AND Logic Based on Suppressor tRNAs Enables Stringent Control of Sliding Base Editors in Pseudomonas putida Authors: Elena Velázquez, Víctor de Lorenzo Published: 2024 ABSTRACT Base editors, e.g., cytosine deaminases, are powerful tools for precise DNA editing in vivo, enabling both targeted nucleotide conversions and segment-specific diversification of bacterial genomes. Yet, regulation of their spatiotemporal activity is crucial to avoid off-target effects and enabling controlled evolution of specific genes and pathways. This work reports a strategy for tight control of base-editing devices through subjecting their expression to a genetic AND logic gate in which two chemical inducer inputs are strictly required for cognate activity. The case study involves an archetypal genetic device consisting of a cytosine deaminase (pmCDA1) fused to a T7 RNA polymerase (RNAPT7), which cause intensive diversification of DNA portions bordered by a T7 promoter and a T7 terminator—but whose activity in vivo has been shown unattainable to govern with standard conditional expression systems. By encoding up to three UAG stop codons into the DNA sequence of the pmCDA1-RNAPT7 fusion, which is transcribed by the 3-methylbenzoate inducible promoter Pm, we first broke the structure of the hybrid protein. Then, to overcome the interruptions caused by UAG codons, we placed transcription of a supF tRNA under the control of a cyclohexanone-dependent system. When tested in the soil bacterium and metabolic engineering chassis Pseudomonas putida KT2440, these modifications changed the performance of the sliding base editor from a flawed YES logic to a precise AND logic. We also showed that such a 2-layer control brings about a minimal background activity as compared to a single-input digitalizer circuit. These results show the ability of suppressor tRNA-based logic gates for achieving stringent expression of otherwise difficult to control devices. The emergence of precision genome editing technologies has ushered in a new era of biological research and therapeutic potential. Techniques like CRISPR-Cas91 or TALENs2 have revolutionized our ability to manipulate DNA with unprecedented precision. Additionally, mutagenesis technologies have boosted methods for directed evolution to identify protein variants with improved properties. Within this framework, although in vitro mutagenesis techniques encompass several advantages, they entail important drawbacks such as the necessity for cloning and transformation into a host cell for expression during selection. These steps demand significant time and labor, particularly during the iterative cycles of mutagenesis and selection. Consequently, in vivo mutagenesis systems are a rising preference because the generation of variants, their expression, and selection can be done in a continuous fashion, thereby accelerating directed evolution.3−5 One approach to continuous, target-specific in vivo mutagenesis involves directing mutagenic enzymes or nucleases to specific regions in DNA, sparing the rest of the genome. Base editors6 are a versatile targeted in vivo mutagenesis tool that performs in different hosts.7−10 The mutagenesis activity of base editors is based on cytosine deaminases (CdA)11 and adenosine deaminases12 coupled with proteins that direct their activity to a particular region in the chromosome. CdAs catalyze the deamination of cytosine, producing uracil in the process, which leads to different types of mutations. For instance, if the generated uracil is not excised by the repair enzyme Uracil DNA N-Glycosylase (UDG13), it will be paired with adenosine during replication, generating a transition mutation from the original C:G pair to the final T:A pair.14 Accordingly, AdAs deaminate adenosine and generate inosine, that is paired with cytosine, thereby causing mutations from A:T to G:C.15 The combination of CdA and AdA with CRISPR-Cas systems6,15 or the T7 RNA polymerase (RNAPT7)16−18 has allowed leveraging base editors for either precise changes or diversification of DNA within specific genomic targets. Alas, despite all this potential, application of base editors and their fusions with other DNA-interacting proteins thereof has been often flawed in bacteria by the lack of adequate conditional expression devices to secure an adequate spatiotemporal manifestation of the activity of interest. Conspicuously, the effector-dependent transcription devices used to control expression of, e.g., sliding base editors (i.e., fusions between a base deaminase and RNAPT7) have been generally—and unpredictably—unable to suppress mutagenesis in the absence of inducers, so leaky expression is one of the bottlenecks in its implementation. This might be caused by the pervasive transcription,19 cryptic promoters,20 internal ribosome entry sites,21 or other signals recognized by host’s expression machinery and which may overrun any engineered regulatory node implanted later. Different strategies can be entertained to solve this drawback, for example, physical decoupling of regulatory elements along with the conditional proteolysis of the target protein,22 engineering super-repressors,23 or engineering digitalizing circuits,24 among others. In the current study, we took inspiration from one of the first logic gates executed with biological parts.25 In our case, this approach—implemented with different genetic components—was instrumental to suppress leaky expression of a sliding base editor composed of the cytosine deaminase PmCDA1 from lamprey Petromyzon marinus(26) fused to the N-terminus of RNAPT7, in the soil bacterium and metabolic engineering chassis Pseudomonas putida KT2440.27−29 As shown below, punctuation of the pmCDA1- RNAPT7 sequence with stop codons and conditional expression in trans of the cognate suppressor tRNA not only changed the logic of the regulatory node from YES to AND but also overran any cryptic expression signals, thereby enabling an unprecedented degree of control of mutagenic intensity in given genomic segments of the host bacterium. Results and Discussion Customary Heterologous Expression Systems Fail to Regulate Activity of Sliding Base Editor pmCDA1-RNAPT7 Genetic devices for engineering controllable expression in prokaryotes typically involve a signal-responsive transcription factor, which binds a cognate promoter, downstream of which a gene of interest is placed.30−32 Out of the many choices available, some have become widely used because of their tight regulation, considerable strength when induced, and excellent ON/OFF ratios. Many of such expression devices widely used in a variety of Gram-negative bacteria include the 3-methylbenzoate (3mBz)-inducible XylS/Pm regulator/promoter pair, which originates from the TOL pWW0 catabolic plasmids of P. putida mt-2.33,34 Since Pm promoter activity in the absence of induction is extremely low,35−37 our first strategy for regulating the activity of the base editor pmCDA1-RNAPT7 in P. putida involved engineering its expression under the control of the XylS/Pm regulatory device.30 The experimental system to this end has been described in detail38 and it is sketched in Supplementary Figure S1A. It involves the use of the P. putida PYRC Δung strain engineered to have its pyrF gene (the prokaryotic counterpart of yeast URA3) flanked by a T7 promoter (PT7) and a T7 terminator. Inactivation of pyrF leads to both uracil auxotrophy and a selectable resistance to antibiotic 5-fluororotic acid (5FOA).39 The lack of the ung gene, encoding the repair enzyme uracil-DNA glycosylase (UDG) in the test strain, is known to increase the frequency of C → T mutations generated by cytosine deaminases.14,40 On this background, the DNA sequence of the pmCDA1-RNAPT7 hybrid was cloned into pSEVA23841 (Table 1), a broad host range vector carrying a standardized XylS/Pm expression module,30 preceded by a canonical ribosomal binding site (5′-GGAGGA-3′). The resulting plasmid named pSEVA238-pmCDA1 (Table 1; Supplementary Figure S1B) was transformed into P. putida PYRC Δung to assess the frequency of appearance of 5FOA-resistant (5FOAR) mutants with and without inducer (3mBz) as a proxy of base changes efficacy through the pyrF DNA sequence. The result of this initial experiment is shown in Supplementary Figure S1C. While cultures of cells bearing pSEVA238-pmCDA1 indeed generated 5FOAR colony counts well above the spontaneous rate of tolerance to the antibiotic,38 addition of 3mBz to the cultures made no difference in the numbers, suggesting that residual expression of the pmCDA1-RNAPT7 saturated the ability of the hybrid to cause mutations through pyrF. Although the specific reason for this phenomenon was not investigated further, this unwelcome behavior could have originated from a minor escape of the Pm promoter.30 In order to tackle this possibility, we upgraded the same expression system by merging it with a digitalizer circuit that brings down transcription below any biologically detectable level. The rationale and components of such a module (explained in detail elsewhere24) secures a nil basal level of expression. This affords toxins like colicin E3—one molecule of which suffices to kill a cell—to be altogether shut down in E. coli CC118 and P. putida KT2440 carrier strains. The construct for digitalization of pmCDA1-RNAPT7 production (outlined in Supplementary Figure S2A) checks the gene of interest at both the transcription and the translation level, thereby lowering its expression to virtually zero under noninduced conditions. Assembly of the DNA parts of the system in vector pSEVA238D (Table 1) originated pSEVA238D-pmCDA1 (Supplementary Figure S2A; Table 1). This plasmid was passed to the same reporter strain P. putida PYRC Δung strain, and the upgraded expression device tested for 3mBz-induction of mutagenic activity of pmCDA1-RNAPT7 on pyrF. As shown in Supplementary Figure S2B, we detected again the same level of 5FOAR mutants in cultures induced or not with 3mBz. This unexpected result suggested that a sufficient number of molecules of the sliding base editor were expressed at all times, activating the system’s mutagenic capacity even without induction of the Pm promoter. Since an active form of pmCDA1-RNAPT7 appears to be produced even without induction of the specific expression system—either on its own or when combined with the associated digitalization circuit—it became plausible that transcription and translation of the hybrid could stem from unidentified cryptic signals within the cognate DNA sequence. This possibility led us to abandon further attempts to control transcription of pmCDA1-RNAPT7 and instead focus on implementing a different regulatory layer aiming at overriding noncanonical expression signals, as explained below. Table 1Strains and Plasmids Used in This StudystraindescriptionreferenceE. coli CC118cloning host. genotype: F- Δ(ara-leu), araD, ΔlacX74, galE, galK, phoA, thi1, rpsE, rpoB, argE (Am), recA1(53)E. coli HB101helper strain for conjugation. genotype: F– λ– hsdS20(rB– mB−) recA13 leuB6(Am) araC14 Δ(gpt-proA)62 lacY1 galK2(Oc) xyl-5 mtl-1 thiE1 rpsL20 gln X44B(AS)B(54)P. putida EM42cell factory strain derived from P. putida KT2440(55)P. putida PYRC ΔungP. putida EM42 PYRc derivative with full deletion of unggene(38)P. putida KT2440wild type strain. P. putida mt-2 derivative cured ofpWW0 TOL plasmid(56)P. putida KT2440 ΔungP. putida KT2440 derivative with full deletion of ung gene.(57)plasmiddescriptionreferencepSEVA238cloning vector; oriV (pBBR1); standard multiple cloning site; KmR(41)pSEVA238-pmCDA1pSEVA238 derivative with the pmCDA1-RNAPT7 fusion from pSEVA221pmCDA1-T7RNAP and a canonical RBS (5′- AGGAGGA 3′) cloned into EcoRI/SpeI sitesthis workpSEVA238DMpSEVA238 derivative including the digitalizer module (lacIq gene, translational coupler, and sRNA targeting lacIq), and msfGFP as the GOI; oriV(pBBR1); KmR(24)pSEVA238D-pmCDA1pSEVA238DM derivative with pmCDA1-RNAT7 as a NheI/EcoRI insert; oriV(pBBR1); KmRthis workpSEVA238D-pm3TpSEVA238DM derivative with pmCDA1-RNAP T7 fusion with 3 amber codons (in place of Y16, Y34 and Y52 amino acids) in the N-terminus region of pmCDA1this workpSEVA238-pmCDA3TpSEVA238 derivative with same insert than pSEVA238D-pm3T and a canonical RBS (5′- AGGAGGA – 3′)this workpSEVA4211-supFpSEVA4211-gfp derivative with the sequence for the expression of the suppressor tRNA supF cloned into EcoRI/SpeI sitesthis workpSEVA651standard SEVA expression vector; oriV(RSF1010); GmR(41)pSEVA651-PT7pSEVA651 derivative with a T7 promoter cloned into PacI/AvrII restriction sitesthis workpSEVA247Rstandard SEVA expression vector; oriV(pRO1600/ColE1); mCherry gene as cargo; GmR(41)pTargetpSEVA651-PT7 derivative with a synthetic T7 terminator (T1644), a canonical RBS (5′-AGGAGGA-3′) and an mCherry gene cloned into HindIII/SpeI restriction sites.this workpSEVA4211-GFPpSEVA427 derivative with ChnR-PChnB repressor/promoter pair cloned as PacI/AvrII insertthis workpT-GFPpTarget derivative. RBS and gfp fragments cloned from pSEVA4211-GFP into EcoRI/XbaI restriction sites of pTarget.this work Strains and Plasmids Used in This Study Amber Suppressor tRNA supF-Mediated Control of Sliding Base Editor Activity A phenomenon long known in prokaryotic genetics is that stop codons UAG (amber), UGA (opal), and UAA (ocre) can be each deciphered instead as amino acid residues by so-called suppressor tRNAs. Such suppressors enable the translation machinery to produce a whole-length protein that would otherwise be halted by the stop signal.42 This concept was harnessed early in the history of synthetic biology for engineering an AND logic gate. Such an earlier design25 relied on effector-dependent expression of an RNAPT7 gene containing two amber stop codons and the separate, inducible transcription of an amber suppressor tRNA. As a result, the RNAPT7 was translated and activated only when both inducers were present, allowing the device to initiate transcription from a PT7 positioned upstream of a reporter gene. That this system was described to display a virtually digital AND behavior inspired us to try a similar approach for precise control of pmCDA1-RNAPT7 activity (Figure 1A). To this end, we first generated the plasmid pSEVA238-pmCDA3T (Table 1; Figure 1B). This construct is identical with pSEVA238-pmCDA1 (Table 1; Supplementary Figure S1A), except that the sequence of the pmCDA1-RNAPT7 gene was punctuated by 3 changes of tyrosine (Y) codons (TAT) to amber stop codons (TAG) at corresponding positions 16, 34, and 52 of the primary sequence (see rationale in the Materials and Methods section), resulting in an inoperative pmCDA1-RNAPT7 (Figure 1B). The second component of the system was plasmid pSEVA4211-supF (Table 1; Figure 1C) which encodes the expression of the amber suppressor tRNA supF upon induction with cyclohexanone (Cyc). Specifically, supF manages to overcome UAG stop codons by introducing Tyr residues at the corresponding protein sites during translation.43 Once compatible plasmids pSEVA238-pmCDA3T and pSEVA4211-supF were ready, they were cotransformed into the reporter strain P. putida PYRC Δung and the frequency of 5FOAR mutants with and without the inducers (i.e., 3mBz and/or Cyc) grossly tested in a drop dilution assay as before. The results shown in Supplementary Figure S3 were the first indication that the thereby engineered AND gate (Figure 2A) could be instrumental in controlling expression of the sliding base editor. Strategy for controlling pmCDA1-RNAPT7 activity in vivo with suppressor tRNAs. (A) Schematic representation of how the RNA of a UAG-punctuated pmCDA1-RNAPT7 gene can be expressed as a full-length protein upon the concourse of a suppressor tRNA (e.g., supF). Note that the lack of translation of the pmCDA1 part of the chimaera may still allow a residual expression of the RNAPT7 due to cryptic internal ribosome entry sites.21 (B) Three tyrosine (Y) codons (TAT) of the pmCDA1-RNAPT7 at positions 16, 34, and 52 of the translated fusion protein were mutated to amber stop codons (TAG) and the DNA segment placed in 3mBz-inducible expression plasmid pSEVA238-pmCDA3T. (C) A second construct (pSEVA4211-supF) was assembled for transcribing suppressor tRNA supF upon induction with cyclohexanone (Cyc). When cells bearing both pSEVA238-pmCDA3T and pSEVA4211-supF are simultaneously exposed to 3mBz and Cyc, supF expression enables translation of full-length pmCDA3T-RNAPT7—otherwise interrupted by the amber mutations. This generates a functional sliding base editor that recognizes the T7 promoter (PT7) to start transcription and cause mutations in its course though the target pyrF gene. Sketch and characterization of the logic AND gate implemented by pSEVA4211-supF and pSEVA238-pmCDA3T to regulate the generation of 5FOAR mutants in P. putida PYRc Δung. (A) Formation of a logic gate with two inputs and one output. (B) The frequency of 5FOA-resistant clones was calculated after 16 h of induction by dividing the CFU count in M9/Uracil/5FOA plates by CFUs in LB plates (viability) and normalizing figures to 109 viable cells. Optimization of In Vivo Activity of pmCDA3T-RNAPT7 Activity In order to identify the circumstances where the regulated mutagenic action of pmCDA3T-RNAPT7 could be unleashed at its best, we next considered the effect of different induction times in the final number of 5FOAR mutants. The outcome of this experiment (Supplementary Figure 3B) showed a remarkable increase in such numbers when the cultures were induced for at least 16 h. The data for that time of induction plotted in Figure 2B not only proved the AND logic of the genetic device, but it also revealed that the total frequency of 5FOAR mutants was almost at the same level as the previously used system with no regulation in the same strain, i.e., 107 vs 5 × 106 5FOAR mutants per 109 viable cells.38 On the other hand, the frequency of mutants with no induction was at the level of the control strain with no plasmids (103 5FOAR mutants per 109 viable cells). That the effect of Cyc addition-alone (inducing supF) was significantly higher than that of 3mBz (for pmCDA3T-RNAPT7 transcription) suggested that—as entertained above—some mRNA of the hybrid gene is generated even in the absence of any upstream promoter. In any case, these results sanctioned the apparent worth of adopting the AND logic strategy to regulate the activity of the sliding base editor. In a further step toward optimization of the system, we pursued making the device simpler through having all functional parts of the logic gate in just one plasmid (instead of having them split into two replicons) by combining supF and pmCDA3T-RNAPT7 expression cassettes in one single plasmid. To this end, the DNA segment of pSEVA238-pmCDA3T containing the sequences of XylS/Pm, RBS and pmCDA3T-RNAPT7 was PCR-amplified and cloned into SwaI-linearized pSEVA4211-supF, thereby generating pSuPm3T (Figure 3A). This plasmid was then transformed into P. putida PYRC Δung to determine whether the system kept the tight regulation of pmCDA3T-RNAPT7 activity observed with the two-plasmid system. The results of Figure 3B showed that while the basal level of mutagenic activity was maintained at the same low levels as before (Figure 2), the frequency of 5FOAR colonies from fully induced cultures dropped considerably but still maintaining a good level of appearance. There might be different explanations for such a loss of activity, including positional effects and a lower-copy number of plasmid pSuPm3T (oriV of RK2) in comparison to the medium-copy number of pSEVA238-pmCDA3T (oriV of pBBR1). One way or the other, implementation of the AND gate in either one or two plasmids kept the stringent regulation of the system while offering a choice of different mutagenesis levels for specific purposes. Implementation of the logic AND gate encoded in single plasmid pSuPm3T to control generation of 5FOAR mutants in P. putida PYRC Δung. (A) Sketch of the organization of plasmid pSuPm3T, with an indication of all functional parts borne by the construct (not to scale). This plasmid was obtained after cloning the expression cassette composed by the pmCDA3T-RNAPT7 fusion, RBS and XylS/Pm promoter from pSEVA238-pmCDA3T into SwaI-linearized pSEVA4211-supF. The resulting plasmid has a low copy-number origin of replication (oriV RK2), a streptomycin-resistant gene (SmR), and the expression cassette composed by a ChnR/PChnB regulator/promoter pair in front of supF, in addition to the previously cloned pmCDA3T-RNAP T7 cassette. Three different terminators (T500, T1TE, and T0) isolate transcription from both expression systems to avoid interference between them. (B) Regulation of mutagenic activity delivered by pSuPm3T. The frequency of appearance of 5FOAR-resistant clones was calculated under the same conditions specified in the legend to Figure 2. Dynamics and Parametrization of pmCDA3T-RNAPT7 Activity Subject to AND Logic The fact that most 5FOAR mutants appeared when cells were well into stationary phase was intriguing, as by that time there are way less resources for transcription and DNA replication. To make sense of this phenomenon, we set out to monitor separately pmCDA3T-RNAPT7 production and mutagenic activity over time. To this end, we picked a GFP reporter as the target gene (instead of pyrF) so that fluorescence emission elicited by the RNAPT7 moiety of the hybrid protein could be followed in single cells along with the mutagenic activity of the pmCDA3T module. For doing this, we constructed plasmid pT-GFP (Table 1; Figure 4A) in which a T7 promoter drives expression of a GFP reporter gene, which becomes thereby the target of the pmCDA3T-RNAPT7 fusion protein. Additionally, the pT-GFP construct included a synthetic T7 terminator44 located downstream of the GFP gene followed by a promoterless mCherry gene preceded by an RBS sequence (Figure 4A). This downstream segment was incorporated for monitoring the activity of pmCDA3T-RNAPT7 beyond the termination signal as well as assessing the potential of off-target mutations in the remainder of the plasmid sequence (see below). The thereby designed pT-GFP plasmid was then cotransformed into P. putida KT2440 Δung with either pSuPm3T or the insetless vector pSEVA238 (Table 1) as a negative control. Mere visual inspection of the cotransformation plates revealed unexpected colony types under blue light. While all cotransformant colonies with pSEVA238+pT-GFP were nonfluorescent, those with pSuPm3T + pT-GFP displayed fluorescence even without any inducer. This necessarily meant that the RNAPT7 moiety of the pmCDA3T-RNAPT7 fusion was being expressed despite being preceded by a base editor punctuated with 3 stop codons. This was likely the result of illegitimate transcription and translation of an active polymerase from signals borne by the nontranslated sequence of the hybrid gene. To examine this possibility, we run an equivalent cotransformation of pT-GFP with pSEVA238-pmCDA1 (Table 1; Supplementary Figure S1), which expresses the unchanged pmCDA1-RNAPT7 under the direct control of the Pm promoter, i.e., subject to a YES logic, not AND. In this case, cotransformants displayed a variety of different types of fluorescence (green, red, none), which kept on diversifying after restreaking individual colonies on new plates (Figure 4B). This behavior suggested that both the RNAPT7 and pmCDA1 components of the protein fusion were expressed and active, even without induction. That most colonies were fluorescent indicated that PT7 was triggered by its cognate RNAPT7. Also, there was significant variability in fluorescence, which plausibly stemmed from the base-editing activity of pmCDA1. These results with the pT-GFP+pSEVA238-pmCDA1 cotransformants were fully consistent with those mentioned earlier (Supplementary Figure S1) on the occurrence of mutagenesis on the pyrF gene of P. putida PYRC Δung transformed with pSEVA238-pmCDA1 without induction with 3mBz. Also, these data abound the notion that expression of the pmCDA1-RNAPT7 hybrid does occur from unknown internal transcription and translation signals borne by the cognate DNA sequence of the hybrid. (A) Architecture of pT-GFP. The T7 promoter (PT7) controls the expression and mutagenesis of a gfp gene by pmCDA3T-RNAPT7. A supereffective synthetic T7 terminator (T16)4 was included to isolate transcription/mutagenesis of the gfp DNA. This is then followed by canonical RBS and an mCherry gene downstream of T16 to measure the efficacy of the terminator. (B) Phenotypes of P. putida KT2440 Δung cells transformed with pT-GFP and pSEVA238-pmCDA1. After cotransformation of pT-GFP and pSEVA238-pmCDA1, erratic fluorescence phenotypes could be identified in colonies on the plate, suggesting continuous variations. Streaked out colonies evolved into different colors and intensities. To track the evolution of both regulated and nonregulated mutagenic devices over time, individual colonies from the cotransformation plates without the inducer discussed above (Figure 4B) were selected for further analyses. These included [i] separate fluorescent and nonfluorescent cotransformants bearing pSEVA238-pmCDA+pT-GFP, thereafter called, respectively, clones F and NF, [ii] a fluorescent cotransformant pSuPm+pT-GFP, and [iii] a control nonfluorescent cotransformant pSEVA238+pT-GFP. These colonies were subsequently investigated using flow cytometry. For this, each of the clones was grown to an OD600 = 0.2 (t = 0 h), after which cultures were split and added or not with 1 mM inducers Cyc and 3mBz. The results of the ensuing cytometry analyses of culture samples at 2 and 20 h after induction are shown in Figure 5. Evolution of mutagenic activity in real time based on fluorescence patterns of single cells of P. putida KT2440 Δung. Results show the fluorescence of cells bearing pT-GFP along with the other plasmids indicated. Note that in the case of cotransformant pT-GFP+ pSEVA238-pmCDA, we analyzed both a nonfluorescent (NF) and a green fluorescent (F) clone isolated from the cotransformation plate. For the experiment shown, cells were grown in M9-citrate medium until OD600 = 0.2 (t = 0 h), after which cultures were split in two and added or not with 1 mM Cyc and 3mBz. Aliquots were taken after 2 and 20 h and analyzed through flow cytometry as indicated in Materials and Methods. (A)–(D) Representative examples of fluorescent emissions of the specified cotransformants counted in approximately 80k cells. (E)–(H) The results of the whole of the samples (n=2 replicates) examined in a numerical form, i.e., percentage of green vs nonfluorescent cells. In the control condition (cotransformant pT-GFP+pSEVA238, Figure 5A), cells showed a very small level of fluorescence regardless of the inducers. In contrast, induced cotransformants pT-GFP+pSuPm3T (Figure 5D) first produced mostly fluorescent cells but later diversified to cells with very different emission levels, including a good share of nonfluorescent bacteria. The same cytometry data are shown in Figure 5E–H in terms of % of fluorescent and nonfluorescent cells. These results were easy to interpret as the result of the mutagenic action of the conditionally expressed pmCDA3T-RNAPT7, and they matched the earlier data generated when examining the emergence of 5FOAR mutants in P. putida PYRC Δung transformed with the same plasmid pSuPm3T (Figure 3). Finally, cells derived from either fluorescent (F) or nonfluorescent (NF) colonies out of the cotransformation of pT-GFP with pSEVA238-pmCDA turned both quickly toward permanent nonfluorescence. This is not unexpected as the constant exposure of the target GFP gene of pT-GFP to the mutagenic action of pmCDA3T-RNAPT7 may quickly inactivate the bulk of the reporter borne by the whole population. In conclusion, it seems that [i] pmCDA1-RNAPT7 manages to get expressed effectively through signals embedded in its cognate DNA sequence, even if lacking upstream transcription; [ii] in the absence of supF tRNA, some polymerase activity is produced without translation—let alone activity—of the leading pmCDA3T domain of the fusion; and [iii] engineering of the thereby described AND gate to regulate expression of the sliding base editor enables effective control of its mutagenic activity in time and space. Efficacy of the Regulated Sliding Base Editor in Hosts with UDG Repair Activity While ung mutants lacking the enzyme uracyl DNA glycosylase (UDG) are known to increase frequency of replacements C → T caused by cytosine deaminases,16,38,45 deletion of the gene in the host strain or coexpression of base editors with UDG-inhibiting peptides add complication to the experimental setup. Given the quite high efficiency of the pmCDA3T-RNAPT7-based diversity generating device described above in the tested Δung strains, we wondered whether we could still get manageable diversification figures in the absence of the repair enzyme. To test this possibility, we generated cotransformants of reporter construct pT-GFP with pSuPm3T, pSEVA238, or pSEVA238-pmCDA, as before, but using the ung+ wild-type strain P. putida KT2440 (Table 1) as the host of the plasmids. The appearance of the cotransformant colonies was identical to those obtained when P. putida PYRC Δung strain was used as the recipient (see above). Namely, all the colonies were nonfluorescent in the control pT-GFP+ pSEVA238 clones; fluorescent with pT-GFP+ pSuPm3T; and a mixture of both fluorescent and nonfluorescent clones was observed when cells were cotransformed with pT-GFP-pSEVA238-pmCDA. As before, we picked for further analyses the same type of colonies from each condition that were analyzed in the Δung background. Cultures of each of them were then set up and processed under identical conditions as those in the legend of Figure 5—excepting by the host strain. The results of Figure 6 exposed again the regulatory effect of the AND gate borne by plasmid pSuPm3T and the autonomous expression of the T7 RNA polymerase moiety of the pmCDA3T- RNAPT7 fusion—the mutagenic activity of which could be seen only upon coinduction with Cyc and 3mBz (Figure 6D,H). Evolution of mutagenic activity on fluorescence patterns of single cells of wild-type ung+P. putida KT2440, endowed with UDG repair activity, i.e., the data shown was generated in an experiment run in the same conditions specified in the legend of Figure 5. Also, while control cultures from clones with pT-GFP+pSEVA238 remained mostly nonfluorescent, both NF and F-derived cultures of pT-GFP+pSEVA238-pmCDA1 clones followed an erratic evolution, quite different from the equivalent cotransformants in the Δung host strain (Figure 5). This was not altogether surprising as the reporter gene in these clones is subject to a continuous, not regulated, mutagenic regime. Additionally, the presence of UDG could repair some of the mutations introduced in the target gene, restoring the fluorescence of the reporter. This could be an explanation of the higher number of fluorescent cells in Figure 6C,G in comparison to Figure 5. The most noteworthy insight from Figure 6 emerges when comparing panels D and H with those in Figure 5. These comparisons reveal that, at the longest induction times, the mutagenic activity of the pmCDA3T-RNAPT7 hybrid is similarly effective in host strains both with (Figure 6) and without (Figure 5) UDG. This characteristic of the system is promising as it enables a simplified application of the described device across a broader range of potential bacterial hosts. Termination pmCDA-RNAPT7 Forward Action with Optimized Sequence T16 As a sidelight of the experiments above, we inspected also the end of the action of the base editors—bearing suppressible amber codons or not—beyond the downstream target gene, i.e., GFP in plasmid pT-GFP caused by the introduction of an optimized synthetic T7 terminator (T1644) just downstream of the sequence of the reporter product. T16 is described to be among the strongest signals able to virtually abolish the transcriptional progression of RNAPT7. Therefore, it became an appealing genetic asset to isolate diversification of target sequences from downstream DNA of the same replicon—whether plasmid-encoded or chromosomally borne. Note that next to the T16 sequence downstream of GFP in pT-GFP (Figure 4A), there is a promoterless mCherry gene, which, due to the architecture of the device, can report cessation or not of the progress of the polymerase beyond such termination. To get some insight into this question, we simply measured the collection of cultures of cotransformants already inspected for green fluorescence shown in Figure 5 (plasmids host ung-minus) and Figure 6 (plasmids hosts ung-plus) under the same growth and induction conditions, but this time recording the red channel of the cytometer for detection of mCherry expression. The results are shown in Supplementary Figure S4 (for cells with a Δung genetic background) and Supplementary Figure S5 (for ung+ clones), which basically behaved the same in this respect. The degree of termination caused by T16 under all conditions was quite good in most cases (see panels A to C and E to G of Supplementary Figures S4 and S5), as red fluorescence levels observed in cotransformants of pT-GFP with pmCDA1-RNAPT7 were indistinguishable of those devoid of RNAPT7 activity. In contrast, cells with pT-GFP and pSuPm3T (panels D and H of Supplementary Figures S4 and S5) indeed showed a transient increase of red fluorescence mostly after 2 h of full induction. Note that under such conditions, the complete functional pmCDA-RNAPT7 is effectively expressed and therefore mutations may occur in the T16 terminator and the mCherry sequence as well. This may explain why, at 20 h, the signal decreases as compared to shorter induction times. In any case, users of pmCDA-RNAPT7 chimeras for targeted DNA diversification (the ones hereby described or any other) might be aware that T7 terminators, even if inserted multiple times,17 do not guarantee complete protection of downstream sequences and other strategies must be considered to bring about such an isolation.16 Conclusions In this study, we document the widespread challenges in controlling the spatial and temporal expression of base editors, particularly those involving a chimeric fusion between cytosine deaminase and T7 RNA polymerase.16 We propose an approach to address this issue by shifting from a YES amplifier logic to an AND logic architecture, where two inputs (e.g., two chemical inducers) are necessary for the effective production of the active hybrid protein. Suppressor tRNAs have previously been used in various ways to eliminate leaky gene expression in bacteria.25,46 In our case, we subjected the expression of the base editor (which contains three amber stop codons) to the conditional production of the supF tRNA. This strategy effectively maintained the mutagenesis activity of pmCDA-RNAPT7 at a background level in the absence of inducers. Throughout this work, we uncovered several insights into the behavior of the system that could provide valuable information for other mutagenesis systems facing similar regulatory nuisances. Specifically, we observed that the DNA sequences of the pmCDA-RNAPT7 variants—or parts of them—exhibited inherent expression capabilities, even without recognizable transcription and translation signals. This included the potential for translation initiation at noncanonical sites, likely due to the presence of cryptic promoters or internal ribosome entry site (IRES)-like elements21 that can interact with the host’s transcription and translation machinery in unexpected ways. Furthermore, our data underscore the difficulty of completely halting a base editor that moves along a DNA sequence with a T7 polymerase, even when specific terminators are present. While these noncanonical expression control mechanisms may help bacteria adapt to diverse environmental conditions, they also complicate the genetic engineering of synthetic biology devices, such as those discussed in this article. In our example, we overcame the uncertainties in expressing the pmCDA-RNAPT7 protein by applying an artificial level of control based on an AND logic gate. We believe that the resulting system, whether in a two-plasmid or single-plasmid format, could be a valuable addition to the toolkit for engineering P. putida and other Gram-negative bacteria. Materials and Methods Bacteria, Growth Media, and General Methods E. coli and P. putida strains used in this work are listed and described in Table 1. P. putida PYRC Δung is a derivative of P. putida EM4247 as reported elsewhere.38 Bacteria were regularly grown in liquid or solid (1.5% agar) LB medium at 37 °C for E. coli and 30 °C for P. putida. When necessary, different antibiotics were supplemented at the following concentrations: kanamycin (Km; 50 μg/mL), gentamycin (Gm; 10 μg/mL for E. coli and 15 μg/mL for P. putida), and/or streptomycin (Sm; 50 μg/mL for E. coli and 100 μg/mL for P. putida). Moreover, cyclohexanone (Cyc) and 3-methyl benzoate (3mBz) were added to liquid media at a final concentration of 1 mM when necessary. In the cases indicated, P. putida was grown in M9 minimal medium48 supplemented with sodium citrate at 0.2% (w/v) as the sole carbon source. 5-Fluoroorotic acid (5FOA, 250 μg mL–1) and/or uracil (ura, 20 μg mL–1) were supplemented to minimal medium plates as required. Plasmids were constructed with either classical procedures (i.e., restriction cloning48) or Gibson assembly.49,50 PCR reactions were run with Q5 High Fidelity polymerase (New England Biolabs, Ipswich, MA, USA) according to the manufacturer’s instructions. Diagnostic PCRs were performed using Green MasterMix (BioTools, Madrid, Spain) and taking fresh single colonies as the starting material. The sequences of all PCR primers are listed in Supplementary Table S1. Triparental matings for mobilization of plasmids from E. coli strains to P. putida were performed with a simplified protocol.51 For CFU counting, culture dilutions were plated on either LB or M9 minimal media with 0.2% citrate and adequate supplements for isolating individual clones, as indicated in each case. Plasmid Construction Plasmids constructed and used in this work are listed in Table 1. For generation of pSEVA238-pmCDA1, the entire segment DNA encompassing the fused DNA sequences of pmCDA1 (cytosine deaminase) and the RNAPT7 was amplified from pSEVA221-pmCDA1-T7RNAP16 with primers pS238-pmCDA-F and pS238-pmCDA-R (Supplementary Table S1). The resulting fragment was then cloned by Gibson assembly49,50 into EcoRI/SpeI-digested pSEVA238,52 thereby producing the pursued plasmid pSEVA238-pmCDA1, in which transcription of the chimeric pmCDA1-RNAPT7 protein is expressed under the control of the 3mBz-inducible XylS/Pm regulatory system.30 As an alternative, the same segment with the fused DNA sequences of pmCDA1-RNAPT7 was amplified from pSEVA221-pmCDA1-T7RNAP with primers pS238D-pmCDA-F and pS238D-pmCDA-R, and the resulting DNA cloned in pSEVA238D24 as the NheI/EcoRI fragment, yielding pSEVA238D-pmCDA1. In this case, pmCDA1-RNAPT7 is still inducible by 3mBz, but the expression is subject to a digitalizer circuit (Supplementary Figure S2A) expected to suppress leaky expression. To build pSEVA4211-supF, a DNA segment with the whole supF sequence along two downstream terminators (T500 and T1/TE) was synthesized by IDT (Integrated DNA Technologies, Inc. Coralville, Iowa, USA), amplified with primers p4211-SupF_fwd and p4211-SupF_rev and cloned into EcoRI/SpeI-digested pSEVA4211-gfp by Gibson assembly—thereby replacing the reporter gene by the supF tRNA sequence. pT-GFP construction involved several steps. First, pSEVA651-PT7 was generated by cloning a canonical T7 promoter (PT7, 5′-TAATACGACTCACTATAGGG-3′) into PacI/AvrII-linearized pSEVA651. Later, the fragments encoding the synthetic T7 terminator T1644 and the mCherry gene along a canonical RBS (5′-GGAGGA-3′) were cloned through Gibson assembly into HindII/SpeI-digested pSEVA651-PT7, yielding pTarget plasmid. Finally, pT-GFP was engineered by first isolating a fragment encoding the RBS and a gfp gene from pSEVA4211-GFP by EcoR/SpeI digestion and then ligating this fragment into EcoRI/XbaI-linearized pTarget with T4 DNA ligase (Roche, Basilea, Switzerland). Engineering an supF tRNA-Suppressible pmCDA1-RNAPT7 Sequence In order to engineer 3 supF tRNA-suppressible amber stop (UAG) signals through the pmCDA1-coding sequence, we first examined the location of the 10 Tyr present in pmCDA1. Four of them turned out to be located in the primary sequence before the catalytic site of the cytosine deaminase (the HAE-PCADC motif26), which spans residues 66–68 and 96–100. On this basis, Tyr codons at positions 16, 34, and 52 were replaced by TAG triplets for preventing translation of the downstream sequence. To build an expression plasmid for such a variant pmCDA13xTAG-RNAPT7, i.e., equivalent to pSEVA238-pmCDA1 but punctuated with 3 amber codons in the said positions (Figure 1B), a DNA segment encompassing the pmCDA1-RNAPT7 fusion but entered with cognate Tyr to stop codon changes was synthesized (Duolix, Venice, Italy). Primers pS238D-pmCDA-F and pPm1TAG-1 were used to amplify the corresponding segment with the three changes indicated before and reassembled with the rest of the wild-type gene fusion in pSEVA238D-pmCDA1. The resulting plasmid (pSEVA238D-pm3T) was PCR-ed with primers pS238-pmCDA-F and pS238-pmCDA-R to amplify pmCDA13xTAG-RNAPT7 cassette, which was then ligated back into EcoRI/SpeI-digested pSEVA238. The result of these operations was pSEVA238-pmCDA3T, which is identical to pSEVA238-pmCDA1 but by the 3 TAG codons that entered in the pmCDA1 portion of the fusion. Finally, plasmid pSuPm3T, with all elements of the engineered logic gate in a single plasmid, was constructed by first amplifying the DNA portion of pSEVA238-pmCDA3T comprising xylS/Pm → RBS→ pmCDA13xTAG-RNAPT7 with primers pSuPm3T_F and pSuPm3T_R and cloning the resulting fragments into SwaI-linearized pSEVA4211-supF through Gibson Assembly. All built plasmids were sequenced for the verification of their accuracy. Mutagenesis Tests Plasmids indicated in each case were separately transformed into the P. putida strains of choice, and the resulting clones were grown overnight at 30 °C in liquid LB with adequate corresponding antibiotics to secure plasmid retention. Samples were then diluted in the same medium to an OD600 of 0.05 and regrown with shaking to an OD600 between 0.2 to 0.3. Then, aliquots of the same volume were split and supplemented with the corresponding inducers (Cyc and/or 3mBz), when stated. Cultures were induced for different times (2, 4, 16, or 20 h) depending on the specific experiment. After induction, cells were washed with PBS 1×, and serial dilutions of 100 μL plated on LB or M9/citrate/uracil medium (for viability) or M9/citrate/uracil/5FOA (for selection of 5FOAR mutants) were performed. After overnight growth, CFUs were counted from each plate, and the mutation rate was calculated by dividing the count of CFU in M9/citrate/uracil/5FOA plates by CFU in the viability plates and normalizing figures to 109 viable cells. Fluorescence Flow Cytometry Cells carrying pT-GFP along pSEVA238, pSuPm3T, or pSEVA238-pmCDA1 were grown on filtered LB supplemented with Gm and Sm (pSuPm3T) or Km (pSEVA238 and pSEVA238-pmCDA1) at 30 °C. Stationary-phase cultures were diluted in fresh-filtered media to an initial OD of 0.05 and incubated until they reached an OD600 of ∼0.2. Then, each culture was split so that one-half could be induced with 1 mM of Cyc and 3mBz at 30 °C. This time point was considered as t = 0 h, and one aliquot was harvested and analyzed to set the basal level of fluorescence of each sample. Finally, 1 mL aliquots were retrieved at two different time points (2 and 20 h). All samples were spun down in a tabletop centrifuge (13,000 rpm for 1 min), washed twice with filtered PBS 1×, and diluted to an OD < 0.1. Cells were immediately analyzed via flow cytometry using a MACS-QuantTM VYB cytometer (Miltenyi Biotec GmbH). GFP was excited at 488 nm, and the fluorescence signal was recovered with a 525/50 nm band-pass filter. mCherry was excited at 561 nm, and the signal was recovered with a 586/615 nm band-pass filter. At least 80,000 events were analyzed for each sample. Cytometry data was processed with FlowJoTM software (www.flowjo.com). GFP and mCherry quantification were tested by first gating cells in a side scatter against forward scatter plot to analyze single events. Then, GFP or mCherry-associated fluorescence was assessed. This experiment was repeated twice with each bacterial background. Graphs were plotted using GraphPad Prism 6 or Microsoft Excel v2405.