2.1 Genetic constructs of deoxyviolacein piggyback transposon vectors

Two plasmids were constructed to introduce deoxyviolacein biosynthetic genes through transposon mutagenesis. The deoxyviolacein operon was obtained from Christopher Anderson (Kittleson et al., 2012) through Addgene (BBa_J72214-BBa_J72090), and was selected based on sufficient pigment production of the initial construct. When utilized, site-directed mutagenesis was performed using the Quikchange II protocol (Agilent, Santa Clara, CA). Plasmids pBB327 and pBB328 were constructed as detailed in Table 1 and primers used to modify these are provided in Table 2. These plasmids are similar to one another, except for different antibiotic selection genes inside the transposon (kanamycin for pBB327 and tetracycline for pBB328). The plasmids contain a transposase, R6K origin, deoxyviolacein expression genes vioABCE, an antibiotic marker, and two repeat sites where the transposase will cut. A map of the plasmids is shown in Figure 1. All plasmid constructs were first introduced into and isolated from Escherichia coli BW25141 and then transformed into E. coli WM3064 for conjugation into recipient strains.

2.2 Growth of donor and recipient bacterial strains

Azotobacter vinelandii DJ (ATCC BAA-1303) was grown on Burk's (B) medium (Dos Santos, 2011). Gluconacetobacter diazotrophicus PA1 5 was grown on GADN medium (Fisher & Newton, 2005; Schwister et al., 2022). Klebsiella grimontii M5aL, Azoarcus olearius DQS-4, Rhodobacter sphaeroides 2.4.1, Marinobacter aquaeolei VT8 and Escherichia coli MG1655 were grown on lysogeny broth (LB) medium. Plates of LB were supplemented with 100 μL of 5 mg/mL 2,6-diaminopimelic acid (DAP) when growing E. coli WM3064. Recipient strains were cultured in liquid medium before the introduction of the transposon.

2.3 Introduction of deoxyviolacein piggyback transposons to model bacteria

Recipient strains were conjugated with transposons following a protocol similar to that used for conjugation with E. coli WM3064 in our laboratory (Barney et al., 2015; Knutson et al., 2021). Recipient strains were first grown in liquid medium for a sufficient time to ensure that cells were in a planktonic form, especially for strains that formed biofilms, hard colonies or extensive exopolysaccharides. E. coli WM3064 containing the desired plasmid was scraped using a loop from a 1-day-old LB plate grown at 30°C containing the specific antibiotic and supplemented with DAP and transferred to a 1.5 mL microfuge tube to obtain ~100 μL volume of cells. The recipient strains were centrifuged for 1 min at ~10,000g to obtain a similar volume of cells. Both cell pellets were resuspended in 0.5 mL of growth medium, and depending on the growth rate of the recipient cells, we mixed different ratios of the two strains in ~100 μL volume. For example, we combined 5 μL of the E. coli WM3064 strain containing plasmid with 100 μL of G. diazotrophicus and mixed thoroughly before spotting them as 9 evenly spaced drops onto GADN medium supplemented with DAP. In each case, we confirmed that the E. coli WM3064 strain would grow on the specific medium before combining the two strains. During the conjugation and outgrowth procedures, no antibiotic was included on the plates. After the cultures had grown overnight, we scraped 3 spots per plate and grew them in 50 mL of the desired liquid medium. The liquid medium lacked DAP, resulting in a counter-selection of the E. coli WM3064 and the enrichment of the recipient strain, and allowed the cells to recover before challenging with antibiotics. We then plated various aliquots of the liquid cultures onto solid medium with antibiotic (but devoid of DAP) for selection. Colonies were screened to determine the number of transposon-integrated colonies and to identify specific colonies with sufficient pigment for the desired applications (Barney et al., 2015; Knutson et al., 2021). The methodology and ratios of donor and recipient strains had to be adjusted on a case-by-case basis for each strain to optimize the conjugation conditions.

2.4 Confirmation of genome integration sites

For specific mutations that yielded a desired amount of pigment, the site where the transposon had integrated and the direction of the insert was determined. This was accomplished by first purifying the individual colony by repeated passage on selective medium. Then genomic DNA was collected using the Quick-DNA Fungal/Bacterial Miniprep Kit (Zymo Research). Genomic DNA was digested using the PstI restriction endonuclease, to cut the genomic DNA into suitable fragments. The DNA was then purified using the DNA Clean and Concentrator-25 (Zymo Research) kit, and ligated to generate circular fragments suitable for PCR. Primers BBP3086 and either BBP1282 (for pBB327) or BBP1176 (for pBB328), were used to perform PCR and confirm a suitable DNA product for sequencing. Positive PCR products were purified using the DNA Clean and Concentrator-25 kit, then sent for Sanger sequencing using primer BBP3086, which provided the insert location and direction in which the transposon had integrated into the genome.

2.5 Assessment of growth rate for specific isolates

To determine if there was a significant growth penalty for selected strains used in further tracer or biosensor studies, strains were streaked pure and several single colonies were used to inoculate each strain into a 50 mL culture of the medium described above for each strain. For each strain, control of the parent strain was also included so that the deoxyviolacein-producing strain could be compared to its parent. Optical density was measured at 600 nm, once the strain had reached a starting density of approximately OD₆₀₀ = 0.1. An additional measurement was made when cells had reached an OD₆₀₀ above 1.0, but while the cells were still in exponential growth. In each case, the analysis was performed in triplicate and was used to determine the growth rate or doubling time. In most cases, colonies were selected with sufficient pigment levels and colony sizes that were comparable to surrounding non-pigmented strains.

2.6 Deoxyviolacein quantification

Deoxyviolacein was quantified using a method similar to what has been previously described (Wang et al., 2009). Five milliliters of liquid culture was pelleted at >10,000g for 1 min and the supernatant was removed. The cell pellet was then rinsed using distilled water and pelleted again at >10,000g for 1 min. The cell pellet was resuspended in 5 mL of ethanol and sonicated for 5 min at 200 W using a Misonix XL-2000 Series sonicator probe, then the cell debris was pelleted again at >10,000g for 1 min and the supernatant was measured for deoxyviolacein at 570 nm.

2.7 Infiltration of leaves and recovery of bacteria

Leaves of 4-week-old Nicotiana benthamiana plants were grown with adequate water and infiltrated with a pigment-producing isolate of Klebsiella grimontii M5aL by a modification of a previously described method (Zhang et al., 2011). Briefly, bacteria were grown overnight in LB medium with 15 µg/mL of tetracycline. Cells from 2 mL of the culture were pelleted by centrifugation at 3500g for 5 min and the cells were resuspended in 20 mL of 0.25X MS medium. The resuspended bacteria were infiltrated into leaves through the stomates of the lower leaf surface using a needle-less 1 mL syringe. To quantify the endophytes, the infiltrated leaves were surface sterilized in 10% bleach and 0.1% Tween 20 for 20 min and washed with distilled water. The infiltrated portion of leaves was ground in 500 µL of sterile ddH₂O using a Bullet Blender to release endophytes. Serial dilutions of the solution remaining following grinding were plated on LB plates and endophytes were grown at 30°C.

2.8 Construction of an improved nitrogen-compound biosensor strain of A. vinelandii

A. vinelandii is a model microbe for the study of aerobic nitrogen fixation. Testing mutants for extracellular nitrogen compound production is important to developing strains with improved biofertilizer potential. A prior effort developed a biosensor strain to indicate the presence of extracellular nitrogen compounds from mutant screens, but required the addition of 5-bromo-4-chloro-3-indolyl β-d-galactopyranoside (X-Gal) to plates (Barney et al., 2015). Developing a deoxyviolacein-based biosensor would alleviate the need to add X-Gal and potentially result in an improved reporter system. A. vinelandii strain AZBB120 lacks urease and the ammonium transporter AmtB (Eberhart et al., 2016). AZBB120 was transformed with pPCRNH3-21 (Barney et al., 2015) to replace the genes coding the nitrogenase regulatory proteins NifL and NifA with a tetracycline cassette, yielding strain AZBB202, which can no longer fix nitrogen (Table 3). AZBB202 was then conjugated with pBB327 to select colonies with elevated deoxyviolacein production, yielding AZBB741. AZBB741 is unable to fix nitrogen diazotrophically or effectively utilize urea, but grows and produces deoxyviolacein pigment in the presence of nitrogen substrates such as nitrate and elevated ammonium, making it suitable to identify other microbes that produce specific extracellular nitrogen compounds.

3.1 Application of the deoxyviolacein-based transposon system

The plasmids pBB327 and pBB328 were constructed with either a kanamycin selection marker or a tetracycline selection marker, respectively. As an initial test case, we conjugated G. diazotrophicus with either plasmid and looked for colonies with a suitable amount of pigment. Both plasmids yielded several colonies (approximately 10 per plate of 500 colonies, Figure 2) with a suitable amount of pigment (sufficient to confirm color development within 3 days of growth on GADN medium). Several colonies were selected and passaged on plates with antibiotics, and then passaged to plates without the antibiotic, to determine if the pigmented phenotype would be lost in the absence of antibiotic. In each case, there was minimal evidence of strain instability during each passage. Genomic locations for the inserts were determined to confirm that the resulting colonies had incorporated the transposon into the genome. Seven colonies were sequenced to determine the location of the TA insertion and were confirmed to contain a genomic insertion located at a specific TA site (insertions found in Gdia_0263, Gdia_0657, Gdia_1202, Gdia_1182, Gdia_2010, Gdia_2094 and Gdia_2169) (Giongo et al., 2010), indicating that the Mariner transposon functioned as designed with the vioABCE operon inserted into a random TA site within the genome. In addition to experiments with G. diazotrophicus, we also introduced the transposon into Escherichia coli, Azotobacter vinelandii, Azoarcus olearius and Klebsiella grimontii. The list of recipient strains is not intended to be an exhaustive selection of potential hosts. Based on the plasmid backbone, we believe it is reasonable to assume that any strain that can be conjugated with the Mariner transposon is also suitable for yielding transposon libraries with these plasmids, and there is a significant body of literature reporting successful transformation with the Mariner transposon for Tn-seq experiments (Christen et al., 2011; Hutchison et al., 2019; Poulsen et al., 2019; Rosconi et al., 2016; van Opijnen et al., 2009; Zomer et al., 2012). While there is some evidence that violacein and deoxyviolacein may have antibacterial properties, this also appears to be strain-specific, and we did not observe evidence of this in the five strains listed above.

Shown in Figure 2 are several representative plates of recipient strains that were tested for transposon mutagenesis and pigment production. The plates were selected to illustrate relative differences in colony pigment levels and are not intended to be comprehensive. In many cases, several plates were prepared from each recipient to select an ideal mutant. For four of the recipient strains shown in Figure 2, only a small number of colonies per plate yielded significant amounts of pigment for further evaluation. However, for A. olearius, a much larger percentage of colonies showed substantial pigment production, indicating that the transposon may give varying degrees of pigment-producing colonies depending on the recipient strain. For the remaining strains, these results confirm that the transposon is not carrying a strong internal promoter system, and is relying on suitable transcriptional initiators that are internal to the genome. Interestingly, when TA sites were located for the specific clones shown in Figure 3 (and listed above for G. diazotrophicus), the upstream element responsible for transcription of the vioABCE operon was not always immediately evident. Differential expression in different strain backgrounds is expected, and users should evaluate for themselves conditional versus constitutive expression in their target strain of interest.

To confirm the stability of individual strains, colonies were passaged on plates for several generations to obtain clean colonies and assess the stability of the pigment-producing phenotype. This was done both in the presence of antibiotic and in many cases, once isolated, without the inclusion of the antibiotic. In some rare cases, a colony would illustrate a propensity to develop a small number of revertants in each passage, manifesting as a loss of pigment in several colonies on the plates. In these cases, the issue was resolved by either selecting another initial colony or by using the alternative plasmid and antibiotic selection to generate alternative transformants. In each of the cases shown in Figure 3, strains could be passaged on fresh plates without antibiotic pressure and exhibited little evidence of revertants or loss of the pigment-producing phenotype. This again must be assessed on a case-by-case basis. Since we generally selected strains that grew at rates similar to the background strain, based on the sizes of the colonies and balanced production of the pigments, we believe that we also selected strains with a lower growth defect and comparable fitness level to the parent strains, as described below. This would indicate a lower selective pressure for the strains to revert or develop additional mutations that disrupted the pigment production.

3.2 Measurement of strain growth rate

As described in the methods, we were careful to select strains of pigmented tracers or biosensors that grew at rates similar to surrounding non-pigmented colonies (as assessed by colony size). For this reason, we assumed that growth rates for these strains would be similar to the starting parental strains. For strain KGBB001 (Table 3), the growth rate was slightly higher (~15%) than for the wild-type K. grimontii strain that served as the parent strain. For strain AZBB741, the growth rate was also slightly higher (~20%) than the parent strain of AZBB202. Both AZBB202 and AZBB741 were grown in B medium supplemented with 20 mM sodium nitrate, as these strains have a disruption in the nifLA operon, resulting in the disruption of the diazotrophic phenotype. Finally, GABB015 had a slightly slower growth rate (~30%) than the wild-type G. diazotrophicus strain that served as the parent strain. In the case of K. grimontii and G. diazotrophicus, these strains were grown on a rich medium that should contain a source of tryptophan, while for A. vinelandii strains AZBB202 and AZBB741, the strains were supplemented with an external nitrogen source, such that none of these three strains were grown under diazotrophic conditions. Additionally, while the growth rates were relatively similar for all three pigment-producing strains, the lag phase for each of these strains was slightly longer.

3.3 Studies of leaf endophytes

Several colonies of the endophytes K. grimontii and G. diazotrophicus with the desired pigment-producing phenotype were selected for studies with various plant hosts and were confirmed to maintain the pigmented phenotype even after several weeks in association with a plant host as an endophyte. Importantly, while in the presence of the plants, no antibiotic pressure was applied to the bacteria. At the end of these experiments, bacterial cells were plated back onto the specific selective medium with the specific antibiotic and screened for pigment production, indicating that these new strains could function effectively as suitable sensors in recovery experiments to distinguish them from other bacteria and determine numbers of viable cells recovered from leaves (Figure 4), roots or other plant organs. Figure 4 shows an example of one such experiment. The pigmented KGBB001 strain could be easily differentiated from other natural endophytes (purple vs. white colonies), which could also be differentiated based on differences in colony morphology.

3.4 Constructing an improved nitrogen-compound biosensor strain

A strain of A. vinelandii containing deletions in the ammonium transporter and urease genes and a disruption in nifLA nitrogen regulator genes (AZBB202) was further conjugated with the deoxyviolacein transposon to isolate a strain with elevated pigment production (AZBB741). This strain is no longer diazotrophic, relying upon exogenous nitrogen sources for growth. This strain was used to plate serial dilutions of the elevated extracellular nitrogen compound-producing strain AZBB163 (Barney et al., 2015) onto a lawn of AZBB741 to rapidly scan individual colonies of AZBB163 from elevated extracellular nitrogen compound production (Figure 4). The strains that are still producing elevated levels of extracellular nitrogen compounds are easily identified by the halo of purple-pigmented AZBB741 growing immediately adjacent to them. This can be used to evaluate strains that are producing elevated ammonium and determine the potential loss of the phenotype in certain populations, if it had occurred.