Gluconacetobacter diazotrophicus Gene Fitness during Diazotrophic Growth

ABSTRACT

Plant growth-promoting (PGP) bacteria are important to the development of sustainable agricultural systems. PGP microbes that fix atmospheric nitrogen (diazotrophs) could minimize the application of industrially derived fertilizers and function as a biofertilizer. The bacterium Gluconacetobacter diazotrophicus is a nitrogen-fixing PGP microbe originally discovered in association with sugarcane plants, where it functions as an endophyte. It also forms endophyte associations with a range of other agriculturally relevant crop plants. G. diazotrophicus requires microaerobic conditions for diazotrophic growth. We generated a transposon library for G. diazotrophicus and cultured the library under various growth conditions and culture medium compositions to measure fitness defects associated with individual transposon inserts (transposon insertion sequencing [Tn-seq]). Using this library, we probed more than 3,200 genes and ascertained the importance of various genes for diazotrophic growth of this microaerobic endophyte. We also identified a set of essential genes.

IMPORTANCE Our results demonstrate a succinct set of genes involved in diazotrophic growth for G. diazotrophicus, with a lower degree of redundancy than what is found in other model diazotrophs. The results will serve as a valuable resource for those interested in biological nitrogen fixation and will establish a baseline data set for plant free growth, which could complement future studies related to the endophyte relationship.

INTRODUCTION

Agricultural yield for many key crops is directly linked to the supplemented nutrients provided through different fertilizers (1). Most conventional fertilizers contain various amounts of the macronutrients nitrogen, phosphorus, and potassium, along with several additional minor macronutrients and a variety of micronutrients (2). The nitrogen component of industrial fertilizers consists of ammonia, nitrate, or urea, which are often applied to soils, where they may undergo transformations to alternative forms that are taken up by the plants (3–5). Applied nitrogen can be lost by wasteful alternative reactions or escape through runoff and is underutilized when overapplied. The production of nitrogen fertilizers at the scales required to feed the world utilizes significant energy and natural gas through the industrial Haber-Bosch process (1, 6–8).

Biological nitrogen fixation (BNF) predates the Haber-Bosch process by at least a billion years (9, 10) and is responsible for more than half of the nitrogen required to produce the biomass on Earth each year (8). Prior to the advent of the Haber-Bosch process, BNF was the predominant means of providing nitrogen to crops. BNF is accomplished by a class of microbes known as diazotrophs that utilize the enzyme nitrogenase to fix nitrogen gas into ammonia (11, 12). Diazotrophs can be found free-living or in many symbiotic associations with other organisms, including higher plants (13). In contrast to nitrogen fixed within the rhizosphere by free-living bacteria, which is susceptible to losses through various routes prior to being acquired by the plant, nitrogen fixed within the plant is less susceptible to loss.

The most recognized nitrogen-fixing association with plants is the symbiotic relationship between bacteria known as rhizobia and legumes, where the rhizobia colonize the roots of the legumes to form nodules (14). This association has been recognized by scientists and farmers for more than a century (15). Other symbiotic relationships are also found with plants, including the actinorhizal bacteria that form relationships with trees such as alders (16). The diazotroph Gluconacetobacter diazotrophicus was discovered in association with sugarcane as an endophyte in 1988 (17). It colonizes the intracellular space of the roots, stems, and leaves and is believed to promote the growth of the plant (18, 19). There is some debate as to how much nitrogen fixed by G. diazotrophicus is transferred to the plant during the endosymbiosis (20). G. diazotrophicus is a Gram-negative member of the Alphaproteobacteria class.

Since this discovery, it has been further demonstrated that G. diazotrophicus will establish associations with a range of additional plants besides sugarcane, including important crop plants such as wheat and coffee (21–23). For this reason, G. diazotrophicus is recognized as having a greater potential than free-living diazotrophs for expanded use with crops that require significant quantities of industrial fertilizers.

Our laboratory has been investigating the potential to improve the model diazotroph Azotobacter vinelandii for application as a biofertilizer, including efforts that increase the production of siderophores, urea, and ammonium (24–30). A. vinelandii is unique from many diazotrophs, as it performs BNF as an obligate aerobe (31, 32). One drawback to this strain is that A. vinelandii is generally found in the rhizosphere and is not known to function as an endophyte with higher plants. We previously studied global transcription of a high-ammonium-secreting strain of A. vinelandii (25) and also created a large transposon library to study gene essentiality in the wild-type A. vinelandii strain (27). In this study, we generated a similar transposon library for G. diazotrophicus to study genes essential to growth on rich medium, and using a reactor system that enabled the growth of G. diazotrophicus under microaerobic conditions, we were able to further study gene fitness for BNF in G. diazotrophicus. The results of this analysis for G. diazotrophicus are discussed in relation to our prior study with A. vinelandii. These results provide a valuable resource to further understand gene fitness in diazotrophs and can serve as a reference point for further future studies to investigate genes important to the endophyte relationship.

RESULTS

Library growth.

Two conditions were selected for comparisons of diazotrophic growth. The first condition utilized a modified Burk’s medium with sucrose as the primary carbon source and which was supplemented with ammonium sulfate. This culture was grown under a standard atmosphere and grew to ~7 generations. The second condition was growth with the same medium in the absence of a supplemented nitrogen source to encourage diazotrophic growth, and the culture was grown in a closed reactor system with a low-oxygen (2.5% by volume) atmosphere, as it was not possible to initiate this diazotrophic growth under a standard atmosphere (33). Under these conditions, the culture grew for ~4 generations before entering the stationary phase. Increasing the oxygen level in the closed reactor could have resulted in denser growth once established, but this would have resulted in an additional variable in the experiment that was undesired for this study. Under both conditions, cultures were grown with citrate as a buffer system to maintain the lower pH preferred by G. diazotrophicus.

Library statistics.

The Mariner transposon integrates at TA dinucleotides. The complete genome of G. diazotrophicus has 58,303 TA sites. Of these sites, 57,574 are within the chromosome (accession no. CP001189.1) and 729 are within the plasmid (accession no. CP001190.1) (JGI assembly [19] used as the reference library and both JGI and Brazil assemblies [18, 19] used for annotation). A total of 8,979 of all TA sites were found to be nonunique, while 3,773 were nonpermissive. In some cases, TA sites were both homologous and nonpermissive. Due to the indiscriminate and restrictive features of these regions, we removed a total of 8,668 TA sites (14.87%): 8,641 from the chromosome and 27 from the plasmid. In addition, we filtered out any TA sites found in the first 5% and last 10% of genes based on concerns that some genes can tolerate insertions in these regions and still retain gene function. While not used in the calculation of fitness for the whole gene, all usable sites are provided in the site fitness data in Data Set S2 in the supplemental material. After this filtering step, 26,349 TA sites were removed (44.43%), leaving a total of 31,954 TA sites remaining (58.81%). It is important to note that this value of TA sites removed includes those found in intergenic regions, though again, these are included in the site fitness data in Data Set S2. The remaining TA sites covered 3,286 genes out of 3,501 (93.9%). There were a total of 215 genes that were unable to be analyzed through this method. Of this total, 164 genes contained homologous or nonpermissive TA sites, while 51 genes did not contain TA sites between the region following the first 5% and prior to the last 10%. As a result of the filters described above, certain genes could not be probed by this technique. This is an inherent limitation in any transposon insertion sequencing (Tn-seq) experiment. However, our library yielded extensive coverage and provides sufficient data to probe 3,286 of the 3,501 genes listed in the JGI assembly and annotation (19).

Major nif cluster genes.

The initial intent of this project was to explore the fitness values of genes related to molybdenum-dependent nitrogen fixation in G. diazotrophicus when grown in the presence of ammonium and under nitrogen fixation conditions. Fitness values for single genes provide a measure of the importance that a gene plays in a given process. Unlike model diazotrophs such as Azotobacter vinelandii, Pseudomonas stutzeri, and Rhodobacter sphaeroides (34), the major genes associated with molybdenum-dependent nitrogen fixation in G. diazotrophicus are arranged in one primary cluster (Fig. 1B) (35). The fitness values of this major gene cluster of nif and nif-related genes are presented in Fig. 1A. The genes were assigned to one of several categories based upon their fitness value. The categories were as follows: genes associated with a large growth defect (purple), moderate defect (blue), and no or minimal defect (green). Genes that were unable to be analyzed based on the lack of sufficient data or the absence of suitable TA sites are shown in white. Similar to previous Mariner Himar1 minitransposons, the design of our transposon system lacks transcriptional terminators such that read-through transcription likely occurs through the transposon (36). Additionally, the tetracycline selection gene represents a strong promoter that may even enhance downstream expression in one direction. Sequencing data from Tn-seq experiments do not allow the user to determine the orientation of any specific insert from the data that is obtained. For this reason, as with any Tn-seq experiment, polar effects cannot be fully discounted, especially for genes that are part of a larger operon (36–38).

FIG 1.

The primary nitrogen fixation cluster of Gluconacetobacter diazotrophicus. Panel A shows gene fitness values for the two growth conditions. Closed circles are without nitrogen grown under microaerobic conditions (N−), while open circles are with nitrogen provided as ammonium (N+). Genes were color coded based on the N− fitness values. Panel B illustrates a gene map of the primary nitrogen fixation cluster.

Under diazotrophic growth, the structural genes encoding nitrogenase (nifH and nifDK) (18) were found to have large fitness losses (Fig. 1). These genes are associated with the α- and β-subunits of the MoFe protein (nifDK) and the Fe (nifH) protein that provides the electrons to the MoFe protein (39, 40). Similar to nifH and nifDK, the genes nifB, nifZ, nifT, nifEN, nifS, nifV, nifW, Gdia_1564, and modA all showed large differential fitness values under diazotrophic growth. Many of these genes have an assigned function in MoFe cofactor biosynthesis (35, 40). Additional genes had fitness losses, but higher variability in the results, including nifU, fdxN, and Gdia_1560. Strains with the four genes of the fixABCX cluster that are part of this larger cluster also showed a pronounced fitness defect under diazotrophic conditions. The rpoN, modC, modB, and fdxB genes were associated with slightly lower gene defects. Genes Gdia_1578 and modD showed minimal differences. Transcriptional start sites and operon structure were evaluated previously by Northern analysis and hybridization studies (35). As mentioned above, some of the fitness losses might be related to potential polar effects. Additionally, it is possible that these genes are important in a competitive environment and may not translate precisely to other conditions. In this regard, Tn-seq provides a potential roadmap for future gene disruptions or modifications that might be independently analyzed.

Nitrogen assimilation and regulation genes.

Beyond the genes within the major nif cluster are those coding for proteins associated with nitrogen assimilation and regulation. These genes range in function from ammonium uptake into the cell to activation of nif gene transcription, and many are not included in the major nif cluster. Rather, these genes are scattered throughout the genome—some organized in operons and some independently transcribed. The fitness values of these genes associated with nitrogen assimilation and regulation are presented in Fig. 2. Like the major nif cluster, these genes also fell into fitness categories consistent with them having a role in growth under diazotrophic conditions.

FIG 2.

Additional nitrogen assimilation and regulation genes of Gluconacetobacter diazotrophicus. Panel A shows gene fitness values for the two growth conditions. Closed circles are without nitrogen grown under microaerobic conditions (N−), while open circles are with nitrogen provided as ammonium (N+). Genes were color coded based on the N− fitness values. Panel B illustrates the relationship in the genome of various gene clusters. Dotted lines represent large gaps between genes.

G. diazotrophicus contains limited functional redundancy in genes related to nitrogen fixation. However, this microbe does display some genetic homologs with redundancy in genes related to nitrogen regulation. G. diazotrophicus contains two homologs of the ammonium transporter gene amtB, termed amtB1 (Gdia_0598) and amtB2 (Gdia_1303), which bring ammonia into the cell (41). While in separate genomic regions, strains with both genes displayed moderate fitness defects under growth with or without nitrogen included (Fig. 2), suggesting a general fitness independent of diazotrophic growth associated with these genes, although this defect increased under diazotrophic growth.

A number of the nitrogen-associated gene disruptions were associated with strong fitness defects, regardless of whether the cultures were provided ammonium or grown diazotrophically. Strains with the four genes gltD (Gdia_0331), gltB (Gdia_0330), glnB (Gdia_1481), and glnD (Gdia_0300) all showed significant defects that increased under conditions of diazotrophic growth. The glnB and glnD genes showed higher variability. In contrast, glnE (Gdia_2987) was associated with a lesser growth defect when the culture was provided with ammonium, but that defect was slightly minimized under diazotrophic growth. Strains with the nifR3 family (Gdia_0487) and ntrBCY genes (Gdia_0484 to -0486) all showed minimal or no defects when provided ammonium, but each had minor improvements under diazotrophic growth, similar to glnE. A number of other genes of interest (especially smaller genes) that might be associated with nitrogen fixation were unable to be analyzed through this approach.

Additional genes associated with diazotrophic growth defects.

In addition to the genes of the major nif cluster, we identified additional genes that contribute to fitness under diazotrophic conditions. Figure 3 shows a selection of genes that were associated with significant growth defects under diazotrophic conditions when these genes were disrupted by transposon insertions. Genes with high variability or substantial fitness losses in medium containing supplemented ammonium were omitted from Fig. 3, although the results for all genes are provided in the supplemental material. Results were ordered in a manner to cluster genes that were in close proximity within the genome. Genes found to be associated with significant defects under diazotrophic growth included clpA (Gdia_2903), genes coding for a YgpP/YjpQ family permease (Gdia_3416 and Gdia_3417), ctrAB (Gdia_0662 and Gdia_0664), and genes coding for the molybdenum cofactor biosynthesis proteins moaBDE (Gdia_1246, -0049, and -0050, respectively), as well as several hypothetical or conserved hypothetical genes.

FIG 3.

Additional genes showing differential fitness values during diazotrophic growth for Gluconacetobacter diazotrophicus. Shown above are a selection of genes associated with minimal fitness defects during growth with supplemented nitrogen (N+; open circles) versus significant growth defects when cultures were grown without provided nitrogen under microaerobic conditions (N−; closed circles). Gene names and locus tags are provided on the y axis. Genes were selected that showed significant differences and low variability. A list of all genes and the recorded fitness values is provided in Data Set S1.

Gene disruptions resulting in fitness improvements.

Similar to a prior study of gene fitness for the diazotroph Azotobacter vinelandii (27), our results revealed a number of gene disruptions that improved fitness of G. diazotrophicus under diazotrophic growth, with ammonium provided, or in both cases (Fig. 4). In Fig. 4, we include a selection of genes associated with considerable fitness increases and minimal variability between replicates. Again, a full assessment of gene fitness for each gene analyzed is provided in the supplemental material. The genes shown in Fig. 4 are arranged in order as they appear in the genome, so that clusters of genes, such as Gdia_0262 and Gdia_0263, Gdia_2284 through Gdia_2294, xagABC, and others, can be compared. Many of these genes resulted in quite dramatic improvement in growth, especially when cultures were grown diazotrophically under microaerobic conditions, similar to the minor differences seen for ntrB in Fig. 2.

FIG 4.

Genes displaying improved differential fitness values during diazotrophic growth for Gluconacetobacter diazotrophicus. Shown above are a selection of genes associated with minimal fitness defects during growth with supplemented nitrogen (N+; open circles) versus significant growth improvements when cultures are grown without provided nitrogen under microaerobic conditions (N−; closed circles). Locus tags are provided on the y axis. Selected genes were associated with significant improvement and low variability. A separate data set for growth on rich medium (GAD; open triangles) is also provided as reference. A list of all genes and the recorded fitness values is provided in Data Set S1.

Essential gene analysis.

One hallmark of Tn-seq studies is the ability to identify essential genes. Assuming that transposon insertion mutations are stable, it is expected that any Tn-seq analysis should yield a set of genes that do not tolerate insertions because these disruptions are too detrimental to the function of the organism under the growth conditions tested. Assessment of gene essentiality requires careful inspection of the results obtained and can be enhanced by factors such as library size, adequate numbers of TA sites within the genes analyzed, and sufficient distribution of mutations across available TA sites. If each of these conditions is met, one can estimate the likelihood that a gene with a sufficient number of TA sites should appear in the data set if it were nonessential. Our library was isolated from GAD medium, which we consider a rich medium, as it contains glucose, yeast extract, and tryptone. It also contained a sufficient number of insertions to perform a reasonable analysis of gene essentiality in G. diazotrophicus.

A first assessment of gene essentiality was done following a protocol similar to that employed previously (42, 43) by taking the average counts for each gene in the t₁ data set and dividing by the number of available TA sites within the gene, then taking the log₂ value of the obtained value. We created a histogram of the distribution of values for each gene. Based on this result, we determined the local minimum between the two peaks and established a cutoff value of log₂ = 2.8. From this analysis, we found 516 genes that were defined as essential based on either no counts or a very low number of counts compared to the population (Fig. S1). We further used the data set of strains grown in GAD medium to assess genes that were stable, but in low abundance, and manually reviewed all of the results. Another 63 genes were defined as nonessential, but displayed low fitness values when cultured in the rich GAD medium (below 0.6 fitness in either replication of the GAD growths). Finally, another 32 genes had low initial counts, but were found to be stable in the GAD medium data set. This left 2,612 genes (79.5% of the 3,286 genes with suitable TA sites) remaining to analyze for fitness in the modified Burk’s medium with and without supplemented nitrogen, which were defined as nonessential.

DISCUSSION

Many genes deemed essential to diazotrophic growth through biological nitrogen fixation (BNF) have been identified in various species through genetic and biochemical studies (12, 27, 39, 44–46). These studies are sometimes convoluted by the existence of homologs that are found in different nitrogenase classes (molybdenum-dependent, vanadium-dependent, and iron-only nitrogenases) or by the repertoire of genes that provide electrons to the nitrogenase catalytic enzymes. As many of the enzymes and proteins associated with nitrogenase and nitrogenase cofactor assembly are oxygen sensitive, many diazotrophs fix nitrogen only under anaerobic conditions, while other diazotrophs function under microaerobic or aerobic conditions, presumably maintaining an anaerobic environment around these oxygen-sensitive proteins. Studies of gene fitness, assessed through the genetic technique of transposon insertion sequencing (Tn-seq), provide an alternative means to measure the importance of specific genes in a global manner by comparing growth rates for a large library of mutants under different growth conditions (36). This technique can provide both significant and subtle differences between growth rates when specific genes are disrupted, providing a clearer picture of the contributions of individual proteins in supporting a specific ability such as diazotrophic growth.

A previous study examined the impact of gene disruptions on diazotrophic growth for Azotobacter vinelandii, a model organism for investigations of aerobic biological nitrogen fixation that has been studied for nearly a century (27). Significantly less is known about Gluconacetobacter diazotrophicus, a diazotroph discovered in 1988 (17) that is generally recognized for its association with sugarcane. G. diazotrophicus is reported to function as an endosymbiont with sugarcane and has also been found to form endosymbiotic relationships with many other plants. It fixes nitrogen under microaerobic conditions and has a smaller genome than A. vinelandii, with less complexity and diversity of nitrogenase systems than the free-living microbe A. vinelandii (18, 19, 32). This contrast between the two microbes and the potential for multiple agricultural applications of G. diazotrophicus encouraged us to pursue Tn-seq studies to compare the BNF-related gene fitness of G. diazotrophicus with that of A. vinelandii.

Since G. diazotrophicus requires microaerobic conditions for diazotrophic growth (33, 47), we cultured this diazotroph in a modified Burk’s medium devoid of a nitrogen source under an artificial atmosphere of 2.5% O₂ and 92.5% N₂ (with the balance argon) to determine the fitness of gene disruptions related to BNF. This was compared to the library grown on the same medium, but with supplemented ammonium under a standard atmosphere. We selected a standard atmosphere as it enabled us to achieve more generations of growth. One caveat of this difference is that it does not allow us to differentiate between genes associated strictly with diazotrophic growth and genes that may be important to microaerobic growth.

As expected, strains with the genes coding for the catalytic nitrogenase enzymes (nifDK for the MoFe protein and nifH for the Fe protein) all showed substantial fitness defects for transposon disruptions (Fig. 1). Additionally, almost all of the genes of this primary nif cluster of genes, starting from rpoN (Gdia_1582) and continuing to fixX (Gdia_1552), were associated with substantial fitness defects when disrupted. It is possible that some of these disruptions may be the result of polar effects, although our transposon should lack terminators and allow read-through, and furthermore results in a strong promoter directed at downstream genes as a result of the strong tetracycline promoter, based on insertions in the correct orientation. However, the potential for polar effects cannot be discounted, and all results should be considered in light of this potential, which is still a topic of debate in interpreting Tn-seq results (37, 38).

G. diazotrophicus has minimal functional redundancy in genes related to nitrogen fixation. While functional redundancy is often considered a hallmark of organisms carrying out niche functions such as BNF, this microbe contains only a single set of genes for the molybdenum-dependent nitrogenase (18, 19). It does not contain genes coding for alternative nitrogenases, such as the vanadium-dependent vnf or iron-only anf gene systems (48, 49), so all nitrogen fixation should be dependent upon Mo-nitrogenase. This is in contrast to other well-studied diazotrophs such as A. vinelandii and Rhodopseudomonas palustris, which each contain all three functional nitrogenases in their genomes (50).

To counter potential molybdenum limitation, G. diazotrophicus employs highly efficient molybdate transporters encoded by modABC (51). These transporters import the bioavailable form of molybdenum, MoO₄²⁻ (52), allowing for FeMoCo assembly. This gene cluster (Gdia_1579 to Gdia_1581) was associated with large fitness deficits under BNF (Fig. 1). However, the modD gene (Gdia_1551) was not associated with differential fitness. It is important to note that the mod genes in G. diazotrophicus, while all found in the primary nif cluster, are not arranged in one consecutive operon. In Escherichia coli, the modABCD cluster is regulated by the transcription factor modE. While modD is separated from modABC in G. diazotrophicus (Fig. 1), it is possible that all four mod genes are likewise regulated by the putative transcriptional regulator gene modE (Gdia_1131), which was associated with a fitness defect under BNF (Fig. 2). G. diazotrophicus additionally possesses molybdenum cofactor biosynthesis protein genes moaABCDE (Fig. 3). In organisms such as E. coli, these proteins are associated with molybdenum accumulation for synthesis of molybdoenzymes other than Mo-nitrogenase (51). However, under BNF conditions, strains with disruptions to moaAB (Gdia_0124 and Gdia_1246) and moaDE (Gdia_0049-Gdia_0050) displayed large fitness losses. These genes, though not located within the primary nif cluster, are still evidently critical for nitrogenase function and BNF and may further facilitate proper management of metals necessary for BNF in G. diazotrophicus. In contrast, our results did not identify any genes associated with iron limitations under diazotrophic growth. This may be a result of the elevated levels of iron provided in the medium, although this was in contrast to what was found for molybdenum.

Some minimal redundancy is found in G. diazotrophicus regarding BNF-related genes, but this redundancy is primarily limited to the dual ammonium transporter amtB homologs and fix gene clusters. This microbe has a primary cluster of nif genes related to nitrogen fixation, with the absence of genes critical in other systems, such as nifJ, nifL, or rnf gene clusters (53–55). This concise arrangement of nitrogen fixation genes allows for the study of a more streamlined system and makes G. diazotrophicus well suited for study through Tn-seq.

Several genes from G. diazotrophicus are associated with electron transport to nitrogenase. BNF is an energetically expensive process, requiring a minimum of 16 mol of ATP and 8 electrons per reduction of 1 mol of N₂. The primary electron donors to nifHDK are reduced ferredoxin and flavodoxin (56). Contained within the main cluster of nitrogen fixation genes, G. diazotrophicus displays two putative ferredoxin protein genes: the 4Fe-4S ferredoxin iron-sulfur binding domain protein gene fdxN (Gdia_1575) and the nif-specific ferredoxin III protein gene fdxB (Gdia_1562). Both proteins displayed differential fitness under nitrogen fixation conditions (Fig. 1). Other ferredoxin proteins are found throughout the genome outside the primary nif cluster. Their genes include the 4Fe-4S ferredoxin iron-sulfur binding domain protein gene fdxA (Gdia_0255) and the 2Fe-2S ferredoxin protein gene fdxE (Gdia_0615). Neither fdxA or fdxE yielded sufficient data to draw conclusions regarding their fitness, likely due to the small size of the genes and limited TA sites for transposon insertions. Both ferredoxins are commonly found in diazotrophs (56).

G. diazotrophicus contains two putative fix clusters—one within the primary nif cluster, and a partial cluster located elsewhere in the genome: fixABCX (Gdia_1552 to Gdia_1555) and fixABC (Gdia_1988 to Gdia_1990). This represents one of the few instances of functional redundancy in BNF genes in G. diazotrophicus. The Fix proteins are thought to provide electrons through the process of electron bifurcation (57). It has been proposed that the diazotrophic taxa acquiring the fix complex are able to generate reduced ferredoxin from NADH/NADPH (58). The fixABCX cluster found in the primary nif cluster displayed significant fitness detriments under BNF conditions (Fig. 1). The second fix homolog cluster was deemed to be essential under all growth conditions and may play a role in other core processes within G. diazotrophicus.

Nitrogen assimilation in G. diazotrophicus, as in other diazotrophs, is carried out by the glutamine synthetase/glutamate synthetase pathway (GS-GOGAT) (59, 60). Glutamine synthase (glnA, Gdia_1482) was determined to be essential by our analysis. The glnA gene is regulated by an adenyltransferase (glnE, Gdia_2947), which was associated with a moderate fitness deficit when the strain was grown on minimal medium with or without fixed nitrogen. A genomic analysis of G. diazotrophicus suggested that G. diazotrophicus contains alternative routes of ammonia incorporation beyond the GS-GOGAT system (18). These potential pathways include four putative proteins: aminomethyltransferase GcvT (Gdia_0534), NAD-synthase NadE (Gdia_1103), histidine ammonia-lyase HutH (Gdia_1458), and d-amino acid dehydrogenase DadA (Gdia_0673). Disruptions to gcvT resulted in slight fitness deficits under nitrogen-sufficient conditions, with a greater fitness deficit under BNF (Fig. 2). The nadE gene was determined to be essential, while strains with dadA and hutH gene disruption did not demonstrate fitness defects. This indicates that gcvT may indeed be involved in the proposed alternative ammonia incorporation routes (18), although the lack of fitness differential for dadA and hutH suggests that these genes are not critical.

G. diazotrophicus contains two amtB gene homologs encoding ammonium transporters. Both genes were associated with growth defects when disrupted under either growth condition, with a slightly greater defect under BNF conditions (Fig. 2). The fitness arising from disruption of either homolog may be mitigated by the presence and activity of the other. Deletion of both genes simultaneously may result in a stronger fitness defect, which we were unable to observe with this data set. Deletion of the single amtB gene in A. vinelandii was found to release sufficient extracellular ammonium to the medium to support the growth of algae in coculture and is of interest to potentially improve the biofertilizer potential of G. diazotrophicus as well (24). Each amtB gene is preceded by a glnK gene encoding a P_II protein. Due to the very small size of these genes, they were not amenable to analysis by our data set. Depending on the nitrogen concentration within the cell, the structure and resulting function of the P_II proteins are modified by uridylylation. The uridylyltransferase or uridylyl-removing enzyme encoded by glnD (Gdia_0300), previously identified in G. diazotrophicus (61), is a critical protein important in the GO-GOGAT regulatory pathway. This gene was associated with a large fitness defect under nitrogen fixation conditions. Prior work indicated that glnD alongside the regulatory P_II proteins regulates expression of nifA (62), the transcriptional regulator of nitrogen fixation (39). A third P_II protein homolog is coded for by glnB (Gdia_1481), which lies upstream of the glutamate synthase gene glnA (Gdia_1482). Single, double, and triple mutants of the glnK and glnB genes in G. diazotrophicus have previously been constructed (62). The function of each of these three P_II protein products were described as nonessential for nif gene expression (62), where the glnK2 protein product was found to act as the nif inhibitor and the proteins encoded by glnK1 and glnB controlled nif expression in response to ammonium availability. Disruptions to glnB exhibited large fitness differentials under both growth conditions (Fig. 2).

The global nitrogen regulatory system is also thought to regulate the GS-GOGAT system in certain diazotrophs (61). The two-component regulatory system NtrBC controls activation and repression of a variety of genes involved in nitrogen regulation. Under nitrogen limitation, NtrC phosphorylates NtrB, activating transcription of genes involved in both nitrogen metabolism and fixation, such as the transcriptional activator nifA (41, 63). In G. diazotrophicus, the ntrBC operon (Gdia_0486-Gdia_0485) also contains the two-component ntrXY operon (Gdia_0483-Gdia_0484) and a nifR3 family protein gene (Gdia_0487). The ntrX gene was determined to be essential by our analysis. Aside from this, the ntrY gene was associated with only a moderate fitness deficit, while ntrB and nifR3 disruptions resulted in slight growth promotion. Previous research on this system in G. diazotrophicus indicated that the ntrX gene is nonessential for nitrogen fixation (64).

Confirmation that the nitrogenase genes nifDK and nifH were found to be essential is an important control for this study. This result confirms that Tn-seq is an effective tool for assessing gene essentiality associated with BNF in G. diazotrophicus. However, the true revelatory benefit of Tn-seq lies in identification of new genes that are not generally thought to be associated with BNF. Fitness-related defects may be strain dependent and may not extend to other model organisms. For example, A. vinelandii accomplishes BNF under standard atmospheric conditions as an aerobic bacterium, while G. diazotrophicus requires microaerobic conditions. The results in Fig. 3 confirmed additional genes related to molybdenum acquisition that were essential to growth under BNF conditions, as described above. This extended analysis revealed additional genes that were important to BNF in G. diazotrophicus, including genes Gdia_3416 and Gdia_3417, encoding members of the YjgP/YjgQ permease family (renamed LptF/LptG in E. coli [65]) and genes associated with the cell envelope-spanning export complex (ctrA and ctrB). The former are implicated in the transport of lipopolysaccharides (65), while the latter are implicated in the transport of polysaccharides (66). Both may play a role in generating an intracellular environment that is amenable to BNF by limiting oxygen access to nitrogenase. The clpA gene has also been implicated as playing a role in exopolysaccharide production, in addition to regulating other processes (67). These results represent a sampling of the genes that were found to cause significant and reproducible fitness defects during BNF. The hemN gene (Gdia_1986), which is important for oxygen-independent biosynthesis of porphyrins (68), was also found to be important during BNF in G. diazotrophicus.

In addition to gene disruptions that result in growth deficiencies, there are also disruptions that resulted in growth promotion. As in any of these experiments, it is possible that growth improvement may be conditionally dependent. Several larger gene clusters showed significant improvement, including the xagABC cluster, which has been associated with biofilm exopolysaccharide composition in Xanthomonas species and resulted in planktonic growth when disrupted (69). Many of the genes that resulted in growth promotion are less-studied genes that require further characterization. However, the beneficial nature to BNF when disrupting these genes warrants further investigation for the reasons behind this phenomenon. Similar improvements during BNF were found for specific genes in A. vinelandii, although the degree of improvement here was higher than what was found in A. vinelandii (27).

Our analysis also provides a list of essential genes for G. diazotrophicus (see Data Set S4 in the supplemental material) when grown in GAD medium under aerobic conditions. Our analysis of G. diazotrophicus found 548 genes to be essential, while our previous study of A. vinelandii found 475 essential genes (27). A detailed comparison of such a large number of genes fell outside the scope of this report, but may be revisited, especially once a similar data set is obtained for an anaerobic diazotroph. In this manner, this data set extends well beyond the figures and results presented here. The full data sets, which are provided in the supplemental material, can be a useful tool in determining whether to target a gene for deletion and for determining the importance of many other genes in the function of the organism. This also provides a baseline analysis that will be complemented in the future by studies of the strain in association with the plant hosts in which the endophyte resides. Our results indeed indicate that G. diazotrophicus is a much more streamlined organism than A. vinelandii in terms of BNF.

MATERIALS AND METHODS

Bacterial strains and growth conditions.

Gluconacetobacter diazotrophicus PA1 5 (ATCC 49037) was obtained from Cedric Owens (Chapman University) and grown aerobically at 30°C on GAD basal medium and modified B medium unless otherwise specified. GAD medium was adapted from DYGS basal medium (33) (glucose, 2 g L⁻¹; yeast extract, 2 g L⁻¹; tryptone, 1.5 g L⁻¹; MgSO₄, 0.5 g L⁻¹; glutamic acid, 1.5 g L⁻¹; adjusted to pH 6.2 with NaOH). The modified B medium was adapted from Burk’s (B) medium (70) (50 mM sodium citrate, plus the following: sucrose, 20 g L⁻¹; MgSO₄ · 7H₂O, 0.2 g L⁻¹; CaCl₂ · 2H₂O, 0.09 g L⁻¹; Na₂MoO₄ · 2H₂O, 0.025 g L⁻¹; FeSO₄ · 7H₂O, 0.05 g L⁻¹; 2 mL 100× phosphate buffer composed of 20 g KH₂PO₄ and 80 g K₂HPO₄ in 1 L distilled water [dH₂O]; adjusted to pH 6.2 with NaOH). The modified B medium was supplemented with 10 mM NH₄(SO₄)₂ for growth with provided nitrogen. For growth under nitrogen fixation conditions, the medium was prepared without a nitrogen source. The GAD medium used to isolate G. diazotrophicus with transposon inserts was supplemented with 100 mg L⁻¹ tetracycline.

Escherichia coli WM3064 was used for conjugation with the plasmid pBB298, which contained the Mariner transposon, and was grown on lysogeny broth (LB) (71, 72). For E. coli strains carrying plasmids, tetracycline was added at 15 mg L⁻¹.

Genetic constructs.

The plasmid pBB298 containing the Mariner transposon was constructed prior to Tn-seq library construction. This plasmid was modified from pEB001 (the reference plasmid) to incorporate two new restriction sites, and then a tetracycline cassette was exchanged for the kanamycin resistance gene. All relevant plasmids (Table 1) and primers (Table 2) are described regarding its construction, and sequences are available upon request. For a nice overview schematic of the Tn-seq workflow, see reference 37.

TABLE 1.

Key plasmid constructs utilized in this work

Plasmida	Process with relevant gene(s) cloned or plasmid(s) manipulated	Parent vector(s)	Source or reference
pBBTET6	Performed site-specific mutagenesis to add second BamHI site to flank both sides of tetracycline selection cassette from pBBTET2	pBBTET2	75
pBB295	Performed site-specific mutagenesis on pEB001 to add additional BamHI sites flanking transposon antibiotic selection marker	pEB001	71
pBB296	Performed site-specific mutagenesis on pEB001 to add additional BamHI sites flanking transposon antibiotic selection marker	pBB295	This study
pBB298	Moved tetracycline resistance cassette from pBBTET6 into pBB296, swapping kanamycin selection for tetracycline	pBB296 and pBBTET6	This study

^aThe sequences of all plasmids in this study are available upon request. The sequence of pBB298 can be found in the supplemental material.

TABLE 2.

Key primers used in this study^a

Primer	Sequence 5′→3′	Purpose
BBP3084	CGCCAAGCTT GCATGGATCC AGGTCGACTC TAGATATC	Add BamHI to pBBTET2
BBP3085	GATATCTAGA GTCGACCTGG ATCCATGCAA GCTTGGCG	Add BamHI to pBBTET2
BBP3091	CTTGACGAGT TCTTCTGAGC GGGATCCTGG GGTTCGCGGA ATTAATTC	Add BamHI to pEB001
BBP3092	GAATTAATTC CGCGAACCCC AGGATCCCGC TCAGAAGAAC TCGTCAAG	Add BamHI to pEB001
BBP3093	GGTTAATTAA GGGCTGCAGG GATCCGATAT CAAGCTTATC G	Add BamHI to pBB296
BBP3094	CGATAAGCTT GATATCGGAT CCCTGCAGCC CTTAATTAAC C	Add BamHI to pBB296
Tn-seq	TCGTCGGCAG CGTCAGATGT GTATAAGAGA CAGNNNCCGG GGACTTATCAT CCAACCT*G	Sleeping Beauty sequencing primer for Mariner transposon

^aBold letters represent bases that were altered in the primers to add restriction sites.

Library construction.

A transposon library was created through introduction of the plasmid pBB298 into G. diazotrophicus by conjugation from E. coli WM3064 containing pBB298. Briefly, ~50 μL (one loop) of G. diazotrophicus cells was removed from a GAD plate and suspended in 0.5 mL LB medium. Similarly, ~50 μL E. coli WM3064 cells containing pBB298 was removed from an LB plate supplemented with 15 μg/mL tetracycline and 100 μM 2,6-diaminopimelic acid (DAP; 50 μL of a 10-mg/mL stock) and suspended separately in 0.5 mL LB medium. From these stocks, 5 μL E. coli WM3064 and 100 μL G. diazotrophicus were combined and mixed with a pipettor before spotting onto GAD plates supplemented with 100 μM DAP. These plates were incubated overnight at 30°C. Each spot on the plate was then transferred into 50 mL GAD medium in a 125-mL flask and incubated overnight at 30°C on a shaker table at 180 rpm. Following growth in GAD medium, 1 mL of each culture was removed and pelleted before removal of 900 μL supernatant. The cells were resuspended in the remaining medium and plated onto GAD medium supplemented with tetracycline. Plates were incubated at 30°C for several days until colonies formed. After sufficient colony growth, the cells were removed and flash frozen in liquid nitrogen before storage at −80°C. This process was repeated until a library of ≥100,000 mutant colonies was created. The final iteration of ~20,000 mutants was not frozen, but instead was inoculated directly into 250 mL GAD medium as described below to account for potential losses associated with genes important to survival during the freezing of cells.

Library growth conditions.

The entire library of cells was inoculated into a single flask containing 250 mL GAD medium. Frozen cells were thawed at room temperature before inoculation. The entire library was combined and mixed in the GAD medium at 28°C with 160-rpm shaking under a standard oxygen atmosphere for 1 h to adequately disperse the library.

Cells from this initial growth of the library were collected through centrifugation at >12,000 × g for 90 s with the subsequent removal of supernatant. The cell pellets were then inoculated in duplicate into 250 mL of each individual medium. The starting optical density at 600 nm (OD₆₀₀) of these cultures was 0.025. For growth with supplemented nitrogen, the medium was incubated at 28°C with 160-rpm shaking under a standard oxygen atmosphere. For growth in medium without nitrogen, bacteria were grown in an enclosed reactor at 28°C with constant stirring in a controlled 2.5% oxygen atmosphere (0.020 L/min argon, 0.370 L/min nitrogen, and 0.010 L/min oxygen). Both cultures were incubated until harvest. The OD₆₀₀ values of each culture and time of harvest are provided in Table S1 in the supplemental material. An additional control used for further assessment of gene essentiality was run by growing the library in GAD medium incubated at 28°C with 160-rpm shaking under a standard oxygen atmosphere.

Preparation of samples for sequencing.

Upon reaching the desired optical density, cells from each growth condition were collected by centrifugation and the supernatant removed. Cells pellets were flash frozen in liquid nitrogen and stored at −80°C until analysis. Genomic DNA from each library was isolated with ZR Fungal/Bacterial DNA MiniPrep kits (Zymo Research) and subsequently stored at −20°C.

Sequencing.

For each sample, 100 ng of genomic DNA was fragmented to a size of 350 bp by an acoustic DNA shearing device (Covaris). NEBNext Tn-seq Illumina libraries were prepared from the sheared DNA though use of a custom method (27). All libraries were pooled and sequenced on one lane of a 2 × 150-bp run on the Illumina NovSeq 6000 system with an SP flow cell type. A transposon-specific (Sleeping Beauty) primer was used for library sequencing.

Analysis of Tn-seq data.

All TA sites within the genome of G. diazotrophicus were located using custom and adapted Python scripts (73). The scripts also located and removed TA sites that were nonunique (the insertion sites are not flanked by unique sequences) and nonpermissive (sites that do not tolerate insertions). Through use of Cutadapt (version 1.18) (74) and Bioawk, raw sequencing reads of the library grown under the differential growth conditions were filtered and trimmed. Reads without the transposon were discarded. The remaining 16-bp sequences were mapped to the full genome (CP001189.1, genome; CP001190.1, plasmid within G. diazotrophicus; JGI assembly [19] used as reference library and annotation in analysis), allowing for identification of TA insertion locations. Frequency of insertion was determined through counting and summing reads mapping to each gene. Fitness (W) was calculated through equation 1 (36), where $N_{t_1}$ and $N_{t_2}$ were the proportions of the gene before (t₁) and after (t₂) growth under selective conditions, and d was the growth expansion factor calculated via optical density readings as OD_600,t2/OD_600,t1:

$$\begin{array}{c}W=\frac{\ln \left(N_{t_2}\times \frac{d}{N_{t_1}}\right)}{\ln \left(\left(1-N_{t_2}\right)\times \left(\frac{d}{1-N_{t_1}}\right)\right)}\end{array}$$ (1)

All subsequent data analysis and visualization were performed in R through the tidyverse, Gviz, forcats, aplot, gggenes and patchwork packages. A link to the workflow can be found at https://github.umn.edu/knuts623/tn-seq.

Data availability.

Raw sequencing reads are available on the NCBI Sequencing Read Archive under BioProject ID PRJNA888400.