Enabling Biological Nitrogen Fixation for Cereal Crops in Fertilized Fields

: Agricultural productivity relies on synthetic nitrogen fertilizers, yet half of that reactive nitrogen is lost to the environment. There is an urgent need for alternative nitrogen solutions to reduce the water pollution, ozone depletion, atmospheric particulate formation, and global greenhouse gas emissions associated with synthetic nitrogen fertilizer use. One such solution is biological nitrogen ﬁ xation (BNF), a component of the complex natural nitrogen cycle. BNF application to commercial agriculture is currently limited by fertilizer use and plant type. This paper describes the identi ﬁ cation, development, and deployment of the ﬁ rst microbial product optimized using synthetic biology tools to enable BNF for corn ( Zea mays ) in fertilized ﬁ elds, demonstrating the successful, safe commercialization of root-associated diazotrophs and realizing the potential of BNF to replace and reduce synthetic nitrogen fertilizer use in production agriculture. Derived from a wild nitrogen- ﬁ xing microbe isolated from agricultural soils, Klebsiella variicola 137-1036 ( “ Kv137-1036 ” ) retains the capacity of the parent strain to colonize corn roots while increasing nitrogen ﬁ xation activity 122-fold in nitrogen-rich environments. This technical milestone was then commercialized in less than half of the time of a traditional biological product, with robust biosafety evaluations and product formulations contributing to consumer con ﬁ dence and ease of use. Tested in multi-year, multi-site ﬁ eld trial experiments throughout the U.S. Corn Belt, ﬁ elds grown with Kv137-1036 exhibited both higher yields (0.35 ± 0.092 t/ha ± SE or 5.2 ± 1.4 bushels/acre ± SE) and reduced within- ﬁ eld yield variance by 25% in 2018 and 8% in 2019 compared to ﬁ elds fertilized with synthetic nitrogen fertilizers alone. These results demonstrate the capacity of a broad-acre BNF product to ﬁ x nitrogen for corn in ﬁ eld conditions with reliable agronomic bene ﬁ ts.

N itrogen is often the most rate-limiting nutrient in agricultural systems. 1,2 Synthetic nitrogen fertilizers address this limitation by providing a reliable source of nitrogen to crop plants.
Fifty percent of the global agricultural productivity gains in the 20th century can be attributed to synthetic nitrogen fertilizers 3 enabling high-yielding crop varieties and irrigation to increase productivity. 4 The benefits of synthetic nitrogen fertilizers are counterbalanced by consistent loss of half or more of the applied nitrogen fertilizers from the field. 2 The resulting consequences have been well documented: water and air pollution, stratospheric ozone depletion, hypoxic dead zones that stretch for thousands of square miles, and the generation of nitrous oxide, a greenhouse gas 300 times more potent than carbon dioxide. 5−9 Synthetic nitrogen fertilizers, from the energyintensive process of synthesis to the inevitable loss of fertilizers from the field, cause an estimated damage of $200B each year. As such, there is a pressing need to mitigate the negative impacts of applied nitrogen fertilizers while increasing nutrient supply to intensify row crop production on existing land to meet the growing demand for food, fuel, and fodder.
One such solution lies with soil microbes known as diazotrophs. These microbes can reduce atmospheric nitrogen to ammonia, a bio-available form of nitrogen, via biological nitrogen fixation (BNF). 10 A protein complex known as nitrogenase enables BNF for microbes. Crop engineering and synthetic plant−microbe symbioses that transfer nitrogenase and the capacity to fix nitrogen to cereals have been widely explored but have proved challenging to implement. 11,12 Developing and commercializing nitrogen fixing bacteria has the potential to combine robust plant−microbe relationships with the capacity to quickly and effectively gene-edit microbes for purpose. BNF for cereal crops is of particular interest, as cereal grains such as rice, wheat, and corn provide 50% of global calories and are the recipients of 45% of global fertilizer applications. 13 However, two major hurdles to the commercialization of BNF microbes for cereal crops exist: the technical challenge of enabling BNF microbes to operate under field conditions, 10,14 and the commercial challenge of successfully bringing such a microbe to market and widespread adoption. 15 BNF is an energetically expensive process requiring high reducing power and a large amount of ATP. As a result, diazotrophs evolved mechanisms which allow them to tightly control the formation and expression of the nitrogenase complex. 16,17 In some farmed soils, the application of synthetic nitrogen fertilizers occurs at concentrations many orders of magnitude greater than the exogenous nitrogen required to suppress nitrogen fixation by the microbes, thereby "switching off" BNF. 18 Despite tight control of nitrogen fixation, diazotrophs are estimated to provide 10% of the nitrogen crop cereal budget. 19 Free living crop-associated diazotrophs capable of providing nitrogen at agriculturally relevant levels as observed by Van Deynze et al. and Ladha et al. 19,20 indicate that this nitrogen source could be developed and optimized for modern agriculture. However, any microbe identified as providing BNF for cereal crops will need to be gene-edited for function in the nitrogen-rich soil conditions which would normally suppress nitrogen fixation. 21 Synthetic biology provides the tools and engineering frameworks to effectively develop microbes for purpose. 22 Advances in genomics enable us to better characterize the myriad members of these often complex families of bacteria. The decreasing cost of DNA sequencing 23 has made it possible to screen the billions of microbes present in soil for specific characteristics and metabolic activities. 24 Improvements in computational power enable fast and reliable distinction between microbes of interest based on genetic variation, and advances in biotechnology, such as precision gene-editing, have made it possible to enhance naturally occurring microbial activities including nitrogen fixation. 25 Sustainable, scalable, drop-in microbial products are not yet widespread across global agriculture. While microbes have long been touted as the clear solution to diverse challenges in sustainable agriculture, only 1% of potentially beneficial bacteria characterized in the laboratory have emerged onto the marketplace. 15 One challenging aspect is that many potentially beneficial microbes found in the nutrient-rich, highly competitive rhizosphere are related to microbes that have the potential to be opportunistic human pathogens, 26 highlighting the need for rigorous assessment of biosafety as a part of any product development effort. A second challenging aspect is that adoption of new agronomic practices and products is frequently slowed by the cost of implementation and uncertain returns on investment. 27 Despite the challenges, the impact of a commercial microbe for BNF in cereal crops could be as profound a shift for agriculture as the discovery of the Haber−Bosch process was over a century ago. Such a product would allow for an immediate and effective change in intensive agriculture practices, meeting our time-sensitive need to scale up alternatives to synthetic nitrogen. 28

■ RESULTS AND DISCUSSION
In this report, we address both engineering and product development challenges to develop the first commercially available nitrogen-fixing microbe for corn. We first identified and characterized an agricultural soil-derived wild-type diazotroph (Klebsiella variicola strain 137; Kv137) through use of computational and synthetic biology tools. We then edited the genome to produce K. variicola strain 137-1036 (Kv137-1036), a non-transgenic strain which fixes nitrogen regardless of exogenous nitrogen levels. We evaluated the nitrogen fixing and excretion capacity of the genetically remodeled strain Kv137-1036 both in vitro and in planta. Upon verification of function, and in partnership with a multitude of collaborators, we addressed the commercial challenges of safety, stability, and efficacy through standard biosafety assay panels, long-term viability assays, and in-field performance across the U.S. Corn Belt over 2018 and 2019 crop years. The result is a microbial product that supplies nitrogen to corn crops in nitrogen-rich soil conditions, with performance suitable for broad acre use in conjunction with standard agricultural practice. Our work introduces an environmentally and economically sustainable alternative nitrogen source for farmers and contributes to a framework for the development and subsequent commercialization of nitrogen-fixing biofertilizer products for agriculture.
Wild-Type Strain K. variicola 137 and Remodeled Strains. Bacterial strain 137 was isolated from the surface of corn roots grown in soil collected near a farm in St. Charles County, Missouri, USA. There are reports of endophytic K. variicola species 29,30 as well as free-living diazotroph species. 31 The Kv137 isolate was derived from plant roots gently rinsed free of bulk excess soil but not surface-sterilized. Subsequent microscopy work (as shown in Figure 2F) revealed the presence of the microbe only on the exterior of corn roots after inoculation, suggesting that Kv137 is a root-associated diazotroph that colonizes the rhizoplane rather than internal plant tissues (data not shown).
The microbe was initially identified as K. variicola at 100% identity through 16S rRNA sequence alignment after genomic sequencing. However, identifying taxa within the genus Klebsiella can be challenging. 32,33 In order to better compare the putative K. variicola 137 strain against a broader cross section of related species, we calculated the percent-average nucleotide identity (ANI) between strain 137 and 16 Klebsiella strains and constructed a phylogenetic tree 34 ( Figure 1A) showing a clear demarcation between the various species of Klebsiella, with strain Kv137 located among other K. variicola species to comprise a monophyletic clade.
Comparisons of Kv137 within the K. variicola clade yielded >98%, well above the standard 95% ANI threshold for species determination ( Figure 1A). Together, the percent ANI and phylogenetic tree analyses support the initial determination of Kv137 within taxon K. variicola. Further genomic analysis revealed the presence of a nif cluster homologous to that described in other plant-associated K. variicola strains. 35,36 The presence of a nif gene cluster in a soil-based microbe, in combination with >98% ANI between Kv137 and K. variicola, a family of microbes known to contain plant-associated diazotrophs, suggested that Kv137 was an intriguing candidate for the development of an agriculturally indigenous diazotroph for BNF in cereal crops.   In order to develop a microbe for use with cereal crops in fertilized fields, we edited the Kv137 genome to decouple regulation of nitrogen fixation from the presence or absence of exogenous nitrogen. 21 Similar to other nitrogen-fixing Gammaproteobacteria, strain Kv137 contains the nif L and nifA genes responsible for control of the nitrogen fixation pathway in an operon driven by a single repressible promoter upstream of nif L 17 ( Figure 1B). We thus created strain Kv137-1036 by replacing nif L of strain Kv137 with an endogenous constitutive promoter consisting of the 500 bp immediately upstream of inf C. 37−40 This substitution removes negative regulation of the NifA protein by eliminating the inhibitor NifL and allowing for the constitutive production of NifA, which we hypothesized would drive nitrogen fixation in the nitrogenreplete conditions of a fertilized field. This edit does not impact the main components of the nif cluster, genes nif HDK, which encode for functional formation of the nitrogenase complex.
To confirm the nitrogenase function in vitro, a nif H knockout mutant strain of Kv137-1036 was also constructed ("Kv137-3738"). This edit used the same mutagenesis approach, introducing the same nif LA edits and also deleting the entirety of the nif H gene. These modifications were carried out by using the guided microbial remodeling methods described in Bloch et al. 28 In Vitro and In Planta Confirmation of Enhanced Nitrogen Fixation Capabilities in Strain Kv137-1036. Nitrogenase Activity In Vitro and Ammonium Excretion Assay. The standard acetylene reduction assay (ARA) was conducted for wild-type and edited strains as previously described. 10,41 Acetylene was injected into the headspace of bacterial cultures, and after incubation, the resulting headspace with ethylene was sampled and quantified as a proxy for nitrogen fixation in nitrogen-free and nitrogen-rich media conditions. In the absence of nitrogen, both wild Kv137 and edited Kv137-1036 produced similar quantities of ethylene. In the presence of 5 mM ammonium, Kv137 showed little measurable acetylene reduction, while Kv137-1036 showed activity similar to the levels of acetylene reduction achieved under nitrogen-free conditions ( Figure 2A). Ethylene production for the parent strain Kv137 in nitrogen-free media was 2.2 × 10 −15 mMol ethylene/CFU h compared to 2.7 × 10 −13 mMol ethylene/CFU h for Kv137-1036, a 122-fold increase in nitrogen fixation. Thus, in the presence of fieldrelevant concentrations of exogenous nitrogen (e.g., fertilized farms), ethylene levels produced by strain Kv137-1036 were significantly higher than parent strain Kv137 ( Figure 2A). As expected, no ethylene production was seen under any condition for the nif H knockout strain Kv137-3738.
BNF for use by cereal crops requires the fixed nitrogen be excreted from the microbe to become available for uptake by the plant root. We measured the ammonium excretion of strains cultured in anaerobic conditions in nitrogen-free media, quantifying the total concentration of ammonium after 3 days ( Figure 2B). Per cell, Kv137-1036 produced an average 23-fold more ammonium than the wild-type strain ( Figure 2B). While no gene edits targeting excretion mechanisms were made, Kv137-1036 releases ammonium into its environment at a significantly higher rate than the wild type or nif H knockout mutant, suggesting that gene edits which optimize for continuous nitrogen fixation generate enough ammonium for passive excretion. These results suggest that Kv137-1036 is able to both fix nitrogen in nitrogen-rich environments such as fertilized fields and transfer a portion of that fixed nitrogen to crops.
Nitrogenase Activity In Planta. To confirm that an increase in nitrogenase expression and activity is sufficient to generate an excess of ammonium ions within Kv137-1036, leading to passive excretion into the rhizosphere, 28 corn seedlings are grown in sterile, transparent plant growth pouches and inoculated with microbial cultures ( Figure 2D). Air was pressed out of pouches prior to sealing; however, the experiment was not conducted under anaerobic conditions. After several days of growth in a growth chamber, the bags were injected with acetylene and exposed overnight, after which the pouches were sampled and analyzed for ethylene. While Kv137 and Kv137-3738 showed little to no detectable ethylene production on corn roots, Kv137-1036 showed clear acetylene reduction compared to controls (p < 0.1) ( Figure  2C). These results show that the increased nitrogen fixation in Kv137-1036 over the wild-type strain translates to the root environment in association with plants. Because no additional carbon source was added at the time of inoculation, these results suggest that Kv137-1036 can use root exudate as a carbon source to fuel nitrogen fixation in planta.
Root Colonization. To effectively deliver fixed nitrogen to plant roots, BNF microbes must be competitive with the existing flora. One aspect of this competitiveness is the ability of the microbe to effectively colonize the rhizosphere, securing access to the root exudates necessary for sustained microbial growth. 42 However, colonization is a complex and poorly characterized process, making rational design difficult. Rather than target colonization mechanisms with gene-editing, we evaluated the microbes for wild colonization competence.
Root colonization was quantified by inoculating corn seeds with approximately 10 7 CFU of either Kv137, Kv137-1036, or E. coli, and harvesting seedling roots after 3 weeks of plant growth for genomic DNA extraction and qPCR analysis targeting the Kv137 genome (Kv137 Probe) or the E. coli genome (E. coli Probe) ( Figure 2E). Both Kv137 and Kv137-1036 colonize corn roots at levels two orders of magnitude higher than the background signal detected in untreated control (UTC) samples and bacterial inoculants without cropspecific associations (e.g., E. coli) ( Figure 2E). The signal detected by the Kv137 probe in UTC samples may represent background Klebsiella species present in the plant growth media as Klebsiella is a common rhizosphere microbe.
To visualize colonization, remodeled strain (Kv137-1595) containing similar nitrogen fixation edits as Kv137-1036 in addition to red fluorescent protein (RFP) fused to nifA was inoculated onto corn seedlings. Fluorescent microscopy showed individual bacterial cells of Kv137-1595 on the exterior surface of the roots ( Figure 2F). Microcolonies along the roots can also be seen. These results indicate that Kv137-1036 can be considered a root-associated diazotroph capable of colonizing corn roots from germination onward and maintaining colonization in the presence of a functional rhizosphere.
Commercial Efficacy of Strain Kv137-1036. Having remodeled a microbe capable of BNF for cereal crops in nitrogen-replete conditions, we validated the commercial potential of formulations containing this microbe, ensuring that any product emerging from this work was both safe and effective. 15 Biosafety Studies. As certain K. variicola strains have been isolated in healthcare environments as opportunistic patho-   Product Stability. A dry powder formulation of lyophilized Kv137-1036 was developed and pre-commercially marketed as Pivot Bio PROVEN in 2018 ( Figure 3D). The dry formulation was suspended and activated in a sterile liquid medium according to packaging instructions. After the cap containing the dry microbial powder was punched in, activating the product, technical and biological replicates were sampled to verify that freeze-dried formulation enabled the temperature and time stability essential for planting. Results indicated that Kv137-1036 is stable in freeze-dried form for several weeks in refrigerated and room temperatures. After 32 weeks, the freezedried powder contained viable cells around 1 × 10 9 CFU/g at 20°C ( Figure 3B). Moreover, the formulated product remained viable for at least 14 days after punch-cap activation ( Figure 3C).
Field Trials. In 2019, a field-ready dry powder formulation containing Kv137-1036 was released commercially as Pivot Bio PROVEN to corn farmers across the U.S. Corn Belt. After activation in accordance with the instructions on the product label, the microbe was applied at planting via in-furrow planting systems, which deposit small quantities of liquid in close proximity to the seed. In 2019, over 2.54 million yield data points were generated from 38 farms growing corn with Pivot Bio PROVEN alongside untreated checks in structured field trials that we designed and commissioned through a third party. In total, 48 large plot trials (without replication) were conducted over 2 years. Trials consisted of two treatments: a control using grower standard practice and that same practice with the addition of Kv137-1036. Table 1 provides a summary The microbial solution will be applied alongside the farmer's standard inputs. (C3) Image of infurrow planting equipment for the delivery of the activated microbial solution onto seed at planting. Simultaneous deposition of seeds and microbes inoculates each corn plant in a field with nitrogen-producing bacteria. (C4) Colonization of corn roots by microbes (red) after germination as described in Figure 2F.  Further information about the dataset is summarized in the Materials and Methods section ( Figure 4). As all trials used the same basic design across geographies and years, a combined site analysis was conducted. 46 Yields of control and inoculated treatments were measured within each site. The percentage of trials showing increased yield due to inoculation, the average yield increase (t/ha), and the 95% confidence interval of this increase as well as the change in CV between treated and untreated were determined. Mean yield and CVs of treated and untreated plots were compared using pairwise Student's t-test statistics at the 95% confidence level.
Inoculation with Kv137 increased maize yield in 35 of 48 trials. In both years, the yield increased significantly at the 95% confidence level across trials, with 12 of 17 (71%) of trials in 2018 and 23 of 31 (74%) of trials in 2019, resulting in yield increases. The CVs of individual farm yield data showed a decrease between treated and untreated plots significant in both years of 25% (p = 0.001) and 8% (p = 0.005) in 2018 and 2019, respectively ( Table 2). The reduction in variability is more completely described in a recently submitted patent application (PIVO-019/00US (316309-2053)).

■ DISCUSSION
Here, we characterize the successful isolation, identification, development, and commercial deployment of a free-living agriculturally relevant diazotroph capable of BNF for corn. Commercialization of Kv137-1036 was made possible by addressing technical challenges, 10 key customer needs (e.g., efficacy; application technologies), and health and safety (e.g., regulatory concerns). 15 Integrating technical improvements with thoughtful investigation of commercial concerns resulted in the first commercial BNF microbe for corn, Kv137-1036.
Despite a robust body of evidence suggesting that free-living microbes have the potential to be a reliable biological source of fixed nitrogen for cereal crops, 12 it is only recently that this hypothesis could be tested effectively. We leveraged decades of research on the regulation of the nif operon and genome editing strategies to improve nitrogen fixation by Kv137 and excretion of bioavailable nitrogen into the rhizosphere for use by the crop. 21 Constitutive expression of the nifA gene in Kv137-1036 resulted in nitrogen fixation in the presence of exogenous nitrogen, making this microbe compatible with modern row crop agriculture. Further work may explore other aspects of the nitrogen fixation pathway either individually or in combination. For example, we recently showed that modifications in genes involved in nitrogen assimilation (glnE) and nitrogen signaling (glnD) significantly increase the amount of ammonium excreted by Kosakonia sacchari PBC6.1, another strain of root-associated diazotroph within the same family as Kv137. 28 The uncertainty surrounding species and taxa in the largely undocumented soil microbiome is a considerable challenge when commercializing soil microbes, and K. variicola variants are no exception. 47−49 In an instance of taxonomic confusion, the well-studied root-associated diazotroph K. variicola strain 342 was initially characterized as K. pneumoniae (Kp342) despite its phylogenetic relationship to K. variicola strain At-22. 50 Kp342 was eventually reclassified as K. variicola strain 342. 51,52 Taxonomic confusion or otherwise, individual species within the K. variicola family have been recognized as opportunistic pathogens. An in silico screening of 31 K. variicola genomes by Martinez and colleagues 32 uncovered a "mosaic distribution" of proteins related to both plant host affinity and potential for virulence. The variety of genes that can be present across K. variicola strains yet not generally universally encoded in the genomes highlight the need to evaluate toxicity empirically for each strain phenotype as part of the commercialization process.
To the authors' knowledge, this is the first paper describing the commercialization of a microbial strain that explicitly investigates biosafety implications. The results of the toxicity and pathogenicity studies documented here indicate that Kv137-1036 is a K. variicola strain without toxicity or infectivity concerns to humans. Together with the genomic analysis of isolate Kv137, these results seem to support the premise that selective pressures result in strain adaptations within the K. variicola family that restrict a given species to either clinical or plant-associated settings (e.g., ref 53). In the years since initial characterization of K. variicola as a distinct Klebsiella species, the molecular framework for distinguishing K. variicola continues to grow. 47 Developing a greater capacity for species identification will enable the commercialization of additional species for agricultural use.
This work also joins the relatively small body of literature that quantifies the impact of microbial inoculants on yield and yield variance in multi-year, multi-geography large plot studies. 46 The consistency of both yield advantage and the reduction of yield variance across 2 years of trials is a promising indication that these free-living diazotrophs show similar BNF performance on corn roots across disparate geographies, possibly due to the consistency of the rhizosphere microclimate. Given that yields in agriculture integrate variables which are largely uncontrolled by the grower, including weather, temperature, soil type, and topography, 54 products which perform reliably have additional market potential.
The potential impact of BNF on cereal crops cannot be fully characterized without an investigation into the nitrogen flux between plant and bacteria. This approach, which may include experiments detailing colonization dynamics and tracing heavy isotopes of nitrogen throughout the system, is challenging but necessary to establish the contribution of microbes to total plant nitrogen. 14,20 More research is needed to determine the precise contribution of nitrogen fixing microbes to the corn plant over the course of the growing season to identify the agronomic equivalent quantity of synthetic nitrogen fertilizer which the microbes can displace.
A novel microbial product such as Kv137-1036 that improves nitrogen use efficiency is one of the few ways that farmers, the fertilizer industry, and the environment all benefit from innovation. 27 This product benefits the grower through  46 BNF for cereal crops not only supports productive yields, but it also has the potential to minimize nitrogen loss to the environment. By fixing nitrogen in close proximity to corn roots, Kv137-1036 can be used to complement existing nutrient management practices to support the "4Rs" of nutrient stewardship: right time, right dose, right place, and right source. 55 Given that even the most highly efficient agricultural systems have an NUE of only 40%, 27 microbial production of nitrogen at the root could signal a sea change in nutrient management.
The economic and environmental benefits of BNF for cereal crops make it a tool uniquely suited for voluntary adoption in efficient agricultural systems. Continued research into both formulations that improve grower access to the product and the performance of microbial biofertilizers across soil types and regions will give growers confidence that BNF for cereal crops will benefit their operation. As a result, farmers will, for the first time, be able to replace and reduce their dependence on synthetic nitrogen fertilizer while maintaining yields. The commercialization of Kv137-1036 marks a turning point for nitrogen, giving growers a much-needed tool to ensure adequate crop nutrition while minimizing nutrient loss to the environment. Hoagland's V2.11-0.5 medium (used for plant assays) contains 9.6 g/L of NH 4 NO 3 , 7.5 g/L of KCL, 3.3 g/L of CaCl 2 , 1.4 g/L of KH 2 PO 4 , 4.9 g/L of MgSO 4 ·7H 2 O, and 0.15 g/L of FeSO 4 ·7H 2 O.
Isolation and Identification of Isolate Strain 137. The microbe of interest was originally obtained from agricultural soil in St. Charles County, Missouri, USA. Soil samples were diluted with 1 mL of PBS and then centrifuged for 1 min at 13,000 rpm, and serial 10 −1 dilutions of the samples were made. Each dilution was plated onto an NFb agar medium supplemented with 0.2% casamino acids. 56 The plates were incubated at 30°C for 4−6 days. The colonies that appeared were tested for the presence of nif H using primers Ueda19f and Ueda407r. 56 To confirm the colonization activity of the microbe, corn seedlings were grown from seed (DKC 66-40, DeKalb, IL, USA) for 2 weeks in a greenhouse environment controlled from 22°C (night) to 26°C (day) and exposed to 16 h light cycles in the same agricultural soil collected from St. Charles County, MO, USA. Roots were harvested and washed with sterile deionized water to remove bulk soil. Root tissues were homogenized, and the samples were centrifuged for 1 min at 13,000 rpm to separate tissue from root-associated bacteria. 21 Positive hits were subsequently purified on the same medium. The presence of nif H was reconfirmed by PCR, and a preliminary identification of the microbe was performed by amplification with 16S rRNA primers 27F and 1492R, 57 followed by Sanger sequencing 58 and NCBI BLAST analysis. 59 Initial NCBI BLAST of the sequence 16S rRNA gene amplicon matched several K. variicola isolates at 100.00% identity. Genomic DNA was prepared with the MagAttract HMW DNA Kit (Qiagen cat no. 67563). PacBio genome sequencing and assembly were performed (SNPsaurus, Eugene, OR), and the resulting genome assembly was annotated with Prokka v. 1.12. 60 For phylogenetic analysis (Figure 1), a total of 19 genome assemblies were reannotated with the most recent version of Prokka (v. 1.13.3) at the time for annotation consistency. These 19 genome assemblies include Kv137, the Kv137-1036 derivative strain (described below), representative genomes for the 16 Klebsiella-type strains with genomes available in the NCBI RefSeq database, and E. coli K-12 substr. MG1655 (for out group comparison). Phylogenetic analysis was performed on a set of 104 conserved protein sequences as described in Parks et al. (2017). Analysis was restricted to proteins that were present annotated in >80% of our assemblies, bringing our protein count from 104 to 100. All 19 genome assemblies contained 80% or more of the 100 remaining protein annotations. Protein sequences were concatenated and aligned with MUSCLE. 61 Any columns of residues present in fewer than 50% of assemblies were considered insufficiently informative and removed before subsequent analysis. Phylogenetic trees were built with FastTree 62 and graphed with FigTree 63 using the E. coli strain as the out group. To more quantitatively determine the identity of Kv137, the ANI of Kv137's genome was compared to the other 18 genomes using Mash at the recommended species cutoff measurement of 95%. 64 Remodeled Strain Description and Genomic Modifications. The genotype of Kv137-1036 is ΔnifL::PinfC, with the nif L locus including a gene deletion and promoter insertion. Kv137-1036 and Δnif H (Kv137-3738) were carried out using the same mutagenesis approach as described in a patent application on our guided microbial remodeling platform. 28 Strains with genomic edits were cured of all plasmids through repetitive sub-culturing for the desired edits.
Acetylene Reduction Assay. A modified version of this standard assay described by Temme et al. 65 was used to measure nitrogenase activity in pure culture conditions. Strains were cultured from single colonies into 4 mL of SOB for 24 h (30°C, aerobic). The growth culture (1 mL) was then added to 4 mL of minimal media or 4 mL of minimal media supplemented with 5 mM ammonium phosphate in airtight culture tubes prepared in an anaerobic chamber and grown for 4 h (30°C, anaerobic). A headspace of 10% was replaced by an equal volume of acetylene and incubation continued for an additional hour. A gas-tight syringe was used to remove 2 mL of headspace in preparation for ethylene production quantification using either an Agilent 6850 or 7890B gas chromatograph equipped with a flame ionization detector (FID). The initial culture biomass was compared to the end biomass by measuring OD590. To establish a negative control, we knocked out the central nitrogenase subunit NifH, deleting the entirety of the nif H gene. The nitrogen fixation − ACS Synthetic Biology pubs.acs.org/synthbio Research Article phenotype was confirmed via ARA prior to use as a negative control. Sterility is maintained throughout this experiment. Ammonium Excretion Assay. Excretion of fixed nitrogen in the form of ammonium was measured using batch cultures in DeepWell plates. Strains were propagated from single colony in 1 mL/well SOB in a 96-well DeepWell plate. The plate was incubated for 24 h (30°C, 200 rpm) and then diluted 1:25 into a fresh plate containing 1 mL/well of growth medium. Cells were incubated for 24 h (30°C, 200 rpm) and then diluted 1:10 into a fresh plate containing minimal medium. The plate was transferred to an anaerobic (Coy) chamber with a gas mixture of >98.5% nitrogen, 1.2−1.5% hydrogen, and <30 ppM oxygen and incubated at 1350 rpm at room temperature for 72 h. The initial culture biomass was compared to the end biomass by measuring OD590. Cells were then separated by centrifugation, and supernatant from the reactor broth was assayed for free ammonium using the Megazyme Ammonia Assay Kit (P/N K-AMIAR) normalized to biomass at each time point. Sterility is maintained throughout this experiment.
Fluorescence Microscopy of Root Surface Colonization. Sprouted corn seeds with roots were half-immersed in a 48-well plate where each well contained 4 mL of sterile Hoagland's V2.11-0.5 solution. The plate was covered with a breathable seal (wells were slit to support the seeds) and incubated in a humidified and temperature-controlled growth room for 72 h. The wells were topped off with fresh Hoagland's solution after 48 h. The remodeled strain Kv137-1595 (genotype: Kv137_glnE_KO2-unintended deletion, ΔnifL-Prm1.2, Prm1.2-RFP-linker-nifA) was inoculated into 5 mL of SOB from a single colony and grown aerobically for 48 h at 30°C. Prior to inoculation, the bacterial culture was diluted to 109 cell/mL. The diluted bacteria culture (10 μL) was added to the roots of each seeding for a final concentration of 107 cells/seedling. The inoculated seedlings were then incubated in the growth room for an additional 24 h. Root sample preparation and staining: Plants were removed from the growth plate, and the roots were rinsed in sterile PBS. Root sections were obtained using a sterile razor blade. The SYTO9 green fluorescent stain (6 μM, Life Technologies P/N S34854) was prepared from a 5 mM stock solution by diluting in sterile DI water. Immediately before imaging, root samples were added to the 6 μM SYTO9 solution and incubated at room temperature for 15 min in the dark. The samples were imaged on a Zeiss LSM710 confocal microscope or a Leica DM IL fluorescence microscope, and image processing was done using the Zeiss Zen Black imaging software and/or Leica LAS X imaging software and ImageJ.
Root Colonization Assay. As described in Bloch et al., a planting medium with minimal background nitrogen was prepared with pure sand autoclaved for 1 h at 122°C, and ∼600 g was measured out into a D40 Deepot (Stuewe and Sons) before planting corn seeds at a depth of ∼1 cm. Seedlings were inoculated with a suspension of cells drenched directly over the emerging coleoptile at 5 d after planting. The inoculum was prepared from 5 mL of overnight culture in SOB, which was spun down and resuspended twice in 5 mL of PBS to remove the residual SOB before final dilution to an OD of 1.0. The plants were maintained under standard growth room conditions using fluorescent lamps and a 16 h day length with a 26°C day temperature and 22°C night temperature. Plants were fertilized twice per week with a modified Hoagland's fertilizer solution containing 2 mM KNO 3 . All pots were watered with sterile deionized H 2 O as needed to maintain consistent soil moisture.
At 3 weeks of growth, three replicate plants were collected, and plant roots were washed and harvested for the preparation of genomic DNA extraction to quantify root colonization using quantitative PCR (qPCR) with primer−probe pairs that targeted genome-specific regions of either Kv137 or the E. coli genomes. The primers were designed by Primer Blast. 66 The Kapa Probe Force kit (Kapa Biosystems P/N KK4301) was used as per manufacturer's instructions, and qPCR reaction efficiency was measured using a standard curve generated from a known quantity of genomic DNA from the target genomes. The data shown in Figure 2E are normalized to genome copies per gram fresh weight using the tissue weight and extraction volume.
ARA In Planta. Corn seedlings were cultured on water in sterile conditions, and three 7 day-old seedlings were placed in each sterile plant growth pouch under aerobic conditions and inoculated with a microbial culture of Kv137, Kv137-1036, or Kv137-3738, which had been centrifuged, the spent medium was removed, and the cells were resuspended in sterile water. All pouches were placed in transparent boxes for growth in a growth chamber with a 16 h day length and controlled humidity for 5−10 days. The pouches with seedlings were then transferred to gastight bags, and acetylene was injected to allow for incubation at 30°C overnight, after which the headspace of the bags was sampled and analyzed by gas chromatography.
Shelf Life and Viability of Freeze-Dried Microbial Powder Formulation. To evaluate the freeze-dried microbes in the punchcap system/to test the stability of cap formulation (dry powder product shelf-life), mean (CFU/g) was sampled over time. The data are an average of triplicate samples from two production batches. Each error bar is constructed using 1 SD error from the mean. To test the product viability after activation, CFU/mL was sampled at various time points after activation (measured in days). Each error bar is constructed using 1 SD error from the mean.
Characterization of the Potential Toxicity, Pathogenicity, and Irritancy of Kv137-1036. A set of six toxicity and pathogenicity studies were conducted by a third-party contract research organization (Product Safety Labs, Dayton, NJ, USA) to characterize the potential toxicity, pathogenicity, and irritancy of a solution containing K. variicola 137-1036 following acute exposure. The test solution contained 1 × 109 CFU/mL K. variicola 137-1036, a concentration higher than a proposed product that would be diluted for use. All studies were conducted under Good Laboratory Practice (GLP). Brief summaries of methods are presented below: • Acute Oral Toxicity (OPPTS 870.1100)An initial limit dose of 2000 mg/kg was administered to one rat.
In the absence of mortality, four additional rats were sequentially dosed at the same level. Since all five rats survived, no additional animals were tested. Individual doses were calculated based on the initial body weights, taking into account the density of the test substance. The test substance was administered directly to the stomach via oral gavage. The animals were returned to their cage and feed was replaced 3−4 h after dosing. All animals were observed daily for 14 days after dosing. Body weights were recorded prior to administration and again on days 7 and 14 (termination of study). Necropsies were performed on all animals at day 14.
ACS Synthetic Biology  67 The animals were observed daily until termination of study. Body weights were recorded prior to exposure and at the end of the study. • Acute Pulmonary Toxicity/Pathogenicity (OPPTS 885.3150)Thirty-eight rats were divided into two study groups. Group A (17 rats of each sex) received a pulmonary dose of 0.15 mL containing 1.0 × 10 1 viable cells of the test microorganism Kv137-1036. Group B (2 rats of each sex) was an UTC. Viable CFUs were enumerated to confirm delivery of greater than or equal to 1 × 10 8 viable cells in 0.1 mL. The test substance was mixed thoroughly and serially diluted with PBS to reach a concentration of 10 8 CFU/mL for dosing. Groups of three rats from Group A were sacrificed at various intervals after dosing to assess the distribution of the microbe in tissues and organs, the potential pathogenicity, and the pattern of elimination of the microbe from the body. Samples of lung, blood, brain, kidney, liver, lymph nodes, spleen, and cecum contents were collected from all animals at their scheduled sacrifice at days 1, 3, 7, 15, and 22/23. Body weights were recorded from surviving animals on days 1, 3,8,15, and prior to sacrifice (day 22/23). At each sacrifice, blood and organ tissues were plated to determine the amount of Kv137-1036 present.
Broad-Acre Field Trial Design and Data Collection. For structured field trials, corn yields produced with the grower's standard nitrogen fertilization practice were compared to corn yields produced with the addition of a commercial formulation of Kv137-1036 (Pivot Bio PROVEN) containing 10 8 CFU/mL to the grower's standard nitrogen fertilization practice. The microbe was applied at a rate of 12.8 fl oz per acre (934 mL/ha) as in-furrow application at planting. Fields were split in half with a Pivot Bio PROVEN treated area on one side of the field and the grower standard practice (GSP) plot on the other. The two treatment zones were identified by digital as-planted (planter monitor) and harvest (combine yield monitor) maps. To obtain customer data from field trials, customers that purchased Pivot Bio PROVEN in 2019 were invited to share their planting, treatment, and yield data with us through an incentive program. The trial layouts were determined by customers for their own fields. Customer fields that were not laid out in a similar fashion to the structured field trials were discarded.
Data Processing of Field Trials. Monitor data for both structured trials and customer data trials were received, cleaned, and processed by a third-party company, IN10T, who performed yield analysis with ArcGIS. Trials were subjected to a further agronomic QC check to screen for serious defects in field conditions, on-farm management, or data collection issues that would skew the results. In order to ensure representative comparisons from both Pivot Bio PROVEN and untreated field regions, IN10T used algorithms to remove unsuitable parts of the field. Header rows, which are typically lower in yield, more prone to damage, and have a varying incident solar radiation profile, were removed from field datasets. The most reliable data from combine harvest monitors occurs in areas where the combine is moving at a steady velocity. Thus, data points were removed with automated filters where the combine was accelerating or where the combine had to slow down to pass obstacles such as field drains or terraces.
To generate Tables 1 and 2, data was separated by year, and the following analyses were performed with Python. The mean of yield within each treatment was calculated for the two treatments within each trial (untreated and treated with Pivot Bio PROVEN), which created 17 pairs of data for 2018 and 31 for 2019. The distribution of the treated and untreated means examined for normality and constant variance. The Box−Cox transformation was applied to 2019 alone. The Shapiro−Wilks test for normality was applied to the resulting data, and the null hypothesis of normality for each dataset was retained. The means were compared using one-sided, paired t-test analysis at the 95% confidence level. The same method was applied for comparing the coefficients of variation between treatments on each trial within years. In this case, the Box−Cox transformation was applied to both years in order to meet the assumptions of normality and constant variance for t-test analysis.