The industrial production and use of nitrogenous fertilizer involves significant environmental and economic costs. Strategies to reduce fertilizer dependency are required to address the world’s increasing demand for sustainable food, fibres and biofuels. Biological nitrogen fixation, a process unique to diazatrophic bacteria, is catalysed by the nitrogenase complex, and reconstituting this function in plant cells is an ambitious biotechnological strategy to reduce fertilizer use. Here we establish that the full array (n=16) of biosynthetic and catalytic nitrogenase (Nif) proteins from the diazotroph Klebsiella pneumoniae can be individually expressed as mitochondrial targeting peptide (MTP)-Nif fusions in Nicotiana benthamiana. We show that these are correctly targeted to the plant mitochondrial matrix, a subcellular location with biochemical and genetic characteristics potentially supportive of nitrogenase function. Although Nif proteins B,D,E,F,H,J,K,M,N,Q,S,U,V,X,Y and Z were all detectable by Western blot analysis, the NifD catalytic component was the least abundant. To address this problem, a translational fusion between NifD and NifK was designed based on the crystal structure of the nitrogenase MoFe protein heterodimer. This fusion protein enabled equimolar NifD:NifK stoichiometry and improved NifD expression levels in plants. Finally, four MTP-Nif fusion proteins (B,S,H,Y) were successfully expressed, demonstrating that multiple components of nitrogenase can targeted to plant mitochondria. These results establish the feasibility of reconstituting the complete componentry for nitrogenase in plant cells, within an intracellular environment that could support the conversion of nitrogen gas into ammonia.

To engineer Mo-dependent nitrogenase function in plants, expression of the structural proteins NifD and NifK will be an absolute requirement. Although mitochondria have been established as a suitable eukaryotic environment for biosynthesis of oxygen-sensitive enzymes such as NifH, expression of NifD in this organelle has proven difficult due to cryptic NifD degradation. Here, we describe a solution to this problem. Using molecular and proteomic methods, we found NifD degradation to be a consequence of mitochondrial endoprotease activity at a specific motif within NifD. Focusing on this functionally sensitive region, we designed NifD variants comprising between one and three amino acid substitutions and distinguished several that were resistant to degradation when expressed in both plant and yeast mitochondria. Nitrogenase activity assays of these resistant variants in Escherichia coli identified a subset that retained function, including a single amino acid variant (Y100Q). We found that other naturally occurring NifD proteins containing alternate amino acids at the Y100 position were also less susceptible to degradation. The Y100Q variant also enabled expression of a NifD(Y100Q)–linker–NifK translational polyprotein in plant mitochondria, confirmed by identification of the polyprotein in the soluble fraction of plant extracts. The NifD(Y100Q)–linker–NifK retained function in bacterial nitrogenase assays, demonstrating that this polyprotein permits expression of NifD and NifK in a defined stoichiometry supportive of activity. Our results exemplify how protein design can overcome impediments encountered when expressing synthetic proteins in novel environments. Specifically, these findings outline our progress toward the assembly of the catalytic unit of nitrogenase within mitochondria. Significance Engineering nitrogenase in plants may help alleviate economic and environmental issues due to the use of nitrogen fertilizer. Mitochondria have shown promise in supporting the function of nitrogenase, including electron donation and metallocluster assembly. Despite these successes, formation of the catalytic unit, NifDK, has proven difficult. Here, we find that when relocated to plant mitochondria, NifD is subject to errant peptidase-based cleavage and is insoluble. Guided by NifD sequence variation amongst bacteria and structural modeling, we designed NifD variants that avoided cleavage and retained function in bacterial assays. Fusion of NifK to degradation-resistant NifD also improved solubility, and the polyprotein retained function in bacterial assays. This work advances efforts to produce crops less reliant on nitrogen fertilizer.

A long-held goal of synthetic biology has been the transfer of a bacterial nitrogen-fixation pathway into plants to reduce the use of chemical fertiliser on crops such as rice, wheat and maize. There are three classes of bacterial nitrogenase, named after their metal requirements, containing either a MoFe-, VFe- or FeFe-cofactor, that converts N₂ gas to ammonia. Relative to the Mo-nitrogenase the Fe-nitrogenase is not as efficient for catalysis but has less complex genetic and metallocluster requirements, features that may be preferable for engineering into crops. Here we report the successful targeting of bacterial Fe-nitrogenase proteins, AnfD, AnfK, AnfG and AnfH, to plant mitochondria. When expressed as a single protein AnfD was mostly insoluble in plant mitochondria, but coexpression of AnfD with AnfK improved its solubility. Using affinity-based purification of mitochondrially expressed AnfK or AnfG we were able to demonstrate a strong interaction of AnfD with AnfK and a weaker interaction of AnfG with AnfDK. This work establishes that the structural components of the Fe-nitrogenase can be engineered into plant mitochondria and form a complex, which will be a requirement for function. This report outlines the first use of Fe-nitrogenase proteins within a plant as a preliminary step towards engineering an alternative nitrogenase into crops.

Engineering nitrogenase directly into crops is a long‐held aspiration in plant biotechnology. Of the three types, the Fe‐only nitrogenase is a promising engineering target, as it has a simpler maturation pathway than the MoFe‐ and VFe‐nitrogenases and does not require any heterometals for its cofactor. However, previously, we have reported that the obligate electron donor of the Fe‐only nitrogenase from A. vinelandii , AnfH, is mostly insoluble when expressed in plant mitochondria. Here, we employed computational methods, the Protein Repair One‐Stop Shop (PROSS) protein optimisation algorithm and Rosetta energy calculations, to design eight variants of AnfH with improved soluble expression. The amino acid substitutions chosen were predicted to lower the free energy of the native state of the protein. All eight AnfH variants, containing between 1 and 11 amino acid substitutions, were more soluble than wild‐type AnfH in plant mitochondria. Of these, three variants were isolated from N. benthamiana leaf, of which AnfH variant 6 (AnfH V6, T200A T228V E241H) had the best features, being approx. 90‐fold more abundant in the soluble fraction. Importantly, AnfH V6 was also functional in ¹⁵ N ₂ reduction with A. vinelandii AnfDKG after [Fe ₄ S ₄ ] cluster reconstitution. These results show that the computational design strategy used here is a powerful approach for engineering nitrogenase into plants and more broadly to plant synthetic biology and recombinant protein production.

Industrial nitrogen fertilizer is intrinsic to modern agriculture yet expensive and environmentally harmful. We aim to reconstitute bacterial nitrogenase function within plant mitochondria to reduce nitrogen fertilizer usage. Many nitrogen fixation (Nif) proteins are required for biosynthesis and function of the mature nitrogenase enzyme, and these will need to be correctly processed and soluble within mitochondria as a pre-requisite for function. Here we present our workflow that assessed processing, solubility and relative abundance of 16Klebsiella oxytocaNif proteins targeted to the plant mitochondrial matrix using an Arabidopsis mitochondrial targeting peptide (MTP). The functional consequence of the N-terminal modifications required for mitochondrial targeting of Nif proteins was tested using bacterial nitrogenase assays. We found that despite the use of the same constitutive promoter and MTP, MTP::Nif processing and relative abundance in plant leaf varied considerably. Assessment of solubility for all MTP::Nif proteins found NifF, M, N, S, U, W, X, Y and Z were soluble, while NifB, E, H, J, K, Q and V were mostly insoluble. Although most Nif proteins tolerated the N-terminal extension as a consequence of mitochondrial processing, this extension in NifM reduced nitrogenase activity to 10% of controls. Using proteomics, we detected a ∼50-fold increase in the abundance of NifM when it contained the N-terminal MTP extension, which may account for this reduction seen in nitrogenase activity. Based on plant mitochondrial processing and solubility, and retention of function in a bacterial assay, our workflow has identified that NifF, N, S, U, W, Y and Z satisfied all these criteria. Future work can now focus on improving these parameters for the remaining Nif components to assemble a complete set of plant-ready Nif proteins for reconstituting nitrogen fixation in plant mitochondria.