection, Cre has the broadest range in detectable tetracycline. Moreover, it was also observed that tetracycline at high dosages negatively affected the reporter gene expression (>12.5 μM). This is most likely due to tetracycline’s inhibition of bacterial protein synthesis by preventing the association of aminoacyl-tRNA with the bacterial ribosome⁶⁷.

We plotted the Receiver Operating Characteristic (ROC) curves and calculated Area Under the Curve (AUC) to quantify assay performance across sensitivity and specificity classification thresholds (Supplementary Fig. 4C). For zinc-induced sensing, we found AUC values of 0.907 for Cre, 0.833 for PhiC, and 1 for Bxb1. For tetracycline-induced sensing, the ROC curve analysis showed AUC values of 0.857 for Cre, 0.730 for PhiC, and 0.841 for Bxb1. These results demonstrate the promising ability of the assay to distinguish both true positive and true negative results.

2-Input-2-Output Logic Gates

The high orthogonality of recombinase and recognition sequences allows for more complex genetic circuits with multiple inputs and outputs to be developed under cell-free conditions, such as the 2-input, 2-output Boolean logic gates (Fig. 4A). We created twelve 2-input Boolean logic gates by placing Cre and PhiC recombination sites, lox66/72, PhiC attB/attP, respectively, by either side of terminator sequences and a mCherry reporter gene (Supplementary Fig. 5). For each 2-input-2-output logic gate, there are four possible input combinations: (0,0), (1,0), (0,1), and (1,1), where ‘1’ represents the presence of an input and ‘0’ indicates its absence. The outputs are also binary, with ‘1’ signifying the presence of an output signal and ‘0’ indicating its absence. In this experiment, the output signal corresponds to mCherry expression. Using the AND gate as an example, mCherry expression is expected only when both Cre (input A) and PhiC (input B) are present. This means that the AND gate will produce an output of ‘1’ (high mCherry fluorescence) only for the (1,1) input condition. For all other input conditions ((0,0), (1,0), and (0,1)), the AND gate will yield an output of ‘0’, resulting in no mCherry fluorescence expression.

Figure 4. 2-input-2-output circuits for two chemical sensing in the cell-free condition.

(A) Schematic for the mechanism of the aTF-implemented 2-input-2-output circuit. Cre is controlled by the tetR system, while PhiC excision circuits are controlled by the smtB system.

(B) Results of 2-input-2-output circuits for detecting tetracycline and zinc. Cre is controlled by tetR, while PhiC is controlled by smtB. To set up the experiment, recombinase plasmids, purified aTFs, water, and sensing molecules are added to the PURE reaction mix. After 30 minutes of incubation at 37°C, reporter plasmids were added to the reaction.

Data shown are the means of technical triplicate or duplicate samples with error bars indicating +1 standard deviation. P-values were calculated as described in the methods.

In the original experimental design, the reporters and recombinase plasmids were premixed and added to the cell-free reaction before incubation. With this design, some circuits do not behave as expected in the logic table, which is mainly due to the delayed shutoff of the reporter fluorescent expression by PhiC recombinase (Supplementary Fig. 5). We reason that this is because of the weak activity of the PhiC recombinase. While Cre is very efficient and active and thus can promptly shut off the reporter expression, the PhiC is much weaker and takes longer to excise the reporter gene. During this time, the reporter expresses and accumulates in the reaction simultaneously, leading to a leaky output signal for the genetic circuits. Therefore, we hypothesize that rapid recombination by PhiC is the key to decreasing the leaky signals, and pre-expression of the recombinases might help to achieve this goal.

In the subsequent experimental design, instead of adding recombinase and reporter plasmids simultaneously to the cell-free reaction, we first pre-incubated Cre and PhiC recombinase plasmids in the cell-free reaction for 30 minutes. Then, we added the reporter plasmids to finish the circuit setup. The 2-input-2-output logic gates yielded intended behaviors (Fig. 4B and Supplementary Fig. 6). This result suggests that timing is also a critical factor in developing and optimizing genetic circuits, and a cell-free platform provides higher control and flexibility of timing over every step of circuits’ execution.

The performance of the circuits was evaluated by a vector proximity (VP) metric measuring the discrepancy between the experimental output and the expected logic table. The expected output indicated by the logic table and the experimental measurements were represented by a truth table vector and a signal vector, respectively, in a four-dimensional space⁴¹ (Supplementary Fig. 7A). The angular difference between the two vectors was measured: a 0° angle indicates that the experimental results represent the intended logic table perfectly, and a 90° angle indicates that the experimental output demonstrates completely wrong output. Among the 11 circuits tested, only 5 circuits (36%) showed a VP angle that was smaller than 15° using the original experimental procedure (Supplementary Fig. 7B). With the modified procedure, the circuits’ performance improved, and the VP angle of all circuits systematically decreased, resulting in 6 of 8 circuits (80%) with VP angle smaller than 15°. Additionally, 2-input-2-output circuits with biosensors yield 7 over 8 circuits (88%) with a VP angle smaller than 15°.

A 2-Input-4-Output BLADE Circuit in Cell-free

The Wong lab has previously developed a 2-input-4-ouptut ‘Boolean logic and arithmetic through DNA excision’ (BLADE) in mammalian cells to exploit the features of site-specific recombinases to enable N-input-M-output combinatorial computation⁴¹. With two inputs, the circuit requires four different outputs, Z = Z_AB = Z₀₀, Z₁₀, Z₀₁, and Z₁₁, to represent 2^N = 2² = 4 (N indicates input number) different states of inputs A and B (Fig. 5A). To investigate the potential of building N-input-M-output in a cell-free environment, we implemented the 2-input BLADE circuits with terminators, fluorescent, and luminescent genes. One key element in the BLADE circuit is the heterospecific site, such as the Cre recombinase’s lox66/72 and lox2272. Although they are only a few base pairs different, lox66/72 and lox2272 can only retain Cre excision capability when paired with their corresponding recognition sites, which means lox2272 can only be paired with lox2272 to trigger Cre recombination activity while lox66 can only be paired with lox72.

Figure 5. 2-input-4-output Decoder.

(A) Schematic for the mechanism of the 2-input decoder. The reporter plasmid expression cassette is separated into four addresses by recombinase recognition sites, resulting in 4 different outputs based on four states of two inputs.

(B) Results of 2-input-4-output decoder for detecting tetracycline and zinc. Cre is controlled by the tetR system, while PhiC is controlled by the smtB system. To set up the experiment, recombinase plasmids, aTFs, water, and sensing molecules are mixed and added to the PURE reaction mix. After 30 minutes of incubation at 37°C, reporter plasmids were added to the reaction.

Data shown are the means of technical duplicate samples with error bars indicating +1 standard deviation. P-values were calculated as described in the methods.

In the initial design of the 2-input-4-output reporter, a low fluorescence signal was observed for GFP and mCherry (Supplementary Fig. 8A). Based on the previous experience of reporter characterization, we hypothesized that this is due to the insert sequences positioning in front of the Z₀₀ addresses. In contrast to the simple excision circuit, the BLADE reporter features increased complexity and more diverse recombinase recognition sites, resulting in a longer insert sequence between the promoter and the expressing gene. It’s worth noting that, in addition to the PhiC attB site and loxP2272 site in the reporter design, an extra 40bp of U1 sequence is present to guide the Gibson reaction during molecular cloning, further elongating the insert in front of the expressing gene. Consequently, we postulated that such an extended sequence in front of the expressing cassette might adversely impact the expression efficiency of the output signals.

To test this hypothesis, three designs of inserts were created and positioned in front of a constitutive mCherry gene (Supplementary Fig. 8B). Design 1 is the insert sequence in the original reporter, Design 2 is made by removing the U1 sequence from Design1 to investigate the effects of U1 on gene expression, and Design 3 is the insert sequence taken from the excision reporter, which was proved to show the high fluorescence output signal in the excision circuit (Supplementary Fig. 8A). Interestingly, Design 2 successfully leads to a significant 18-fold increase in output expression by eliminating the 40bp U1 and substantially shortening the distance between the promoter and the expressing gene. Moreover, Design 3 also demonstrated a 2-fold higher output expression than Design 2, even with only a 1bp difference in their length (Supplementary Fig. 8B). This suggests that, in addition to insert length, the sequence design itself may also influence gene expression. One hypothesis of such observation is that mRNA secondary structure might play a vital role in determining the expression efficiency of the reporter gene.

With the promising data shown, we replaced the insert in the BLADE reporter with the Design 3 insert to develop a BLADE reporter version 2 (Supplementary Fig. 8C). The experimental data showed that the fluorescence signal largely increased with the optimized reporter version 2 (Supplementary Fig. 8C–D). However, leaky expression is observed for all three addresses, indicating that the terminator sequence at the end of each address is too weak. Both version 1 and 2 BLADE reporter designs employed the T7 terminator at the end of each address to simplify the molecular cloning process, as the synTerm posed challenges for PCR and Gibson Assembly due to its repetitive sequences. Unfortunately, the T7 terminator proved inadequate for the BLADE circuit, leading to leaky expression.

To address the leaky expression issue, we replaced all T7 terminators with synTerm, creating BLADE reporter version 3. This modification proved successful, providing the optimal performance for the BLADE circuit (Supplementary Fig. 8E). The VP angle metric for the BLADE circuits was evaluated to be less than 25% for all fluorescent and luminescent outputs (Supplementary FIG. 7B). Furthermore, the BLADE circuits, when combined with aTFs-based biosensors, can be used for environmental surveillance for small molecules and heavy metal ions such as tetracycline and zinc (Fig. 5B). The VP angle metric for the BLADE circuits was evaluated to be less than 25%, for all fluorescent and luminescent outputs (Supplementary FIG. 7B).

Cell-Free Biological Memory Storage with CRIBOS

The capability to record crucial biological memory shapes research and extends to diverse applications in disease diagnostics and medical therapeutics^68–72. For example, immunological memory enables the development of innovative therapeutic interventions such as vaccines and immunotherapy^73,74. Additionally, DNA barcoding plays a pivotal role in detecting invasive organisms and uncovering novel species^75–77. These applications underscore the far-reaching impact of biological memory storage in advancing various fields of science and technology.

DNA has been used to transmit genetic information to the next generations because of its high stability, storage capacity, small volume, and inheritance. Moreover, employing DNA as the substrate for memory storage facilitates direct interfacing with sensors capable of detecting significant biological and environmental signals. This integration allows for the activation of biological programs in response to various environmental changes and bridges the gap between sensing and biological responses. Site-specific recombinase permanently modifies DNA sequence with high fidelity and efficiency, serving as a powerful writer to encode biological information into the DNA-based memory storage medium. Thus, with the success of building recombinase-based Boolean logic gates in a cell-free environment, we created a portable, robust, and stable biological memory storage device with the CRIBOS platform.

To evaluate the portability potential, we apply CRIBOS in a paper-based setting. One benefit of the PURE system is that it can be lyophilized on a sterile and abiotic paper-based platform. The lyophilized paper-based cell-free reaction is stable at room temperature for over a year and can be activated through rehydration^78,79. In contrast, liquid-based reactions must be stored at −80°C to maintain enzyme activity.

In addition to device portability, memory stability is critical for a memory storage device. DNA memory can not only be effectively modified on paper through recombinase circuits but can also be stably preserved under room temperature for over four months upon activation (Fig. 6A and Fig. 6B). Paper-based CRIBOS with a plasmid excision circuit reporter DNA and linear recombinase expressing gene was activated on day 0 through rehydration, and information was stored on the paper. At different time points post-activation, memory stored in the paper-based device was retrieved and analyzed. The reporter DNA with biological memory was retrieved through elution, and then collected DNA was used to transform bacteria to express mCherry reporter output (Fig. 6A). The recombinase excision rate was evaluated based on the percentage of mCherry+ bacteria colonies. High excision efficiency in the reporter+Cre condition indicates that the information was well-preserved at room temperature, and no DNA mutation is observed based on the sequencing results (Fig. 6B and Supplementary Fig. 9).

Figure 6. Paper-based recombinase circuits and memory recording platform.

(A) Schematic of the experimental process for paper-based cell-free recombinase setup. Reporter plasmid and linear recombinase DNA are mixed with PURE reaction mix and then freeze-dried onto a paper disc for stable storage under room temperature. Then, the reactions are activated through rehydration. Modified DNA-based memory was stored on the paper disc for different lengths of duration, ranging from 1 day to 4 months. On the day of analysis, DNA-based memory is retrieved from a paper disc and transformed into bacteria for signal output expression.

(B) Memory stability evaluation for paper-based memory storage devices. Paper-based CRIBOS was activated through rehydration on day 0. DNA-based memory was collected on different days up until four months.

Data shown are the means of technical triplicate samples with error bars indicating +1 standard deviation. P-values were calculated as described in the methods.

CRIBOS memory stability was also measured in 2-input-2-output (Fig. 7A) and the BLADE circuit (Fig. 7B). For the 2-input-2-output circuits, plasmid DNA of NOR gate or AND gate was lyophilized on paper discs along with linear Cre and linear PhiC expressing genes. The circuits were activated on day 0 through rehydration, and information was stored on the paper. At Day 7, 14, 21 and 28 post-activation, memory stored on the paper-based device was retrieved and analyzed with the same method as the paper-based Cre excision circuit. The recombinase excision rate of the NOR and AND gate indicates that the 2-input-2-output circuits can perform well in a paper-based environment and the information was well-preserved under room temperature for at least a month (Fig. 7A).

Figure 7. Paper-based multi-input multi-output circuits.

(A) Memory stability evaluation for paper-based 2-input-2-output memory storage devices. Paper-based CRIBOS was activated through rehydration on day 0. DNA-based memory was collected on 1, 7, 14 and 28 days after activation.

(B) Memory stability evaluation for paper-based 2-input-4-output memory storage devices. Paper-based CRIBOS was activated through rehydration on day 0. DNA-based memory was collected on 1, 7, 14 and 28 days after activation.

Data shown are the means of technical triplicate samples with error bars indicating +1 standard deviation. P-values were calculated as described in the methods.

The BLADE reporter was lyophilized on paper discs along with linear Cre and PhiC expressing genes for the paper-based BLADE circuit. The circuit was activated on day 0 through rehydration, and information was stored on the paper. At Day 7, 14, 21, and 28 post-activation, memory stored on the paper-based device was retrieved and transformed into BL21(DE3) bacteria for fluorescent and luminescent expression. The Normalized Translation Output (NTO) of transformed bacteria demonstrates that the BLADE circuit can perform well in a paper-based environment and the information was well-preserved under room temperature for at least a month (Fig. 7B).

Discussion

CRIBOS is a powerful recombinase-based, multiplex genetic circuit platform for cell-free biosensing, combining transcription factor-based sensing with recombinase-mediated signal processing. CRIBOS’s unique strength derives from its modularity, memory storage capability, and complex logic processing. Our results show that six different recombinases can function in cell-free conditions with high activity, with over a 20-fold difference between with and without recombinases. The BLADE circuit previously developed in mammalian cells has been successfully refactored for the cell-free environment. By implementing aTFs-based small molecule sensors, the BLADE circuit can be induced by multiple chemical inducers in parallel. Besides aTFs-based sensors, recombinase genetic circuits are compatible with chemical-induced dimerization (CID)-based sensors⁸⁴, enabling CRIBOS to detect changes in temperature, fluorescence, luminescence, and the presence of a broader range of chemicals in the future. Furthermore, recombinase-based logic allows for the irreversible recording of transient biological events and the detection of sequences of events.

These features from CRIBOS enable applications that span multiple fields. They are particularly valuable for long-term environmental monitoring, disease diagnostics, and quality control in biomanufacturing. For instance, in environmental surveillance, it can be used to detect contaminants such as heavy metals, antibiotics, and toxins in water sources. In medical diagnostics, CRIBOS offers a promising tool for recording biomolecular markers, aiding in infection detection and immune response monitoring. In biomanufacturing, the system can be applied for real-time monitoring of metabolite levels or process conditions, ensuring stable and efficient bioproduction. Furthermore, in synthetic biology research, CRIBOS provides a versatile platform for designing programmable biological circuits, which can be used for controlled gene expression, biosensing, and bio-recording applications. Additionally, unlike conventional biosensing methods that rely on live cells, CRIBOS can be freeze-dried on paper discs, significantly increasing its shelf life and making it more practical for point-of-care diagnostics and on-site environmental monitoring without cold storage.

Through rigorous characterization, we discovered multiple design rules for recombinase-based Boolean logic gates in the cell-free environment. First, due to the strong activity of the T7 promoter, a comparably strong terminator is required to suppress the leaky recombinase expression of recombinases and reduce background fluorescence. Second, the binding affinity between the recombinase and its recognition site post-excision also significantly impacts circuit performance. The recognition sites must be carefully calibrated to ensure that recombinases properly disengage post-recombination, allowing the reporter expression. Third, the insert length between the promoter and the recombinase recognition sites affects the genetic circuit performance, with a 10bp insert providing the optimal balance for effective recombination and robust output expression. These insights will aid in the rational design of future cell-free genetic circuits.

The challenges encountered during the optimization and troubleshooting process illustrate that applying genetic circuits developed in cells to cell-free is not trivial. Our design rules, therefore, will facilitate future recombinase circuit developments. Although cell-free is a natural environment for circuit development and can provide benefits and convenience that cellular platforms and animal models cannot provide, it poses challenges and issues not present in other platforms. For instance, while living organisms can offer genetic circuits with continuous and durable energy sources and building blocks, a cell-free environment has limited means to regenerate high-energy compounds and starting materials needed for protein synthesis and circuit components. This illustrates that proper resource allocation is critical for cell-free operations. Also, cell-free systems are more diluted than the cell lysate, not just in the concentration of the transcription and translation machinery but also in the degree of macromolecular crowding^80–83. Such differences in spatial density can influence the biochemical reaction rates and equilibria, changing the efficiency and competitive effect of protein-protein and protein-DNA binding. The biochemical characteristics of recombinase-recognition sites binding might also be affected in the cell-free reaction, as suggested by the complete shutoff of the loxP reporter signal by Cre recombinase due to the high Cre-loxP binding affinity post recombination. While lox66/72 sites were previously used for fixing the directionality of the recombination, few studies have reported its functions in improving circuit output signals in cells or animal models because the native Cre-loxP system can function properly and efficiently in those platforms, and there is no need to replace the loxP sites into lox66/72 sites. These discrepancies between cell and cell-free systems will be critical considerations for building complex genetic circuits and expanding the cell-free bio-computation toolkits.

While CRIBOS exhibits strong potential for multiplexing with multiple recombinases and aTFs, several critical factors must be addressed when scaling up. Expanding the multiplexing capacity relies on the availability of additional orthogonal recombinases and aTFs with minimal crosstalk. Although previous studies have successfully expanded the recombinase toolkit to encompass dozens of orthogonal recombinases, their functionality in cell-free environments remains relatively underexplored^40,85. However, our study shows that a standardized characterization procedure can be applied across different recombinases, enabling their efficient function in the cell-free system. This finding suggests that integrating additional recombinases into CRIBOS is a manageable challenge, overcoming a key limitation often encountered in other genetic circuit designs. Similarly, although various aTFs and biosensors have been discovered or engineered, their performance in cell-free conditions remains unknown^52,54,56. Thus, expanding the space of usable biosensors in CRIBOS also requires extensive characterization. Moreover, increasing the number of inputs and outputs also raises concerns about circuit complexity and resource competition, as cell-free systems have limited energy regeneration and translation capacities and, therefore, require additional validation to ensure compatibility and functionality^86,87. Despite these challenges, the modularity of CRIBOS provides a promising foundation for systematically integrating additional recombinases and sensors, enabling the future construction of higher-order circuits capable of more advanced logic and sensing tasks.

Lastly, we demonstrated that CRIBOS can be applied in a paper-based setting to improve the portability and stability of the genetic circuit platform. Paper-based CRIBOS is the world’s first genetic circuit with memory lasting over four months with minimal resource, energy cost, and maintenance required. A method to encode, store, and retrieve memory from the paper-based CRIBOS was developed and tested at room temperature, demonstrating its potential for broader field applications, such as continuous environmental monitoring and pollution detection. While the current method for memory retrieval is cost- and resource-effective, more sensitive and rapid methods for detection out in the field can also be explored in future studies by combining with next-generation sequencing and genetic engineering tools. CRIBOS memory devices can also be tested and characterized under more diverse reaction conditions, such as under extremely high or low temperatures. In addition, although we only applied CRIBOS in a paper-based setting, cell-free reactions can function in various settings, such as microfluidics and flow strips^88–91. Thus, in the future, we can combine CRIBOS with microfluidics devices to further generate a portable memory storage device that allows for continuous and multiplex environmental sensing under a low-resource setting.

Methods

Strains and Growth Medium.

NEB 5-alpha Competent E. coli (New England Biolabs, C2987U) was used for routine molecular cloning. E. coli strain Rosetta 2 (DE3) pLysS (Novagen, 71401) was used for recombinant protein expression. BL21(DE3) Competent E. coli (New England Biolabs, C2527H) was used for recombinase circuit testing. LB broth (Thermo Fisher Scientific, BP1426–2) supplemented with appropriate antibiotic(s) (100 μg/ml carbenicillin (Thermo Fisher Scientific, 10177012), 50 μg/ml kanamycin (Bio Basic, KB0286–25) and/or 25 μg/ml chloramphenicol (Sigma-Aldrich, C0378–25G)) was used as a bacterial growth medium. Bacterial transformations were performed based on the manufacturer’s protocols.

Bacterial assay for paper-based BLADE circuit.

DNA from paper-based BLADE circuits were eluted from the paper disc using 10uL elution buffer. Then 2.5uL of eluted DNA was transformed to 50uL of BL21(DE3) Competent E. coli (New England Biolabs, C2527H) based on manufacturer’s protocol of Mix & Go! E.coli Transformation Kit (Zymo Research, T3002). After transformation, 50uL of pure LB was added to the transformed bacteria and then cultured in 37°C shaking incubator. After 1hr recovery, 25uL of transformed bacteria was cultured in 200uL with carbenicillin overnight. Next day, the overnight bacterial culture was diluted 1:100. After 2hr shaking incubation at 37°C, the bacteria was induced with 100 μM IPTG. After 3hr of induction, mCherry, GFP and NanoLuc expression was evaluated using a Plate Reader (SpectraMax M5, Molecular Devices).

Molecular Cloning.

DNA oligonucleotides for cloning and sequencing were synthesized by ThermoFisher Scientific. Genes encoding fluorescent, luminescent proteins and 2X terminators were synthesized as gBlocks (Integrated DNA Technologies). All constructs were created using standard restriction enzyme digestion, Gibson isothermal assembly, and Unique Nucleotide Sequence (UNS) Guided assembly procedures into a pET15b plasmid. UNS Guided assembly utilizes a mechanism similar to Gibson’s isothermal assembly but with standard homology sequences that were computationally optimized. Constructed plasmids were transformed and maintained in NEB 5-alpha (New England Biolabs (NEB) C2987H) Escherichia coli competent cells at 37°C overnight before miniprep (Epoch Life Sciences, 2160250).

Protein Production and Purification.

All aTFs expression plasmids were purchased from Addgene of Jung JK, Alam KK, Verosloff MS, et al²². For aTFs production, plasmids were transformed into Rosetta 2 (DE3) pLysS E.coli strain. A single colony of transformed bacteria was cultured in liquid LB media overnight. The next day, overnight cultures were diluted 1:100 into 200mL-500mL culture media. OD of culture media was measured by a plate reader every 30 minutes. When OD reached 0.5–0.8, culture was induced with 100 μM Isopropyl β-D-thiogalactoside (IPTG) (Sigma-Aldrich, I6758–5G) overnight at 30°C. After overnight IPTG induction, bacterial culture was pelleted through a centrifuge at 3500Xg for 20 minutes. Every 1g of bacterial pellet is resuspended in 10ml SoluLyse lysis buffer (Amsbio, L200500) supplemented with 1 tablet of protease inhibitor cocktail (Sigma-Aldrich, 11836170001), 40mM imidazole (Fisher Scientific, O3196–500), 200uL Lysozyme (Sigma-Aldrich, L4919–500MG) and 0.5uL Benzonase (EMD Millipore, 71205–3). Lysozyme was pre-resuspended to 10mg/mL in water and stored at −80°C upon arrival. The resuspended cell pellet was placed on a mixer for 20 minutes and centrifuged at 4000g for 10 minutes at 4°C to remove bacterial debris post-incubation. The bacterial supernatant was then sterilely filtered with a 0.22μm filter (EMD Millipore, SCGP00525) to remove bacterial precipitate and avoid clogging in FPLC columns.

Clarified supernatants were purified using His-tag affinity chromatography with a HisTrap column (Cytiva, 17–0406-01), followed by Heparin chromatography with a heparin column (Cytiva, 17–5247-01) using an AKTAxpress fast protein liquid chromatography system. The fractions collected from the FPLC were buffer-exchanged and concentrated utilizing a protein concentrator (Thermo Scientific, 88515). Protein concentrations were evaluated using Nanodrop. The purity and size of proteins were validated with SDS-PAGE and western blot. Purified proteins were stored in PBS at 4°C.

In Vitro Protein Synthesis.

NEB PURExpress In Vitro Protein Synthesis Kit (New England Biolabs, E6800S) supplemented with RNase Inhibitor (Sigma-Aldrich, 3335402001) was used to generate data for cell-free experiments based on the manufacturer’s protocol. The concentrations of plasmids and linear DNAs are listed in the Supplement. The plasmids, DNAs, aTFs, and small molecules are first premixed before combining with the cell-free reaction mix. Then, 5μL of cell-free circuits were incubated in each well of a 384-well clear bottom black plate (Corning, 3542) at 37°C for 5hr. An endpoint measurement, or dynamic measurement, was performed on a plate reader (SpectraMax M5, Molecular Devices) with an excitation wavelength of 580nm and emission wavelength of 611nm for mCherry, an excitation wavelength of 485nm and emission wavelength of 525nm for GFP, and an excitation wavelength of 381nm and emission wavelength of 445nm for BFP. Nano-Glo Luciferase Assay (Promega, N1110) was used for NLuc luminescence detection. Nano-Glo Luciferase Assay Substrate was diluted 1:50 by Nano-Flo Luciferase Assay Buffer. The luminescence level was measured immediately after a volume of the diluted substrate equal to 6X the volume of the cell-free reaction was added to each well.

A linear DNA template with a T7 promoter, followed by a recombinase-expressing gene and a T7 terminator, was PCR (CoWin Biosciences CW2965F) amplified from the corresponding recombinase plasmids using primers listed in the Supplement. Amplified DNA was purified with a PCR cleanup kit (Epoch Life Science 2360250), and band size and sequence were verified through gel electrophoresis and sequencing.

Concentrations of plasmids DNA, linear DNA, purified proteins and chemical inducers used in the cell-free reaction are listed in the Supplementary Tables 1–32.

Freeze-drying.

Filter paper (Cytiva, 1442–070) for depositing and freeze-drying cell-free reactions was first treated with 5% bovine serum albumin (Sigma-Aldrich, A3059–100G) overnight at 37°C. The paper was then cut into 2-mm-diameter paper discs with a biopsy punch and placed at the bottom of a 96-well PCR plate. Before lyophilization, 2μL of cell-free reaction mix was applied directly onto the paper disc, and the PCR plate was sealed by porous AeraSeal Sealing Films (Excel Scientific, B-100). Post-freezing by liquid nitrogen, the PCR plate was immediately sent for lyophilization overnight. After lyophilization, the paper discs were inserted into the bottom of the 384-well plate and activated by adding 2μL of autoclaved DI water.

Paper-based recombinase circuits.

Paper discs with lyophilized CRIBOS system were placed into the wells of a 384-well clear-bottom black plate (Corning, 3542). On Day 0, 2 μL of pre-warmed autoclaved DI water was added directly to the paper disc in each well to activate the cell-free reaction. The 384-well plate was then kept at room temperature with the plate lid covered. At different time points post reaction activation, DNA stored on the paper discs was eluted using 5 μL of pre-warmed miniprep elution buffer (Epoch Life Sciences, 2160250). The eluted DNA was subsequently diluted 1:2 with an equal volume of elution buffer. To retrieve biological information, 2.5 μL of the diluted DNA was used to transform 50 μL of bacteria, following the transformation methods described above.

Plate Reader Quantification and MEF Standardization.

A Texas Red Dye (Invitrogen, D1828) was used to convert arbitrary mCherry fluorescence intensity into nanomolar equivalent Texas Red Dye (labeled as normalized translational expression). A serial dilution was prepared with PBS from a 500 μM stock solution. The samples were prepared with triplicates, and fluorescence values were measured at an excitation wavelength of 580nm and emission wavelength of 611nm on a plate reader (SpectraMax M5, Molecular Devices). Fluorescence for a concentration in which a single replicate saturated at the plate reader was excluded from the analysis. The rest of the measurements were averaged and formed a linear regression line, which is used to convert the measured arbitrary fluorescence value to the concentration of the Texas Red Dye (Supplementary Fig. 7). Standard curves were created with the same process for each plate reader for data collection to normalize the results. The same process was used to convert GFP fluorescence intensity into nanomolar equivalent Fluorescein (Sigma-Aldrich, F6377) and NanoLuc luminescence intensity into nanomolar equivalent NanoLuc purified protein (Promega, E499A). All fluorescence and luminescence outputs are normalized to the nanomolar equivalent of corresponding dye or proteins and then present on the figure as “Normalized Translational Output (NTR).”

Diagnostic Analysis.

For each inducer and recombinase combination, the limit of blank (LoB) was calculated as the mean of the blank added to three times the standard deviation of the blank. The limit of detection (LoD) was calculated as the LoB plus three times the standard deviation of low concentration sample. The dynamic range was computed as the mean expression of highest output for tetracycline and zinc positive samples over the mean expression of negative samples. False positive rate and true positive rate were used to construct the receiver operating characteristic (ROC) curve and compute the area under the curve (AUC).

Vector Proximity Computational Analysis.

Each 2-input-2-output or 2-input-4-output Boolean function corresponds to a logic table with four rows and one output column or four rows and four output columns, correspondingly, the output column in each Boolean function was mapped to a 4-dimensional vector a, called the “truth table vector”. The fluorescent reporter output signal of each genetic circuit measured from experimental implementation was also mapped to a 4-dimensional vector b, called the “signal vector.” We evaluated the accuracy of a genetic circuit by computing the angle θ between the vector a and b using the formula:

$$\mathit{\theta }={\mathit{c}\mathit{o}\mathit{s}}^{-1}\left(\frac{\mathit{a}·\mathit{b}}{\left|\mathit{a}\right|\left|\mathit{b}\right|}\right)$$

When computing θ, we capped the signal values in b to a maximum of 400nM Texas Red, 60nM Fluorescein, and 400nM NanoLuc purified protein. The angular difference ranges from 0° (best) to 90° (worst).

Statistical Analysis.

All cell-free in vitro experiments involved setting up PURE in vitro protein synthesis expression of DNA into n = 2 or 3 reactions. Fluorescence and luminescence intensities for each reaction were averaged, and the standard deviation was taken. To evaluate statistical significance, data between groups were evaluated with unpaired two-tailed t-tests. P values are reported (not significant = p > 0.05, * = 0.01 < p < 0.05, ** = 0.001 < p < 0.01, ∗∗∗ = 0.01 < p < 0.001, **** = 0.001 < p < 0.0001).

Resource Availability

Lead Contact:

Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Wilson Wong (wilwong@bu.edu).

Materials Availability:

Plasmids generated in this study are available from the lead contact upon reasonable request.

Data and Code Availability:

• Source data statement: data have been deposited at Zenodo and are publicly available as of the date of publication. Accession numbers are listed in the key resources table.

• Code statement: This paper does not report code.

• Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

Supplementary Material

SUPPLEMENTAL INFORMATION. Document S1. Figures S1–S10 and Table S1–S32

Abbreviation:

CRIBOS

Cell-Free Recombinase Integrated Boolean Output System

PURE

Protein Synthesis Using Recombinant Elements

BLADE

Boolean Logic and Arithmetic through DNA Excision

aTFs

Allosteric Transcription Factors