sign in sign up
<<
Home , Ferredoxin, Nitrogenase, Nitrogenase (flavodoxin), Photosystem I

Modular electron-transport chains from eukaryotic function to support activity

Jianguo Yanga,1, Xiaqing Xiea,1, Mingxuan Yanga, Ray Dixonb,2, and Yi-Ping Wanga,2

aState Key Laboratory of and Gene Research, College of Life Sciences, Peking University, Beijing 100871, China; and bDepartment of Molecular Microbiology, John Innes Centre, Norwich, NR4 7UH, United Kingdom

Edited by Robert Haselkorn, University of Chicago, Chicago, IL, and approved January 12, 2017 (received for review December 7, 2016) A large number of genes are necessary for the reduced by a variety of systems, depending on the and activity of the nitrogenase to carry out the process physiology of the host (13–17). of biological fixation (BNF), which requires large amounts A number of studies have suggested , root plastids, or of ATP and reducing power. The multiplicity of the genes involved, mitochondria as suitable locations for expression of nitrogenase in the sensitivity of nitrogenase, plus the demand for (18–20). These energy-conversion organelles can po- and reducing power, are thought to be major obstacles to tentially provide reducing power and ATP required for the nitro- engineering BNF into cereal crops. Genes required for nitrogen gen-fixation process. Diverse reduction reactions carried out in fixation can be considered as three functional modules encoding these organelles rely on different electron-transport chains (21). electron-transport components (ETCs), required for Multiple gene copies of have been identified in all cluster biosynthesis, and the “core” nitrogenase apoenzyme, re- , including photosynthetic or nonphotosynthetic ferredoxins spectively. Among these modules, the ETC is important for the mainly expressed in chloroplasts or root plastids, respectively; and supply of reducing power. In this work, we have used Escherichia -like adrenodoxins located in the mitochondria (21, 22). coli as a chassis to study the compatibility between The major function of the photosynthetic ferredoxins is to transfer and the -only with ETC modules from target electrons from I to NADPH, catalyzed by leaf-type plant organelles, including chloroplasts, root plastids, and mito- ferredoxin–NADPH oxidoreductase (LFNR) (23). In addition, chondria. We have replaced an ETC module present in diazotrophic photosynthetic ferredoxins work to distribute reducing power de- – with genes encoding ferredoxin NADPH rived from the photosynthetic process to several ferredoxin- (FNRs) and their cognate ferredoxin counterparts from plant or- dependent for nitrogen and assimilation (24). – ganelles. We observe that the FNR ferredoxin module from chlo- Electron transfer between root-type ferredoxin–NADPH oxidore- roplasts and root plastids can support the activities of both types ductase (RFNR) and ferredoxin in the root plastid is reversed, with of nitrogenase. In contrast, an analogous ETC module from mito- NADPH generated in the oxidative pentose- pathway chondria could not function in electron transfer to nitrogenase. being used to reduce RFNR and, in turn, ferredoxin (25). In mi- However, this incompatibility could be overcome with hybrid tochondria, adrenodoxin serves to transfer electrons from NADPH- modules comprising mitochondrial NADPH-dependent adreno- dependent adrenodoxin oxidoreductase (MFDR) to the doxin oxidoreductase and the ferredoxins FdxH or desulfurase Nfs1 to participate in the biosynthesis of the biotin (26). FdxB. We pinpoint endogenous ETCs from plant organelles as Recently, we successfully reassembled the Klebsiella oxytoca power supplies to support nitrogenase for future engineering of (Ko) MoFe (27) and the “minimal” vinelandii (Av) diazotrophy in cereal crops. FeFe (28) nitrogenase systems in Escherichia coli (Fig. 1). From the synthetic point of view, these two nitrogenase systems | electron transport | plant organelles | nitrogenase can be divided into three functional modules: the electron- engineering transport component (ETC) module, the metal cluster biosynthesis

itrogen is one of the primary nutrients limiting plant pro- Significance Nductivity in agriculture (1). Industrial nitrogen fertilizers are used to circumvent this limitation, but have resulted in environ- Engineering nitrogenase into cereal crops requires detailed under- mental pollution and expensive economic costs, especially in de- standing of the components required for efficient nitrogen fixation. veloping countries (2, 3). These factors have potentiated a renewed We have used a synthetic biology modular approach to evaluate focus toward engineering biological nitrogen fixation (BNF) in ce- components from , root plastids, and mitochondria that real crops. BNF, the process that converts gaseous nitrogen to function as electron donors to both conventional Mo nitrogenase by nitrogenase enzymes, contributes >60% of the total and the alternative Fe nitrogenase systems. The knowledge atmospheric N2 fixed in the biogeochemical (4). obtained in this study not only identifies electron-transfer compo- Nitrogenases are a family of metalloenzymes that consist of two nents from plant organelles that can be used to support nitroge- separable components, dinitrogenase (Fe protein) and nase activity, but also is likely to enable reduction of the number of dinitrogenase (XFe protein, where X is equivalent to Mo, V, or Fe, target genes required to engineer nitrogen fixation in plants. depending on the heterometal composition of the co- Author contributions: J.Y., R.D., and Y.-P.W. designed research; J.Y., X.X., and M.Y. performed factor) (Fig. 1 and refs. 5 and 6). All three nitrogenases catalyze the research; J.Y., R.D., and Y.-P.W. analyzed data; and J.Y., X.X., R.D., and Y.-P.W. wrote the paper. biological reduction of N2 according to the following equation: N2 + + − The authors declare no conflict of interest. (6 + 2n)H + (6 + 2n)e → 2NH + nH (n ≥ 1) (7–9). In this 3 2 This article is a PNAS Direct Submission. process, electrons are first transferred to the Fe protein, which, in Freely available online through the PNAS open access option. turn, donates electrons to the XFe protein with of two See Commentary on page 3009. ATP per electron (Fig. 1) (10–12). Although Fe protein 1J.Y. and X.X. contributed equally to this work. is the obligate electron donor for XFe protein in all characterized 2To whom correspondence may be addressed. Email: [email protected] or wangyp@pku. nitrogenase systems, the in vivo electron donor for Fe protein is less edu.cn. stringently conserved (9). Direct electron donors to Fe protein are This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10. either reduced flavodoxin or reduced ferredoxin, which, in turn, are 1073/pnas.1620058114/-/DCSupplemental.

E2460–E2465 | PNAS | Published online February 13, 2017 www.pnas.org/cgi/doi/10.1073/pnas.1620058114 Downloaded by guest on September 29, 2021 (Zm; ZmFDI and ZmFDIII), Oryza sativa (Os; OsFD1 and PNAS PLUS OsFD4), and Triticum aestivum (Ta; TaFD) were selected for fur- ther study. These ferredoxin-encoding genes were codon-optimized for E. coli (Dataset S1), and expressed from the inducible PLteto-1 promoter (Fig. 2A;detailsareprovidedinSI Materials and Methods). The fdxH gene from As was also introduced as a control to verify effectiveness of the inducible system. To assay electron transport by plant ferredoxins, the fla- vodoxin encoded by nifF in the Ko NifJ–NifF module was replaced by coexpression of the respective plant gene and the resultant activity of the reassembled MoFe (27) or the minimal SEE COMMENTARY FeFe (28) nitrogenase was analyzed by the method of reduction. Although significant background activity was observed in the absence of the NifJ–NifF module, all hybrid ETC modules stimulated nitrogenase activity by both the MoFe and FeFe systems, to varying extents (Fig. 2 B–E). Stimulation of activity by plastid ferredoxins was dependent on the presence of NifJ, in- dicating that reducing power is provided by the pyruvate oxido- reductase activity of this electron donor (Fig. S2). Interestingly, values >100% were observed for the FeFe nitrogenase system when NifF was replaced with the ferredoxins from As (FdxH), Cr (PETF), or Os (FD1), respectively (Fig. 2B and Table S1). This phenomenon suggests that the AvAnfH protein in the hybrid Fig. 1. (Upper) Modular arrangement of genes required for MoFe and the minimal FeFe nitrogenase (28) may prefer ferredoxin, rather than minimal FeFe nitrogenase systems. Letters within the arrows represent the flavodoxin, as an electron donor. All of the chloroplast ferredoxins corresponding nif genes or the anfHDGK structural genes encoding FeFe ni- could restore ∼100% activity for the FeFe nitrogenase system, trogenase. (Lower) Schematic pathways for electron donation to nitrogenase with the exception of the NifJ–AtFD2 hybrid module, which and electron transfer within nitrogenase. Structures of representative proteins ∼ B – are shown. PFOR (NifJ), pyruvate–ferredoxin (flavodoxin) oxidoreductase showed 76% activity (Fig. 2 ). In contrast, only the NifJ [ (PDB) ID code 1B0P]; Rnf complex, NADH–ferredoxin oxi- CrPETF and NifJ–TaFD hybrid modules could restore >90% doreductase [the structure shown in gray is a homology model based on the activity to the MoFe nitrogenase system, whereas the NifJ–AtFD2, + Nqr complex (Na -translocating NADH–quinone oxidoreductase from Vibrio NifJ–ZmFDI, and NifJ–OsFD1 hybrid modules exhibited <70% alginolyticus; PDB ID code 4P6V) using the online software from https:// activity (Fig. 2C). All root-plastid ferredoxin-derived hybrid swissmodel..org]; FNR (PDB ID code 1QUE); NifF, flavodoxin (PDB ID modules showed lower nitrogenase activities compared with their code 2WC1); FdxN, 2[4Fe–4S]-type ferredoxin (PDB ID code 2OKF); FdxH, – chloroplast ferredoxin counterparts derived from the same or- [2Fe 2S]-type ferredoxin (PDB ID code 1FRD); Fe protein, dinitrogenase re- ganism (Fig. 2 B–E). It is possible that the different activities ductase (PDB ID code 1G5P); and XFe protein (where X refers to Mo, V, or Fe), dinitrogenase (PDB ID code 3K1A; MoFe nitrogenase). The cofactors of the observed on substitution of ferredoxins from various plant origins Fe and XFe proteins are shown as ball-and-stick models. Atom colors are Fe in could result from variations in expression levels. However, we rust, S in yellow, C in gray, O in red, and heterometal X in purple. were unable to confirm this possibility because we could not detect ferredoxin proteins by Western blotting using an anti-His mono- clonal antibody (details are provided in SI Materials and Methods). module, and the “core” enzyme module (Fig. 1). In the present Together, these results indicated that hybrid ETC modules formed study, ETC modules from plastids and mitochondria, representing with the NifJ protein, and plastid ferredoxins can direct the potential locations for nitrogenase in plants, were used to test their transfer of electrons to nitrogenase. capability to support the activity of either MoFe or FeFe nitroge- Because mitochondria represent another potential location for nase in E. coli. Our results indicate that intact ETC modules from nitrogenase in plants, the capability of mitochondrial ferredoxins to the chloroplast and root plastid, or hybrid modules from mito- support nitrogen fixation was also investigated in E. coli.Thesame chondria, can functionally support nitrogenase activity. Therefore, strategy was used to clone the mitochondrial adrenodoxin coding our results unravel the requirements for electron-transport com- genes from At (AtMFD1 and AtMFD2) as described above. When – ponents for engineering diazotrophy in different plant organelles. mitochondrial ferredoxin-derived hybrid ETC modules (NifJ AtMFD1 or NifJ–AtMFD2) were introduced into E. coli,nores- Results toration of activity was observed compared with the nifF-deficient F G Hybrid ETC Modules Consisting of the NifJ Protein and Plastid Ferredoxins MoFe or FeFe nitrogenase systems (Fig. 2 and ). To exclude Can Functionally Support Nitrogenase Activity. In many , the possibility that the GroESL proteins are potentially required for proper and efficient folding of mitochondrial ferredoxins in the direct electron donor to nitrogenase is either reduced ferredoxin E. coli or reduced flavodoxin, which have been demonstrated to receive (26), a high-copy plasmid carrying the GroESL-encoding electrons from the pyruvate–flavodoxin (ferredoxin) oxidoreduc- genes was cotransformed with MoFeorFeFenitrogenasesystems carrying either the NifJ–AtMFD1 or NifJ–AtMFD2 hybrid mod- tase, encoded by nifJ,insomecases(13–15). Most plants are known ules. Similar negative results were obtained (Fig. S3). Interestingly, to have multiple copies of ferredoxins located in different organelles phylogenetic analysis showed that the mitochondrial ferredoxins (21). Through preliminary sequence analysis, we found that chlo- are not clustered with any of the well-defined electron donors for roplast and root-plastid ferredoxins from plants show high sequence Anabaena fdxH nitrogenase (Fig. S4). Overall, these results suggest that the mito- identity with the sp. PCC 7120 (As) gene product chondria ferredoxins cannot couple with the NifJ protein to form (Fig. S1), which is the primary electron donor for nitrogenase in functional ETC modules for nitrogenase systems. (29). To investigate whether hybrid ETC modules formed by the NifJ protein and plastid ferredoxins could support Influence of Hybrid ETC Modules Consisting of Plant-Type Ferredoxin– nitrogenase activity in E. coli, coding sequences of several repre- NADPH Oxidoreductases and Well-Defined Electron Donors on

sentative plastid ferredoxins from Chlamydomonas reinhardtii (Cr; Nitrogenase Activity. In plants, three different types of the ferre- MICROBIOLOGY CrPETF), (At; AtFD2 and AtFD3), Zea mays doxin–NADPH oxidoreductases (FNRs) are identified and exist

Yang et al. PNAS | Published online February 13, 2017 | E2461 Downloaded by guest on September 29, 2021 Fig. 2. Influence of hybrid ETC modules consisting of the NifJ protein with ferredoxins (FDs) from different plant organelles on nitrogenase activity in E. coli. (A) Schematic diagram for electron transport between hybrid ETC modules and nitrogenases. (B–G) In each case, nifF was replaced by FDs to form hybrid modules consisting of NifJ with chloroplast FDs (B and C), NifJ with root-plastid FDs (D and E), or NifJ with mitochondrial FDs (F and G). In all cases, cultures were assayed for acetylene reduction either in the absence (filled bars) or presence (open bars) of the appropriate inducer required to express heterologous ETCs as described in SI Materials and Methods and Fig. S9. The activity of FeFe or MoFe nitrogenases when expressed in the presence of the NifJ–NifF module from native nif promoters represents 100% activity in each case (in the absence of added inducer). FeFe represents the minimal FeFe nitrogenase system, and MoFe represents the reassembled MoFe nitrogenase system. Assembly of these nitrogenases requires both the “metal cluster biosynthesis module” and the “core enzyme” module, respectively, as shown in Fig. 1. As, Anabaena sp. PCC 7120; At, Arabidopsis thaliana;Cr,Chlamydomonas reinhardtii;Os,Oryza sativa; Ta, T. aestivum; Zm, Zea mays. Error bars indicate the SD observed from at least three independent experiments.

in different organelles. All of these FNRs function to mediate plastid FNRs (Fig. S5). In addition, no flavodoxins have been electron transfer between ferredoxins and NADPH (23, 25, 26). identified in plants (31). Therefore, the plant-type FNRs do not From phylogenetic analysis, we found that the chloroplast FNRs appear to have coevolved with the flavodoxins. form a subgroup with the FNR proteins from cyanobacteria, whereas the root-plastid FNRs form another subgroup with the Intact ETC Modules from Chloroplasts and Root Plastids Can Functionally FNR proteins from (Fig. S5). In contrast, the mitochon- Support Nitrogenase Activity. After evaluating the function of the drial MFDRs diverged away from these two subgroups during hybrid modules, further experiments were carried out to investigate early evolution (Fig. S5). whether intact ETC modules, consisting of FNRs and their cognate To investigate whether hybrid ETC modules consisting of the ferredoxin counterparts from plant organelles, could support ni- plant-type FNRs and well-defined electron donors (KoNifF, trogenase activity. By combining the Ptac-controlled FNRs with AsFdxH, and AsFdxB) could direct electron transfer to nitroge- PLtetO-1-controlled ferredoxins (details are provided in SI Materials E. coli nase in , well-characterized chloroplast or root-plastid FNRs and Methods), two chloroplast ETC modules, CrFNR–PETF and from Cr and Zm, plus the At mitochondrial MFDR, were selected ZmLFNR–FDI; one root-plastid ETC module, ZmRFNR–FDIII; to verify this assumption. These hybrid modules were transformed – E. coli and one mitochondrial ETC module, AtMFDR MFD1 were con- into the , and their activities were assayed by the acetylene structed. Because it is known that the AsPetH–FdxH module from reduction method. None of the hybrid ETC modules consisting of cyanobacteria can function to support nitrogen fixation in its orig- the plant-type FNRs and NifF could stimulate acetylene reduction inal host, this module was also constructed and used as a control. by either the MoFe or FeFe nitrogenase systems (Fig. S6). In The ability of the intact ETC modules to support nitrogenase contrast, all hybrid modules formed with the plant-type FNRs and activity as replacement for the NifJ–NifF module was assayed by AsFdxH could partially restore nitrogenase activity to both the 15 B C both acetylene reduction and N-assimilation methods. With the MoFe and FeFe nitrogenase systems (Fig. 3 and ). However, – only MFDR from mitochondria could support electron transfer to exception of the AtMFDR MFD1 module from mitochondria, all other ETC modules conferred ability to partially restore acetylene the nitrogenases when coupled to AsFdxB (Fig. S6). 15 The fact that ETC modules formed with the plant-type FNRs reduction and N assimilation to both MoFe and FeFe nitroge- and NifF could not functionally support nitrogenase activity is nases (Fig. 4 B–E). In contrast, no obvious stimulation of activity not surprising, because FNRs known to reduce flavodoxins are could be observed, when either of the two components from each limited to a few bacteria, such as Fpr from E. coli and FNRs from of the modules was expressed individually, demonstrating that the some species of cyanobacteria, which are usually related to the plant-type ferredoxins do not function in electron transport in the biosynthesis of P450 (30). Microbial FNRs showed a absence of their cognate FNR and, conversely, that the plant distant evolutionary relationship with the chloroplast and root- FNRs cannot donate electrons to nitrogenase in the absence of a

E2462 | www.pnas.org/cgi/doi/10.1073/pnas.1620058114 Yang et al. Downloaded by guest on September 29, 2021 system (33). These findings have highlighted the difficulties in PNAS PLUS minimizing nif gene sets for synthetic biology. One approach to reducing the number of genes required for nitrogenase activity is to make use of host-encoded ETCs. Bio- informatic analysis indicated that ETC modules from plant or- ganelles show close evolutionary relationships with nitrogenase ETCs from nitrogen-fixing cyanobacteria (Figs. S4 and S5). Therefore, it is reasonable to question whether these ETCs can replace the NifJ–NifF module and support nitrogen fixation. We have used the modularity concept to analyze compatibility be- SEE COMMENTARY tween ETCs from different plant organelles with nitrogenase en- zyme modules. Our results indicate that plastid ferredoxins can efficiently participate in electron transfer to nitrogenase from ei- ther the nif-specific pyruvate–flavodoxin (ferredoxin) reductase encoded by nifJ or from its FNR counterparts in plants. Overall, we observed little difference in the response of the MoFe or FeFe nitrogenases to the various ETC modules we analyzed, suggesting that plant ETCs can function with either enzyme. We found that almost all plant-originated ferredoxins used Fig. 3. Influence of hybrid ETC modules consisting of the cyanobacterial in this study (except adrenodoxins from mitochondria) could ferredoxin AsFdxH combined with FNRs from different plant organelles on nitrogenase activity in E. coli.(A) Schematic diagram for electron transport between hybrid ETC modules and nitrogenases. (B and C)TheNifJ–NifF ETC module was replaced by hybrid modules consisting of plant-type FNRs and AsFdxH. Experimental details and abbreviations are the same as in Fig. 2. Error bars indicate the SD observed from at least three independent experiments.

coexpressed ferredoxin component (Fig. S7). Thus, both members of the plant ETC pairs are necessary for functionality. Two of the chloroplast modules, CrFNR–PETF and ZmLFNR–FDI, showed comparable amounts of activity to those observed with the AsPetH–FdxH module from cyanobacteria (∼45% acetylene reduction and ∼30% 15N assimilation) with both FeFe and MoFe nitrogenases (Fig. 4). However, the ZmRFNR–FDIII module from the root plastid was less active than its corresponding chloroplast module from the same plant with respect to both nitrogenases (Fig. 4 B–E). Interestingly, weak complementation was observed with the AtMFDR–MFD1 module (11% 15N assimilation activity) when combined with MoFe nitrogenase, compared with the NifJ–NifF-deficient neg- ative control (6% 15N assimilation activity) (Fig. 4E). However, this phenotype was not observed with FeFe nitrogenase. (Fig. 4D). Because multiple copies of ferredoxins exist in E. coli,itis possible that the enhanced activity of MoFe nitrogenase resulted from the contribution of hybrid modules formed between AtMFDR and those from E. coli ferredoxins. This effect may not be observed with the FeFe nitrogenase system because of its higher background activity in the absence of nifF and nifJ (Fig. 4D). Together, these results demonstrate that intact ETC mod- ules from plastids, but not their equivalents from mitochondria, are capable of transferring electrons to nitrogenases. Discussion It is generally accepted that several factors may limit successful engineering of nitrogenase in cereal crops, including provision of the appropriate physiological environment for the enzyme [e.g., availability of energy (ATP), reducing power, and a low-oxygen environment], in addition to the relative large number of nif genes required for biosynthesis of nitrogenase itself. Attempts Fig. 4. Influence of intact ETC modules consisting of FNRs from different plant have been made to reduce the number of genes required for both organelles with their cognate ferredoxins (FDs) on nitrogenase activity in E. coli. MoFe (32) and FeFe nitrogenase biosynthesis (28) by using (A) Schematic representation for electron transport between intact plant ETC E. coli nif modules and nitrogenases. (B–E) Nitrogen fixation by FeFe or MoFe nitroge- as a chassis. However, in the case of genes originating 15 Paenibacillus nases was assayed either by acetylene reduction (B and C; black bars) or N from sp., it was observed that nitrogenase activity assimilation (D and E; striped bars). Experimental details and abbreviations are decreased drastically when the number of genes were decreased the same as in Fig. 2. Error bars for the acetylene reduction assay indicate the SD 15 to nine for MoFe nitrogenase (32), but activity could be re- observed from at least three independent experiments. Error bars for Nas- MICROBIOLOGY covered when additional genes were recruited back into the similation indicate the SD observed from at least two independent experiments.

Yang et al. PNAS | Published online February 13, 2017 | E2463 Downloaded by guest on September 29, 2021 functionally substitute for NifF for both FeFe and MoFe nitro- redoxins. We propose that this electron-transfer pathway can be genases (Fig. 2). This finding implies that the interface between used to provide reducing power for nitrogenase in the absence of these ferredoxins and NifH/AnfH are competent for electron the nif-specific NifJ–NifF ETC module (Fig. 5). transfer. In many cases, the previously reported potentials of Although chloroplasts are potentially problematic for nitro- these ferredoxins (Table S1) are higher than those of the Fe genase engineering in terms of oxygen evolution, it is possible to protein NifH (−412 mV) (34), which disfavors electron transfer to consider solutions based on mechanisms of oxygen avoidance nitrogenase in vitro (35). Furthermore, the redox potential of the + used by cyanobacteria. The electron-transfer route to LFNR in NADH/NAD couple (midpoint potential −380 mV) is, in theory, chloroplasts involves activation of II and I too high to drive electron donation to nitrogenase by FNRs, al- and the subsequent reduction of ferredoxin, which feeds elec- though NAD(P)H-dependent electron-transfer chains apparently trons into LFNR. The reaction catalyzed by LFNR is reversible; support nitrogenase activity in vivo in aerobic diazotrophs. To whereas reduction by ferredoxin favors biosynthesis of NADPH account for this conundrum, it has been proposed that either (23), the reverse reaction uses NADPH to reduce oxidized fer- motive force or electron bifurcation drives the reversed redoxin (40). In this case, LFNR can be considered to be anal- electron flow required to overcome these bioenergetic constraints ogous to a battery, which accumulates reducing equivalents in in vivo (36, 37). Because multiple factors, including expression the light and discharges this power (as NADPH) in the dark to levels, interface compatibility, and protein ratios, may affect fuel redox reactions (Fig. 5). Because the chloroplast ETC electron transfer to nitrogenase, it is not possible to determine modules studied in this work function with and whether ferredoxin redox potentials relate to the efficiency of can support nitrogenase activity in E. coli (Fig. 4), it is possible to electron transfer in our experimental system. Overall, our results lay a solid foundation for the use of extant envisage a scenario for energetic coupling of with ETC modules from plant organelles to engineer BNF in cereal nitrogen fixation, with temporal separation of their activities in crops. The observation that ETC modules from both chloroplasts and root plastids can functionally support nitrogenase activity implies that engineering diazotrophy in plastids does not require insertion of an ETC module. For instance, for the FeFe nitroge- nase system, of the minimal 10 genes required (28), 2 of these function as the bacterial ETC module and presumably would not be essential for engineering nitrogenase activity in plants. To further reduce the number of nif/anf genes required, plant genes that can substitute for components of the “metal cluster bio- synthesis” module (Fig. 1), such as those required for iron–sulfur cluster assembly, could also be considered. Two recent studies have shown that the [Fe–S] cluster biosynthesis components from eukaryotic organelles can substitute for NifU and NifS to incorpo- rate the [4Fe–4S] cluster into the Fe protein of nitrogenase (19, 20). Accumulated mutational studies indicate that many diazotrophs have multiple pathways that mediate electron flow to nitrogenase. These pathways include multiple electron-transfer proteins (ferre- doxins and flavodoxins) and multiple oxidoreductases, including py- ruvate–flavodoxin (ferredoxin) oxidoreductase, FNR, and the Rnf complex [a membrane-bound ion-translocating NADH–ferredoxin oxidoreductase, first identified in Rhodobacter capsulatus by Schmehl et al. (16)] (Fig. 1). The putative membrane-associated electron- Fig. 5. Schematic model illustrating potential routes for electron transfer to transferring flavoprotein complex encoded by the fixABCX operon nitrogenase in engineered plant organelles. The diagram depicts an artificial has also been proposed to transfer electrons to nitrogenase (38). Such plant cell in which a chloroplast and root plastid coexists in the same cell. The main components or processes for generation of reducing power are shown. redundancy has endowed nitrogenase with the ability to cope with – Components within organelles with solid outlines represent existing proteins heterogenous ETC modules. Conversely, the [2Fe 2S]-type ferre- present in plants, whereas components with dashed outlines represent those doxin is the most extensively usedelectronshuttle,whichismain- required to engineer nitrogenase activity as suggested by our results. The tained throughout the tree of life and is involved in a plethora of red arrow in the root plastid represents RFNR–ferredoxin-mediated electron metabolic, regulatory, dissipative, and developmental processes (31). transfer from NADPH to nitrogenase, with reducing equivalents being sup- To satisfy the requirements for electron transfer to diverse proteins, plied by the oxidative pentose phosphate pathway (OxPPP). In mitochondria, ferredoxins have evolved multiple interfaces to cope with different the red arrow indicates the election transfer pathway from MFDR to nitro- metabolic partners (39). This interface flexibility may provide another genase, which can function if a heterologous ferredoxin such as AsFdxH or explanation for the compatibility of the nitrogenase system with het- AsFdxB is introduced. NADPH can be supplied either through glycolysis or via the oxidative TCA cycle by isocitrate (ICDH). In chloroplasts, erogenous ETC modules from plant chloroplasts and root plastids. light-activated photosystem II (PSII) extracts electrons from water and Combining the results obtained from this study, a schematic transfers them to plastoquinone (PQ), and then through cytochrome b6f

model for electron transfer to nitrogenase in organelles of engi- (Cytb6f) to (PC), which then feeds electrons to the light-oxi- neered plants is proposed (Fig. 5). Our results imply that the en- dized photosystem I (PSI). PSI-derived electrons are used to reduce ferre- dogenous ETC module present in mitochondria is not competent to doxin, which then transfers reductant to either LFNR, for NADPH production, support nitrogenase activity (Fig. 4), reflecting the requirement for or to nitrogenase, for nitrogen fixation. The NADPH generated can also an additional ETC component, such as AsFdxH or AsFdxB, that promote reverse electron transfer to nitrogenase via ferredoxin, catalyzed can function as a hybrid module for direct electron transfer to ni- by LFNR. The bottom lighter half of the chloroplast represents the light “ ” trogenase (Fig. 3). In this scenario, NADPH generated either by condition, with the blue arrow representing the photo-coupled charging process for accumulating NADPH; the darker half of the chloroplast repre- glycolysis or by isocitrate dehydrogenase can provide reducing sents the dark condition, with the yellow arrow representing the LFNR– equivalents for electron transfer to nitrogenase through the hybrid “ ” – ferredoxin mediated discharging process for transferring electrons from MFDR ferredoxin pathway (Fig. 5). In root plastids, the NADPH NADPH to nitrogenase (N2ase). As the major site for ammonia assimilation, generated by degradation of glucose through the oxidative pentose ammonia produced in the chloroplast can be immediately assimilated by phosphate pathway is used by RFNR to catalyze reduction of fer- synthetase (GS) (42). Glc-6P, glucose-6-P; Rib-5P, ribose-5-P.

E2464 | www.pnas.org/cgi/doi/10.1073/pnas.1620058114 Yang et al. Downloaded by guest on September 29, 2021 the light and dark periods. In nonheterocystous cyanobacteria, scribed in SI Materials and Methods. Each of the constructs was confirmed by PNAS PLUS nature has been able to reconcile the processes of nitrogen fix- DNA sequencing before any further experimentation. ation and oxygenic photosynthesis through temporal separation, Acetylene Reduction Assay. The C2H2 reduction method was used to assay in which photosynthetic CO2 fixation occurs in the light and nitrogen fixation is carried out in the dark (41). According to our nitrogenase activity as described (28). Details are provided in SI Materials and Methods. Data presented are mean values based on at least three replicates. model, electron transfer from the ferredoxin to LFNR would occur during the light period, resulting in accumulation of excess 15 15 N2 Assimilation Assay. To detect N2 assimilation, E. coli JM109 derivatives NADPH. This reductant could be used in the dark to drive the expressing reassembled nitrogenase systems were grown as described (28). reverse-catalytic reaction of LFNR, enabling reduction of ni- More details are provided in SI Materials and Methods. Data presented are trogenase by reduced ferredoxin (Fig. 5). mean values based on at least two replicates. SEE COMMENTARY

Materials and Methods ACKNOWLEDGMENTS. This work was supported by National Science Founda- Bacterial Strains and Growth Medium. Bacterial strains used in this study are tion of China (NSFC) Grant 31530081; 973 National Key Basic Research Program listed in Table S2. Media and antibiotics were used as described (28) and are in China Grant 2015CB755700; China Postdoctoral Science Foundation Grant 2015M580014; and State Key Laboratory of Protein and Plant Gene Research detailed in SI Materials and Methods. Grant B02. Y.-P.W. is recipient of NSFC National Science Fund for Distinguished Young Scholars Grant 39925017. R.D. was supported by U.K. Biotechnology and Construction of Recombinant Plasmids. Plasmids used in this study are listed in Biological Sciences Research Council Grant BB/J004553/1. J.Y. was supported in Table S2 and Fig. S8. Plasmids were constructed by using procedures de- part by the Postdoctoral Fellowship of Peking-Tsinghua Center for Life Sciences.

1. Xu G, Fan X, Miller AJ (2012) Plant nitrogen assimilation and use efficiency. Annu Rev 27. Wang X, et al. (2013) Using synthetic biology to distinguish and overcome regulatory Plant Biol 63(1):153–182. and functional barriers related to nitrogen fixation. PLoS One 8(7):e68677. 2. Galloway JN, et al. (2008) Transformation of the nitrogen cycle: Recent trends, 28. Yang J, Xie X, Wang X, Dixon R, Wang Y-P (2014) Reconstruction and minimal gene questions, and potential solutions. Science 320(5878):889–892. requirements for the alternative iron-only nitrogenase in Escherichia coli. Proc Natl 3. Tilman D, Balzer C, Hill J, Befort BL (2011) Global food demand and the sustainable Acad Sci USA 111(35):E3718–E3725. intensification of agriculture. Proc Natl Acad Sci USA 108(50):20260–20264. 29. Masepohl B, Schölisch K, Görlitz K, Kutzki C, Böhme H (1997) The -specific 4. Canfield DE, Glazer AN, Falkowski PG (2010) The evolution and future of Earth’sni- fdxH gene product of the cyanobacterium Anabaena sp. PCC 7120 is important but trogen cycle. Science 330(6001):192–196. not essential for nitrogen fixation. Mol Gen Genet 253(6):770–776. 5. Schindelin H, Kisker C, Schlessman JL, Howard JB, Rees DC (1997) Structure of ADP x 30. Sancho J (2006) Flavodoxins: Sequence, folding, binding, function and beyond. Cell AIF4(-)-stabilized nitrogenase complex and its implications for signal transduction. Mol Life Sci 63(7-8):855–864. Nature 387(6631):370–376. 31. Pierella Karlusich JJ, Lodeyro AF, Carrillo N (2014) The long goodbye: The rise and fall 6. Tezcan FA, et al. (2005) Nitrogenase complexes: Multiple docking sites for a nucleo- of flavodoxin during plant evolution. J Exp Bot 65(18):5161–5178. tide switch protein. Science 309(5739):1377–1380. 32. Wang L, et al. (2013) A minimal nitrogen fixation gene cluster from Paenibacillus sp. WLY78 7. Eady RR, Richardson TH, Miller RW, Hawkins M, Lowe DJ (1988) The ni- enables expression of active nitrogenase in Escherichia coli. PLoS Genet 9(10):e1003865. trogenase of . Purification and properties of the Fe protein. 33. Li X-X, Liu Q, Liu X-M, Shi H-W, Chen S-F (2016) Using synthetic biology to increase Biochem J 256(1):189–196. nitrogenase activity. Microb Cell Fact 15:43. 8. Schneider K, Gollan U, Dröttboom M, Selsemeier-Voigt S, Müller A (1997) Compara- 34. Deistung J, Thorneley RN (1986) Electron transfer to nitrogenase. Characterization of tive biochemical characterization of the iron-only nitrogenase and the molybdenum flavodoxin from Azotobacter chroococcum and comparison of its redox potentials nitrogenase from Rhodobacter capsulatus. Eur J Biochem 244(3):789–800. with those of flavodoxins from and Klebsiella pneumoniae 9. Halbleib CM, Ludden PW (2000) Regulation of biological nitrogen fixation. J Nutr (nifF-gene product). Biochem J 239(1):69–75. 130(5):1081–1084. 35. Braaksma A, Haaker H, Grande HJ, Veeger C (1982) The effect of the redox potential 10. Duval S, et al. (2013) Electron transfer precedes ATP hydrolysis during nitrogenase on the activity of the nitrogenase and on the Fe-protein of Azotobacter vinelandii. . Proc Natl Acad Sci USA 110(41):16414–16419. Eur J Biochem 121(3):483–491. 11. Hageman RV, Burris RH (1978) Nitrogenase and nitrogenase reductase associate and 36. Haaker H, Veeger C (1977) Involvement of the cytoplasmic membrane in nitrogen dissociate with each catalytic cycle. Proc Natl Acad Sci USA 75(6):2699–2702. fixation by Azotobacter vinelandii. Eur J Biochem 77(1):1–10. 12. Hu Y, et al. (2006) Nitrogenase Fe protein: A molybdate/homocitrate insertase. Proc 37. Buckel W, Thauer RK (2013) Energy conservation via electron bifurcating ferredoxin + Natl Acad Sci USA 103(46):17125–17130. reduction and proton/Na( ) translocating ferredoxin oxidation. Biochim Biophys Acta 13. Shah VK, Stacey G, Brill WJ (1983) Electron transport to nitrogenase. Purification and 1827(2):94–113. characterization of pyruvate:flavodoxin oxidoreductase. The nifJ gene product. J Biol 38. Edgren T, Nordlund S (2004) The fixABCX genes in encode a Chem 258(19):12064–12068. putative membrane complex participating in electron transfer to nitrogenase. 14. Schmitz O, Gurke J, Bothe H (2001) Molecular evidence for the aerobic expression of J Bacteriol 186(7):2052–2060. nifJ, encoding pyruvate:ferredoxin oxidoreductase, in cyanobacteria. FEMS Microbiol 39. Kameda H, Hirabayashi K, Wada K, Fukuyama K (2011) Mapping of protein-protein Lett 195(1):97–102. interaction sites in the plant-type [2Fe-2S] ferredoxin. PLoS One 6(7):e21947. 15. Yakunin AF, Hallenbeck PC (1998) Purification and characterization of pyruvate oxi- 40. Onda Y, et al. (2000) Differential interaction of maize root ferredoxin:NADP(+) oxi- doreductase from the photosynthetic bacterium Rhodobacter capsulatus. Biochim doreductase with photosynthetic and non-photosynthetic ferredoxin isoproteins. Biophys Acta Bioenergetics 1409(1):39–49. Plant Physiol 123(3):1037–1045. 16. Schmehl M, et al. (1993) Identification of a new class of nitrogen fixation genes in 41. Bothe H, Schmitz O, Yates MG, Newton WE (2010) Nitrogen fixation and Rhodobacter capsulatus: A putative membrane complex involved in electron trans- metabolism in cyanobacteria. Microbiol Mol Biol Rev 74(4):529–551. port to nitrogenase. Mol Gen Genet 241(5-6):602–615. 42. Lam HM, Coschigano KT, Oliveira IC, Melo-Oliveira R, Coruzzi GM (1996) The mo- 17. Neuer G, Bothe H (1985) Electron donation to nitrogenase in of cyano- lecular-genetics of nitrogen assimilation into amino acids in higher plants. Annu Rev bacteria. Arch Microbiol 143(2):185–191. Plant Physiol Plant Mol Biol 47(1):569–593. 18. Cheng Q, Day A, Dowson-Day M, Shen G-F, Dixon R (2005) The Klebsiella pneumoniae 43. Gatti-Lafranconi P, Dijkman WP, Devenish SR, Hollfelder F (2013) A single mutation in nitrogenase Fe protein gene (nifH) functionally substitutes for the chlL gene in the core domain of the lac repressor reduces leakiness. Microb Cell Fact 12:67. Chlamydomonas reinhardtii. Biochem Biophys Res Commun 329(3):966–975. 44. Montoya JP, Voss M, Kahler P, Capone DG (1996) A simple, high-precision, high- 19. Ivleva NB, Groat J, Staub JM, Stephens M (2016) Expression of active subunit of ni- sensitivity tracer assay for N(inf2) fixation. Appl Environ Microbiol 62(3):986–993. trogenase via integration into plant genome. PLoS One 11(8):e0160951. 45. Chen Y-J, et al. (2013) Characterization of 582 natural and synthetic terminators and 20. López-Torrejón G, et al. (2016) Expression of a functional oxygen-labile nitrogenase com- quantification of their design constraints. Nat Methods 10(7):659–664. ponent in the mitochondrial matrix of aerobically grown yeast. Nat Commun 7:11426. 46. Hurley JK, et al. (1997) Structure-function relationships in Anabaena ferredoxin: 21. Hanke G, Mulo P (2013) Plant type ferredoxins and ferredoxin-dependent metabo- Correlations between X-ray crystal structures, reduction potentials, and rate constants + lism. Plant Cell Environ 36(6):1071–1084. of electron transfer to ferredoxin:NADP reductase for site-specific ferredoxin mu- 22. Fukuyama K (2004) Structure and function of plant-type ferredoxins. Photosynth Res tants. Biochemistry 36(37):11100–11117. 81(3):289–301. 47. Terauchi AM, et al. (2009) Pattern of expression and substrate specificity of chloro- 23. Kurisu G, et al. (2001) Structure of the electron transfer complex between ferredoxin plast ferredoxins from Chlamydomonas reinhardtii. J Biol Chem 284(38):25867–25878. and ferredoxin-NADP(+) reductase. Nat Struct Biol 8(2):117–121. 48. Matsumura T, et al. (1999) Complementary DNA cloning and characterization of 24. Knaff DB, Hirasawa M (1991) Ferredoxin-dependent chloroplast enzymes. Biochim ferredoxin localized in bundle-sheath cells of maize leaves. Plant Physiol 119(2): Biophys Acta Bioenergetics 1056(2):93–125. 481–488. 25. Green LS, et al. (1991) Ferredoxin and ferredoxin-NADP reductase from photosyn- 49. Akashi T, et al. (1999) Comparison of the electrostatic binding sites on the surface of + thetic and nonphotosynthetic tissues of tomato. Plant Physiol 96(4):1207–1213. ferredoxin for two ferredoxin-dependent enzymes, ferredoxin-NADP( ) reductase 26. Picciocchi A, Douce R, Alban C (2003) The plant biotin synthase reaction. Identification and . J Biol Chem 274(41):29399–29405.

and characterization of essential mitochondrial accessory protein components. J Biol 50. Hanke GT, Kimata-Ariga Y, Taniguchi I, Hase T (2004) A post genomic characterization MICROBIOLOGY Chem 278(27):24966–24975. of Arabidopsis ferredoxins. Plant Physiol 134(1):255–264.

Yang et al. PNAS | Published online February 13, 2017 | E2465 Downloaded by guest on September 29, 2021