Hypermutation of specific genomic loci of Pseudomonas putida for continuous evolution of target genes.

The ability of T7 RNA polymerase (RNAPT7 ) fusions to cytosine deaminases (CdA) for entering C➔T changes in any DNA segment downstream of a T7 promoter was exploited for hyperdiversification of defined genomic portions of Pseudomonas putida KT2440. To this end, test strains were constructed in which the chromosomally encoded pyrF gene (the prokaryotic homologue of yeast URA3) was flanked by T7 transcription initiation and termination signals and also carried plasmids expressing constitutively either high-activity (lamprey's) or low-activity (rat's) CdA-RNAPT7 fusions. The DNA segment-specific mutagenic action of these fusions was then tested in strains lacking or not uracil-DNA glycosylase (UDG), that is ∆ung/ung+ variants. The resulting diversification was measured by counting single nucleotide changes in clones resistant to 5-fluoroorotic acid (5FOA), which otherwise is transformed by wild-type PyrF into a toxic compound. Although the absence of UDG dramatically increased mutagenic rates with both CdA-RNAPT7 fusions, the most active variant - pmCDA1 - caused extensive appearance of 5FOA-resistant colonies in the wild-type strain not limited to C➔T but including also a range of other changes. Furthermore, the presence/absence of UDG activity swapped cytosine deamination preference between DNA strands. These qualities provided the basis of a robust system for continuous evolution of preset genomic portions of P. putida and beyond.

Introduction

Generation of genotypic and phenotypic variability in bacteria typically involves adoption of in vivo DNA variation methods which – in contrast to other schemes – enable mutation, expression and selection steps to be run as a continuous process (Simon et al., 2019). Whether occurring spontaneously or stimulated by exogenous agents (Miller et al., 1977; Brouwer et al., 1981; Akkaya et al., 2018) or endogenous mutation triggers (Muteeb and Sen, 2010), modifications of the DNA habitually occur throughout the whole genome of the organism at stake. While this can be beneficial for evolving complex phenotypes (Ibarra et al., 2002; Klein‐Marcuschamer et al., 2009; Liu et al., 2020), most haphazard mutations in a bacterial genome are either neutral or plainly detrimental. Instead, in order to focus in vivo diversification into a limited number of sites/genes of the genome, special methods need to be implemented. A successful one is that named MAGE (multiple automated genome engineering) which is based on the Red‐recombination system of lambda phage and the delivery of mutagenic ssDNA oligonucleotides to Escherichia coli cells (Wang et al., 2009; Isaacs et al., 2011; Ronda et al., 2016). This system has been later adapted for segment‐specific protein diversification (Nyerges et al., 2018; Al‐Ramahi et al., 2021) and applied to a suite of non‐E. coli bacteria thereof (van Kessel and Hatfull, 2007; van Pijkeren and Britton, 2012; Aparicio et al., 2016). MAGE and related methods rely on recurrent cycles of ssDNA transformation for raising a large enough library of variants, and they thus still depend on user's action for bringing about the desired site‐specific diversification. MAGE alternatives include adoption of E. coli strains expressing an error‐prone DNA polymerase I which enables continuous mutagenesis of genes cloned in plasmid vectors with a ColE1‐type origin of replication (Camps et al., 2003). Conceptually, similar approaches using other genetic devices have also been successfully implemented in eukaryotic cells (Crook et al., 2016; Arzumanyan et al., 2018; Ravikumar et al., 2018). However, strategies for all in vivo continuous evolution of defined segments of bacterial genomes remained parted until the discovery that fusions of cytosine deaminases (CdA) to the bacteriophage T7 RNAP (RNAPT7) caused high mutagenic rates C➔T (and accordingly G➔A) in any DNA sequence located downstream of a T7 promoter (Moore et al., 2018; Alvarez et al., 2020; Chen et al., 2020; Cravens et al., 2021; Park and Kim, 2021). On this basis, a number of genetic platforms exploiting different base editors fused to RNAPT7 have been developed in which the bases of the ssDNA that become exposed along transcription become the substrate of the deamination reaction (Molina et al., 2022). This occurrence has been exploited to set off evolution of delimited DNA segments in vivo without affecting the rest of the genome. Given the imperfect efficacy of T7 phage transcription terminators (T7T) to make RNAPT7 to come off the engaged DNA (Moore et al., 2018; Tarnowski and Gorochowski, 2022), one useful strategy to define the downstream limit of the mutagenized region involves expression of gRNA‐dCas9 complexes for blocking the progression of the polymerase (Alvarez et al., 2020). While the available wealth of results accredits the mutagenic activity of CdA‐RNAPT7 fusions in E. coli, whether same is applicable to other bacterial species and strains of biotechnological interest remains unknown.

In this work, we have investigated the applicability of such a diversity‐generating device to the Gram‐negative bacterium and metabolic engineering platform Pseudomonas putida KT2440 (Nelson et al., 2002; Bitzenhofer et al., 2021). This strain is a pWW0 plasmid‐free derivative of the soil isolate P. putida mt‐2 (Worsey and Williams, 1975), which naturally holds a suite of qualities that make it an adequate host for running strong redox and harsh reactions with a value for industrial and environmental biotechnology (Nikel et al., 2014; Ankenbauer et al., 2020; Weimer et al., 2020). One important characteristic in this respect is the abundance of genes encoding oxido‐reductases able to generate high NAD(P)H, a metabolic trait that facilitates growth in a wide range of aromatic substrates (dos Santos et al., 2004; Kim and Park, 2014) and endows a superior tolerance to different types of stress. Owing to this, the robust redox metabolism of P. putida KT2440 facilitates heterologous expression of biochemical routes that are hardly supported by other bacteria (Zobel et al., 2017). On a practical side, P. putida KT2440 is certified to be a safe and non‐pathogenic bacterium for recombinant DNA experiments (Kampers et al., 2019). Finally, the number of molecular tools available for all types of genetic manipulations in this bacterium is comparable – and in some cases exceeding – to those existing for E. coli (Martin‐Pascual et al., 2021). Having such desirable properties, it is no surprise that P. putida KT2440 and its derivatives have become chassis of choice for a wide range of metabolic engineering undertakings (Martinez‐Garcia et al., 2014; Dvorak and de Lorenzo, 2018; Sanchez‐Pascuala et al., 2019; Martinez‐Garcia et al., 2020).

The work below documents the ability of CdA‐RNAPT7 fusions to boost DNA variability in a preset genomic segment of a P. putida KT2440 derivative delimited by a T7 promoter and a T7 terminator, the efficacy and types of mutations caused under various conditions, and the confinement of the diversification regime to the desired DNA portion of the cell's genetic complement. The results not only verify the performance of the approach in this bacterium, but also provide a complete genetic platform for continuous evolution of specified sectors of the P. putida KT2440 chromosome.

Results and discussion

Rationale for benchmarking activity of CdA‐RNAP fusions in vivo

In order to implement the DNA segment‐targeted mutagenic device shown in Fig. 1 in P. putida and – if effective – parameterize its performance, we started by tailoring dedicated strains for reporting important characteristics of the method. The starting isolate was not strain KT2440 but the genome‐edited variant P. putida EM42 (Martinez‐Garcia et al., 2014; Table S2) that is deleted of the flagellar machinery, which results in higher endogenous levels of NAD(P)H. Furthermore, P. putida EM42 lacks a number of instability determinants, that is the four prophages and the Tn7‐like transposon otherwise encoded in the wild‐type genome. The 702 bp genomic segment bearing the pyrF gene (PP1815, genomic coordinates 2 040 625–2 041 326 in P. putida KT2440; www.pseudomonas.com) was then edited as shown in Fig. S1 to place [i] a T7 promoter (P ) downstream of the pyrF oriented opposite to the course of native transcription and [ii] a T7 terminator (T7T) upstream of the innate promoter of the gene (Fig. 2A). pyrF encodes the essential enzyme orotidine‐5′‐phosphate decarboxylase, which is the prokaryotic counterpart of yeast URA3 (Galvao and de Lorenzo, 2005). The presence of the enzyme can be selected both positively (growth in minimal medium without uracil) and negatively, as it converts 5‐fluoroorotic acid (5FOA) into 5‐fluorouracil, a toxic compound (Boeke et al., 1984). The resulting test strain was called P. putida PYRC (Fig. 2A). One further derivative was then constructed (Fig. 2B) by deleting the ung gene as described in Experimental Procedures. This gene encodes uracil‐DNA glycosylase (UDG), the enzyme that initiates base repair when U residues accidentally occur in the chromosome (Lindahl et al., 1977; Parikh et al., 1998; Schormann et al., 2014). Since cytosine deaminases convert C to U, ung is a key determinant of the mutagenic efficiency of these enzymes (Di Noia and Neuberger, 2002; Rada et al., 2002). The growth in minimal media, 5FOA sensitivity and phenotypic characteristics of the resulting strains such as colony size (P. putida PYRC and P. putida PYRC ∆ung) were indistinguishable from those of P. putida EM42, indicating that the genomic edits had no influence in the physiology of the strains (Fig. S2).

Fig. 1

Schematic representation of the segment‐specific mutagenic regime caused by CdA‐RNAPT7 fusions in vivo.

A. The basic arrangement involves one or more gene of interest –or any DNA sequence thereof – flanked by a T7 promoter and a T7 terminator. In the sketch, the direction of transcription from P is opposite to native gene reading, but can be otherwise if desired. The CdA‐RNAPT7 fusion is then expressed in trans, bound to P and transcription initiated by the polymerase activity of the fusion.

B. As CdA‐RNAPT7 proceeds downstream of the P , the cytosine deamination moiety may find C residues and change them to U counterparts.

C. Mutation occurs as the ssDNA of the transcription bubble exposes C bases, which become substrates of CdA. If the resulting U is not corrected by the repair machinery of the cell, it will base pair with A during the replication process, generating a change from the original C:G pair to a mutated T:A pair. Note that changes can occur, albeit with different frequencies, in the two strands of the DNA. [Colour figure can be viewed at wileyonlinelibrary.com]

Fig. 2

Genetic parts and devices for monitoring action of CdA‐RNAPT7 fusions on the pyrF gene of P. putida.

A. Organization of the genomic regions of interest in test strain P. putida PYRC.

B. Same, P. putida PYRC ∆ung.

C. Organization of pSEVA221 [RNAPT7] alone. The plasmid is endowed with a low‐copy number origin of replication oriV (RK2), an oriT to mediate conjugation, a KmR gene and the TetR/P repressor/promoter pair for transcription of downstream genes.

D. Expression plasmids of the two fusion proteins used in this work: rAPOBEC1‐RNAPT7 and pmCDA1‐RNAPT7. Each cytosine deaminase was fused to the N‐terminus domain of RNAPT7 through a 28 aa flexible linker (84 nt). [Colour figure can be viewed at wileyonlinelibrary.com]

In order to assess the mutagenic action of CdA‐RNAPT7 fusions on the thereby arranged pyrF sequence, strains P. putida PYRC and P. putida PYRC ∆ung were separately transformed with plasmids expressing two different cytosine deaminases fused to the N‐terminus of RNAPT7 through a flexible linker (G3S)7 (Alvarez et al., 2020). Specifically, the CdA moieties of the hybrid proteins were those of rat (rAPOBEC1, low‐activity) and lamprey (pmCDA1, high‐activity). These were borne by pSEVA221 [rAPOBEC1‐RNAPT7] and pSEVA221 [pmCDA1‐RNAPT7], which are described in (Alvarez et al., 2020) and sketched in Fig. 2D. Note that the cargoes inserted in the low‐copy number KmR vector pSEVA221 consist of DNA segments for expression of each of the fusions under the control of a tetR/P system (Bertram and Hillen, 2007). As we pursued a constant supply of the mutagenic source in vivo, we preemptively inspected the behaviour of the transcriptional device in P. putida by means of a gfp reporter gene, which fortunately turned out to be constitutive (Fig. S3). Finally, strains P. putida PYRC and P. putida PYRC ∆ung were transformed also with empty vector pSEVA221 as well as with control pSEVA221 [RNAPT7], which expressed the RNAPT7 under the control of the same tetR/P as in the other plasmids (Fig. 2C). These constructs provided a reference for diagnosing merely spontaneous mutations in the first case and the sheer effect of counter‐transcribing pyrF DNA with an intact RNAPT7 in the second. The strain collection resulting in these procedures shaped the genetic platform on which the site‐focused diversity‐generating device of Fig. 1 was tested in P. putida as referred to below.

Setting the baseline for CdA‐unrelated mutagenesis of

In order to distinguish the onset of 5FOAR clones due to the genuine action of CdAs from those arising from the inborn variability frequency of P. putida and/or high transcription levels from an active T7 promoter, we first focused on strains P. putida PYRC and P. putida PYRC ∆ung transformed respectively with control plasmids pSEVA221 and pSEVA221 [RNAPT7]. These were plated on M9‐citrate media supplemented with uracil (to assess viability) and M9‐citrate supplemented with uracil and 5FOA (to evaluate pyrF‐related mutation frequencies). The same strains were also plated on LB supplemented with rifampicin (Rif) for measuring mutation frequencies of the off‐site gene rpoB (Campbell et al., 2001; Jatsenko et al., 2010). As shown in Fig. 3A, the wild‐type strain with the insert‐free vector pSEVA221 presented mutant frequencies 2.1 × 103 5FOAR CFU per 109 viable cells of the P. putida test strain (Galvao and de Lorenzo, 2005; Aparicio et al., 2018). This occurrence increased by ~ two‐fold in the PYRC ∆ung strain (4.7 × 103 5FOAR CFU; Fig. 3A). This was not unexpected, as UDG is part of the general DNA repair machinery and such a moderate rise of basal mutation levels in ∆ung mutants has been observed in P. putida (Algar et al., 2020). These figures thus set the extent of naturally occurring – and somehow anticipated –numbers for the reporter system. Expression of RNAPT7‐alone in P. putida PYRC and P. putida PYRC ∆ung delivered, however, a different scenario (Fig. 3A). While the wild‐type strain gave frequencies of appearance of 5FOAR clones in the range of those observed in the strains with the empty vector (1.5 × 103 5FOAR CFU), the same increased by > 10‐fold in the PYRC ∆ung host (3.1 × 104 5FOAR CFU). This was not necessarily disadvantageous, but how could it happen? It is known that NH2‐containing bases of the ssDNA transiently formed during transcription become highly vulnerable to spontaneous hydrolytic deamination (Lindahl and Nyberg, 1974; Shen et al., 1994). Given the high transcriptional activity of RNAPT7‐alone, it is plausible that the lifetime of ssDNA during transcription is increased in comparison with other RNAPs (Golomb and Chamberlin, 1974; Studier and Moffatt, 1986), but mutations can be fixed in most cases. However, deletion of ung is likely to amplify occurrence of C deamination, thereby increasing the eventual pyrF inactivation rates in respect to the background (see below for analyses of different types of mutants).

Fig. 3

Frequency of P. putida clones resistant to either 5FOA (putatively in pyrF) or rifampicin (putatively in rpoB).

A. 5FOAR mutants of wild‐type strain P. putida PYRC and its ∆ung derivative transformed with either empty pSEVA221 vector or expressing RNAPT7.

B. Rif R mutants of the same strains. Note very different scales of the Y axis.

C. 5FOAR mutants of P. putida PYRC and P. putida PYRC ∆ung transformed with plasmids expressing fusions of CdA variants rAPOBEC1 and pmCDA1 fused to RNAPT7.

D. RifR mutants of the same strains. The mean (bars), standard deviation (lines) and single values (dots) of four (5FOAR) or two (RifR) independent biological replicates are shown in each case. [Colour figure can be viewed at wileyonlinelibrary.com]

When same strain set was tested for resistance to rifampicin instead of 5FOAR, a different pattern became apparent (Fig. 3B). First, the frequency of appearance of RifR mutants in the wild‐type hosts was lower than equivalent 5FOAR clones. This was not surprising as RifR clones are expected to stem from changes in specific segments of the rpoB gene (encoding the essential β subunit of the housekeeping RNAP; Jatsenko et al., 2010), and therefore, the number of targets resulting in the resistance phenotype throughout the corresponding gene sequence ought to be lower. Second, the loss of UDG significantly increased such frequencies, in particular in the PYRC ∆ung strain with pSEVA221 [RNAPT7]. We entertain that such rise probably reflected the inherent mutagenic effect of the ung mutation (Duncan and Weiss, 1982; An et al., 2005; Algar et al., 2020), which could be exacerbated by the stress caused by the high expression of RNAPT7. Although the specific mechanisms behind the totals shown in Fig. 3A and B remain speculative, the numbers given by these experiments provided a robust baseline for appraising the effect of combining RNAPT7‐driven transcription with C deamination as addressed next.

CdA‐RNAP fusions boost diversification of DNA downstream of a T7 promoter

In order to inspect the consequences of directing CdA‐RNAPT7 towards the edited pyrF‐spanning genomic region shown in Fig. 2A and B, strains P. putida PYRC and P. putida PYRC ∆ung expressing either rAPOBEC1‐RNAPT7 or pmCDA1‐RNAPT7 were passed through the same tests as above for quantifying emergence of both 5FOAR and RifR mutants. As shown in Fig. 3C, the effect of fusions was different in each case. The wild‐type strain expressing rAPOBEC1‐RNAPT7 gave levels of 5FOAR CFU indistinguishable from those generated by RNAPT7 alone. Yet, the same parameter increased ~ 60‐fold when pSEVA221 [rAPOBEC1‐RNAPT7] was placed in the PYRC ∆ung host. While this result hinted that many of the mutants originated in C➔T changes, the net increase in 5FOAR clones was relatively small as compared to the equivalent system devoid of CdA activity, that is expressing RNAPT7 only. The situation changed significantly when the rat CdA was replaced in the constructs by the lamprey deaminase pmCDA1. As shown in Fig. 3C, placing pSEVA221 [pmCDA1‐RNAPT7] in the PYRC strain already delivered 5FOAR mutants at the same frequencies than the strain P. putida PYRC ∆ung with pSEVA221 [rAPOBEC1‐RNAPT7]. Furthermore, when pmCDA1‐RNAPT7 was tested in the PYRC ∆ung host, the frequency of mutants boosted to ~ 1.76 × 107 5FOAR CFU per 109 viable cells, that is ~ 1% of the whole population and > 500‐fold increase relative to the numbers corresponding to the identical strain with RNAPT7‐only (ca. 3.1 × 104 5FOAR CFU per 109 viable cells). Interestingly, as shown in Fig. 3D, the number of RifR clones detected in strains carrying the CdA variants samples changed minimally in respect to the controls (void vector and sole expression of RNAPT7). This suggested that introduction of the CdAs in the diversity‐generating platform of Fig. 2 increased changes in pyrF well above the background mutagenesis caused by the other components of the genetic device. In summary, these results indicated that [i] rAPOBEC1 and pmCDA1 fused to RNAPT7 enhanced diversification of target DNA sequences placed downstream of a T7 promoter, that [ii] this activity is checked by the UDG activity and that [iii] pmCDA1 was way more efficient than rAPOBEC1 as a mutagenic driver – a feature already reported when the system was tested in E. coli (Alvarez et al., 2020). One practical consequence of this is that the action of the hyperactive lamprey's variant in the wild‐type host (Fig. 3C) was equivalent to that of the rat's protein in the ∆ung genetic background – an issue not devoid of interest for general applicability of the method (see below). The results above provided an estimation of the efficacy of the different arrangements for site‐focused mutagenesis of the P. putida genome. Yet, they still say little on the origin and scope of the mutations. To examine this issue, the type of changes caused by each of the CdA‐RNAPT7 was inspected next.

Mutation profiles caused by CdA‐RNAP fusions

Figure 4 summarizes the range of mutations incorporated to the pyrF DNA in 5FOAR clones of P. putida PYRC ∆ung expressing rAPOBEC1‐RNAPT7 and pmCDA1‐RNAPT7. To generate the data shown, the genomic segment with the reporter construct engineered in the strain, that is the intervening DNA between the T7T terminator and the P promoter, was PCR‐amplified with primers out_TS1pyrF and out_TS2pyrF covering the entire DNA portion and the cognate sequences determined (see Experimental procedures). As a reference, the pyrF segments of 5FOAR colonies of the same strain transformed with either vector‐alone pSEVA221 (12 clones) or pSEVA221 [RNAPT7] (27 clones) were inspected as well. Changes in the pyrF DNA were not detected in 5FOAR mutants arising from the strain with the void vector, suggesting that the mutations mapped in another part of the genome (e.g. those determining transport or others). In contrast, we did notice a variety of changes in the same genomic segment exposed to the transcribing action of RNAPT7, including deletions, insertions, transitions and transversions (Fig. 4A). Despite the relatively small number of clones analysed in these controls, the results implied that transcription with RNAPT7 has a low‐level but still significant mutagenic effect on the DNA involved (Beletskii et al., 2000; Kar and Ellington, 2018; Alvarez et al., 2020). This is a relevant phenomenon for optimization of a general method of continuous evolution (see below).

Fig. 4

Characterization of pyrF mutations found in 5FOAR colonies expressing CdAs.

A Total number and types of mutations associated to each construct borne by P. putida PYRC ∆ung, indicating the base substitutions found.

B. Distribution and number of mutations throughout the pyrF segment of P. putida PYRC ∆ung (sketched on top) in 5FOAR clones carrying the construct indicated and the number of clones analysed in each case. The boundaries of the pyrF DNA sequence, the T7 promoter (P ) and T7 terminator (T7T) are indicated along the location of the putative pyrF promoter. Note very different scales of the Y axis. The base changes are tagged into the coding sequence of pyrF. Mutations types are indicated with the same colour codes. Extension of deletions are indicated with the boundaries of the nucleotides erased from the cognate DNA.

C. Average number of mutations per clone found in the 5FOAR colonies analysed for each of the CdA‐RNAPT7 fusions. Single values are represented with red dots and means and standard deviations with black lines. Final figures are highlighted in each case. [Colour figure can be viewed at wileyonlinelibrary.com]

The same type of analysis was run next on 34 5FOAR clones derived from the reporter PYRC ∆ung strains expressing each of the CdAs under examination fused to RNAPT7. Regardless of the CdA, all changes detected corresponded to C to T or G to A transitions occurring in both DNA strands. Interestingly, no other modifications were found, thereby enabling us to trace all changes to the action of the CdAs. The strain expressing pmCDA1‐RNAPT7 accumulated the highest number of transitions: a total of 1334 in all 5FOAR clones, with an average of ~ 39 mutations per clone. In contrast, the strains with rAPOBEC1‐RNAPT7 displayed ~ 25‐fold less mutagenic activity (48 transitions detected in the clones) with just above 1 mutation per 5FOAR isolate (Fig. 4B and C).

One detail of interest was that regardless of the CdA type fused to RNAPT7, G➔A transitions were detected more frequently than C➔T in the pyrF coding strand (44 vs 4 for rAPOBEC1 and 1202 vs 132 for pmCDA1 (Fig. 4A). This indicated that there is a higher mutagenic activity of the fusions on the non‐coding strand of pyrF in the reporter system of Fig. 2. Such a mutational bias favours changes in the non‐template strand of transcription from P by ~ 90% vs. the template counterpart – a phenomenon observed in E. coli as well (Beletskii et al., 2000; Kar and Ellington, 2018; Moore et al., 2018; Alvarez et al., 2020). This preference might be caused by a higher exposure of the C residues of this strand to the deamination activity of the CdAs during transcription. Yet, 10% is still a notable frequency, a circumstance that plays in our favour for bringing about continuous evolution of the DNA segment of interest, as eventually every C change is in principle possible independently of the DNA strand.

Finally, in order to evaluate the efficacy of the T7 terminator (T7T) engineered in the reporter system for circumscribing the mutational effects of CdA‐RNAPT7 to the DNA segment of interest, we inspected the occurrence of C➔T and G➔A transitions beyond the nominal termination site. To this end, the pyrF upstream regions (i.e. beyond T7T) of a sample of 5FOAR clones already analysed (11 from strain with pmCDA1‐RNAP T7 and 17 with rAPOBEC1‐RNAPT7) were sequenced with primers out_TS1pyrF and pyrF_R for detection of changes indicative of CdA‐RNAPT7 trespassing the T7T site. In the case of rAPOBEC1 fusion, no mutations beyond pyrF were found in the adjacent ~ 1 kb region (data not shown). On the contrary, a significant number of changes did occur in samples expressing the more active pmCDA1‐RNAPT7 hybrid, as shown in Fig. S4. Specifically, the pyrF upstream region kept an average number of mutations per clone and DNA length that was about half of those counted in the reporter gene. These figures suggested that just one T7T may suffice to hinder the gross progression of the lower‐activity CdA‐RNAPT7 further than a cognate termination signal, but was not enough to do entirely the same with the more active counterpart (Fig. S4). This did not come as a surprise, as not less than four terminators are necessary to stop altogether transcription initiated in a T7 promoter (Moore et al., 2018). Regardless of the mechanisms behind differential termination activity of the T7T signal with either CdA‐RNAPT7, these results exposed one more parameter to consider for choosing a diversity‐generating option according to specific needs (see below). But, aside of the CdA type and termination efficacy, other preferences need to be fixed as explained in the next sections.

UDG swaps DNA strand preference for cytosine deamination

We observed a significant level of 5FOAR mutants in the wild‐type ung + strain carrying the pmCDA1‐RNAT7 fusion which promoted us to inspect in more detail the type of mutations that emerged in such conditions. Since pmCDA1 has been described as a high‐activity CdA (Lada et al., 2011), this outcome could simply reflect that a higher C to U deamination efficiency in DNA titers out the correction mechanism mediated by the housekeeping UDG activity. But also, UDG action causes excision of U residues (Lindahl et al., 1977; Parikh et al., 1998; Schormann et al., 2014), leaving transient abasic sites likely to generate random replacements in the spot of the deaminated C. While this might be an unwanted occurrence when using CdAs for gene editing, it can play in our favour in applications where a large diversity of changes is pursued. With these notions in mind, we sequenced the pyrF DNA region of 60 5FOAR clones derived from ung + strain P. putida PYRC transformed with pSEVA221 [pmCDA1‐RNAPT7]. Note that, as indicated above (Fig. 3), the net frequency of resistant clones was > 60‐fold lower in this UDG+ strain as compared to the ∆ung equivalent (5.68 × 104 CFU vs 1.76 × 107 CFU), but still considerable. Analyses of the resulting clones expectedly revealed a lower number of mutations per clone, although similar to those caused by rAPOBEC1‐RNAPT7 fusion in the ∆ung strain (compare Fig. 6 vs. Fig. 4). Unexpectedly, however, the vast majority of mutations found were C➔T transitions that had taken place in the pyrF coding strand. Given the arrangement of the reporter construct (Fig. 2A and B), this indicated a higher deamination activity in the template strand for RNAPT7 transcription. This result is exactly opposite to what was found in the ∆ung background with the same pmCDA1‐RNAPT7 fusion, where most changes were G➔A (Fig. 4). In addition, we found a few transversions C➔A, an occurrence that stems from filling abasic sites after cytosine deamination. Such changes were not found in any other 5FOAR colony sequenced from any previous construct. Although the spatial architecture of the complex between the transcribing pmCDA1‐RNAPT7 and the components of the repair system is unknown, it is plausible that UDG changes the exposure of one DNA strand or the other to the deaminating activity of the fusion protein or that U residues in the template strand are less accessible to the activity of the glycosylase, but regardless of the precise process, this result adds a significant value to using pmCDA1‐RNAPT7 fusion in the ung + wild‐type strain, as a high variability can be obtained without a need to inactivate UDG. We did not fail to notice that this particular setup delivers an average of one detectable mutation per clone (Fig. 6C), what appears as a good balance between variability and function stability in the search of new phenotypes.

Fig. 6

Characterization of pyrF mutations found in 5FOAR colonies of UDG+ strain P. putida PYRC expressing the pmCDA1‐RNAPT7 fusion.

A. Total number of mutations identified in the 60 clones analysed, with an indication of the base substitutions found.

B. Average number of mutations identified per sequenced clone. Single values are represented with red dots and means and standard deviations with black lines.

C. Type and position of each mutation in the pyrF gene identified in the 60 5FOAR clones sequenced. The boundaries of the T7 promoter (P ), T7 terminator (T7T) and the putative pyrF promoter are indicated. The base changes shown correspond to the coding sequence of pyrF. The different types of mutations labelled with the same colour codes as in (A). [Colour figure can be viewed at wileyonlinelibrary.com]

Transient inhibition of native UDG boosts CdA‐RNAP‐mediated mutagenesis

As shown above, the lack of UDG is key for enabling C deamination to turn into conversion to a T residue in DNA. In pragmatic terms, this means that the hosts for optimal operation of CdA‐RNAPT7 fusions should lack the cognate ung gene. In most cases, deletion of the corresponding DNA can be easily entered in the genome of the strain at issue, but in other instances, the loss of UDG can interfere with the outcome of the pursued diversification. In these instances, it is possible to inhibit the enzyme in vivo through co‐expression of an uracil‐DNA glycosylase inhibitor (UGI) protein. The most used UGI comes from the bacteriophage PBS2 (Katz et al., 1976; Cone et al., 1980; Wang and Mosbaugh, 1989), and it consists of a thermostable and highly acidic small peptide (84 amino acids, 9.4 kDa), which binds reversibly to UDG in 1:1 molar stoichiometry (Wang and Mosbaugh, 1989). To leverage this UGI for our purposes, the cognate DNA sequence was optimized for expression in Pseudomonas (Gallagher et al., 2021) and cloned in medium‐copy number SmR/SpR vector pSEVA4311 (see details in Experimental Procedures). The resulting UGI+ plasmid was then passed through conjugation to ung + strain P. putida PYRC already bearing either pSEVA221 [rAPOBEC1‐RNAPT7] or pSEVA221 [pmCDA1‐RNAPT7], and the exconjugant strains were assayed for emergence of 5FOAR colonies as before. The results (Fig. 5) showed that incorporation of pSEVA4311 [UGI] to PYRC strain increased mutagenic efficiency of rAPOBEC1‐RNAPT7 on pyrF by ~ 250‐fold (ca. 4.3 × 105 CFU) and that of pmCDA1‐RNAPT7 by ~ 1000‐fold (ca. 3.1 × 107 CFU) in respect to the conditions lacking UGI (1.9 × 103 CFU and 5.7 × 104 CFU, respectively). Inspection of the data of Fig. 5 in fact suggests that ectopic expression of UGI brought about the same mutant frequencies as those observed in PYRC ∆ung strains (i.e. with an ung deletion). It is thus feasible to induce a transient UDG‐minus phenotype by addition of compatible construct pSEVA4311 [UGI] to the cells already loaded with plasmids expressing CdA‐RNAPT7 variants for the sake of boosting diversification of the target gene. To test whether the mutation profile of each DNA strand involved was the same as in the PYRC ∆ung strain, a number of 5FOAR clones of the P. putida PYRC reporter strain expressing both pmCDA1‐RNAPT7 and the UGI peptide were sequenced. The results shown in Fig. S5 indicated that – similarly to the situation in a genetic context lacking UDG –deamination occurred preferentially in the non‐template strand. This occurrence widens the range of conditions and strains that can be chosen for particular DNA diversification assignments.

Fig. 5

Effect of pSEVA4311[UGI] in mutagenic efficacy of CdA‐RNAPT7.

A. Drop assay showing the frequency of 5FOAR colonies of ung + strain P. putida PYRC with plasmids expressing separately rAPOBEC1‐RNAPT7 or pmCDA1‐RNAPT7 and added or not with compatible plasmid pSEVA4311[UGI] as indicated in each case.

B. Schematic representation of the functional parts of pSEVA4311 [UGI], not to scale. This is a medium‐copy number plasmid with the origin of replication of pBBR1, an oriT for conjugative mobilization, an SmR/SpR gene for selection and a ChnR/P expression system driving transcription of the UGI gene.

C. Frequency of 5FOAR clones generated from P. putida expressing the CdA‐RNAPT7 variants indicated and added with of pSEVA4311[UGI]. Compare these figures with those of the same strains without UGI shown in Fig. 3C. Adding or not the ChnR inducer cyclohexanone to the cultures had no significant effect on the reported frequencies (not shown). The mean (bars), standard deviation (lines) and single values (dots) of two independent biological replicas are shown. [Colour figure can be viewed at wileyonlinelibrary.com]

Conclusion

The work presented in this article provides the genetic tools and the variables to consider for setting continuous diversification of any DNA segment of the genome of P. putida. Since all constructs are borne by broad‐host range plasmids, it is plausible that the very same platform can be effectively reused in other Gram‐negative bacteria. Specifically, by using a pyrF‐based reporter system, the efficacy of each CdA‐RNAPT7 fusion as mutagenic agent has been determined, the role of UDG in the process settled and the efficacy of a T7 terminator to restrain DNA diversification beyond a prefixed site documented. Given the degeneracy of the genetic code and that not every base change necessarily originates an inactive pyrF mutant, it is likely that the actual DNA diversification figures are at least threefold higher than the ones shown throughout this work. As indicated above, the desired mutagenesis levels can be modulated at will by picking one of the fusions available and playing with having either ung or ung + strains as hosts of the process – or engineering a transient UDG‐minus phenotype upon addition of a compatible plasmid expressing the inhibitory peptide UGI. Although not tested directly in this work, the length of the genomic segment to be diversified could in principle be limited by engineering a series of terminators or a dCas9‐based device (Alvarez et al., 2020) to inhibit the advancing CdA‐RNAPT7. In other cases, such rowing base editors can be let to proceed through longer DNA segments in the genome if desired. In sum, we believe that the hereby described devices may become a phenomenal addition to the already rich toolset available for many types of bioengineering endeavours in P. putida (Martínez‐García and de Lorenzo, 2017; Martin‐Pascual et al., 2021).

Experimental procedures

Strains, plasmids, growth conditions and general techniques

The list of bacterial strains and plasmids used in this study can be found in Tables S1, S2 and S3. The P. putida strains employed in this work were derived from variant EM42 (Martinez‐Garcia et al., 2014) a genome‐edited derivative of reference wild‐type isolate KT2440 (Nelson et al., 2002). E. coli strains DH5α, CC118 (Hanahan and Meselson, 1983; Manoil and Beckwith, 1985) and their λpir derivatives were used as hosts of intermediate constructs. The presence of λpir allows proliferation of plasmids with an R6K oriV origin of replication (e.g. pEMG). 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. Where indicated, plates of M9 minimal medium (Sambrook et al., 1989) were supplemented with 0.2% (w/v) sodium citrate as the sole carbon source and added with 5‐fluoroorotic acid (5FOA, 250 μg ml−1) and/or uracil (ura, 20 μg ml−1) as required. Antibiotics were added to liquid or solid media at various concentrations: streptomycin (Sm, 50 μg ml−1 for E. coli and 100 μg ml−1 for P. putida), kanamycin (Km, 50 μg ml−1) and rifampicin (Rif, 100 μg ml−1). DNA manipulations followed standard laboratory techniques (Sambrook et al., 1989). Plasmids were constructed with either classical cloning procedures or Gibson assembly (Gibson et al., 2009; Gibson et al., 2010). PCR reactions were run with Q5 High Fidelity polymerase (New England Biolabs, Ipswich, MA, USA) according to manufacturer's instructions. Diagnostic PCRs were performed using Green MasterMix (BioTools, Madrid, Spain) taking fresh single colonies as the starting material. P. putida cells were transformed as described by Choi and Schweizer (2006)) in a Gene Pulser/Pulse Controller (Bio‐Rad, Hercules, California, USA) system with 2.5 kV, 25 μF, 200 Ω. Tri‐parental matings (Kessler et al., 1992) for mobilization of plasmids from E. coli strains to P. putida were performed with a simplified protocol (Sambrook et al., 1989). 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.

Construction of deletion and insertion mutants in P. putida strains

The method for inserting the reporter PYRC segment into the genome of P. putida was adapted from Martinez‐Garcia and de Lorenzo (2011); Fig. S1). In brief, a ~ 1 kb DNA fragment flanked by homology regions surrounding the native pyrF locus was synthesized (Genecust™, Ellange, Luxemburg) and added to pEMG as an EcoRI and BamHI insert, thereby originating pEMG [PYRC]. This plasmid was then transformed to strain P. putida EM42 ∆pyrF and consequently plated on LB agar plates supplemented with Km to isolate co‐integration events. Then, pSW‐I plasmid controlling expression of I‐SceI endonuclease was also electroporated into selected clones. This enzyme, induced by addition of 3‐methyl benzoate, triggered the second recombination event by cleaving the chromosome at its specific targets included in the pEMG sequence. Finally, Km‐sensitive clones were isolated and checked with colony PCR with primers out_TS1pyrF (GCCGCTGTTCGCCAGTCTGTCG) and out_TS2pyrF (GATCGACTACAAGGCCGAAGACGTGC) to verify the correct modification in the genome vs. reversion to the wild‐type genotype. pSW‐I plasmid was then cured by several growth passages with no antibiotic pressure. The ensuing strain was called P. putida PYRC. Deletion of the ung gene of P. putida was done as explained in (Martinez‐Garcia and de Lorenzo, 2011) and verified with diagnostic PCR oligos ung‐junction‐F (CATCGCGGGCATTGATCG) and ung‐junction‐R (GAAGTGCGTC AGCGGTCC).One correct clone was also cured of pSW‐I plasmid and named P. putida PYRC ∆ung. All plasmids used in these procedures are listed in Table S3.

Expression of cytosine deaminases fused to RNAP fusions and mutagenesis tests

Plasmids encoding the two CdA‐RNAPT7 used in this work have been described elsewhere (Alvarez et al., 2020). In them, either rat apolipoprotein B mRNA editing enzyme 1 (rAPOBEC‐1) or Petromyzon marinus (lamprey) cytidine deaminase 1 (PmCDA‐1) is respectively fused to the N‐terminus domain of the RNAPT7 by a (G3S)7 linker (28 aa in total). In addition, they both bear a T7 tag (an epitope composed of an 11‐residue peptide from the leader sequence of the T7 phage gene 10) added to the N‐terminus of each of the CdA sequences and the fusions are transcribed through a TetR/PtetA expression device (Lederer et al., 1996). Control plasmids, empty pSEVA221 and pSEVA221 [RNAP ] are described in Table S3. Plasmids were separately entered into reporter P. putida, and strains with the different constructs grown overnight at 30°C in liquid LB with Km. Cultures were then diluted in the same medium to an OD600 of 0.02 and regrown with shaking to an OD600 of 0.4 to 0.7. Cells were then washed with PBS 1x and serial dilutions of 100 μL plated on M9 medium with Ura (for viability) and with Ura + 5FOA for selecting 5FOAR mutants. After overnight growth, CFU were counted from each plate and the mutation rate was calculated dividing the count of CFU in M9/Ura/5FOA plates by CFU in M9/Ura plates and normalizing figures to 109 viable cells. The off‐target activity of the system was calculated by plating the same cells on LB/Rif media, and RifR clones were also normalized to 109 viable cells.

Sequencing and analysis of cytosine deaminases‐induced mutations

5FOAR colonies were used as starting material for colony PCR reactions with primers out_TS1pyrF and out_TS2pyrF (see above) which amplified a ~ 2.2 kb fragment covering the reporter pyrF region engineered in P. putida PYRC (Fig. 2) and ~ 500 bp upstream, thereby encompassing also ORFs PP1813 and PP1814. Amplified DNAs were directly purified with ExoSap‐IT™ PCR Product Cleanup Reagent (Thermo Fisher Scientific, Applied Biosystems) and submitted to Sanger sequencing with primers pT7p (TAATACGACTCACTATAGGG) and pT7t (ACCCC TCAAGACCCGTTTAG) to address the type of mutations in the pyrF locus or with primers out_TS1pyrF (GCCGCTGTTCGCCAGTCTGTCG) and pyrF_R (TCACGCCAATCAGCAACG) to analyse the mutations upstream of T7T. Sequences were analysed manually with SeqMan Pro™ DNASTAR software and compared against the sequence of the modified pyrF segment as reference. Results were plotted using GraphPad Prism 6 (GraphPad Software Inc., San Diego, CA, USA).

Construction and utilization of UGI expression plasmid

In order to build a construct for entering UGI activity in P. putida, the ChnR‐P promoter was excised from pSEVA2311 as a PacI/AvrI fragment and inserted into the same sites of pSEVA427 resulting in pSEVA4211‐gfp (Table S3). The DNA sequence encoding UGI was PCR‐amplified from pPSV39‐UGI with oligonucleotides p4211‐UGI‐F (CGCGAATTCCACGGGAGGAAAGATGACG) and p4211‐UGI‐R (CTATCAACAGGAGTCCAA GACTAGTTTAGAGCATCTTGTTTTGTTCTC) and cloned through Gibson assembly into EcoRI/SpeI‐digested pSEVA4211‐gfp, yielding pSEVA4211[UGI]. Then, the pBBR1 oriV from pSEVA131 was isolated as a FseI‐AscI fragment and cloned into similarly digested pSEVA4211 [UGI] to generate the SmR plasmid pSEVA4311 [UGI], which is compatible with pSEVA221 [rAPOBEC1‐RNAPT7] and pSEVA221 [pmCDA1‐RNAPT7]. pSEVA4311 [UGI] was then passed to the P. putida strains indicated in each case through tri‐parental mating as explained above. In order to test the effect of UGI expression on mutagenic efficacy of the CdA‐RNAPT7 fusions, cultures of the thereby constructed strains were grown overnight at 30°C in LB supplemented with Km and Sm to secure retention of both plasmids. Each of the mutagenic fusions cells were regrown in fresh media, let grow to mid‐exponential phase and processed and plated on selective media to assess viability and emergence 5FOAR CFU as explained before.

Conflict of interest

The authors declare no conflict of interest.

Table S1. E. coli strains used in this work.

Table S2. Pseudomonas putida strains used in this work.

Table S3. Plasmids used and constructed in this work.

Fig. S1. Refactoring the pyrF region of P. putida as a reporter of mutagenic activity of CdA‐RNAP T7 fusions. (A) Organization of the genomic region of interest in the starting strain P. putida EM42 ΔpyrF 2018. (B) Arrangement of delivery plasmid pEMG [PYRC] (not to scale). Vector pEMG (Martinez‐Garcia and de Lorenzo, 2011) was inserted with a synthetic DNA fragment composed by the P. putida pyrF gene bordered by a T7 terminator (T7T) at 5’ and a T7 promoter (P) at 3’ along with flanking 500 pb of DNA homologous to either side of the native genomic sequence. The cassette was then introduced in P. putida EM42 ΔpyrF by recombination following the protocol described in Experimental Procedures. (C) After resolution of the cointegrate, the resulting strain was named P. putida PYRC, which carried the PYRC reporter segment in its genome as indicated at the bottom.

Fig. S2. Characterization of P. putida reporter strains. Viability of reporter and parental strains in minimal medium M9/Citrate, M9/Citrate supplemented with uracil and M9/Citrate supplemented with uracil and 5FOA. Each culture was grown overnight and then normalized to OD600 = 1 in PBS 1x. Series of ten‐fold dilutions of each culture were prepared and 5 μl drops of each dilution were plated.

Fig. S3. Performance of the tetR/P expression device in P. putida EM42. (A) Fluorescent cell cytometry of E. coli CC118 and P. putida PYRC bearing plasmid pS221 Ptet‐GFP1. The pictures show2 the distribution of fluorescence in populations under non‐inducing conditions (t = 0 h and No aTc) and after induction at the indicated time points with 0.5 μM aTc (t = 1 h and 2 h). The region considered negative for the fluorescence signal is marked with a grey dashed line, as assessed by control cells carrying an empty pSEVA221 plasmid (purple plot). At least 80.000 events were analyzed in each sample. (B) Visual inspection of GFP fluorescence signal in the same samples under blue light. Note a regulated, aTc‐inducible expression of GFP in E. coli in contrast with a virtually constitutive expression in P. putida. Anhydrotetracycline (aTc); arbitrary units (a.u.)

Fig. S4. Mutagenic activity of advancing CdA‐RNAP T7 fusions beyond a T7 termination signal. The figure summarizes the characterization of mutations upstream of the pyrF gene of P. putida PYRC Δung borne by 5FOAR colonies expressing the pmCDA1‐RNAPT7 fusion. (A) Number and frequency of mutations found through upstream region of the PYRC cassette of 11 5FOAR colonies. Adjacent PP1813 and PP1814 genes are shown. Different types of mutations are indicated with a color code. (B) Frequency of transitions per clone and nucleotide caused by pmCDA1‐RNAPT7 on the PYRC segment proper (on target) and the upstream region beyond the terminator. (C) Average number of mutations per clone found in the upstream region of the PYRC segment of the 5FOAR colonies. Single values are represented with red dots and means and standard deviations with black lines. (D) Total number of mutations and base substitutions found in the upstream region shown.

Fig. S5. Non‐template strand preference is recovered in UGI expressing clones. (A) Total number and types of mutations borne by P. putida PYRC expressing pmCDA1‐RNAPT7 and UGI. (B) Average number of mutations per clone found in the 5FOAR colonies analyzed. Single values are represented with red dots and means and standard deviations with black lines. (C) Distribution and number of mutations throughout the pyrF segment of P. putida PYRC in 28 5FOAR clones carrying the construct indicated previously. The boundaries of the pyrF DNA sequence, the T7 promoter (P) and T7 terminator (T7T) are indicated along the location of the putative pyrF promoter. The base changes are tagged into the coding sequence of pyrF. Mutations types are indicated with the same color codes.

Click here for additional data file.