The entire GI can be excised from the chromosome of P. putida KT2440 via functional replacement with either tmk from E. coli or the endogenous PP_1964 gene

Based on the experimental evidence described above, it appeared that dNMPK _Pp plays an essential function encoded in the GI gene clusters of P. putida. To further substantiate this idea, the entire tmk-prophage gene cluster was replaced with the tmk gene from E. coli MG1655 (tmk_Ec ), yielding P. putida NTW1804 (Table 1). Such functional replacement (Δtmk ^GI::tmk_Ec , Fig. 1D) succeeded readily, indicating that TMPK is indeed the essential function encoded within the viral GI region. In this strategy, holB was translationally coupled to the tmk_Ec gene via the aforementioned genetic configuration (5′-TGATG-5′). In a previous study by Páez-Espino et al. (28), deletion of the ars1 operon, encoded within the tmk ^GI region, amplified the sensitivity of the mutant strain to arsenate (AsO₄ ³⁻) salts. We adopted the As(V)-tolerance phenotype as a proxy to explore if the insertion of the synthetic Δtmk ^GI::tmk_Ec construct brings about any noticeable physiological response to a KH₂AsO₄ challenge. When subjected to increasing concentrations of KH₂AsO₄, up to 300 mM, the growth profiles of the parental P. putida strain were contrastingly different from those of its Δtmk ^GI::tmk_Ec derivative, P. putida NTW1804 (Fig. S1 in the Supplemental Material). P. putida NTW1804 exhibited significantly reduced growth rates and prolonged lag times at concentrations equal to and exceeding 25 mM KH₂AsO₄, while the two strains displayed identical growth at As(V) concentrations up to 5 mM. These observations validated the importance of the ars1 gene cluster in As(V) tolerance, and demonstrated that the entire GI can be functionally replaced with a gene encoding the TMPK enzyme from E. coli without affecting the overall growth phenotype of P. putida KT2440. Furthermore, when P. putida NTW1804 was cultivated in de Bont minimal (DBM) medium under glycolytic (30 mM glucose) or gluconeogenic (30 mM citrate) conditions, or in complex LB medium, the growth phenotype was identical to that of the parental KT2440 strain (results not shown). Building on these preliminary results, we implemented the same functional replacement approach in a genome-reduced variant of P. putida KT2440, strain SEM10 (37). The resulting strain, P. putida NTW1282 (Table 1) exhibited the same growth phenotype as strain SEM10 and the engineered P. putida NTW1804 variant across a range of culture conditions (i.e., complex LB medium and DBM medium added with acetate, glucose or citrate; results not shown), further validating the overall strategy.

Adopting the same framework, we studied the next obvious candidate that could provide the TMPK activity. As indicated above, the very last CDS in the genes clusters encoded in the GI of strain KT2440, PP_1964, is predicted to encode a deoxynucleotide monophosphate kinase (i.e., dGMP/dTMP/dihydroxy-CMP kinase, dNMPK _Pp ). Interestingly, the entire GI could also be successfully removed by leaving only PP_1964 under the original transcriptional and translational control sequences of the tmk.A fragment and translationally coupled to holB (Table 1). Furthermore, the resulting strain, P. putida SEM11, showed an almost identical growth phenotype to its parental strains EM42 and SEM10 under various culture conditions (LB medium, DBM medium supplemented with glucose, citrate, or acetate; results not shown).

In particular, using a plate reader experiment in a 24-well plate, cultures of P. putida KT2440 and SEM11 were grown in DBM medium with 30 mM glucose. The engineered strain displayed faster growth (10% increase in the μ _max values) and higher biomass yields (ca. 13% increase) compared to its KT2440 ancestor (Fig. 2A). Taken together, these results also provide evidence that PP_1964 (i.e., dNMPK _Pp ) is the bona fide TMPK activity of P. putida.

Open in a separate window

Fig 2

Assessing the impact of genetic modifications on thymidylate kinase function in P. putida KT2440. (A) Growth curves (smoothing splines) of the parental strain and genetic variants with a deletion in the tmk locus and an orthogonally regulated copy of PP_1964. Cells were cultured in 24-well plates in DBM medium supplemented with 30 mM glucose; cell growth was estimated as the optical density measured at 600 nm (OD₆₀₀). Results represent average values ± standard deviations from three independent biological replicates. The yellow-shaded area indicates sampling at OD₆₀₀ values between 0.5 and 1.2 to perform quantitative proteomics. (B) dNMPK _Pp abundance in the parental and engineered P. putida strains determined via targeted quantitative proteomics. Protein abundance values were normalized to the average dNMPK _Pp content obtained for strain KT2440; individual measurements are indicated for three independent biological replicates and the error bars indicate 95% confidence intervals.

Fine-tuning dNMPK _Pp levels through synthetic transcriptional control systems reveals a high degree of buffering capacity in the nucleotide metabolism of P. putida KT2440

We reasoned that rendering the expression of PP_1964 titratable through synthetic transcriptional control systems could shed light on essentiality via artificially modulating bacterial growth. Hence, the gene was placed under the control of several orthologous, inducible expression control systems. The first strategy consisted of regulating PP_1964 expression via the XylS/Pm system (44), induced by 3-methylbenzoate (3-mBz). Following a two-step allelic exchange approach, a synthetic construct [containing the T1 terminator from phage λ (i.e., λT1), the xylS gene that encodes the 3-mBz-responsive regulator, the Pm promoter and the PP_1964 gene] was chromosomally inserted into an intergenic region ~5.3 kb downstream of the tmk ^GI locus, followed by removal of the entire cluster but leaving holB intact. When cultivated in DBM medium containing 30 mM glucose, the resulting mutant strain, P. putida NTW1797 (Table 1), had wild-type-like growth patterns even in the absence of any chemical inducer (Fig. 2A). Indeed, the addition of 3-mBz at concentrations as low as 0.1 mM resulted in a reduced biomass yield (i.e., 21% lower final OD₆₀₀ values), and supplementing 3-mBz at 1 mM negatively affected the final cell density (reduced by 26%) and μ _max (with an 11% slower growth than in the control experiment without 3-mBz supplementation, Fig. S2A in the Supplemental Material). Since 3-mBz is not expected to mediate toxicity in the concentration range used in these experiments (45), the detrimental effect observed could be attributed to the expression of PP_1964. Furthermore, considering that XylS/Pm is known to be a leaky expression system depending on the growth conditions (46), we followed the same allelic replacement strategy, but placed PP_1964 under the transcriptional control of the tight L-arabinose-inducible AraC/P_BAD system (47). The resulting mutant strain (P. putida NTW1788, Table 1) was again able to grow in DBM medium under glycolytic conditions in the absence of any inducer (Fig. 2A). The addition of L-arabinose at 0.2% (wt/vol), however, did not result in any significant increase in the μ _max and caused only minor increased in biomass yield (7%) of this engineered strain (Fig. S2B in the Supplemental Material).

Assuming that our prior attempts of making PP_1964 conditionally dependent on the addition of inducer compounds failed due to the leakiness of the synthetic systems employed, we sought to control transcription levels via activators provided in trans. To this end, and concomitant with the removal of the tmk ^GI locus, we placed PP_1964 downstream of an element composed of the T1 terminator of phage λ and the Pm promoter (λT1-Pm). The rationale behind this design was preventing the activation of the Pm promoter in the absence of XylS through regulator-promoter decoupling (48). The last step of our gene editing method, resolving the previously co-integrated suicide plasmid, adopted the suicide plasmid pQURE6, which carries a copy of xylS and two copies of Pm that control the expression of the I-SceI meganuclease gene as well as the replication protein gene trfA [i.e., making plasmid replication dependent on the addition of 3-mBz (37)]. We hypothesized that if there is no transcriptional activity on PP_1964 in the absence of XylS, the resulting mutant strain would be unable to lose pQURE6 after the procedure. However, the engineered strain (P. putida NTW1835, Table 1) readily lost plasmid pQURE6 regardless of 3-mBz addition and showed no growth defects (Fig. 2A).

To further understand the observed phenotypes, a quantitative, whole-proteome analysis was performed on the previously described strains in the absence of any chemical inducer (Fig. 2B). The leakiness of the chromosomally-encoded XylS/Pm system resulted in a threefold increase of the intracellular dNMPK _Pp abundance in P. putida NTW1797 as compared to that in strain KT2440. The AraC/P_BAD system was indeed the tightest among the tested expression constructs, with an intracellular dNMPK _Pp concentration of only ~20% compared to that in the parental P. putida strain. Surprisingly, even in the absence of the XylS activator (strain NTW1835), PP_1964 expression from Pm reached ~80% of the value in the control (Fig. 2B), which may be due to transcriptional read-through from upstream CDSs. Strain SEM11 had dNMPK _Pp levels comparable to that of P. putida KT2440 (Fig. 2B). By integrating the information about dNMPK _Pp abundance and strain phenotypes gathered thus far, the following picture emerged: (i) deletion of the tmk ^GI locus and the 60 CDSs encoded therein increases biomass yields, (ii) the metabolism of P. putida KT2440 can tolerate a significant reduction in dNMPK _Pp availability without compromising growth and (iii) a drastic increase in dNMPK _Pp formation leads to reduced μ _max values and biomass yields in glucose cultures.

In a final attempt to make the formation of dNMPK _Pp controllable, PP_1964 was placed under the control of the Pm promoter in a different configuration without introducing xylS into the chromosome. In addition, because the λ T1 terminator was suggested to have limited functionality in P. putida (39), the effective transcriptional termination sequence T^S was placed upstream of Pm→PP_1964 to prevent eventual RNA polymerase read-through. Despite these efforts, strain NTW1834 also readily lost plasmid pQURE6 (carrying xylS) in the absence of 3-mBz and, while its growth rate was impaired (25% lower μ _max), grew to the same maximum cell densities as the parental strain (Fig. S2C in the Supplemental Material). In fact, if the cells were forced to retain pQURE6 through the addition of gentamicin as well as 3-mBz, the growth rate was further reduced (results not shown). In the absence of the XylS activator and the exclusion of transcriptional read-through, residual expression of PP_1964 can be attributed to unspecific activation of Pm by BenR (PP_3159), which is encoded in strain KT2440 as part of the benzoate degradation pathway (49 – 51).

Thus far, the implementation of various genetic architectures around the tmk ^GI locus (Fig. S2D in the Supplemental Material) demonstrated that this large viral DNA segment contains a single essential function, providing TMPK activity with high metabolic buffering capacity. A CRISPR-Cas9-based gene disruption approach through base-editing at the single-nucleotide resolution provided definite evidence for the essentiality of dNMPK _Pp in P. putida KT2440, as disclosed in the next section.

CRISPR-Cas9-base-editing and complementation assays expose the essentiality of dNMPK _Pp and its role as the sole thymidylate kinase of P. putida KT2440

As illustrated by the genetic designs attempted thus far, the entire GI of P. putida KT2440 (including the two tmk gene moieties) could be successfully removed as long as either PP_1964 was preserved or its function was complemented by a copy of the tmk gene from E. coli. Any attempts to delete PP_1964 via allelic exchange were unsuccessful (details not shown). However, the failure of implementing these genetic manipulations is not a definite proof of gene essentiality (52). Thus, a more reliable strategy was employed to provide conclusive evidence of the role of PP_1964 as the sole source of TMPK activity in strain KT2440. To this end, tmk_Ec was cloned into the expression vector pSEVA4418, in which tmk_Ec was placed under the transcriptional control of the L-rhamnose-inducible RhaRS/P_rhaBAD system (Table 1). In addition, plasmid pMBEC6∙PP_1964 was constructed to introduce premature STOP (pmSTOP) codons via multiplexed cytidine base-editing (40). This base-editing plasmid encodes three synthetic guide RNAs that target protospacers distributed across the PP_1964 coding sequence (Table 1) and are suitable to introduce pmSTOP codons via C-to-T transitions (Fig. 3A and B). Next, the base-editing plasmid pMBEC6∙PP_1964 was delivered into P. putida SEM11 (Δtmk ^GI::PP_1964, Table 1) harboring either pSEVA4418 (empty vector, used as a control) or pSEVA4418∙tmk_Ec (carrying the TMPK-encoding gene from E. coli). After transformation with pMBEC6∙PP_1964, recovery of the cells in liquid LB medium for 24 h and isolation of individual colonies on LB medium plates without any additives, the PP_1964 locus was PCR-amplified and sequenced for 96 randomly picked colonies of each strain (Fig. 3C). All colonies that had received tmk_Ec in trans exhibited at least one pmSTOP codon in PP_1964, while none of the tested colonies without tmk_Ec showed any null mutations in any of the targeted protospacer sequences. These findings provide strong evidence that the dNMPK _Pp activity encoded in PP_1964 provides the essential TMPK function in P. putida KT2440, as the CDS could only be inactivated if it was simultaneously complemented by tmk_Ec in trans.

Open in a separate window

Fig 3

Exploring the essentiality of PP_1964 in P. putida KT2440 and assessing thymidylate kinase activity of dNMPK _Pp . Plasmid pMBEC6∙PP_1964, encoding a CRISPR-Cas9-guided cytidine base-editing system as well as three synthetic guide RNAs targeting PP_1964, was delivered into strain SEM11 to inactivate the dNMPK _Pp gene. (A) Physical map indicating the locations of the three targeted protospacers within the PP_1964 coding sequence. Target cytosine nucleotides that yield premature STOP codons (pmSTOP) upon editing and transition to thymine are highlighted in magenta. Protospacer adjacent motif (PAM) sequences, 5′-NGG-3′ (where N indicates any nucleotide), are drawn in turquoise. Codon triplets are separated by vertical gray lines. (B) Functional principle of the CRISPR-Cas-guided cytidine base-editing system. A nicking Cas9 variant (nCas9, carrying the D10A mutation) fused to a cytidine deaminase is guided to a target locus via a synthetic guide RNA. The cytidine deaminase causes C→T transitions within a defined range in the non-target DNA strand, allowing for the introduction of pmSTOP codons within the open reading frame (43). (C) Fraction of P. putida SEM11 clones harboring pmSTOP codons in PP_1964 in the presence (+) or absence (–) of the E. coli tmk gene (tmk_Ec ) after the base-editing procedure. Bar plot show the percentage of clones harboring at least one pmSTOP codon out of 96 sequenced P. putida SEM11 colonies with plasmid pRHA·tmk_Ec (Table 2) or the corresponding empty vector. (D) Droplet assay with E. coli strain JHR114 (Δtmk) harboring a chromosomally encoded tmk gene from Mycobacterium tuberculosis under transcriptional control of the AraC/P_BAD expression system (Table 2). This Tmk-deficient E. coli strain was transformed with derivatives of plasmid pRHA encoding either tmk_Ec , Pseudomonas aeruginosa tmk (tmk_Pa ) or PP_1964 controlled by the L-rhamnose-inducible RhaRS/P_rhaBAD system. Overnight cultures, grown in tryptone medium supplemented with 0.2% (wt/vol) L-arabinose, were serially diluted with 0.9% (wt/vol) NaCl and 8 µL droplets were spotted onto solid tryptone medium supplemented with 25 µg mL^–1 streptomycin as well as 0.2% (wt/vol) L-arabinose and 0.2% (wt/vol) L-rhamnose as indicated. Bacterial colonies were photographed after incubating the plates at 37°C for 16 h.

To further demonstrate a role for dNMPK _Pp in providing the TMPK function, a complementation assay was performed using E. coli strain JHR114 (Table 1), which harbors a chromosomally-encoded tmk gene from Mycobacterium tuberculosis (tmk_Mt ) under the control of the AraC/P_BAD expression system, with its native tmk copy deleted (33). Hence, the growth of this strain is dependent on the induction of tmk_Mt via the addition of L-arabinose, making it suitable to screen for thymidylate kinase activities provided in trans. JHR114 was transformed with derivatives of pRHA, harboring either tmk_Ec, tmk from Pseudomonas aeruginosa (tmk_Pa , sharing an 80.5% amino acid sequence identity with tmk from P. putida KT2440) or PP_1964 under the transcriptional control of the L-rhamnose-inducible RhaRS/P_rhaBAD system. The resulting E. coli strains (i.e., JHR114, and its derivatives carrying plasmids pRHA·tmk_Ec , pRHA·tmk_Pa or pRHA·PP_1964, Table 1) were pre-grown in liquid tryptone medium with L-arabinose, serially diluted and spotted onto tryptone agar with or without L-arabinose or L-rhamnose (Fig. 3D). As expected, all strains were able to grow in the presence of L-arabinose, and strain JHR114 without any plasmid was only able to grow in the presence of L-arabinose. Functional complementation of the TMPK activity by tmk_Ec , tmk_Pa and PP_1964 could be verified in the presence of L-rhamnose. Surprisingly, the strains transformed with pRHA·tmk_Ec and pRHA·PP_1964 grew even in the absence of inducers, while JHR114 carrying pRHA·tmk_Pa did not. In the genetic architecture of vector pRHA, tmk_Ec , tmk_Pa , and PP_1964 display almost identical translation initiation rates [computed and predicted with the RBS Calculator tool (53)]—hence, different protein levels should not be the reason behind these phenotypic differences. Thus, Tmk _Ec and dNMPK _Pp appear to have higher TMPK activities than Tmk _Pa, and the leaky transcriptional levels afforded by P_rhaBAD sufficed to recover growth. When cultured in M9 minimal medium supplemented with 20 mM glucose, strain JHR114 carrying pRHA·tmk_Pa was also able to grow, albeit with a severely reduced μ _max (Fig. S3 in the Supplemental Material). Addition of L-rhamnose, in contrast, restored growth for all E. coli strains harboring a plasmid. Interestingly, the induction of tmk_Ec expression resulted in a significant growth impairment which was not observed by overexpressing PP_1964 or tmk_Pa . The results obtained from the complementation experiments in E. coli confirm the role of dNMPK _Pp as a TMPK and expose a remarkable metabolic buffering capacity of this enzyme in bacterial hosts other than P. putida.

Structural alignment and phylogenetic analysis of nucleoside monophosphate kinases

The experimental evidence presented in the preceding sections indicates that the essential PP_1964 (dNMPK _Pp ) gene is the only source of TMPK activity in P. putida KT2440. In order to assess the evolutionary relationship of dNMPK _Pp with other (deoxy)nucleoside monophosphate kinases, a phylogenetic analysis was conducted across annotated bacterial and viral genomes. To this end, we performed a structural alignment of 200 bacterial adenylate kinases (AMPKs), 200 bacterial guanylate kinases (GMPKs), 200 bacterial TMPK, 200 bacterial uridylate kinases (UMPKs), and 300 enzymes annotated as deoxynucleoside monophosphate kinase in bacteria and 81 dNMPKs found in viruses—representing a sampling space large enough to establish relationships between these functions. Next, this multi-sequence alignment was used to construct a phylogenetic tree (Fig. 4) using FastTree (54) based on the structure-based PROMALS3D alignment (55). Based on the framework proposed by Leipe et al. (56), we assumed that TMPKs belong to the heritage of a common ancestor of P-loop kinases—as this enzyme is the only (d)NMP kinase ubiquitous in Bacteria, Archaea and Eukaryota. Deoxynucleoside monophosphate kinases are closely related to the TMPK group, exemplified by the dNMPK enzyme from λ phage T4 (56 – 58). Furthermore, AMPKs and GMPKs were proposed to share a common bacterial ancestry distinct from the archaeal-eukaryotic origin of CMPK (59).

Open in a separate window

Fig 4

Phylogenetic tree illustrating the distribution of nucleoside monophosphate kinases in bacteria and viruses. The tree was constructed using FastTree (46) based on a structural PROMALS3D alignment (47) of 200 bacterial adenylate kinases (AMPKs), 200 bacterial guanylate kinases (GMPKs), 200 bacterial TMPKs, 200 bacterial uridylate kinases (UMPKs), and 300 enzymes annotated as (putative) deoxynucleoside monophosphate kinases in bacteria and 81 dNMPKs found in viruses. Specific cases, including enzymes of P. putida KT2440, are indicated with arrows; relative phylogenetic distances are indicated in the branches of the tree when relevant.

The phylogenetic tree underscores a parallel evolution of AMPK and GMPK in bacteria with a similar distance to the TMPK group—although the precise evolutionary relationships of the nucleotide kinases cannot be inferred from the analysis (60). The UMPK group, which shares no significant sequence similarity to other nucleoside monophosphate kinases (61), is clearly positioned as an outlier outside the remaining members of the tree. The enzyme encoded by PP_3363, annotated as thymidylate kinase, is even more isolated—with no apparent relationship to either the TMPK or any other group. Interestingly, the phylogenetic analysis presents a different perspective on the relationship of dNMPKs to the remaining groups, placing them in a distinct evolutionary domain. dNMPKs from both bacteria and viruses are situated closer to bacterial CMKs, which share conserved structural features with other members of the NMP kinase family (62, 63). The dNMPK sub-tree encompassing dNMPK _Pp also harbors enzymes from the two archetypal E. coli bacteriophages T4 [i.e., Tequatrovirus T4, which recognizes dGMP, dTMP and 5-hydroxymethyl-dCMP (64)] and T5 [i.e., Tequintavirus T5, which acts on all four canonical dNMPs (65)].

Based on these observations, and to further explore the prevalence, function and evolutionary origin of enzymes in the dNMPK group of the bacterial domain, we conducted a BLASTp search (66) against non-redundant protein sequences (nrPS) using dNMPK _Pp as the query and setting a maximum E value of 0.00001. This search identified 871 proteins with the following annotations: adenylate kinase (14 cases), deoxynucleoside monophosphate kinase (667 cases), hypothetical protein (186 cases), nucleoside triphosphate hydrolase, phosphomevalonate kinase, nucleoside triphosphate hydrolase and choline dehydrogenase or related flavoprotein (1 nrPS of each of these categories). The majority of these hits (622 nrPS) were found within the Pseudomonadales order, with additional members of the γ-proteobacteria class (totaling 696 nrPS) encountered in isolates from the Moraxellales, Enterobacterales, Oceanospirillales and Alteromonadales orders. Furthermore, orthologs of dNMPK were identified in β-proteobacteria (93 nrPS), primarily within the large Burkholderiales order (83 nrPS). Additionally, 44 homologs were identified in bacterial phages—predominantly those infecting bacteria within the Enterobacteriaceae family.

To gain an even broader understanding of the distribution and function of dNMPK enzymes, we also examined the sequenced genomes of a diverse sample of organism groups. Our analysis led to several key findings. In γ-proteobacteria, the tmk gene is typically found adjacent to holB—a genetic architecture epitomized in the P. putida KT2440 chromosome (Fig. 1D). We also observed that the disruption of the tmk gene by a GI containing a dNMPK homologue occurred in several members of γ-proteobacteria, including Pseudomonas, Halomonas, Acinetobacter and Psychrobacter—with the size of these GIs varying considerably across isolates. However, despite substantial differences in the cluster length and functional content, a strong conservation of four prophage-associated genes flanking dNMPK was found (Fig. S4 in the Supplemental Material). Moreover, we found that some strains of E. coli, Shigella flexneri or Citrobacter freundii contained prophages carrying potential homologs of dNMPK _Pp as well as functional gene sequences for TMPK activities. These prophages were fully integrated into the genomes of their respective bacterial hosts and displayed various levels of decay. We further discovered both an intact tmk gene (adjacent to holB) and an unannotated DNA sequence in a strain of Pseudomonas parafulva (NS96), which, upon translation, showed 44% identity to dNMPK _Pp . Interestingly, no evidence for the presence of a prophage could be found in the surrounding genomic context in this case in particular. Finally, we identified dNMPK _Pp homologs located adjacent to the holB gene, with sequence identities between 40% and 49% to dNMPK _Pp in some β-proteobacteria members, e.g., Parapusillimonas granuli strain DSM 18079 and Alcaligenes faecalis strain JQ135, and also in the γ-proteobacterium Methylomicrobium lacus LW14 (Fig. S4). However, no gene encoding dTMP kinase was detected in these organisms. Additionally, in these cases, no genetic elements associated with mobile genetic elements could be found in the vicinity of the dNMPK gene. The genomic position of the CDS, together with the absence of a tmk gene, suggests that the dNMPK enzymes have adapted to perform the essential TMPK function in these strains—possibly as a result of a unique evolutionary path (67) or a gene-loss event (68). This evolutionary adaptation bears important implications for understanding the functional versatility of dNMPK enzymes and the extent to which they can substitute for other (d)NMP kinases in different microorganisms. Interestingly, our phylogenetic tree reconstruction highlights a close relationship between dNMPK _Pp and dNMPK from A. faecalis (dNMPK _Af , Fig. 4), implying that both enzymes could display similar substrate specificities.

Drawing from the close structural relationship between dNMPK from bacteriophage T4 of E. coli (acting, as indicated above, on dNMP, dTMP and 5-hydroxymethyl-dCMP) and the enzyme of P. putida KT2440, the next section focuses on the structural aspects of dNMPK _Pp , with emphasis on homology modeling and substrate-binding pocket analysis. On this basis, this part of the study aims at elucidating the potential function of the enzyme of strain KT2440 as a TMPK activity and to better understand its unique biochemical characteristics.

Cloning procedures and plasmid construction

All plasmids used in this work are listed in Table 1. All genetic manipulations followed protocols previously established for Pseudomonas species (97, 98). Uracil-excision (USER) cloning (99, 100) was used for the construction of all plasmids in this work, except for pMBEC∙PP_1964, which was assembled via Golden Gate cloning (101). Oligonucleotides used to construct the plasmids and bacterial strains are listed in Table S1 in the Supplemental Material. DNA fragments employed in assembly reactions were amplified using Phusion U high-fidelity DNA polymerase (Thermo Fisher Scientific) according to the manufacturer’s specifications. The identity and correctness of all plasmids and DNA constructs were confirmed by Sanger sequencing (Eurofins Genomics, Ebersberg, Germany). For genotyping experiments after cloning procedures and genome manipulations, colony PCRs were performed using the commercial OneTaq master mix (New England BioLabs, Ipswich, MA, USA), according to the manufacturer’s instructions. E. coli DH5α λpir (Table 2) was employed for cloning procedures. Chemically competent E. coli cells were prepared and transformed with plasmids according to standard protocols (94); P. putida was rendered electro-competent following the method by Choi et al. (102). The different strategies followed for allelic replacement in the GI locus are summarized in Fig. 1D. In the “stitching” approach, the tmk gene of P. putida KT2440 shared an identical sequence with 33 CDSs encoding TMPK activities in γ-proteobacteria (accessible at the NCBI website with accession number WP_020192642.1).

Mass spectrometry-aided targeted proteomics

Overnight pre-cultures of P. putida strains were grown in DBM medium supplemented with 30 mM glucose. The main cultures, consisting of 25 mL of the same medium, were prepared in 250 mL Erlenmeyer flasks and inoculated with an initial OD₆₀₀ of 0.05. During the mid-exponential phase (Fig. 2A), samples were collected and cells were harvested through centrifugation at 17,000 × g for 2 min at 4°C. Following supernatant removal, the cell pellets were frozen and stored at –80°C until quantitative proteomic analysis, conducted as previously described (103, 104). A protein database comprised of the P. putida reference proteome [UP000000556 (20)] was utilized to assign the detected peptides to their respective functions, along with heterologously expressed proteins. Proteomic data obtained as described above were analyzed using a customized R script within RStudio (version 2021.09.2). Initially, only entries with complete values were retained, filtering for proteins detected in every replicate of at least one condition. The abundance values were log₂-transformed and normalized via variance stabilization normalization using the versatile vsn package (105). Subsequently, the dNMPK _Pp abundance in the experimental strains tested was extracted from the data set and represented graphically after normalization against the protein content of P. putida strain KT2440 (104).

Phylogenetic analysis and construction of phylogenetic trees

Amino acid sequences of enzymes belonging to the group of nucleoside monophosphate kinases were extracted from Uniprot (106). Queries were used with the respective protein EC numbers and “Bacterium” or “Viruses” as taxonomy identifiers. Entries for AMPK, CMPK, GMPK, TMPK and UMPK were filtered for “reviewed” entries. A filter was applied to retain only proteins with a length between 180 and 300 amino acids; proteins with identical sequences were removed. Then, a sample of proteins for each functional class was randomly drawn as indicated in the text and in the figure captions. After this sampling, the sequences of proteins found in P. putida KT2440 were added to the data sets. A single FASTA file, containing all sampled protein sequences, was used to perform a structural alignment via Promals3D (55) using the standard settings. The alignment in FASTA format was trimmed to remove gaps not present in at least 10% of all sequences using trimAl version 1.2 (107). The phylogenetic tree was constructed with FastTree version 2.1.12 (54) using the command (FastTree.exe -quote -wag -cat 20 alignment_trimmed.fasta).

The authors are indebted to Prof. V. de Lorenzo and Dr. A. Sánchez-Pascuala (CNB-CSIC, Spain) for insightful discussions on early experiments that led to seminal observations relevant to the present study. The financial support from The Novo Nordisk Foundation through grants NNF20CC0035580, LiFe (NNF18OC0034818), and TARGET (NNF21OC0067996), and the European Union’s Horizon 2020 Research and Innovation Program under grant agreement no. 814418 (SinFonia) to P.I.N. is gratefully acknowledged.

N.T.W. and P.I.N. conceived the idea; N.T.W. executed most of the experimental plan and outlined the manuscript structure; K.R. performed molecular biology work; N.T.W. and A.D. wrote the initial draft; N.T.W., A.D., and P.I.N. contributed to writing and editing and finalized the manuscript. All authors have made direct and intellectual contributions to the editing before final submission and approved the manuscript.

The authors declare no conflicts of interest that might be perceived as affecting the objectivity of this paper.