Construction of pSAM_Rl and transposition frequency within the Rhizobiaceae

The MmeI-adapted mariner transposon suicide delivery vector pSAM_Rl was constructed from a previously described MmeI-adapted mariner delivery vector pSAM_Bt [4]. To do so, the 974 bp Ery^R in pSAM_Bt was replaced with a 979 bp Neo/Kan^R resistance cassette from pSC189 and the 304 bp Bacteroides thetaiotaomicron rpoD promoter region was replaced with a 366 bp RLV3841 rpoD promoter region. A map of the pSAM_Rl construct is shown in Figure 1, with restriction enzyme sites used for cloning indicated. The pSAM_Rl construct was maintained in E. coli strain SM10λpir as this strain harbours both the λpir gene, required for the replication of R6Kƴ oriR, and chromosomally integrated tra genes, required for conjugative transfer via the plasmid borne RP4-oriT. Analysis of the transposition frequency of the pSAM_Rl mariner transposon was evaluated via conjugative transfer of the pSAM_Rl suicide vector from E. coli SM10λpir into R. leguminosarum, S. meliloti, and A. tumefaciens. Transposition frequency was highest in RLV3841, yielding an average of 2.01 × 10⁻⁴ transposon mutants per recipient cell. The frequencies of transposition in A. tumefaciens and S. meliloti were observed to be 8.04 × 10⁻⁵ and 2.54 × 10⁻⁵ mutants per recipient cells, respectively.

Analysis of pSAM_Rl transposition in the RLV3841 genome

Distribution of phenotypic classes throughout the RLV3841 genome

Each RLV3841 replicon was analyzed using the Tn-HMM python module [10] to classify genes based on the observed density of Tn insertions in each gene within the mixed mutant cell population. For example, a gene which had no detectable Tn insertion sites following DNA sequencing from the pooled mutant DNA suggests cells carrying a mutation in this gene were not maintained in the population. These genes are described as conferring an essential phenotype. The procedure is described in greater detail in the Materials and Methods.

Specific essential genes within the RLV3841 genome

Transposon mutagenesis with pSAM_Rl in RLV3841

We have successfully demonstrated the implementation of a mariner class transposon to mutagenize selected species within the family Rhizobiaceae. We found that the MmeI-adapted mariner transposon harbored on pSAM_Rl could mutagenize R. leguminosarum, A. tumefaciens, and S. meliloti at a high frequency. In the case of RLV3841, it was observed that this high frequency of mutagenesis was dense enough to generate saturating libraries of transposon insertion mutants. We were able to generate transposon mutant libraries that saturated the RLV3841 genome with an insertion density of 0.88, which is higher then the insertion density of the data set used for validation of the HMM used for analysis [10]. This suggests that the combination of mariner based transposon insertion sequencing with the Bayesian based HMM analysis would yield accurate and full genome level results. Furthermore, our analysis used approximately 2.3 M reads. And increased read depth could help increase confidence in the analysis, particularly in resolving the phenotypic classifications of essential and growth defective.

We also observed no bias of mariner transposon insertion across the RLV3841 chromosome and plasmids, suggesting that the transposon inserts randomly and therefore should allow for reliable whole-genome screening approaches. Analysis of read depth showed that there was also little bias in mean read-count per insertion, except in the case of pRL7, which was slightly higher than the other replicons. We suggest that this is a result of pRL7 being maintained at a higher copy number than the other megaplasmids and chromosome, and so sampling of the mutant pool DNA would results in a higher sampling of pRL7 transposon insertion tags than the other plasmids, or chromosome. When the increase in mean read counts is taken into consideration with the insertion density, it appears that although the higher copy number of pRL7 resulted in a higher mean read depth, it did not increase the insertions density, suggesting that the saturation of pRL7 had reached a plateau and an increased presence of pRL7 did not result in a corresponding increase in the number of unique insertions sites.

Analysis of RLV3841 INSeq using a hidden Markov model

Analysis of the TY INSeq data set with the Tn-HMM analysis package assigned accurate phenotypic classification to several genes thought to be essential housekeeping genes necessary for growth under normal conditions. The analysis showed that insertions in genes required for ATP synthesis were absent in the mutant pools. These genes are expected to be essential due to their central role in metabolism and their designation as essential in this assay supports the validity of using the MmeI-adapted mariner in conjunction with the HMM analysis as our INSeq methodology. Furthermore, visual investigation of the transposon insertion density around the region encoding the ATP synthase genes revealed a high insertion density leading up to and after the genes, further supporting the conclusion that the technique is robust and can discern regions of essentiality from those of other states at a high resolution.

In a few instances, a gene expected to be essential was observed to be neutral. RL4697 is annotated as a putative DNA polymerase III alpha sub-unit, and is therefore predicted to be required for proper DNA polymerase function; however, the gene was classified in the neutral category with 96.1% of all potential insertion sites in the locus observed to have insertions. We suggest that RL4697 may be misanotated, as RL1723, another DNA polymerase III alpha subunit, was observed to be essential (Table 3). This highlights another potential use of INSeq in the Rhizobiaceae for validation and quality improvement of genome annotations.

Five of the megaplasmids in the RLV3841 genome were observed to have a set of 3 plasmid replication genes that the INSeq approach identified as essential for plasmid replication and maintenance. The exception was pRL7 which has two sets of replication genes [41] and therefore functional redundancy may have complicated the the classification of the pRL7 rep genes into the phenotypic classes. The classification of the rep genes on each plasmid as essential provides validation of the INSeq method. Tn insertion within a replication locus would result in the loss of the plasmid from the mutant cell populations harvested for DNA extraction and INSeq DNA sequencing. This result highlights the value of the INSeq approach in the genetic characterization of novel plasmids, as the method is able to identify plasmid encoded genes that are required for plasmid replication and maintenance. Furthermore the method can identify plasmid-encoded genes that provide a fitness advantage to the host under specific growth conditions based on an observed low Tn insertion density.

The genes from previously described mutants with growth defects on TY medium were also examined. Previous work identified [38] several genes that are important for growth on TY medium. When we compared those results with the results produced by the INSeq analysis of RLV3841 grown on TY medium we observed good concordance with these previously published results. In our results, the four TY related genes were all observed to result in a growth-defective phenotype after ~16 generations of growth, which is in agreement that the interruption of these genes via mutagenesis will result in impaired growth on TY.

Bacterial strains, growth conditions and plasmids

Construction of pSAM_Rl

Plasmid DNA was isolated using GenElute™ Plasmid Miniprep Kit (Sigma-Aldrich). The Neo/Kan^R cassette within pSC189 [48] was PCR amplified using primers Tn189KmR_Fwd_XhoI and Tn189_Rev_XbaI (Additional file 1: Table S1) such that the XhoI and XbaI restriction enzyme sites were introduced on the 5’ and 3’ end, respectively. The 991 bp Neo/Kan^R PCR product was subsequently cloned into the pGEM®-T Easy Vector System creating plasmid pGEM::189Km^R. The Neo/Kan^R cassette in pGEM::189Km^R was digested with XhoI and XbaI restriction enzymes creating a 979 bp fragment that was directionally cloned into the pSAM_Bt mariner transposon. The resulting plasmid pSAM_Km was maintained in the E. coli strain PIR1(Invitrogen), which allows for high copy number maintenance of R6Kƴ-oriR plasmids.

Cloning of the RLV3841 rpoD promoter region was carried out by PCR amplifying a 366 bp region upstream of the rpoD gene start codon using primers Rlv_rpoD_Pro_Fwd and Rlv_rpoD_Pro_Rev (Additional file 1: Table S1). The rpoD promoter PCR product had a 5’ BamHI and 3’ NdeI restriction enzyme site introduced, and was subsequently cloned into the pGEM®-T Easy Vector System to create the plasmid pGEM::rpoD. The RLV3841 rpoD promoter region was then excised from pGEM::rpoD using NdeI and BamHI, and directionally cloned into pSAM_Km to create the vector pSAM_Rl (Figure 1). The new pSAM_Rl suicide vector carried an MmeI-adapted mariner transposon harbouring a Neo^R/Kan^R cassette, and had the himar1C9 transposase transcriptionally fused to a RLV3841 rpoD promoter. For use in transposon mutagenesis, the pSAM_Rl construct was electroporated into E. coli SM10λpir (obtained from Dr. Peter Howard, University of Saskatchewan).

Testing pSAM_Rl transposition frequency

Transposition mutagenesis using pSAM_Rl was done in triplicate. Donor and recipient cells were grown in broth culture to an OD₆₀₀ of approximately 0.8 and were pooled in a ratio of 1000 μL recipient to 500 μL of donor in a 1.5 mL microcentrifuge tube. The conjugation mixture was pelleted at 12,500 rpm for 3 min, washed once with 1000 μL 1X PBS, and resuspended in approximately 100 μL 1X PBS. The cell suspensions were then spotted onto pre-warmed TY agar plates and incubated at 30°C overnight. Conjugation spots were scraped and resuspended in 1000 μL of 1X PBS. Enumeration of transposon mutants was done using TY agar supplemented with 50 μg/mL Neo and the appropriate Rhizobiaceae counter selectable antibiotic (Table 4). Enumeration of total Rhizobiaceae was done on TY agar with the species specific selectable antibiotic.

Generating transposon mutant libraries for sequencing

Six independent conjugations of pSAM_Rl into RLV3841 were conducted on TY agar as described above. After 24 hours incubation at 30°C each of 6 mating spot was scraped and resuspended in 1 mL of 1X PBS and then pooled together in a final volume of 6 mL. For selection on TY agar, 1000 μL of resuspended cells were plated across 2 separate 245 × 245 mm² (Corning) Neo + Str TY agar plates, in triplicate and incubated for ~48 hour at 30°C.

Following incubation, a faint film like growth was scraped off each plate and resuspended in 1 mL of 1×PBS, vortexed for 1 minutes, and then pelleted at 15, 000 RPM for 10 minutes. The supernatant of each resuspension was very viscous and still contained cells, it was equally aliquoted into 2 × 1.5 mL microcentrifuge tubes. The original pellet, and two tubes of supernatant, were then brought up to a final volume of 1000 μL with 1 M NaCl, vortexed thoroughly, and incubated on ice for 1 h to disrupt the viscous exopolysaccharide diffuse capsule to better collect the cells. The NaCl cell suspensions were then pelleted at 15, 000 RPM for 10 minutes, and the pellets from each replicate were pooled independently and resuspended in 1000 μL of TE buffer (pH 8.0). The resulting 3 mutant pools were used for independent DNA isolation and downstream library preparation.

Preparing sequencing libraries and DNA sequencing

Transposon insertion tags (Tn-tags) consisted of 53 bp of pSAM_Rl transposon sequence, including the 27 bp inverse repeat sequence, and 15–16 bp of adjacent genomic DNA. Library preparation was carried out independently for each of the 3 collected Tn-mutant pools. Tn-tags were prepared for DNA sequencing using a modified version of the INSeq method [49] to make the sequencing process amendable to the Ion Torrent PGM sequencing platform. Linear PCR products were amplified using the primer Ion Torrent BioSAM (Additional file 1: Table S1), with an annealing temperature of 58.6°C and 500 ng of template DNA. Linear PCR products were purified using a QIAquick PCR Purification Kit (Qiagen) according to the manufacturer's recommended protocol. The biotinylated linear PCR products were then bound to Pierce Streptavidin Magnetic Beads (Thermo Scientific) and enzymatic steps during library preparation were performed as described [49] with the substitution of Klenow (New England Biolabs), Random Primer 6 (New England Biolabs) and T4 DNA Ligase (New England Biolabs). Additionally, a custom library adapter, INSEQ_Adpt, was used in the adapter ligation step. The final PCR amplification of sequencing template was accomplished using fusion primers designed in accordance with Ion Amplicon Library Preparation (Fusion Method, Pub#: 4468326 Rev. C), using the PCR amplification conditions described in the INSeq methodology [49]. The forward fusion primers IT_A_FP_1, IT_A_FP_2, and IT_A_FP_3 included IonXpress barcode sequences 1, 2, and 3 respectively, for downstream sequence separation. The reverse primer IT_trP1_FP was used in conjunction with a forward primer to introduce the trP1 sequencing adapter. The final sequencing template prepared from the Ion Amplicon library preparation was 187 bp in length. Sequencing template was gel purified using the Invitrogen E-Gel® SizeSelect™ system, and was analyzed using a Bioanalyzer High Sensitivity DNA Chip (Agilent Technologies) prior to sequencing for quality and molarity. The three technical replicates had a final concentration of 1.37, 1.82, and 1.23 μg/ul of sequencing library after size selection, respectively.

DNA sequencing was performed on the Ion Torrent PGM using 200 bp sequencing chemistry and a 316v2 sequencing chip. The total raw sequencing output of the Ion Torrent was 1.25, 1.73, and 1.74 million reads for each of the 3 replicates. The raw sequencing reads were then pooled for downstream data extraction an analysis.

Data extraction and transposon insertion analysis

Quality trimming to Q20 and trimming of adapter sequences was performed using cutadapt [50] and the final 16–15 bp tn-tags were checked for a leading 'TA' motif using a custom python script. The resulting 3,192,486 transposon insertion tags were mapped to the R. leguminosarum bv. viciae 3841 reference genome (RefSeq:NC_008378.1 to NC_008384.1) [51] using the Bowtie short read aligner [52], allowing for no mismatches in the alignment, and only reporting insertion tags that mapped to a single unique location. Short read alignment resulted in 2,319,239 unique transposon insertion tags mapping to the RLV3841 reference genome, after 131,676 reads were ommitted due to multiple alignments, and 509,341 reads failed to align. The .sam output file from the Bowtie alignment was converted into a .bam format using Samtools [53], and was then converted to .bed format using bedtools. Transposon insertion reads were grouped by specific RLV3841 replicons for downstream analysis. The .bed files of the aligned transposon insertion tags were converted to .wig format using a custom python module developed in house. The .wig formatted INSeq data sets generated for each of the 7 replicons in the RLV3841 genome were then analyzed independently using the Tn-HMM python module [10]. Briefly, the python module used a HMM as described in [10], in conjunction with the Viterbi algorithm to calculate the state of each 'TA' insertion site within the genome, independent of gene boundaries. Next, the computer module analyzed the state of successive 'TA' sites within gene boundaries to assign a state for the gene as a whole (See Additional file 2 for the RLV3841 chromosomal output). Four phenotypic classifications are possible: essential, growth defective, neutral, and growth advantage. Figure 2 provides a visual example of the four phenotypic classifications found within a selected region of the RLV3841 genome.