Abstract

1. Introduction

Given the world's finite amount of fossil fuels, the need for renewable sources of fuels and commodity chemicals is becoming increasingly important. To address this issue, much work has gone into the production of these chemicals from biological sources. The majority of research to date has focused on the conversion of cellulosic biomass from plants into value added chemicals. However, recent reports have shown that biorefineries will need to valorize lignin streams to be economically viable (Davis et al., 2013). Lignin comprises 15–40% of terrestrial plant biomass (Ragauskas et al., 2014). Unlike cellulosic biomass, which is largely comprised of homogenous mixture of polysaccharides, lignin is a heterogeneous and interlinked polymer comprised of aromatic compounds which has limited its use in biorefineries. Currently, lignin streams in most biorefineries are slated for combustion for the generation of process heat and electricity. Several groups have used a combination of chemical and/or biological means to depolymerize lignin, and some have taken advantage of aromatic-consuming bacteria, such as Pseudomonas putida species, to transform lignin streams into value added chemicals (Linger et al., 2014, Rahimi et al., 2014, Vardon et al., 2015, Zakzeski et al., 2012).

A major limiting factor for rapid metabolic engineering of Pseudomonas putida is the availability of tools for sophisticated genetic engineering. The primary tools currently used for directed genetic engineering in P. putida KT2440, replicating plasmids such as derivatives of the pBBR family of plasmids (Kovach et al., 1995) and homologous recombination-based gene replacement technologies (Johnson and Beckham, 2015, Schafer et al., 1994), have many shortcomings. Replicating plasmids typically require constant selection for maintenance, have highly variable copy number (Lin-Chao and Bremer, 1986, Paulsson and Ehrenberg, 2001), and often come with a fitness cost (Mi et al., 2016). Current homologous recombination-based technologies rely on inefficient plasmid integration by host DNA repair enzymes, leading to low transformation rates. Recombination of these plasmids by single cross-over events leads to genome structures that are inherently unstable, allowing deletion or integration of DNA sequences, but simultaneously makes them unsuitable for rapid prototyping. In an attempt to address some of these shortcomings, P. putida genetic engineering tools have been expanded to include oligonucleotide (Aparicio et al., 2016), λ Red/Cre (Luo et al., 2016), and meganuclease (Martinez-Garcia and de Lorenzo, 2011) recombineering technologies.

Site-specific recombinases are enzymes that catalyze recombination between two specific sequences of DNA (for a review see (Brown et al., 2011)). Recombinases Flp and Cre are commonly used in molecular genetics, but these enzymes recognize identical sites and perform reversible recombination between the sites. Serine recombinases such as ΦC31 integrase, on the other hand, natively catalyze unidirectional phage genome integration through the recombination between attP (phage genome) and attB (bacterial genome) sequences, generating distinct attL and attR sequences in the process. Unlike the unidirectional λ phage integrase from E. coli that requires the E. coli protein IHF for recombination, these serine recombinases do not require any host factors and have been used for chromosomal insertion of heterologous DNA in organisms across the tree of life (Brown et al., 2011, Guss et al., 2008, Keravala et al., 2006, Siuti et al., 2014, Thomson et al., 2012, Xu and Brown, 2016). In addition to ΦC31 integrase, other serine recombinases have been identified and characterized, including the determination of the corresponding attB and attP sites enabling these systems to also be developed as genetic tools (Brown et al., 2011).

In other organisms, tuning protein expression can be very important for achieving increased yields from engineered pathways (Alper et al., 2005, Jones et al., 2015, Xu et al., 2013), but there are no well-characterized promoter or RBS libraries in P. putida to enable rational tuning of protein expression. Previous engineering efforts in P. putida have primarily relied on the lac family promoters (lac, lacUV5, tac, trc) from E.coli (Borrero-de Acuna et al., 2014, Johnson and Beckham, 2015, Meijnen et al., 2008, Meijnen et al., 2009, Nielsen et al., 2009, Vardon et al., 2015), with a few using native promoters such as Pm (Gemperlein et al., 2016), rrn (Wang et al., 2010), or PP_1099 promoter (Lin et al., 2016). The relative expression of the lac-family promoters in P. putida remains unknown, and the use of native promoters increases the likelihood of cryptic promoter regulation. To address these gaps in genetic tools, we tested serine recombinases to develop a high efficiency integration system to characterize a library of genetic elements to enable rational tuning of protein expression in P. putida.

2. Materials and methods

2.2. Pseudomonas putida transformation and strain construction

2.2.1. Construction of hsdR deletion and integrase replacement strains by homologous recombination

P. putida KT2440 was used as a wild-type parent strain for all strains. hsdR deletion and integrase replacement strains were constructed using the pK18mob-sacB kanamycin resistance/sucrose resistance selection/counter-selection marker system (Marx, 2008) as described in detail previously (Johnson and Beckham, 2015). Briefly, transformations were carried out by electroporation of ~500–1000 ng of plasmid DNA into KT2440 cells. Transformed colonies were selected by growth overnight at 30 °C on LB agar medium with 50 µg/mL kanamycin sulfate. Resulting transformed colonies were single colony purified and incubated overnight at 30 °C on LB agar medium containing 50 µg/mL kanamycin sulfate to eliminate residual untransformed cells from being transferred. For counter-selection, colonies were streaked onto YT+25% sucrose plates (10 g/L yeast extract, 20 g/L tryptone, 250 g/L sucrose and 18 g/L agar) and incubated overnight at 30 °C. Resulting colonies typically have recombined to remove sacB from the genome, as its expression causes a significant growth defect in the presence of sucrose. However, cells containing sacB can grow very slowly in the presence of sucrose, so colonies were streaked for isolation again on YT+25% sucrose plates and incubated at 30 °C overnight to reduce the possibility of transferring cells in which sacB remains in the genome. The resulting colonies (typically 20) were cultured in LB broth overnight at 30 °C and screened for either the deletion of hsdR or replacement of hsdR with a P_tac-integrase/attB cassette by colony PCR and kanamycin sensitivity. P. putida integrase strains are listed in Table 1.

Table 1

P. putida strains constructed for bacteriophage serine integrase testing.

Strain	Integrase/attB	Genotype
JE1646	None	hdsR
JE1644	FBT1	hdsR::Ptac:ΦBT1int-attB
JE1645	FC1	hdsR::Ptac:ΦC1int-attB
JE1643	RV	hdsR::Ptac:RVint-attB
JE90	BxB1	hdsR::Ptac:BxB1int-attB

Open in a separate window

2.2.2. Plasmid transformation assays

Plasmid transformation assays were performed using a variation of the transformation procedure described in Section 2.2.1. 600 ng of each plasmid was transformed into either wild-type, JE90, JE1643-1646 (Table 1) electrocompetent cells. Following electroporation, cells were resuspended in 950  μl SOC, transferred to a microfuge tube, and incubated at 30 C to allow recovery. Due to the low efficiency of homologous recombination, all of the recovery volume was plated for pΔPP_0545. For the remaining plasmids, several fractions of the recovery cultures were plated and enumerated. For Fig. 1 assays, all samples were plated on LB agar supplemented with 50 μg/mL kanamycin sulfate, incubated for 24 h at 30 °C, then enumerated. For Fig. 3 assays, samples were plated on LB agar media supplemented with 15 or 50 μg/mL kanamycin sulfate, incubated for 16 h at 30 °C, enumerated, incubated for an additional 24 h (total 40 h) again at 30 °C, and enumerated a second time. Correct insertion of DNA into the attB site was confirmed using multiplex PCR as described in Supplemental Figs. 2 and 3. Analogous to the E. coli CRIM system (Haldimann and Wanner, 2001b), multiplex PCR using a 4 primer set (primers 1–4 Supplemental Figs. 2 and 3) were used to verify plasmid integration. Primers 1 and 2 flank the attB and primers 3 and 4 flank the attP. After site-specific integration, primers 1 and 4 flank the resulting attL and primers 2 and 3 flank the resulting attR. The sizes of the attB/attP/attL/attR PCR products are all different and can be resolved on an agarose gel. If three bands representing attL, attR, and attP are present, it indicates that in addition to integration via the phage integrase, there is an additional copy of the plasmid present. This second copy likely inserted via homologous recombination into the first insertion or integration of a concatamer from E. coli, preserving an attP site.

Open in a separate window

Fig. 1

Serine integrase activity in Pseudomonas putida KT2440. (A) Graphic representation of integrase-catalyzed between the BxB1-attP (P) of pPolyAttP and the BxB1-attB (B) site in the genome of ΔhsdR::BxB1int-attB, with arrows representing putative promoter sequences. (B) Chart of plasmid transformation efficiency in 6 strains with a replicating (black), suicide (white), or integrase target plasmid (gray), with error bars representing the standard deviation in 3 replicate assays.

Open in a separate window

Fig. 3

Design and testing of improved integrase target plasmids. (A) Graphic representation of integrase-catalyzed site-specific recombination between the BxB1-attP (P) of 5 plasmids and the BxB1-attB (B) site in the genome of ΔhsdR::BxB1int-attB using glyphs from the Synthetic Biology Open Language Visual graphical notation. Numbers above primer arrows represent the primers used for screening as described in Supplemental material. Red arrow (KanR) and Pink (KanR2) represent kanamycin resistance markers from pUC57 and pK18mobsacB, respectively, expressed from the original promoters from the source vectors. Green arrow (mNeon) represents mNeonGreen ORF. Directionality of terminators is not indicated, but double terminators are in opposing orientations to insulate elements. (B & C) Number of colony forming units observed after 1 (light gray) or 2 (dark gray) days of incubation following transformation of ΔhsdR::BxB1int-attB with the replicating pBBR1-MCS-2 plasmid or 5 BxB1-attP containing plasmids. All error bars represent standard deviation in colony counts from 3 independent transformation experiments. Results are charted for selection on (B) 50 mg/L and (C) 15 mg/L kanamycin sulfate. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article).

2.2.3. Construction fluorescent reporter strains

Fluorescent reporter strains for promoter and ribosomal binding site characterization were constructed by a variant of the transformation and strain construction procedure described previously (Johnson and Beckham, 2015). Electrocompetent JE90 cells were transformed with 200–400 ng of plasmid DNA. Following recovery outgrowth for 1 h, cells were plated onto LB agar medium supplemented with 15 μg/mL kanamycin sulfate. Plates were incubated at 30 °C and colonies that appeared overnight were marked. The largest of these colonies were often found by colony PCR to have integrated plasmid concatemers (or other forms of multiple copy plasmid integrants). Plates were further incubated overnight at room temperature (~24 °C) and moderately sized isolated colonies were patched to LB agar plates and incubated overnight. Colonies were screened by colony PCR to confirm single copy plasmid integration events.

3. Results

3.2. Utilization of BxB1 integrase to rapidly test protein expression in P. putida

We used the BxB1 serine integrase to rapidly construct strains containing a promoter library based upon the tac promoter. Wild type P. putida does not encode the transcription factor LacI, preventing regulation via the lacO operator sequence. Specifically, we characterized the rates of mNeonGreen production with 36 constitutive σ⁷⁰ promoters, derived from all combinations of 6 distinct −35 and −10 box sequences (Fig. 2A, Supplemental Table S4), during mid-logarithmic growth. Promoters are named JE[1][2][3][4][5][6] (see Fig. 2A), where values in the [2] and [4] positions represent −35 and −10 variants respectively. Values in the [1, 3, 5], and [6] positions refer to variants of upstream (of −35) sequences, sequences between −35 and −10 elements, downstream (of −10) sequences, and further downstream operator/5′UTR sequences respectively. Sequences used here for −35 and −10 variants were derived from their respective sequences in a small lac promoter variant library used to tune expression of LacZ in E. coli (Brewster et al., 2012). Promoters were cloned upstream of an insulated mNeonGreen expression cassette in the plasmid pJE483 (Fig. 2A), which contains a kanamycin resistance cassette and BxB1 attP site. Plasmids were transformed into the P. putida strain JE90 (KT2440 ΔhsdR::BxB1int-attB), and single copy integration at the BxB1 attB site was verified by colony PCR (data not shown). To check for read-through transcription, we compared strains harboring integrated pJE482 (no mNeonGreen gene) and pJE483 (no promoter) plasmids. Fluorescence was not significantly different between the two strains, demonstrating the insulation of the reporter gene (data not shown). Background fluorescence was accounted for by subtracting fluorescence of the promoterless reporter strain from all samples. In addition to the tac promoter (Fig. 2B – dark gray bar), we used the E. coli lac (Fig. 2B – white bar) and lacUV5 (Fig. 2B – light gray bar) promoters as references for promoter strength. Unsurprisingly, the tac promoter, which contains consensus −35 and −10 sequences for σ⁷⁰ promoters but non-consensus spacing of 16 bases rather than 17, produced the most mNeonGreen protein. Similar to results in E. coli, expression of the reporter with the tac promoter is 5.3-fold higher than with the lacUV5 promoter (de Boer et al., 1983) and 40-fold higher than with the lac promoter. These promoters cover a 72-fold range of expression, with the tac promoter being highest and the JE121511 promoter being lowest (Fig. 2B – blue bars, Supplemental Table S6).

Open in a separate window

Fig. 2

mNeonGreen production assays to determine promoter and ribosomal binding site (RBS) strength in Pseudomonas putida KT2440 strain ΔhsdR::BxB1int-attB. (A) Graphic representation of reporter plasmid integration with a diagram of promoter and RBS components used in the study. (B) mNeonGreen production by strains harboring mNeonGreen driven by members of the −35 and −10 variant constitutive promoter library. The −35 and −10 variant sequences are indicated above the chart, and particular variants in each promoter indicated beneath the chart. Positions [1, 3, 5], and [6] of the promoter are all variant 1. (C) mNeonGreen production by strains harboring one of three −35 variants with positions [3, 4, 5], and [6] of the promoter all variant 1, and one of five position [1] (UP-element) variants. (D) mNeonGreen production by strains harboring one of three −35 variants with positions [1, 3, 4, 5], and [6] of the promoter all variant 1, and one of 11 RBS variants each. (B-D) All mNeonGreen production values represent the average of three independent experiments, with error bars representing the standard deviation of mNeonGreen production. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article).

In addition to the −35 and −10 boxes, some promoters contain elements upstream of the −35 box, called UP-elements (Fig. 2A), which can further increase transcription rates from σ⁷⁰ promoters (Ross et al., 1993). UP-elements are typically species-specific and are commonly found in ribosomal RNA promoters. These elements are thought to function by increasing binding affinity for the alpha subunit of RNA polymerase. In an attempt to further increase the range of expression in the library, we incorporated 4 similar, but distinct, UP-elements from the ribosomal RNA promoters in KT2440 upstream of 3 constitutive promoters from the library (Supplemental Table S4). Generally, when one is using an UP-element, it is to increase transcription rates, so we chose three high (JE111111) to medium-high (JE121111 & JE151111) strength promoters to use as a test platform for incorporating UP-elements into synthetic promoters. Incorporation of rRNA promoter UP-elements increased expression from the JE111111 (tac promoter) by ~1.7 to 2-fold and JE121111 promoter by ~4 to 5-fold (Fig. 2C, Supplemental Table S7). Unexpectedly, incorporation of the same UP-elements significantly decreased expression from the JE151111 promoter by ~12 to 23-fold.

Protein expression can also be tuned post-transcriptionally by many factors including start codon sequence (O'Donnell and Janssen, 2001), ribosomal binding site sequence and spacing (Salis et al., 2009), RNA secondary structure (Mortimer et al., 2014), and translation inhibiting RNA-protein interactions (Moreno et al., 2007). The original 5′ UTR and RBS sequence used is a derivative of the constitutive expression cassette in pTAC-MAT-Tag-2 (Sigma Aldrich). To further tune protein expression, we generated 10 RBS variants by mutating 4 nucleotides of the core ribosomal binding site (Supplemental Table S5) in plasmids containing promoters JE111111 (tac), JE121111, and JE151111 (Fig. 2D, Supplemental Table S8). The RBS variants utilized with each promoter mostly clustered into two groups, with RBS JER07 being in between, and protein production levels varied by ~6–7-fold from the highest to lowest strength RBS.

4. Discussion and conclusions

Rapid metabolic engineering of organisms such as P. putida will be critical for efforts to create organisms that can add value to the lignin fraction of plant biomass. The BxB1 integrase system developed here for P. putida will dramatically increase the speed of the build phase in the iterative design-build-test-learn cycle, allowing constructs to be ready for testing in two days. This approach is substantially faster than using homologous recombination to stably “knock-in” genes into the chromosome, which takes a minimum of five to six days. The unidirectional site-specific recombination also eliminates the instability and copy number issues associated with autonomously replicating plasmids. Further, replicating plasmids are generally unsuitable for industrial use as they require constant selection, which is cost-prohibitive; our integration system should provide more reliable predictive value for chromosomal protein expression because all constructs are at single copy. Additionally, we have routinely integrated 6 kb plasmids with the Bxb1 integration system, each containing a 3.5 kb pathway, without a noticeable drop in transformation efficiency (unpublished data, Elmore & De Capite). We expect that any size limitations for plasmid integration will be determined by the ability to physically transfer large DNA molecules into the cell. While the first generation plasmid designs worked best with the use of lower kanamycin concentrations or longer incubation times, the development of a second generation of plasmids eliminated these drawbacks and will make this toolset easier to use.

The utility of serine integrases has been demonstrated in a broad range of organisms, including mammals (Keravala et al., 2006), plants (Thomson et al., 2012), yeast (Xu and Brown, 2016), archaea (Guss et al., 2008), and bacteria (Siuti et al., 2014). Many other serine integrases are used as genetic tools in other organisms (Brown et al., 2011, Keravala et al., 2006, Xu and Brown, 2016), and many of them may be suitable for expanding the repertoire of genetic tools in P. putida. Similar to the CRIM system in E. coli (Haldimann and Wanner, 2001a), the development of a temperature-sensitive replicating plasmid for P. putida would also expand the utility of serine integrases and serve as a platform for many other recombineering tools. To our knowledge, no temperature sensitive plasmids have yet been developed for P. putida, but this advance would add further value to the serine recombinase tools developed here by enabling transient expression during the initial integration as well as unidirectional removal of antibiotic resistance genes in a manner similar to the bidirectional Flp-frt and Cre-lox systems. This system could also be expanded by using other serine recombinases for which attB and attP sites have been characterized (Brown et al., 2011).

Many enzymes and pathways require a delicate balance of subunits (Davis et al., 2000) and pathway components for optimal activity (Jones et al., 2015). Here we have begun the development of a suite of genetic elements to tune expression in P. putida. For example, the JE711111 promoter (tac + rRNA UP-element) would be suitable for very high expression of rate limiting cytosolic enzymes. A medium-high expression promoter, such as JE131111, could be used for non-rate limiting enzymes to reduce the metabolic load of an exogenous pathway. A medium-low promoter like JE131311 may be more suitable for circumstances where very strong expression could be inappropriate, such as for transcription factors or membrane proteins. Interestingly, not all promoters performed as expected. In particular, variants of JE151111 with the addition of any of the four UP elements significantly reduced rather than enhanced mNeonGreen production. Because of the enigmatic nature of this decrease, we would recommend avoiding these four promoter variants. Ultimately, this library is a starting point, and we expect that additional tuning of RBS, UP, −35 box, and −10 box elements will allow researchers to further broaden the dynamic range for rationally designed protein expression. Furthermore, we expect the knowledge gained from tuning the −35 and −10 box elements in the tac promoter will be useful for rationally tuning expression in other σ⁷⁰ promoters of interest, such as those with useful transcription factor operator sequences. The mNeonGreen reporter gene construct will also serve as an ideal testing device for the characterization of other genetic parts, such as terminators or more complex regulatory circuits.

In conclusion, the development of the BxB1 integrase system and promoter library will significantly alter the landscape of what is possible in P. putida engineering. The decreasing cost of synthetic DNA (Kosuri and Church, 2014), affordable high-throughput DNA sequencing technologies (Goodwin et al., 2016), automated high-throughput assays, and efficient genome editing technologies has allowed high-throughput engineering efforts to become a reality in model organisms such as Escherichia coli and Saccharomyces cerevisiae. With the genome editing technology developed here, P. putida is much closer to fully exploiting the same technological advances.

Declarations of Interest

We wish to confirm that there are no known conflicts of interest associated with this publication and there has been no significant financial support for this work that could have influenced its outcome.

Acknowledgements (includes funding sources)

Footnotes

References