Gene Digital Circuits Based on CRISPR-Cas Systems and Anti-CRISPR Proteins

CRISPR-Cas systems and anti-CRISPR proteins were integrated into the scheme of Boolean gates in Saccharomyces cerevisiae. The new small logic circuits showed good performance and deepened the understanding of both dCas9/dCas12a-based transcription factors and the properties of anti-CRISPR proteins.

Synthetic gene Boolean gates and digital circuits have a broad range of applications, from medical diagnostics to environmental care. The discovery of the CRISPR-Cas systems and their natural inhibitors-the anti-CRISPR proteins (Acrs)-provides a new tool to design and implement in vivo gene digital circuits. Here, we describe a protocol that follows the idea of the "Design-Build-Test-Learn" biological engineering cycle and makes use of dCas9/dCas12a together with their corresponding Acrs to establish small transcriptional networks, some of which behave like Boolean gates, in Saccharomyces cerevisiae. These results point out the properties of dCas9/dCas12a as transcription factors. In particular, to achieve maximal activation of gene expression, dSpCas9 needs to interact with an engineered scaffold RNA that collects multiple copies of the VP64 activation domain (AD). In contrast, dCas12a shall be fused, at the C terminus, with the strong VP64-p65-Rta (VPR) AD. Furthermore, the activity of both Cas proteins is not enhanced by increasing the amount of sgRNA/crRNA in the cell. This article also explains how to build Boolean gates based on the CRISPR-dCas-Acr interaction. The AcrIIA4 fused hormone-binding domain of the human estrogen receptor is the core of a NOT gate responsive to β-estradiol, whereas AcrVAs synthesized by the inducible GAL1 promoter permits to mimic both YES and NOT gates with galactose as an input. In the latter circuits, AcrVA5, together with dLbCas12a, showed the best logic behavior.

In 2011, researchers proposed a computational method and developed a corresponding piece of software for the automatic design of digital synthetic gene circuits¹. A user had to specify the number of inputs (three or four) and fill in the circuit truth table; this provided all the necessary information to derive the circuit structure using techniques from electronics. The truth table was translated into two Boolean formulae via the Karnaugh map method². Each Boolean formula is made of clauses that describe logic operations (sum or multiplication) among (part of) the circuit inputs and their negations (the literals). Clauses, in their turn, are either summed up (OR) or multiplied (AND) to compute the circuit output. Every circuit can be realized according to any of its two corresponding formulae: one written in POS (product of sums) form and the other in SOP (sum of products) representation. The former consists of a multiplication of clauses (i.e., Boolean gates) that contain a logic sum of the literals. The latter, in contrast, is a sum of clauses where the literals are multiplied.

Electric circuits can be realized, on a breadboard, by physically wiring different gates together. The electric current permits the exchange of signals among gates, which leads to the computation of the output.

In biology, the situation is more complex. A Boolean gate can be realized as a transcription unit (TU; i.e., the sequence "promoter-coding region-terminator" inside eukaryotic cells), where transcription or translation (or both) are regulated. Thus, at least two kinds of molecules establish a biological wiring: the transcription factor proteins and the non-coding, antisense RNAs¹.

A gene digital circuit is organized into two or three layers of gates, namely: 1) the input layer, which is made of YES (buffer) and NOT gates and converts the input chemicals into wiring molecules; 2) the internal layer, which consists of as many TUs as there are clauses in the corresponding Boolean formula. If the circuit is designed according to the SOP formula, every clause in the internal layer will produce the circuit output (e.g., fluorescence) in a so-called distributed output architecture. If the product of sum (POS) formula is used, then a 3) final layer is required, which will contain a single multiplicative gate collecting the wiring molecules from the internal layer.

Overall, in synthetic biology, many different schemes can be designed for the same circuit. They differ in the number and the kind of both TUs and wiring molecules. In order to choose the easiest solution to be implemented in yeast cells, each circuit design is associated with a complexity score S, defined as

where A represents the number of activators, R represents the number of repressors, and a is the amount of antisense RNA molecules. If either activators or repressors are absent from the circuit, their contribution to S is zero. Therefore, it is more difficult to realize a circuit scheme in the lab (high S) when it requires a high number of orthogonal transcription factors. This means that new activators and repressors shall be engineered de novo in order to realize the complete wiring inside the digital circuits. In principle, novel DNA-binding proteins can be assembled by using Zinc Finger proteins³ and TAL effectors⁴ as templates. However, this option appears too arduous and time-consuming; therefore, one should rely mostly on small RNAs and translation regulation to finalize complex gene circuits.

Originally, this method was developed to fabricate digital circuits in bacteria. Indeed, in eukaryotic cells, instead of antisense RNAs, it is more suitable to talk of microRNAs (miRNAs) or small interfering RNAs (siRNAs)⁵. However, the RNAi pathway is not present in the yeast S. cerevisiae. Hence, one should opt for fully transcriptional networks. Suppose that a circuit needs five activators and five repressors; its complexity score would be S = 32. Circuit complexity can be reduced by replacing the 10 transcription factors with a single dCas9⁶ (nuclease deficient Cas9) fused to an activation domain (AD). As shown in⁷, dCas9-AD works as a repressor in yeast when binding a promoter between the TATA box and the TSS (transcription start site) and as an activator when binding well upstream of the TATA box. Thus, one can replace 10 transcription factors with a single dCas9-AD fusion protein and 10 sgRNAs (single guide RNAs) for a total complexity score of S = 11. It is quick and easy to synthesize ten sgRNAs, whereas, as previously commented, the assembly of 10 proteins would demand much longer and more complicated work.

Alternatively, one might use two orthogonal dCas proteins (e.g., dCas9 and dCas12a): one to fuse to an AD, and the other bare or in combination with a repression domain. The complexity score would increase by only one unit (S = 12). Hence, CRISPR-dCas systems are the key to the construction of very intricate gene digital circuits in S. cerevisiae.

This paper deeply characterizes the efficiency of both dCas9- and dCas12a-based repressors and activators in yeast. Results show that they do not demand a high amount of sgRNA to optimize their activity, so episomal plasmids are preferentially avoided. Moreover, dCas9-based activators are far more effective when using a scaffold RNA (scRNA) that recruits copies of the VP64 AD. In contrast, dCas12a works well when fused to the strong VPR AD directly. Furthermore, a synthetic activated promoter demands a variable number of target sites, depending on the configuration of the activator (e.g., three when using dCas12a-VPR, six for dCas9-VP64, and only one with dCas9 and a scRNA). As a repressor, dCas12a appears more incisive when binding the coding region rather than the promoter.

As a drawback, however, CRISPR-dCas9/dCas12a do not interact with chemicals directly. Therefore, they might be of no use in the input layer. For this reason, alternative Boolean gate designs containing anti-CRISPR proteins (Acrs) have been investigated. Acrs act on (d)Cas proteins and inhibit their working⁸. Hence, they are a means to modulate the activity of CRISPR-(d)Cas systems. This paper thoroughly analyzes the interactions between type II Acrs and (d)Cas9, as well as type V Acrs and (d)Cas12a in S. cerevisiae. Since Acrs are much smaller than Cas proteins, a NOT gate responsive to the estrogen β-estradiol was built by fusing the hormone-binding domain of the human estrogen receptor⁹-HBD(hER)-to AcrIIA4. Besides, a handful of YES and NOT gates that expressed dCas12a(-AD) constitutively and AcrVAs upon induction with galactose were realized. At present, these gates serve only as a proof of concept. However, they also represent the first step toward a deep rethinking of the algorithm to carry out the computational automatic design of synthetic gene digital circuits in yeast cells.

1. Design and construction of the sgRNA/crRNA expression cassette

NOTE: There are two kinds of sgRNA/crRNA expression cassettes: one-termed SNR52¹⁰-is composed of the RNA polymerase III-dependent SNR52 promoter, the sgRNA/crRNA sequence, and the SUP4 terminator; another-abbreviated as RGR¹¹-consists of the RNA polymerase II-dependent ADH1 promoter, the RGR (ribozyme-guide RNA-ribozyme) structure that contains two ribozymes (a hammerhead ribozyme-HH, and a hepatitis delta virus ribozyme-HDV) and the sequence of the sgRNA/crRNA in-between, and the ADH1 terminator. The sgRNA guiding Cas9 homologs are made of a spacer sequence and the characteristic direct repeat¹², whereas the crRNA for Cas12a proteins comprises the direct repeat followed by the spacer sequence^13,14 (see Supplementary Table 1 for all DNA sequences used in this study).

1. Design the spacer sequence for Cas9/Cas12a-mediated transcriptional activation.
   1. Harness the bacterial lex operator sequence (named lexOp) to be the target site^15,16 and insert it into the CYC1 promoter that drives the expression of the yeast-enhanced green fluorescent protein (yEGFP)¹⁷. Hence, the spacer sequence is defined by and complementary to the inserted lexOp.
2. Check the orthogonality of the spacer sequence via the CRISPRDIRECT tool¹⁸.
   1. Paste the lexOp sequence flanked by the PAM sequence in the text field, define the PAM sequence as NRG for dCas9 and TTTV for dCas12a, and specify the species from the drop-down list as Budding Yeast (Saccharomyces cerevisiae) S288C Genome. Click on Design. Make sure there is no matching target site in 20mer+PAM nor in 12mer+PAM search.
3. Construct the sgRNA/crRNA expression cassette.
   1. Use touchdown PCR to amplify the DNA sequences of standard biological parts, like promoters, coding sequences, and terminators.
      1. Prepare a reaction mixture containing: 20-40 ng of DNA template, 1 µL of 10 µM forward primer (i.e., ot25, sgRNA/crRNA expression plasmid construction), 1 µL of 10 µM reverse primer (i.e., ot26, sgRNA/crRNA expression plasmid construction), 5 µL of 2.5 mM dNTP mix, 0.5 µL of DNA polymerase, 10 µL of 5x DNA polymerase reaction buffer, and double distilled water (ddH₂O) up to a total volume of 50 µL.
         NOTE: Refer to Supplementary Table 2 for a list of primers used in this study.
      2. Run the touchdown PCR program on a thermocycler:
         Stage 1: 98 °C for 30 s.
         Stage 2 with 10 cycles: 98 °C for 10 s, 68 °C for 20 s, and 72 °C for 15 s.
         Stage 3 with 25 cycles: 98 °C for 10 s, 59 °C for 20 s, and 72 °C for 15 s.
         Stage 4: 72 °C for 2 min.
         Finally, hold at 4 °C until starting the follow-up experiments.
         ​NOTE: The 68 °C in stage 2 and 59 °C in stage 3 depend on the melting temperatures of both forward and reverse primers, varying from different pairs of primers. The time at 72 °C in stages 2 and 3 is determined by the PCR product's length and the DNA polymerase's speed.
   2. Isolate the PCR products via gel electrophoresis (100 V, 30 min). Elute the DNA sequences from the agarose gel via a DNA gel extraction kit (see Table of Materials).
      NOTE: A 0.8% agarose gel is required for fragments longer than 500 nt, a 1.5% agarose gel for fragments between 100 nt and 500 nt, and a 2% agarose gel for fragments shorter than 100 nt.
   3. Insert the TU expressing sgRNA/crRNA into a pRSII404/424 shuttle vector¹⁹, which contains the ampicillin resistance gene and the yeast-selectable auxotrophic marker gene-TRP1.
      1. Digest the shuttle vector at 37 °C for 1 h with the two restriction enzymes SacI and Acc65I. Prepare the reaction mixture by adding 5 µg of the shuttle vector, the recommended amounts of enzymes, digestion buffer (according to the enzyme instructions), and ddH₂O up to a total volume of 30 µL.
      2. Verify the shuttle vector digestion via gel electrophoresis (see substep 1.3.2). Then, inactivate the two enzymes at 65 °C for 20 min.
      3. Use the Gibson assembly method²⁰ to insert the purified PCR products into the cut-open shuttle vector by letting in the equimolar DNA mixture at 50 °C for 1 h.
      4. Transform Escherichia coli DH5α competent cells with the above Gibson reaction mixture via the heat shock transformation method²¹. Transfer the transformed E. coli cells onto Luria-Bertani (LB) agar plates containing ampicillin (0.1 g/L). Put the plates into the incubator at 37 °C and let the cells grow overnight.
      5. Pick four colonies from the LB agar plate and culture them separately overnight at 37 °C in LB solution containing ampicillin (0.1 g/L). Then, use the mini-preparation procedure to extract plasmids from the E. coli cells²².
      6. Use the primers ot18 and ot19 (see Supplementary Table 2 for oligo sequences) to sequence and confirm the inserted transcription unit via the Sanger method²³.
         ​NOTE: In later experiments, the constructed and confirmed plasmids will be inserted into the yeast genome via the PEG/LiAc protocol²⁴.

2. Design and construction of the scaffold RNA expression cassette

NOTE: The scaffold guide RNA (scRNA) is composed of the sgRNA sequence and the MS2 hairpin structures²⁵. Two kinds of MS2 hairpin structures are used in this work: wild-type MS2 hairpin-wt, and f6 MS2 coat protein (MCP) aptamer-f6.

1. Synthesize an scRNA template that is able to accommodate different spacer sequences (i.e., pSNR52-spacer_DR(SpCas9)-2×MS2(wt+f6)-SUP4t).
   ​NOTE: In this study, the scRNA template was synthesized by a gene synthesis company (see Table of Materials).
2. Design proper primers (see Supplementary Table 2) to run PCR on the necessary spacer sequences.
3. Follow the procedures in step 1.3 to construct an scRNA expression cassette.

3. dSpCas9 engineering and expression plasmid construction

1. Obtain the plasmid pTPGI_dSpCas9_VP64 (see Table of Materials).
2. Construct the acceptor vector pRSII406-pGPD-ATG-XbaI-XhoI-CYC1t, based on the pRSII406 shuttle vector, via touchdown PCR and the Gibson assembly method (refer to step 1.3). The plasmid provides a strong constitutive promoter-pGPD, and a terminator-CYC1t.
3. Digest the pTPGI_dCas9_VP64 plasmid and the newly constructed acceptor vector (5 µg for 1 h or 10 µg for overnight-see step 1.3.3.1 as a reference) with XbaI and XhoI at 37 °C. Separate and purify the insert fragment and acceptor vector as in step 1.3.2.
4. Ligate the purified insert fragment and the acceptor vector with T4 DNA ligase at 16 °C for 8 h. Prepare the ligation solution by adding 50-100 ng of the purified acceptor vector, purified target fragments in equimolar amount with the acceptor vector, 1.5 µL of T4 buffer, 0.5 µL of T4 ligase, and ddH₂O up to a total volume of 15 µL.
5. Follow steps 1.3.3.3 and 1.3.3.4. Then, confirm that the newly constructed plasmid is correct by digestion with XbaI and Xhol (37 °C, 1 h) and gel electrophoresis (see step 1.3.2).

4. dCas12a engineering and plasmid construction

1. Construct the plasmids expressing dCas12a-AD.
   1. Synthesize two yeast codon-optimized dCas12a proteins (denAsCas12a and dLbCas12a) flanked by BamHI and Xhol restriction enzyme sites.
      NOTE: In this study, the two yeast codon-optimized dCas12a proteins were synthesized by a gene synthesis company (see Table of Materials).
   2. Construct the acceptor vector pRSII406-promoter-ATG-NLS-GS-HIStag-GS-BamHI-sp-XhoI-AD-NLS-TAA-mTGUO1 via touchdown PCR and the Gibson assembly method (see step 1.3), where the "promoter" is either pGPD or pGAL1, "sp" represents a short random sequence, and "AD" is either VPR or VP64.
   3. Insert each dCas12a protein into the two newly constructed acceptor vectors via digestion with BamHI and XhoI and ligation with T4 DNA ligase (refer to steps 3.3 and 3.4).
2. Construct the plasmids expressing a bare dCas12a.
   1. Construct an acceptor vector for the galactose-inducible expression cassettes of dCas12a as pRSII406-Acc651-pGAL1-ATG-NLS-GS-HIStag-GS-BamHI-sp-XhoI-GS-NLS-TAA-CYC1t using touchdown PCR and the Gibson assembly method (see step 1.3).
   2. Digest the plasmids containing the dCas12a proteins and the above acceptor vector with BamHI and Xhol, then ligate them with T4 DNA ligase to get the plasmid pRSII406-pGAL1-Acc651-ATG-NLS-GS-HIStag-GS-BamHI-dCas12a-XhoI-GS-NLS-TAA-CYC1t (refer to steps 3.3 and 3.4).
   3. Digest the plasmid obtained in step 4.2.2 with Acc65I and BamHI, and then replace pGAL1 with pGPD via touchdown PCR and the Gibson assembly method to construct pRSII406-pGPD-ATG-NLS-GS-HIStag-GS-BamHI-dCas12a-XhoI-GS-NLS-TAA-CYC1t.

5. Anti-CRISPR protein engineering and plasmid construction

NOTE: Three kinds of promoters have been employed to drive Acrs expression: an inducible promoter-pGAL1, four yeast constitutive promoters-pGPD, pACT1, pTEF1, and pTEF2, and a synthetic constitutive promoter-genCYC1t_pCYC1noTATA²⁶.

1. Obtain the plasmids containing the sequences of type II Acrs (AcrIIA2, AcrIIA4²⁷, and AcrIIA5²⁸) and type V-A Acrs (AcrVA1, AcrVA4, and AcrVA5²⁹) from a gene synthesis company.
2. Construct the plasmids based on the pRSII403 shuttle vector to express Acrs.
   1. Construct AcrIIA expression cassettes.
      NOTE: Use touchdown PCR to amplify four different promoters (i.e., pGPD, pACT1, pTEF2, and genCYC1t_pCYC1noTATA), the three kinds of AcrIIA, and two terminators (ADH1t and CYC1t). Build a series of TUs expressing AcrIIAs, under different promoters, via the Gibson assembly method (refer to step 1.3).
   2. Construct AcrVA expression cassettes.
      1. Synthesize the acceptor sequence: ATG-FLAGtag-GS-BamHI-sp-XhoI-NLS-GS-NLS-TAA-Tsynth6, where "sp" is a random sequence that will be replaced by AcrVAs later.
         NOTE: In this study, the acceptor sequences were synthesized by a gene synthesis company (Table of Materials).
      2. Assemble the acceptor vectors pRSII403-promoter-ATG-FLAGtag-GS-BamHI-sp-XhoI-NLS-GS-NLS-TAA-Tsynth6, where the "promoter" is: pGAL1, pTEF1, and genCYC1t_pCYC1noTATA. Use the Gibson assembly method (refer to step 1.3).
      3. Insert each of the three AcrVAs into the acceptor vector described in step 5.2.2.2 (via touchdown PCR and the Gibson assembly method [see step 1.3]) to build a set of plasmids producing AcrVAs.
3. Further engineer AcrIIA4 by extending its sequence with the sequences of the GS linker and HBD(hER). This permits the building of a β-estradiol-responsive circuit.
   1. Use touchdown PCR to obtain GS-HBD and AcrIIA4 parts separately (see step 1.3.1).
   2. Put AcrIIA4, GS-HBD, and the shuttle vector into the Gibson mixture and construct the complete plasmid via the Gibson method (see step 1.3.3).

6. crRNA detection: RT-qPCR and primers' design

NOTE: crRNA detection was achieved via RT-qPCR, which is organized in three steps.

1. Perform the RNA extraction and purification from yeast cells via an RNA kit.
   1. Culture yeast cells overnight in 2 mL of synthetic defined complete medium (SDC, 1 L volume: 20 g of Glucose, 2 g of AA mix, 6.7 g of YNB, 396 mg of Leucine, 79.2 mg of Tryptophan, 79.2 mg of Histidine, and 79.2 mg of Uracil) by using a 24-well plate (240 rpm, 30 °C).
   2. In the morning, dilute the cell culture (1:100) into 2 mL of fresh SDC and keep growing the yeast cells at 30 °C, 240 rpm, for another 4 h.
   3. Harvest the whole 2 mL of the cell solution and centrifuge at 20,238 x g for 2 min. Remove the supernatant carefully since the cell pellet is small.
   4. Extract the RNA from yeast cells using an RNA kit.
   5. Check the RNA quality.
      1. Prepare a 1% agarose gel. Mix 5 µL of each RNA sample with 1 µL of DNA loading dye. Then load the mixture on the gel and run it.
      2. Ensure that two clear bands at ~4,000 nt and ~2,000 nt, corresponding to ribosomal RNA (25S/18S), are present on the gel. A further blurred band can be seen at ~80 nt for tRNA.
   6. Use the RNA samples immediately for cDNA synthesis or store them at -80 °C for future use.
2. Reverse transcription: Use the stem-loop method³⁰ to form the first strand of cDNA corresponding to the crRNA (40 nt). For the reverse transcription of the sgRNA (almost 100 nt), the procedure is the same as that for the cDNA synthesis of the reference gene ACT1.
   NOTE: The method for the reverse transcription of the crRNA is different from that used with the sgRNA and the ACT1 mRNA. Since the crRNA is very short, it was treated as a microRNA and used the microRNA reverse transcription method (stem-loop approach) to obtain the corresponding cDNA. Two cDNA synthesis kits (a stem-loop kit for the crRNA and a usual reverse transcription kit for the ACT1 gene) were used in crRNA quantification. The same amount of RNA was used in the two kits (see Table of Materials) to make the experiment results comparable. The primer used with the stem-loop kit was designed according to the stem-loop sequence and the last six nucleotides at the 3' end of the crRNA (for the stem-loop and primer sequence, see Supplementary Table 2).
   1. Stem-loop method for crRNA reverse transcription
      1. Take the RNA template, primer, and buffers out of the freezer and let them melt on ice.
      2. Genomic DNA removal: First, prepare 10 µL of the reaction mixture according to the kit instructions. Put the mixture in a thermal cycler at 42 °C for 2 min. Finally, transfer it onto ice.
      3. Synthesis of the first cDNA strand: Prepare 20 µL of the reaction mixture by adding 10 µL of the mixture from step 6.2.1.2, 1 µL of stem-loop primer (2 µM concentration), 2 µL of 10x RT reaction buffer, 2 µL of reverse transcription enzyme mix (containing the reverse transcriptase), and 5 µL of RNase-free H₂O.
      4. Place the reaction mixture in a thermal cycler and run the following program: 25 °C for 5 min, 50 °C for 15 min, and 85 °C for 5 min. Use the product for the qPCR reaction immediately or store it at -80 °C.
   2. sgRNA and ACT1 mRNA reverse transcription
      1. First reaction: Prepare a 13 µL mixture consisting of the primer mix, dNTP mix, RNA template (50 pg-5 µg), and RNase-free water (apart from the RNA template, all provided by the kit), according to the kit instructions. Use 1 µg of RNA template. Put the mixture into a thermal cycler at 70 °C for 10 min.
      2. Second reaction (cDNA synthesis): Prepare the reaction mixture by adding the reagents (as described in the kit instructions) to the 13 µL of the first reaction solution up to a final volume of 20 µL. Place the mixture in a thermal cycler for 15 min at 50 °C and then for an additional 5 min at 85 °C. Use the product for the qPCR reaction immediately or store it at -80 °C.
3. SYBR kit for qPCR: Ct value detection
   NOTE: The reverse primer used in the qPCR of the crRNA is fixed because it corresponds to the reverse complement of the stem-loop sequence (see Supplementary Table 2). The forward primer, in contrast, is variable and depends on the sequence of the crRNA. The forward and reverse primers for the sgRNA and ACT1 mRNA qPCR are designed at https://www.ncbi.nlm.nih.gov/tools/primer-blast/. Two primers are selected when the difference between their melting temperatures is no greater than 2 °C (see Supplementary Table 2). Each sample is measured in three replicates.
   1. Prepare the qPCR reaction mixture according to the manufacturer's instructions for the SYBR kit.
   2. Set the following qPCR program in a real-time PCR machine.
      Preincubation: 10 min at 95 °C.
      PCR stage: 15 s at 95 °C, followed by 34 s at 55 °C. Set the cycle of the PCR stage to 45 times. Melting stage: 10 s at 95 °C, followed by 60 s at 65 °C, and finally 1 s at 97 °C.
   3. Calculate the relative mRNA expression values via the Pfaffl formula³¹.

7. Immunofluorescence to detect Cas proteins

NOTE: Cas proteins (CasP) are fused to a His_tag.

1. Yeast cell preparation
   1. Pick some yeast cells using a sterile loop and culture them in 5 mL of YPD-rich medium overnight at 240 rpm at 30 °C. In the morning, add 500 µL of the cell solution to 20 mL of fresh YPD and grow them at 240 rpm at 30 °C until OD₆₀₀ reaches 0.6.
   2. Take 5 mL of the culture and mix it with 500 µL of 37% formaldehyde. Let the mixture stay at room temperature (RT) for 10 min. Harvest the cells by centrifugation at 1,500 x g for 5 min.
   3. Remove the supernatant and resuspend the cells with 1 mL of fixation buffer (0.1 M KH₂PO₄, 0.5 M MgCl₂, 3.7% formaldehyde, pH = 6.5). Keep the cell solution at RT for 20 min.
   4. Centrifuge the cell solution at 1,500 x g for 5 min. Discard the supernatant and resuspend the cells in 1 mL of washing buffer (0.1 M KH₂PO₄, 1.2 M sorbitol, pH = 6.5) supplemented with 4 µL of beta-mercaptoethanol and 4 µL of lysate solution (5 mg/mL). Put the cell solution into an incubator at 37 °C for 20 min.
   5. Centrifuge the cell solution at 1,500 x g for 5 min and discard the supernatant. Wash the cell pellet twice with 1 mL of PBS by centrifugation (1,500 x g for 5 min).
   6. Resuspend the cells in 100 µL of PBS plus 0.05% Tween 20 and add 4 µL of BSA solution (10 mg/mL). Keep the cell solution at RT for 20 min.
2. Incubation with primary antibody
   1. Add the Anti-His tag primary antibody to the mixture in step 7.1.6 at a 1:400 dilution. Keep the solution at RT for 2 h.
   2. Centrifuge the mixture in step 7.2.1 at 1,500 x g for 5 min and remove the supernatant. Add 1 mL of PBST and centrifuge (1,500 x g) for 5 min to wash the cell pellet. Repeat this operation twice. Finally, discard the supernatant and suspend the cells in 100 µL of 1x PBST.
3. Microscopy cell detection
   1. Mount the cells on a slide; take 2 µL of the cell solution from step 7.2.2 and place it on a glass slide. Cover it with a coverslip.
   2. Observe the cells under a fluorescence microscope. Turn on the fluorescent light source, microscope, and computer. Write down the fluorescent light source number and open the microscope software on the computer.
   3. Put the slide on the microscope stage. Choose the 40x objective lens and observe the cells under the green light (550 nm). Move the coarse focus knob until the contour of yeast cells appears. Move the fine focus knob to focus the cells.
   4. Detect the cells with the microscope software. Close the microscope field of view and switch to the computer screen. Click on Live, wait for 3-5 s, and click on Capture to take a picture. Save the picture.
   5. Turn off the computer, microscope, and fluorescent light source.

8. Data acquisition: FACS

NOTE: Green fluorescence is detected via flow cytometry (i.e., fluorescence-activated cell sorting [FACS] measurements). Yeast cells are cultured, in general, at 30 °C and 240 rpm to run FACS experiments. However, cells might demand some precautions depending on their genetic content. Cells that contain the dCas12a-VPR gene (controlled by the GPD constitutive promoter) must be grown for 24 h in the SDC solution. Afterward, cells are diluted at a ratio of 1:100 in fresh SDC and grown for another 12 h before measuring the fluorescence intensity. Cells modified with the AcrIIA4-HBD(hER) gene demand dilution as well. Moreover, OD₆₀₀ needs to be controlled. First, the cells are allowed to grow in SDC overnight (over 14 h). In the morning, OD₆₀₀ is measured. Then the culture is diluted in SDC, supplied with a diverse concentration of β-estradiol, down to OD₆₀₀ = 0.1. Before FACS experiments, the cells are grown for another 7 h such that OD₆₀₀ reaches 0.8-1.0. Cells expressing dCas9-VP64 or dCas12a-VP64 are grown in SDC for 20-24 h without dilution and further growth before the measurements at the FACS machine.

1. Switch on the FACS machine 20 min before the measurements to warm up the laser.
2. Prepare the samples (dilution): mix 20 µL of the cell culture with 300 µL of ddH₂O.
3. Run the FACS software on the computer connected to the FACS machine and create a new experiment. Set the measurement parameters (i.e., FSC(SSC)-A/H/W and histogram).
4. Select the filter according to the excitation and emission wavelengths of the samples. For example, the target protein here is yEGFP, whose excitation and emission wavelengths are 488 nm and 507 nm, respectively. So, select the FITC or GFP filter (excitation wavelength: 488 nm; emission wavelength: 527/32 nm). Set the acquisition cell number to 10,000.
5. Adjust the FITC filter voltage by measuring the intensity of fluorescent beads. Ensure that the relative difference in the intensity of the beads between two consecutive experiments does not exceed 5%.
6. Wash the machine with ddH₂O for a few seconds to remove any possible excess beads.
7. Measure the sample fluorescence intensity. Click on Preview and wait for 3-5 s for sample injection stability. Finally, click on Acquire.
8. Measure the beads again at the end of the experiment. The voltage is the one that was used during the initial beads' measurement (see step 8.4, 438-441 V). Check whether the relative difference between the two beads' measurements exceeds 5%.
9. Export the FACS data as FCS files.
10. Analyze the FCS files with R Studio software.

9. Data analysis

NOTE: Use the Flowcore R Bioconductor package³² within R studio. The FCS files were analyzed using a script written in R language.

1. Open R studio and load the script Bdverse analysis.R to analyze the FCS files.
2. Set the experiment name (ename), the directory (path) where the FCS files are
   Stored (dir_d), and where the result files will be created (dir_r).
3. Set the fluorescence channel. For instance, select_ch = "GPA-A" if green fluorescence was measured.
4. Set the number of samples that were measured (s_num).
5. Set the dimensions of each dot plot (sxlim, sylim). Set the maximal length of the x and y axis for bar plots and box plots (xlimit, ylimit). xlimit must be greater than or equal to s_num.
6. Choose the gating method by removing the # from the corresponding lines.
   NOTE: morphGate is an automatic gating method carried out by the script (i.e., the region of the dot plots where the cells are denser is recognized and selected by the program). polygonGate and rectangleGate demand to look at the dot plots and define either the vertices of a polygon or the side of a rectangle that embraces the zone where most of the cells lie.
7. Select the flowSet object gated, corresponding to the chosen gating method. Select the dot plot resolution (res; should be at least equal to 256).
8. Uncomment flt_low <- filter_low(sp) to remove measurements where fluorescence is negative. Uncomment flt_sp <- filter(flt_lw) to remove outliers due to other experiments.
9. Press Source and run the script. All the files containing the results from the analysis are created in dir_r.

sgRNA/crRNA expression by an RNA polymerase III-type promoter
First, this work addressed the engineering of the transcriptional activation circuit (circuit 1) shown in Figure 1A. It contained three basic components: 1) the gene encoding for yEGFP (the reporter), which was preceded by a series of different synthetic promoters that provided target sites for dCas9/dCas12a-AD; 2) a yeast codon-optimized version of dCas9 or dCas12a fused to an activation domain (VP64 and VPR, respectively) and containing either one or two nuclear localization sequences (NLSs). Both dCas proteins were produced by a strong constitutive promoter-pGPD; and 3) an sgRNA/crRNA sequence that guided dCas9/dCas12a-AD to the target sites. The activation efficiency of dCas9/dCas12a-based activators was visualized and reflected by the fluorescence intensity of the reporter (measured via FACS experiments).

One dCas9 protein (dSpCas9) and two dCas12a proteins (denAsCas12a and dLbCas12a) were tested. A 3.36-fold activation was achieved with dSpCas9 extended, at its C terminus, with the VP64 AD and binding a synthetic promoter-upstream of yEGFP-containing six copies of the lexOp target site. The sgRNA was placed into an integrative shuttle vector and transcribed by the RNA polymerase III-dependent SNR52 promoter (SNR52i configuration, see Figure 1B). In the case of dCas12a, denAsCas12a-VPR returned the highest activation (4.45-fold) from a synthetic promoter with three operators when the crRNA was expressed via the SNR52i configuration (Figure 1C). Under the same conditions, dLbCas12a-VPR achieved its best fluorescence enhancement (3.21-fold) (Figure 1D). It should be noted that the comparison term, in every experiment, was a circuit whose sgRNA/crRNA could not bind the lex operators.

Multicopy plasmids are not necessary
The SNR52i sgRNA expression cassette was replaced with an RGR structure expressed by a moderately strong promoter-pADH1. However, in both dCas9 and dCas12a cases, activation in the presence of an sgRNA/crRNA generated by the self-cleavage of RGR appeared comparable to or even lower than that achieved with an sgRNA/crRNA produced via SNR52i, despite the fact that pSNR52 was considered as a weak promoter (see Figure 1C,D for the first results obtained with dCas12a).

To further explore the connection between sgRNA/crRNA amount and activation efficiency, the two sgRNA/crRNA expression systems were inserted into an episomal plasmid, which can be taken up by the cell in 10-40 copies and generate a greater quantity of sgRNA/crRNA. As shown in Figure 2A, the activation by crRNA located on an integrative plasmid (either SNR52i or RGRi) was 1.4- to 2.4-fold higher than that achieved when the same crRNA was expressed by an episomal plasmid (either SNR52m or RGRm). The trend was confirmed by the sgRNA. In this case, the integrative plasmid guaranteed a 1.1- to 1.5-fold higher activation (Figure 2B). To exclude that the results were caused by a loss of the episomal plasmids, RT-qPCR was carried out to quantify the sgRNA/crRNA relative abundance in vivo. The results, in Figure 2C,D, verified that the episomal vector produced a much higher level of sgRNA/crRNA than the integrative vector, no matter the expression system (RGR or SNR52). These results showed that the SNR52 system could work even better than the RGR system, and a higher amount of sgRNA/crRNA in the cell did not guarantee a higher activation by the CRISPR-Cas system. Therefore, episomal plasmids should not be employed in the construction of gene digital circuits where dCas9/dCas12a-AD are used.

Scaffold RNA engineering
An scRNA was engineered by extending the sequence of an sgRNA with MS2 hairpin structures that permitted to recruit VP64 AD when fused to the MS2 coat protein (MCP, see Figure 3A). In this way, no engineering of, nor modifications on, dCas9 were necessary. Two MS2 variants were tried: wt and f6. The scRNA containing the single MS2 hairpin-1×MS2(wt) and 1×MS2(f6)-gave a 5.27-fold and a 4.34-fold activation, respectively. However, the scRNA with a combination of the two hairpins-2×MS2(wt+f6)-returned the overall highest activation in this study (7.54-fold, see Figure 3B). These results demonstrated that engineering an scRNA was far more effective than fusing any activation domains to dSpCas9 directly.

Boolean gate based on Acr proteins
To further control and tune transcriptional activation by CRISPR-Cas systems, circuit 1 was modified with the insertion of a fourth TU for the expression of anti-CRISPR proteins (see Figure 4A for AcrIIAs and Figure 5A for AcrVAs). After showing that Acrs were effective, in S. cerevisiae, in contrasting the activation due to dCas9/dCas12a-AD, a new circuit was built (see Figure 5D) to test AcrVA action on dCas12a-based repressors (a previous work³³ had already shown that AcrIIAs could inhibit dCas9-based gene down-regulation). The new, Acr-containing small networks worked as simple Boolean gates (YES and NOT), which might lead to a restyling of the input layer of more complex synthetic gene digital circuits.

AcrIIA4 is a strong inhibitor of dSpCas9
Four different promoters with diverse strengths were used to drive the expression of three kinds of AcrIIs-AcrIIA2, AcrIIA4, and AcrIIA5. The results showed that the three AcrIIs worked in a dose-dependent manner in S. cerevisiae. When expressed by a strong promoter-pGPD-they reduced the fluorescence level reached by dSpCas9:scRNA_2×MS2(wt+f6)-MCP-VP64 to 0.21, 0.11, and 0.13 of its value, respectively (Figure 4B). Since AcrIIA4 was the only one that caused high inhibition of fluorescence expression even when produced by a weak, synthetic promoter-genCYC1t_pCYC1noTATA-we could infer that AcrIIA4 was the strongest inhibitor among the three AcrIIs. Next, the HBD(ER) was fused to the C terminus of AcrIIA4 to build a β-estradiol sensing device (see Figure 4C). In the presence of the estrogen β-estradiol, AcrIIA4-HBD(ER) could translocate into the nucleus and then neutralize the function of the dSpCas9-based activator. The titration curve in Figure 4D shows that the circuit behaves like a NOT gate with an ON/OFF ratio approaching 2.3.

AcrVAs are repressors of dCas12a proteins
A NOT gate was designed and built by inserting the pGAL1-driving AcrVA expression cassette into circuit 1. In this way, AcrVA synthesis, and the consequent repression of dCas12a-AD, could be induced by galactose (Figure 5A). As shown in Figure 5B,C, AcrVA1 hindered both denAsCas12a and dLbCas12a as activators by reducing fluorescence expression from 19% to 71%, depending on the circuit scheme. AcrVA4 and AcrVA5 cannot exert any action on denAsCas12a³⁴. However, they performed a strong inhibition of dLbCas12a-based activators by reducing fluorescence expression up to 84% (AcrVA5) and 82% (AcrVA4). On the whole, AcrVA5 turned out to be the most reliable, among these three AcrVAs, in inhibiting dLbCas12a-based activators because it guaranteed, in different circuits, higher than 70% repression.

AcrVA1 action is concentration-dependent
The relation between AcrVAs' in vivo concentration and their inhibitory effects on both activators and repressors based on denAs/dLbCas12a was also studied. To this aim, each AcrVA was expressed under different promoters: the strong pGAL1, the medium-strength pTEF1, and the weak genCYC1t_pCYC1noTATA. As illustrated in Figure 5E, AcrVA1 displayed big fluctuations in its performance depending on the promoter that led its synthesis. AcrVA1 worked reasonably well only when produced by pGAL1. Under pTEF1 and genCYC1t_pCYC1noTATA, AcrVA1 showed some repression on the bare dLbCas12a only. AcrVA4 and AcrVA5, in contrast, appeared to be less sensitive to their concentration, especially when interacting with the bare dLbCas12a.

This data showed that AcrVA4 and AcrVA5 generally performed better than AcrVA1 in inhibiting dLbCas12a-based transcription factors in S. cerevisiae. It should be noted that AcrVA5 has a peculiar working mechanism, since it acts as an enzyme that modifies LbCas12a permanently. However, as mentioned above, AcrVA5 (together with AcrVA4) cannot interact with denAsCas12a.

When AcrVAs were expressed under pGAL1, circuits become either YES gates (dCas12a was fused to an AD) or NOT gates (bare dCas12a). The former looked all highly performant, while the latter appeared to work better in the presence of AcrVA4 or AcrVA5.

Figure 1: Transcriptional activation mediated by dCas9/dCas12a-AD. (A) Circuit 1 diagram. The synthetic promoters of yEGFP contained six (6x) and three (3x) copies of target sites for dCas9 or dCas12a, respectively. After that, dCas9/dCas12a-AD combines with sgRNA/crRNA; the synthetic promoter is targeted and activated due to the presence of the activation domains. (B) The best activation efficiency of dSpCas9-VP64 was achieved when there were six lexOp target sites on the synthetic promoter, and the sgRNA was transcribed by SNR52i. (C,D) The highest activation efficiency of dCas12a-VPR was obtained when three copies of lexOp were inserted into the synthetic promoter, and the crRNA was generated by SNR52i. SNR52i means that the sgRNA/crRNA was produced by pSNR52, and the expression cassette was placed inside an integrative shuttle vector. RGRi means an integrative plasmid hosting the RGR cassette to express sgRNA/crRNA was used. The negative control, "-", represents a sgRNA/crRNA containing a scrambled spacer sequence that does not match with the lexOp site nor with any sequence along the yeast genome. "bA" indicates that the sgRNA/crRNA binds the antisense strand of target DNA, while "bS" stands for binding the sense strand. Each fluorescence level represents the mean value from at least three independent experiments (i.e., carried out on different days). Error bars are the standard deviation of the mean. Please click here to view a larger version of this figure.

Figure 2: The comparison of integrative and episomal plasmid-producing sgRNA/crRNA. (A) Activation efficiency of dLbCas12a-VPR:crRNA when targeting a single site on the promoter upstream of yEGFP. (B) dSpCas9-VP64:sgRNA activated n× synthetic promoter ("n" stands for the number of lexOp target sites). (C,D) Normalized sgRNA/crRNA expression level^15,16. "i" means that the sgRNA/crRNA expression cassette was placed into an integrative shuttle vector, while "m" stands for multicopy (i.e., episomal) plasmid. "bA"/"bS" indicates that the sgRNA/crRNA binds the antisense/sense strand of the DNA. "ctrl" is a negative control, where a scrambled sgRNA/crRNA was expressed. Each fluorescence level represents the mean value from at least three independent experiments (i.e., carried out on different days). Error bars are the standard deviation of the mean. Please click here to view a larger version of this figure.

Figure 3: The activation efficiency of the bare dSpCas9 in complex with an scRNA. (A) The schematic diagram of the interactions among scRNA, MCP-VP64, and the bare dSpCas9¹⁶. The cap-like structure in purple represents MCP (MS2 coat protein). One MS2 hairpin (the purple structure in the scRNA) can recruit and bind two copies of MCP. Thus, an scRNA enables not only DNA-binding by dSpCas9 but also activation of gene expression through the recruitment of MCP-VP64. (B) The activation efficiency of dSpCas9:scRNA-MCP-VP64. Three kinds of scRNA were tested: one carrying the wild-type MS2 hairpin-1×MS2(wt), another engineered with the f6 MCP aptamer-1×MS2(f6), and the last one containing both hairpins-2×MS2(wt+f6), which turned out to be the most performant. Each fluorescence level represents the mean value from at least three independent experiments (i.e., carried out on different days). Error bars are the standard deviation of the mean. Please click here to view a larger version of this figure.

Figure 4: AcrIIA-related circuits and results. (A) The AcrIIA expression cassette was inserted into circuit 1. This additional TU includes pGPD leading the expression of AcrIIAs to counteract dSpCas9:scRNA_2×MS2(wt+f6)-MCP-VP64. (B) The inhibition efficiency of AcrIIAs on the best dSpCas9-based activator in Figure 3B. The black dashed line represents the fluorescence in the presence of the dSpCas9-based activator. The figures above each column show the inhibition efficiency calculated as the OFF/ON ratio (i.e., the fluorescence level in the presence of the AcrIIA divided by the fluorescence in the absence of any AcrIIA). In the legend, the strength of the four constitutive promoters gradually increases from up to down. (C) Diagram of the β-estradiol sensing device (NOT gate) expressing AcrIIA4-HBD(hER). (D) Titration curve of the circuit in (C)¹⁶. The green curve refers to the change in fluorescence in the functional circuit. The black curve was derived from the strain without the expression of AcrIIA4-HBD(hER)-the negative control. The dashed line in gray marked the fluorescence plateau at the equilibrium. It was calculated as the mean of the fluorescence values at concentrations of β-estradiol not lower than 125 nM. Each fluorescence level represents the mean value from at least three independent experiments (i.e., carried out on different days). Error bars are the standard deviation of the mean. Please click here to view a larger version of this figure.

Figure 5: AcrVA-related circuits and results. (A) The AcrVA expression cassette was inserted into circuit 1. The new TU contains the inducible promoter pGAL1 upstream of the AcrVA genes that neutralize the working of dCas12a-AD. The new circuit is a NOT gate regulated by galactose. (B,C) Results from the galactose-responsive NOT gate in (A). Here, pGAL1 drives the synthesis of AcrVAs, which then interact with dCas12a-AD¹⁵. The relative fluorescence corresponds to the OFF/ON ratio. (D) Galactose-responsive YES gate. It employs AcrVAs under the control of pGAL1 and the bare dCas12as that represses the synthesis of yEGFP. (E) Comparison of the inhibition efficiencies of AcrVAs expressed by promoters of different strength¹⁵. The groups "AcrV effects on repression" and "bare dLb" refer to the circuit in (D). The groups "AcrV effects on activation" and "dLb-VPR" are the results of the NOT gate in (A). Each fluorescence level represents the mean value from at least three independent experiments (i.e., carried out on different days). Error bars are the standard deviation of the mean. Please click here to view a larger version of this figure.

Supplementary Table 1: A list of all DNA sequences used in this study. Please click here to download this File.

Supplementary Table 2: A list of primers used in this study. Please click here to download this File.

Supplemental Coding Files: The R studio script to analyze the FCS files. Please click here to download this File.

The protocol showed a possible complete workflow for synthetic gene digital circuits, following the "Design-Build-Test-Learn" (DBTL) biological engineering cycle and concerning both dry-lab and wet-lab experiments. Here, we focused on the CRISPR-Cas system, mainly dSpCas9, denAsCas12a, dLbCas12a, and the corresponding anti-CRISPR proteins, by designing and building in S. cerevisiae small transcriptional networks. Some of them mimicked Boolean gates, which are the basic components of digital circuits. All circuits described here permitted us to depict the properties and features of CRISPR-associated and anti-CRISPR proteins in S. cerevisiae. These results are essential to include these proteins in the scheme of gene digital circuits.

The DBTL concept provides a framework in synthetic biology, while many optimizations and improvements shall be made after testing a new artifact. For instance, in circuit 1, there was initially only one target site (one copy of lexOp) for dCas9/dCas12a-AD on the synthetic promoter upstream of yEGFP. After testing that circuit configuration, we found that it could achieve no more than a twofold activation^15,16. Then, we assumed that by increasing the number of copies of lexOp, as in⁷, we could reach a higher transcriptional activation. Indeed, a higher fluorescence level was obtained by using three to six lexOp sites (Figure 1). Moreover, we further improved the performance of the circuits hosting dSpCas9 by engineering an scRNA, which is easier than fusing one or more ADs to a big protein like dSpCas9 (Figure 3). Additionally, by using a promoter of different strengths to produce the three AcrIIAs that we chose, we concluded that AcrIIA4 was the strongest inhibitor among them. Thus, we built a new NOT gate responsive to β-estradiol by fusing the HBD(ER) to AcrIIA4 and exploiting the strong repression of AcrIIA4 on our best dSpCas9-based activator (Figure 4).

Similarly, we deeply characterized both the working of denAsCas12a and dLbCas12a in yeast and their interactions with three AcrVAs (Figure 5). For each dCas12a-AcrVA pair, we built a NOT (dCas12a was fused to an AD) and a YES (bare dCas12a) gate responsive to galactose. On the whole, dLbCas12a, together with AcrVA5, resulted in the best system to compute simple logic functions.

The method described here presented some critical steps. All protein DNA sequences were yeast codon-optimized to ensure higher expression in S. cerevisiae. To avoid unspecific targets of dSpCas9/denAsCas12a/dLbCas12a in the S. cerevisiae genome, we selected a bacterial operator such as lexOp. Moreover, strains containing the GAL1 promoter showed a significant growth delay that could limit the applicability of pGAL1 to synthetic gene circuits¹⁵.

Some modifications could also be brought to some steps in the overall method. In order to improve the efficiency of the digestion-ligation procedure, it is preferable to digest 10 µg (overnight) of both the insert-containing plasmid and the acceptor vectors, rather than only 5 µg in 1 h. In this way, a higher DNA concentration is reached after the elution step. The time for T4-ligation should be extended from 1 h (manufacturer's protocol) to 8 h. Finally, strains containing the dCas12a-VPR fusion protein should be diluted after a 24 h culture and grown for a further 12 h before running a FACS experiment. Under this condition, the variability among the fluorescence levels from different cells is no longer too high, and an acceptable standard deviation accompanies the mean value of the fluorescence intensity over a cell population.

In summary, this protocol explained how to simplify the design of gene digital circuits by making use of dCas proteins and, possibly, anti-CRISPR proteins. More importantly, we showed in detail how these families of proteins work in S. cerevisiae and which of them are the most promising for a future usage inside digital networks. An unsolved problem is the coupling of CRISPR-dCas/anti-CRISPR systems and chemicals, which represent the circuit inputs and cannot directly bind dCas proteins or anti-CRISPRs. Here, we bypassed the problem by using either the inducible GAL1 promoter or the HBD(ER) attached to AcrIIA4. However, a way to generalize the architecture of the circuit input layer is necessary to design synthetic gene digital circuits for different bioengineering areas such as metabolic engineering, biosynthesis, biosensing, biodiagnostics, and bioremediation.

The authors declare no competing financial interest.

We want to thank all the students of the Synthetic Biology lab-SPST, TJU-for their general help, together with Zhi Li and Xiangyang Zhang for their assistance in FACS experiments.