Potential complications when developing gene deletion clones in Xylella fastidiosa

Background The Gram-negative xylem-limited bacterium, Xylella fastidiosa, is an important plant pathogen that infects a number of high value crops. The Temecula 1 strain infects grapevines and induces Pierce′s disease, which causes symptoms such as scorching on leaves, cluster collapse, and eventual plant death. In order to understand the pathogenesis of X. fastidiosa, researchers routinely perform gene deletion studies and select mutants via antibiotic markers. Methods Site-directed pilJ mutant of X. fastidiosa were generated and selected on antibiotic media. Mutant cultures were assessed by PCR to determine if they were composed of purely transformant cells or included mixtures of non-transformants cells. Then pure pilJ mutant and wildtype cells were mixed in PD2 medium and following incubation and exposure to kanamycin were assessed by PCR for presence of mutant and wildtype populations. Results We have discovered that when creating clones of targeted mutants of X. fastidiosa Temecula 1 with selection on antibiotic plates, X. fastidiosa lacking the gene deletion often persist in association with targeted mutant cells. We believe this phenomenon is due to spontaneous antibiotic resistance and/or X. fastidiosa characteristically forming aggregates that can be comprised of transformed and non-transformed cells. A combined population was confirmed by PCR, which showed that targeted mutant clones were mixed with non-transformed cells. After repeated transfer and storage the non-transformed cells became the dominant clone present. Conclusions We have discovered that special precautions are warranted when developing a targeted gene mutation in X. fastidiosa because colonies that arise following transformation and selection are often comprised of transformed and non-transformed cells. Following transfer and storage the cells can consist primarily of the non-transformed strain. As a result, careful monitoring of targeted mutant strains must be performed to avoid mixed populations and confounding results.


Background
Xylella fastidiosa is a Gram-negative, xylem-limited, insect-vectored bacterium that is a causal agent of many economically important plant diseases, including Pierce's disease of grapevines [1]. When infected vector insects probe plant tissues in search of the vascular xylem elements and sap contents, X. fastidiosa is subsequently transmitted to healthy plants. Once in the plant xylem, X. fastidiosa is postulated to migrate, attach, aggregate, and form biofilm that clogs the vessels leading to disease development.
To study the roles of X. fastidiosa genes and their encoded proteins, researchers traditionally delete X. fastidiosa genes using transposons or directed deletion with antibiotic-resistant markers [8,10]. These processes rely on identifying the mutant strain through antibiotic selection. X. fastidiosa Temecula 1 is sensitive to ampicillin, chloramphenicol, gentamicin, kanamycin, novobiocin, rifampin, and tetracycline [11]. Therefore plating transformants on these antibiotics should theoretically provide appropriate selectable markers for differentiation between wild-type X. fastidiosa and mutant strains. We recently discovered that nontransformed X. fastidiosa strains can survive on selectable medium, presumably due to spontaneous antibiotic-resistant mutants and/or extensive bacterial aggregation between transformed and non-transformed strains. As a result, strains presumed to be mutant clones are often a mixture of mutant and non-transformed X. fastidiosa, and over time, the nontransformed X. fastidiosa can become a significant population within a mixed sample.

Construction of X. fastidiosa pilJ mutant
The pilJ mutant was constructed by double cross over recombination resulting in replacement of the pilJ gene with a kanamycin cassette as preciously described [10]. Approximately 500 bp (base pair) were amplified upstream and downstream of the pilJ gene using primers pilJA/pilJB and pilJC/pilJD respectively (Table 1). All primers were purchased from Sigma or Integrated DNA Technologies (Coralville, IA). The polymerase chair reaction (PCR) conditions were as follows: denaturation at 95°C for 2 min., 35 cycles of denaturation at 95°C for 30 sec., annealing at 55°C for 45 sec., and extension at 72°C for 1 min., followed by 72°C for 3 min. A 1-kb fragment was generated from the upstream and downstream fragment using splice extension overlap PCR using the conditions mentioned above with some modifications. Only the pilJA/pilJD primers were used with an annealing temperature of 63.5°C. The 1-kb fragment was cloned into pUC19 plasmid (Invitrogen) to generate pUC19-pilJ. A kanamycin cassette cloned from Topo vector pCR2.1 (Invitrogen) was excised from pGEM T-Easy (Promega, Madison, WI) using flanking the AscI restriction sites. The pUC19-pilJ plasmid was digested with AscI and the kanamycin cassette inserted into the 1 kb fragment. The presence of the pilJ deletion construct in pUC19 was confirmed by PCR. One microliter of the deletion construct was transformed into electro-competent X. fastidiosa [14]. Transformed cells were incubated in 1 mL PD2 broth for 24 hr before being plated onto PW agar plates amended with kanamycin (10 μg/mL) for 7-10 days. Target gene deletion was verified by PCR, using primers pilJA/pilJD or pilJE/pilJF, and the strain with pilJ gene deleted was designated XfΔpilJ.

PCR amplification of DNA to confirm deletion of pilJ gene
The PCR mix included 100 ng of DNA, 200 mM dNTP (deoxyribonucleotide triphosphates), 2 mM MgSO 4 , 0.5 U Platinum Taq (Invitrogen), and 40 nM each of primer (pilJA/pilJD or pilJE/pilJF) in a 25 μL reaction mixture. PCR conditions were as follows: denaturation at 95°C for 2 min., 35 cycles of denaturation at 95°C for 45 sec., annealing at 60°C for 15 sec., and extension at 72°C for 2 min. and 30 sec., followed by 72°C for 6 min. PCR fragments were separated by gel electrophoresis and visualized using the Bio-Rad GelDoc XR system (Bio-Rad, Hercules CA).

Real-time (RT) PCR amplification to confirm deletion of pilJ gene
The real time PCR mix included 12.5 μL SybrGreen real-time PCR mix (Bio-Rad) and 40nM of each primer RST33 CACCATTCGTATCCCGGTG Xylella fastidiosa detection [17] in a total of 25 μL. PCR conditions include denaturation at 95°C for 3 min., and 35 cycles of 95°C for 10 sec., 50°C for 5 sec., 72°C for 25 sec. The melt curve was calculated at 76-95°C with 0.5°C increments for 5 sec.

Bacterial aggregation
Bacteria, Escherichia coli or X. fastidiosa, were grown on Luria Bertani (LB) or PW agar plates. Cells were removed from the plates and suspended in succinate-citratephosphate (SCP) buffer [15] to an OD 600 of 0.10 (4×10 7 CFU/mL). The cells were suspended vigorously by vortex mixer (Fisher Scientific, Springfield, NJ) for 5 minutes at maximum speed and by pipetting before being observed for aggregates. Cells were examined on a Axioskop 2 Plus microscope (Carl Zeiss Microscopy, Thornwood, NY) with a QImaging Retiga Exi camera (QImaging, Surrey, Canada) at 40X using QCapture 2.9.13 software (QImaging).

Mutant and wild-type X. fastidiosa on antibiotics
Wild-type X. fastidiosa and the XfΔpilJ mutant cells were grown to an OD 600 of 0.10 in PD2 liquid media. The XfΔpilJ mutant was a pure mutant having undergone multiple rounds of isolation followed by RT-PCR confirmation of not containing mixed populations. Wild-type bacteria, the XfΔpilJ mutant, or equal concentrations of both were suspended in PD2 media and incubated at 28°C for 24 hr, as occurs during a transformation [14]. After incubation, 100 μL was plated onto PW agar plates containing 0, 4, 10, 25, or 50 μg/mL kanamycin, and plates were incubated for 7-10 days at 28°C until growth was visible. Bacteria were scraped and collected from each plate and conventional PCR was conducted, as previously described. The PCR fragments obtained were analyzed by gel electrophoresis and visualized using the Bio-Rad GelDoc XR system (Bio-Rad). The experiment was performed three times.

Results and discussion
The pilJ gene encodes a putative chemotaxis receptor of interest [16]. The gene was deleted from X. fastidiosa Temecula 1 using site directed replacement with a kanamycin resistant marker [10]. Transformants were selected on antibiotic plates at 10 μg/mL since the minimum inhibitory concentration of kanamycin for X. fastidiosa Temecula 1 is 4 μg/mL [11]. All subsequent work with the transformed cells (XfΔpilJ mutants) were performed with 50 μg/mL kanamycin. Deletion of the pilJ gene was confirmed by PCR using multiple primer sets ( Figure 1). The pilJA/pilJD (AD) primers amplified a 3082 bp band from wild-type control bacteria and a 2200 bp band from the deletion plasmid pUC19-pilJ and the XfΔpilJ strain ( Figure 2). The pilJE/pilJF (EF) primers are complementary to sequences within the pilJ gene producing a 2030 bp band for wild-type control cells and no fragments for the XfΔpilJ bacteria or plasmid control. The X. fastidiosa-specific RST31/33 primers [17] confirmed that the bacteria were X. fastidiosa. As expected, these primers failed to amplify a band from the pUC19-pilJ plasmid. The XfΔpilJ strain was subsequently tested in a number of behavioral assays to explore the role of the PilJ protein (data not shown). The XfΔpilJ strain was placed in storage at -80°C in PD2 with 7% DMSO. The XfΔpilJ strain was streaked onto PW agar plates amended with kanamycin after -80°C storage. The XfΔpilJ mutant was observed to have behavioral phenotypes different from that previously observed for the mutant before storage but similar to wild-type X. fastidiosa (data not shown). The genotype of the XfΔpilJ mutant was therefore assessed by PCR. XfΔpilJ mutant directly from the -80°C stock was streaked onto PW agar plates with kanamycin to obtain single colonies and assessed by PCR. Of the twelve single colonies analyzed with the EF primers, 11 gave bands suggestive of non- Figure 1 Orientation of primers for Xylella fastidiosa pilJ gene deletion. Location of binding sites for PCR primers and length of resulting PCR products for transformed XfΔpilJ strains and for wild-type control or non-transformed cells. RST31/33 are primers specific to X. fastidiosa and used for bacteria confirmation. transformed cells (Figure 3). Colony 12, which lacked a fragment with the EF primers, suggesting it was the XfΔpilJ mutant strain, was streaked onto PW-kanamycin for a second round of single colony isolation. Subsequently, 32 colonies from round two were transferred to new PW plates with kanamycin and used for PCR with the AD and EF primer sets (Figure 4). Five colonies (colonies 1, 2, 4, 15, 17) appeared to be the XfΔpilJ mutant strain, as they did not give a band with the EF primers and had a 2200 bp band with the AD primers, while 11 colonies exhibited results typical of non-transformed cells, as they gave bands with the EF primers and a 3082 bp band with the AD primers. Mixed colonies of non-transformed and transformed cells were also observed (colony 22); where the EF primers amplified a band indicating the non-transformed strain was present, and the AD primers amplified a mutant size band indicating that the XfΔpilJ mutant was also present. Samples that appeared to be non-transformed or failed to amplify a band with AD primers, whether they amplified a band with EF primers or not, were not further analyzed. Two of the five transformed colonies were streaked onto PW agar plates with kanamycin to obtain single colonies for a third round of isolation. Of the 16 colonies examined in round three, 13 gave the mutant phenotype with the EF and AD primers ( Figure 5). Again, those colonies appearing to be non-transformed, failing to amplify a band with AD primers, or giving a very small band with the AD primers were not further examined. Four of the XfΔpilJ mutants were restreaked onto PW with kanamycin and assessed by real-time PCR for the presence of the pilJ gene (data not shown). None of these colonies were positive for the pilJ gene, therefore the samples were stored at -80°C in PD2 with 7% DMSO. This phenomenon of contamination by non-transformed cells was not limited to the XfΔpilJ mutant strain, but observed with a number of our X. fastidiosa deletion mutants (data not shown).
Mixtures of constructed kanamycin-resistant XfΔpilJ mutant and non-transformed strains may have occurred due to spontaneous antibiotic resistant mutation in wild-type cells, high aggregation rate of X. fastidiosa, or a combination of both events. The ability of bacteria to develop spontaneous resistance to antibiotics is a well known  phenomenon [18]. Sub-optimal antibiotic conditions provide particularly favorable conditions for these mutant strains to emerge. Of note, X. fastidiosa mutants were grown at standard, not sub-optimal, kanamycin concentrations [11]. Mixed aggregates of susceptible wild-type with kanamycin-resistant constructed XfΔpilJ mutant strains may also explain the findings. X. fastidiosa characteristically and spontaneously forms aggregates that are not easily dispersed, compared to bacteria such as Escherichia coli, even after vigorous resuspension by mixing and pipetting ( Figure 6). Presumably both aggregated and planktonic cells were transformed during mutant generation, and mixed aggregates of transformed and non-transformed cells formed subsequently. The aggregate formation may result in decreased susceptibility to antibiotics by the non-transformed bacteria [19,20]. After transformation when plating onto selective media (kanamycin) non-transformed bacteria may be "protected" from antibiotics by antibiotic resistant bacteria present in the aggregate, as is found in biofilms [21].
We tested the importance of mutants protecting wildtype bacteria from antibiotics by growing cells (wild-type X. fastidiosa, XfΔpilJ mutant strain, or an equal mixture) for 24 hours and plating them on PW with various concentrations of kanamycin, as is done for transformations. The wild-type-only X. fastidiosa sample gave the expected bands with all the primers tested when grown on 0 μg/mL of kanamycin (Figure 7). At higher concentrations of antibiotics no bacterial growth was observed on agar plates, therefore PCR could not be conducted. The XfΔpilJ-only mutant sample grew on all concentrations of antibiotics tested and gave the expected fragments with the AD and RST primers, while the EF primers did not produce a fragment, as expected. In one of three trials, PCR analysis of the plated mixed colonies of wild-type and the XfΔpilJ mutant showed the presence of both strains; AD primers amplified the XfΔpilJ mutant 2200 bp band, and the EF primers amplified a wild-type 2030 bp band at all concentrations of antibiotics tested except 10 μg/mL. These results indicate the presence of a small number of wild-type bacteria even on PW agar plates containing 50 μg/mL of kanamycin. While spontaneous antibiotic resistance cannot be ruled out, our results suggest that protection of the wild-type strain in mixed cell populations readily occurs, particularly as no wild-type-only samples grew on any kanamycin concentrations. In addition, the fact that the  wild-type EF band was faint and not observed on one kanamycin PW plate (10 μg/mL concentration) may indicate that in our original findings ( Figure 2) the number of wild-type cells in the mixed population may have been below detection level by conventional PCR.
Overall, it is possible that following transformation, colonies on selection plates may not have arisen from a single cell but from an aggregate containing a mix of the transformed and non-transformed cells. The presumed XfΔpilJ mutant clone was then stored in PW amended with 7% DMSO. After thawing and refreezing the nontransformed bacteria present may grow and attain populations that can affect the overall population dynamics and skew results of phenotypic assays as observed after recovery of stored XfΔpilJ, suggesting they were more fit for the freeze/thaw process.
Various aspects of X. fastidiosa aggregation have been reported. X. fastidiosa was found to aggregate into starlike clusters in microfluidic chambers under conditions of xylem flow [5]. The extent of aggregation is known to be dependent on media [22], xylem fluid source [23], and grapevine xylem chemistry [24]. Calcium can increase the ability of the bacterium to form aggregates while calcium chelators, such as ethylene glycol tetraacetic acid (EGTA) and 1,2-bis(o-aminophenoxy)ethane-N,N,N′,N′-tetraacetic acid acetoxymethyl ester (BAPTA/AM), cause a decrease in aggregation [25]. It would be interesting to determine if growth in media that promotes planktonic suspensions, followed by transformation, would reduce the possibility of mixed clones when generating X. fastidiosa mutants.

Conclusions
While it is possible to make targeted deletions in X. fastidiosa, complications may arise due to spontaneous antibiotic resistance and/or cell aggregates formed by the bacteria. The presence of aggregates may allow non-transformed bacteria to survive on PW kanamycin agar plates. At least three passages of single colony isolation followed by PCR may be required to minimize the amount of non-transformed cells contaminating the transformants. It may also be prudent to perform single colony isolation after retrieval from storage before use of mutant strains in assays.  Figure 7 Protection of wild-type Xylella fastidiosa from antibiotic selection pressure. Wild-type X. fastidiosa (wt), XfΔpilJ mutant (J), or an equal mixture of both (M) were grown in PD2 liquid media before being plated onto agar plates with 0, 4, 10, 25, or 50 μg/mL kanamycin, and tested by PCR. The pilJA/pilJD (AD) primers amplified a 3082 bp band for wild-type cells and a 2200 bp fragment from the XfΔpilJ strain. The pilJE/pilJF (EF) primers amplified a 2030 bp band for wild-type bacteria and no equivalent bands for the XfΔpilJ mutant strains. For mixed samples, the AD primers amplified a Xf pilJ strain fragment and the EF primers amplified a wild-type band. The RST31/33 (RST) primers confirmed that the bacteria were X. fastidiosa. Wells of each cell type and kanamycin concentration condition are numbered as follows: (1) AD amplification, (2) EF amplification, and (3) RST amplification.