Multiple iso-proteins of FNR in Arabidopsis: evidence for different contributions to chloroplast function and nitrogen assimilation


Guy Hanke Fax: + 81 6 6879 8613; e-mail:


In chloroplasts ferredoxin:NADP(H) oxidoreductase (FNR) enzymes oxidize the final reduced product of the photosynthetic electron transport chain, ferredoxin (Fd), to reduce NADP+, and play a role in cyclic electron transport. Oppositely, in non-photosynthetic plastids FNR oxidizes NADPH to provide reduced Fd for enzymes of bioassimilation and biosynthesis. These separate plastid types predominantly contain different iso-proteins, with distinct leaf FNR (LFNR) and root FNR (RFNR) features. Genomic and transcript information has identified multiple isoforms of both LFNR and RFNR in several species. We have used a technique for rapidly purifying Fd-interacting proteins from Arabidopsis thaliana to identify the two LFNR and two RFNR proteins encoded in the genome. Analysis of purified LFNRs revealed variation in pI and in abundance between stromal and thylakoid fractions of chloroplasts. Transcript and protein levels of the two LFNRs were similar in leaves, but varied in relative abundance between stems and siliques and in response to different nitrogen growth regimes. Relative transcript accumulation and protein abundance of the two RFNR isoforms varied between organs and in response to different nitrogen growth regimes. These results show that the multiple FNR iso-proteins of A. thaliana have variable metabolic roles and contribute differentially to nitrogen assimilation.


Ferredoxin (Fd) receives 1 electron from PSI and transfers it to chloroplast enzymes requiring reducing power. The enzyme Fd:NADP(H) oxidoreductase (EC, FNR) accepts two electrons from Fd in order to photoreduce NADP+ for use in carbon fixation. Other Fd-dependent enzymes include nitrite reductase (NiR) and glutamine-oxoglutarate aminotransferase (GOGAT), which are both involved in the assimilation of nitrogen. Under non-photosynthetic conditions (such as in root plastids, or chloroplasts in the dark) supply of reduced Fd to such enzymes can be maintained by reversing the FNR reaction, using catabolically generated NADPH to reduce Fd. Reflecting this functional difference, photosynthetic and non-photosynthetic tissues predominantly contain different subgroups of both Fd and FNR iso-proteins (Hanke et al. 2004a; Ceccarelli et al. 2004). For ease of comparison with previous work, proteins with highest homology to FNRs from photosynthetic tissues will be referred to here as leaf FNR (LFNR), whereas those with highest homology to FNRs from non-photosynthetic tissues will be called root FNR (RFNR). There is considerable evidence for isoform-specific Fd:FNR interactions. Leaf type Fd and FNR isoforms in combination are more efficient at reducing NADP+ (Hase, Mizutani & Mukohata 1991; Hanke et al. 2004b), whereas root type Fd and FNR isoforms in combination are more efficient at reducing Fd (Onda et al. 2000).

Even before the availability of genomic information it was known that some species, such as maize (Kimata & Hase 1989), have multiple isoforms of both root type and leaf type Fd. Ever increasing transcript and genomic information indicates that this abundance of Fd iso-proteins occurs in all plants and extends to both LFNR and RFNR types. In previously published work, we examined different Fds represented in the Arabidopsis thaliana genome (Hanke et al. 2004b), and the principal focus of this paper will be on FNRs from the same source.

In addition to linear photosynthesis, a role for LFNR has been proposed in cyclic electron transport based on association with both the cytochrome b6/f complex (Zhang & Cramer 2004) and NADP(H) dehydrogenase complex (Quiles & Cuello 1998), and on measurements of FNR-dependent quinone reduction (Bojko, Kruk & Wieckowski 2003). LFNR is also proposed to function in refolding of imported redox proteins, based on localization at the plastid inner envelope membrane (Kuchler et al. 2002). Co-suppression of one LFNR transcript in Arabidopsis resulted in a dwarfed pale green phenotype, but it is unclear whether transcript reduction is restricted to a single isoform, and the effects on specific metabolic processes were not measured (LeClere & Bartel 2001). It is not known whether these multiple proposed functions are performed by specific iso-proteins, or whether the separate LFNRs encoded in available genomes represent redundant capacity.

We have partially purified both LFNR iso-proteins encoded in the Arabidopsis genome, and examined their physical and functional properties and subchloroplast location. In addition we have identified two Arabidopsis RFNR genes. Tissue expression of all Arabidopsis FNRs and Fds (as the functional electron transfer partners of FNR) is presented.

The essential action of FNR and Fd in nitrogen assimilation has been well documented (see Crawford 1995; Stitt 1999). The bulk of nitrogen is assimilated from nitrate and ammonium taken up by the roots. In root cells, nitrate is reduced to nitrite in the cytosol. Nitrite is toxic and immediately imported to the plastid where Fd-dependent NiR reduces it to ammonium (FNR catabolically reducing Fd to fuel the process). The bulk of ammonium fixation into amino acids is then performed by NADH-dependent GOGAT. Nitrate and ammonium are also transported through the vasculature to the leaves. Here, the malate–oxaloacetate shuttle transfers photosynthetic reducing power (all electrons having passed through Fd and FNR) to the cytosol for nitrate reduction. Nitrite is then imported into the chloroplast and reduced by Fd-dependent NiR to ammonium which is in turn fixed by Fd-dependent GOGAT. Assimilation of nitrate is therefore more energetically expensive than ammonium, and in leaves nitrate reduction results in alkalization of the leaf, whereas ammonium reduction results in acidification (Stitt 1999). Therefore, growing plants on solely nitrate or ammonium creates variation in reductive demands and pH regulation. We have exploited this to separate the physiological roles of FNR iso-proteins by examining expression patterns on different growth regimes.


Plant materials and growth conditions

All Arabidopsis plants were Columbia ecotype. Those used for generating RNA samples used in tissue-specific detection of transcript, and in purification of LFNRs were grown as described by Hanke et al. (2004b). Plants on varied nitrogen regimes were grown hydroponically as described by Orsel et al. (2004) and harvested following 3 weeks growth. Sampling in all cases took place at the onset of the light period.

FNR isolation and separation

For FNR isolation, Fd-sepharose columns were prepared according to Onda et al. (2000) from AtFd2 and AtFd3 proteins recombinantly expressed and purified as described by Hanke et al. (2004b). All chromatography was performed using an ÄKTA primeTM. system (Amersham Biosciences, Tokyo, Japan). Crude Arabidopsis shoot and root protein fractions were prepared as described by Hanke et al. (2004b), except that following separation of proteins soluble and insoluble in 70% saturated ammonium sulphate, the insoluble pellets were reserved. These were re-suspended in 50 m m Tris HCl pH 7.5, 1 m m MgCl2 and desalted over a column made of Sephadex G-25. The resultant shoot and root protein fractions were loaded onto AtFd2 and AtFd3 columns, respectively, and extensively washed with the same buffer before elution of Fd-binding proteins over a gradient from 0 to 500 m m NaCl. When anion-exchange was performed, Fd-binding proteins were then desalted and loaded onto a RESOURCE Q anion-exchange column (Amersham Biosciences). Eluent fractions were collected over a gradient of 0–1 m NaCl.

Electrophoresis and immunodetection

Protein fractions were treated with sodium dodecyl sulphate (SDS) buffer as described by Hanke et al. (2004b), separated by SDS-polyacrylamide gele electrophoresis (PAGE) on a 12.5% gel and either stained with Coomassie Brilliant Blue, or blotted to Polyvinylidene defluoride (PVDF) membrane followed by immunodetection with antibodies specific for maize leaf Fd, maize root Fd (as described in Hanke et al. 2004b), maize leaf FNR and maize root FNR (as described by Onda et al. 2000) or spinach NiR (1 : 50 000 dilution).

Mass spectrometry and protein sequencing

Fractions containing the separated FNR proteins (approximately 100 pmol each) were reduced and carboxymethylated essentially as described by Crestfield, Moore & Stein (1963), dialyzed against 50 m m NH4HCO3 and freeze dried. From this point, samples were prepared for mass spectrometry essentially as described by Shen et al. (2002), except that proteins were digested for 12 h at 21 °C in 100 µL MilliQ water containing 1 m unit activity Achromobacter lysyl endopeptidase (Wako Pure Chemical Industries Ltd, Osaka, Japan). Trifluroacetic acid (TFA) was added to 10 µL of each sample to give a 1% (v/v) concentration and the sample was desalted using a C18TM. zip tip (Millipore, Billericay, MA, USA) according to the manufacturer's instructions, washed with 0.1% TFA and eluted in 1 µL 50% (v/v) acetonitrile, 0.1% TFA. After mixing with the matrix (α-cyano-4-hyroxycinnamic acid) samples were air-dried and matrix-assisted laser-desorption time of flight (MALDI-TOF) mass spectrometry performed essentially as described by Shen et al. (2002). Sequence database searches with Mascot software (Matrix Science, London, UK) identified specific protein sequences. Proteins for sequencing were separated by SDS-PAGE, blotted to a PVDF membrane and visualized with Ponceau-S (Nacalai tesque, Kyoto, Japan) so that bands could be excised, washed extensively in distilled water and sequence determined to the tenth residue by automated Edman degradation using an ABI 491 cLC protein sequencer (Applied Biosystems, Foster City, CA, USA).

Enzymatic analysis

Activity of FNR enzymes in reduction of Fd was measured with a cytochrome c reduction assay, as described by Hanke et al. (2004b). FNR concentrations in purified fractions were estimated by Coomassie staining of bands separated by SDS-PAGE. FNR was added to assays at approximately 20 n m.

Chloroplast isolation and localization of LFNR

Chloroplasts were prepared essentially as described by Mach (2002), and then ruptured in 50 m m Tris-HCl pH 7.5, 1 m m MgCl2. Following centrifugation at 4 °C 15 000 rpm in a bench top centrifuge for 5 min, the supernatant was reserved as a stromal fraction. The pellet was twice re-suspended in the same buffer and centrifuged as above to remove any residual stromal proteins. The final pellet was re-suspended in the same buffer with the addition of 0.1% Triton X-100, centrifuged as above and the soluble fraction reserved as peripheral thylakoid proteins. Fd-binding proteins were separately isolated from these soluble and thylakoid fractions, before LFNR iso-proteins were separated by anion exchange chromatography as described previously.

Semi-quantitative reverse transcription-polymerase chain reaction

Total RNA was prepared using an RNeasy Plant Mini Kit (Qiagen K.K., Tokyo, Japan) according to the manufacturer's instructions, except with a duplication of the DNase step to digest any genomic DNA. RNA concentration was determined spectrophotometrically, and 2 µg was used to manufacture cDNA using the OmniscriptTM. kit (Qiagen) according to the manufacturer's instructions, except in the case of silique samples where 0.4 µg was used due to low RNA concentrations. Primers for polymerase chain reaction (PCR) were as follows: AtFd1 5′-CGCAATCGTAAG CACCTCTT- 3′ and 5′-AAGCCACACAGGTCAAGAC A-3′, AtFd2 5′-CTTCATTCATCCGTCGTTCC-3′ and 5′-AGGGTAAGCAGCACAAGTGA-3′, AtFd3 5′-CCTCC ACTAGCAT GACCAA-3′ and 5′-TGGGTAAGCCACA CAAGTCA-3′, AtFd4 5′-TGGATCAAGTACTCTACTC CTCTTACA-3′ and 5′-AAGATCGGATTGTTTGTGAG TGT-3′, AtLFNR1 5′-TATCGCGAGTAGTGCCATTG-3′ and 5′-TGTCTGGGTTCTTCTCCTTCA-3′, AtLFNR2 5′-CAGCCATCAAGGAGAAATCC-3′ and 5′-GGTGCC AAATCACACAAGAA-3′, AtRFNR1 5′-GGGAAGGA CAAAGCTATGGA-3′ and 5′-GCAAGTCCGTCAAA CTTGAAA-3′, AtRFNR2 5′-TACCCGAGCACGACAA TATG-3′ and 5′-CGGGCTTTGAATCACATAGG-3′.

Each PCR was performed with 1/10, 1/100, 1/1000 and 1/10 000 dilutions of cDNA, to ensure that none of the reactions were saturated with template. PCR was hot started at 98 °C and carried out for 30 cycles of 98 °C for 30 s, 30 s annealing temperature (see later), and 72 °C for 45 s. Where transcript was compared between samples, an initial PCR with QuantumRNATM. 18S standards (Ambion, Austin, TX, USA) was used to confirm equivalent cDNA template was present (according to manufacturer's instructions). To ensure maximum primer specificity, PCR annealing temperatures were as high as would allow consistent amplification, as follows: all ferredoxins 65 °C, AtLFNR1 and AtRFNR2 66 °C, AtLFNR2 and AtRFNR1 68 °C. Products were visualized by ethidium bromide staining following agarose gel electrophoresis.

Accession numbers

Gene bank accession numbers for Arabidopsis Fd genes are AtFd1 (At1g10960), AtFd2 (At1g60950), AtFd3 (At2g27510), AtFd4 (At5g10000), and for Arabidopsis FNR genes are AtLFNR1 (At5g66190) AtLFNR2 (At1g20020) AtRFNR1 (At4g05390) AtRFNR2 (At1g30510). FNR proteins used in the alignment to generate the phylogenetic tree in Fig. 1b are maize leaf (BAA88236), maize root (AAB40034), rice leaf (XP_463801), rice root (XP_476624), tobacco leaf (O04977), tobacco root (O04397). Accession numbers of the proteins in the phylogenetic tree in Fig. 6 are given in Table 4

Figure 1.

Identification of Arabidopsis FNR genes. (a) Plant type FNR genes encoded in the A. thaliana genome were identified by blastp searches for their protein products of AGI proteins in the TAIR database (, using maize LFNR and maize RFNR. Proteins were aligned using ClustalW ( Residues common to at least four sequences are shown as black on white, residues common only to maize LFNRs are highlighted in green and those common only to maize RFNR are highlighted in yellow. Only those residues following the transit peptide cleavage point of maize LFNR (.) and maize RFNR (:) are highlighted. (b) Mature sequences of A. thaliana FNRs were aligned with known leaf and root FNR proteins using ClustalW (, and the resulting tree was drawn using software at the phylodendron website ( Transit peptide cleavage points for AtLFNRs were determined by N-terminal sequencing, N-terminal sequences of RFNRs were estimated based on comparison with known N terminal sequenced RFNR proteins.


There are two LFNR and two RFNR proteins encoded in the Arabidopsis genome

Six genes have previously been described as FNR-like proteins in the Arabidopsis genome (Wang et al. 2003). One of these (At4g32360) has been subsequently identified and characterized as a mitochondrial FNR (Takubo et al. 2003). Comparison of the translated amino acid sequences with those of known FNRs reveals that one sequence (At5g66810) has a very low homology and so was not studied further. The four others can be considered classical plant type FNR proteins and the phylogenetic tree in Fig. 1b shows relationships between the mature sequences of these proteins and those of known FNRs from other species. Based on this analysis, two are LFNRs (above 80% identity to maize leaf FNRI) and two are RFNRs (above 80% identity to maize root FNR).

An alignment of these predicted protein sequences with maize LFNR and maize RFNR is shown in Fig. 1a. LFNR genes were labelled FNR-1 and FNR-2 in a previous study (Frisco et al. 2004) and so are correspondingly referred to here as AtLFNR1 and AtLFNR2, respectively. RFNR genes were arbitrarily named AtRFNR1 and AtRFNR2 (accession numbers in methods).

AtLFNR1 and AtLFNR2 proteins are detected in Arabidopsis shoots

Ferredoxin binding proteins were isolated from Arabidopsis shoots and roots (Fig. 2a). Following Western blotting, a single shoot protein band of the approximate size of FNR (32 kDa) reacted with a LFNR specific antibody (Fig. 2b). Anion exchange chromatography was performed to separate Fd-interacting proteins. Fractions were Western blotted as before, and the putative LFNR resolved into two bands, corresponding to the greatest elution peaks (Fig. 2c). These bands were cut out and subjected to MALDI-TOF mass spectrometry and N-terminal sequencing (Table 1). Both methods identified the high-salt eluting protein as AtLFNR1 and the low-salt eluting protein as AtLFNR2. Diagnostic, isoform-specific peaks from mass spectrometry are given in Table 2.

Figure 2.

Identification of Arabidopsis FNR proteins. (a) Fd chromatography was performed with crude protein fractions prepared from 100 g Arabidopsis shoots (left panel) and 20 g Arabidopsis roots (right panel). Extracts were desalted and applied to 20 mL sepharose-AtFd2 (left panel) and sepharose-AtFd3 (right panel) before washing and elution over a 0–1 m NaCl salt gradient (shown as concentration entering the column). Protein elution was followed by absorbance at 280 nm. Fractions above the solid bar were reserved as Fd-binding proteins. (b) Immunodetection with specific antisera raised against maize LFNR or maize RFNR following SDS-PAGE and blotting of Fd-interacting proteins from total Arabidopsis shoots (S) or roots (R). The position of the nearest molecular weight marker is given on the right. (c) Separation of LFNR iso-proteins. Upper panel shows anion exchange chromatography of Fd-interacting proteins from total Arabidopsis shoots (protein content of eluent monitored at 280 nm) with elution over a salt gradient (NaCl concentration of eluent monitored by conductivity). Fractions indicated on the graph by vertical bars were separated by SDS-PAGE and stained with Coomassie Brilliant Blue (lower panel). The position of the nearest molecular weight marker is given on the right. Separated proteins were used for MALDI-TOF mass spectrometry or N-terminal sequencing.

Table 1.  Properties of the separated AtLFNR proteins
Anion exchangeHigh salt eluting proteinLow salt eluting protein
  1. AtLFNR proteins separated by anion exchange chromatography were identified by mass-spectrometry and N-terminal sequencing. pI of mature proteins was calculated using the predictive program at ExPASy ( Separate fractions containing AtLFNR1 and AtLFNR2 were used in cytochrome c-based Fd-reduction assays with indicated Arabidopsis Fds to calculate Km.

MALDI-TOF identificationAt5g66190 (AtLFNR1)VTTDTTEAPGAt1g20020 (AtLFNR2)ITTETDTPTP
N-terminal sequence
calculated pI5.54  6.19  
cytochrome c reduction assay Kmm)AtFd1AtFd2AtFd3AtFd1AtFd2AtFd3
3.5 ± 0.32.5 ± 0.24.6 ± 0.24.3 ± 0.33.5 ± 0.25.4 ± 0.3
Table 2.  Identification of separated LFNRs by mass spectrometry
High salt eluted proteinLow salt eluted protein
peakPutative AtLFNR1 fragmentpeakPutative AtLFNR2 fragment
  1. The mass of fragments from lysyl endopeptidase digestion of two putative AtLFNR proteins (separated by anion exchange chromatography) were established by MALDI-TOF MS. Diagnostic peaks are presented with corresponding predicted digestion fragments of AtLFNR1 and AtLFNR2 of the same mass.


AtLFNR1 and AtLFNR2 vary in pI and affinity for Fd

N-terminal sequencing indicates the cleavage point of the chloroplast transit peptide occurs at an equivalent position in both AtLFNR iso-proteins. N-terminal sequence also allows accurate prediction of the pI of the mature protein. AtLFNR1 is significantly more acidic (Table 1), consistent with its elution from the anion exchange column at higher salt concentrations. To compare interaction with Fd (and by analogy, effectiveness in photosynthetic NADP+ reduction), we measured the enzyme activity of fractions containing AtLFNR1 and AtLFNR2. The Km values in Table 1 show that for all Fd isoforms examined the affinity of AtLFNR1 is slightly, but significantly, higher than AtLFNR2. The kcat values for these reactions were similar between the two AtLFNRs (in the order of 50–100 s−1), but estimates of FNR protein concentration were not accurate enough for a comparison between the two.

Two Arabidopsis RFNR proteins are differentially distributed between shoots and roots

Fd-interacting proteins from Arabidopsis shoots and roots were Western blotted using antisera specific to maize RFNR. Two bands, of different mobility on SDS-PAGE, were detected in both fractions (Fig. 2a). Both were of a higher molecular weight than the LFNR bands. Both LFNR and RFNR antisera were purified against recombinant enzymes to be highly specific (Onda et al. 2000), and non-specific interaction with bands corresponding to RFNR and LFNR, respectively, was negligible. Both RFNRs were detected at equal intensity in shoot Fd-binding proteins, but in root Fd-binding proteins the lower band was far more abundant. This difference was also observed when total root and shoot protein extracts were used (data not shown). Following separation by SDS-PAGE, the lower, more abundant protein band was cut from the gel and identified by MALDI-TOF mass spectrometry as the product of the gene AtRFNR2 (diagnostic peaks are shown in Table 3). At this time, sufficient protein for mass spectrometry could not be obtained for the upper band.

Table 3.  Identification of root abundant RFNR by mass spectrometry
peakPutative AtRFNR2 fragment
  1. The mass of fragments from a lysyl endopeptidase digestion of the most abundant AtRFNR iso-protein in roots (separated by SDS-PAGE) were established by MALDI-TOF MS. Diagnostic peaks are presented with corresponding predicted digestion fragments of AtRFNR2 of the same mass.


Relative expression of FNR and Fd genes depends on tissue

Tissue specific expression of AtLFNR and AtRFNR isoforms, and of the different isoforms of Fd (their electron transfer partners) was compared by semi-quantitative reverse transcriptase (RT)-PCR (Fig. 3). Expression pattern of the two leaf type AtFds (AtFd1 and AtFd2) was almost identical, and restricted to green tissues. This was also true of both LFNRs. A notable variation in expression pattern is that transcripts of AtLFNR1 appeared relatively more abundant in siliques, and those of AtLFNR2 relatively more abundant in stems. Root type Fd (Fd3), and both RFNRs were expressed in all tissues examined except flowers. By comparison with leaf type FNRs and Fds, expression of these genes in the stems was greatly increased relative to leaves, reflecting a greater metabolic role in tissues with reduced photosynthetic capacity. Interestingly, relative expression of AtRFNR1 in roots was low in comparison to AtRFNR2. This suggests AtRFNR2 may be the predominant root type isoform, while AtRFNR1 is more constitutive. Western blots in Fig. 2b show relative abundance of the lower RFNR band in shoots and roots reflects the transcript levels of AtRFNR2, while the upper band reflects that of AtRFNR1. Based on this we tentatively assume the upper, more constitutive band is AtRFNR1. Previously we reported the presence of a high redox potential Fd, AtFd4, in Arabidopsis (Hanke et al. 2004b). Despite very low transcript levels, which prevented detection at comparable cDNA concentrations (limited by low RNA recovery from siliques), we did unambiguously detect AtFd4 transcript levels in total shoot and root extracts (data not shown). This protein does not exchange electrons with FNR and so will not be considered in this paper as further research is required to establish its metabolic role.

Figure 3.

Relative tissue distribution of Arabidopsis FNR and Fd expression. Transcripts were detected by semi-quantitative PCR with indicated primers, as described in Methods and Materials, following RNA isolation from different Arabidopsis tissues.

Chloroplast thylakoid and stroma predominantly contain different AtLFNR iso-proteins

In addition to NADP+ photoreduction, recent studies have implicated LFNR in cyclic electron transfer, partly by virtue of association with different thylakoid membrane protein complexes. As an initial step in establishing if these different functions could be isoform specific, we checked for variation in subchloroplast location. When chloroplasts isolated from Arabidopsis shoots were separated into stromal and thylakoid fractions (shown in the Western blots in Fig. 4a) the majority of FNR was detected in the membrane fraction, but some protein was soluble. Detection of the soluble enzyme NiR confirmed clear separation of soluble and membrane-bound forms. On treatment with 0.1% Triton X-100 the majority of LFNR protein moved to the soluble phase.

Figure 4.

Sub-chloroplast location of AtLFNRs. (a) Arabidopsis chloroplasts were separated into equivalent volumes of soluble and membrane fractions, either in buffer, or with the addition of 0.1% Triton-X 100 (as indicated). Fractions were separated by SDS-PAGE, blotted to membranes and probed with antibodies raised against LFNR or NiR. M, thylakoid membrane fraction; S, stromal fraction. (b) Total AtFd2 interacting proteins were isolated from thylakoid membrane (solubilized by 0.1% Triton X-100) and stromal Arabidopsis chloroplast fractions. These were eluted from a Resource Q anion exchange column over a salt gradient. Eluted fractions were separated by SDS-PAGE and stained with Coomassie Brilliant Blue.

Fd-binding proteins were prepared from both the stromal fraction and a Triton X-100 solubilized membrane fraction. LFNR isoforms were separated by anion exchange and Coomassie staining in Fig. 4b allows comparison of different AtLFNR iso-proteins at the thylakoid membrane and in the stroma. Although both proteins are present in both fractions, AtLFNR1 is more abundant in the membrane fraction, and AtLFNR2 is more abundant in the soluble fraction. We confirmed that this difference reflects iso-protein abundance, rather than differential Fd interaction, by repeating the experiment with total thylakoid and stromal proteins and detecting LFNRs by western blotting (data not shown). The gentle rupturing of chloroplasts in our method should maintain FNR–thylakoid interactions, and FNR in the thylakoid fraction remained associated after several washes, suggesting that there really is a dual distribution of these proteins within the chloroplast.

Relative expression of Fd and FNR isoforms is partly determined by nitrogen supply

The involvement of Fd and FNR in nitrogen assimilation is well documented, and previous microarray studies have indicated that transfer from ammonium to nitrate growth conditions induces expression of both AtRFNRs, but neither AtLFNR (Wang et al. 2000, 2003). To further characterize isoform-specific involvement in nitrogen assimilation, we have compared the steady-state transcript and protein levels of different FNRs and Fds between plants grown under different nitrogen conditions. By contrast to rapid nitrate induction (as measured previously in microarrays), a comparison of relative steady-state transcript abundance could reveal differences between the physiological roles of isoenzymes, dependent on the variable energetic demands on photosynthesis and on pH regulation, caused by continual growth under different nitrogen regimes. Figure 5 shows transcript and protein levels of FNRs and Fds in Arabidopsis plants grown under constant ammonium (5 m m), limiting nitrate (0.1 m m) and high nitrate (6 m m). RT-PCR revealed several striking differences between expression of different isoforms. Only AtFd1 of the leaf type Fds is expressed at equivalent levels in all nutrient regimes tested. Transcript level of AtFd2 is much higher in shoots of ammonium grown plants than those of nitrate grown plants, and lowest under high nitrate concentrations. AtFd1 and AtFd2 can be separated by SDS-PAGE (Hanke et al. 2004b) and western blots show that by contrast to transcripts, protein level abundance does not vary remarkably (Fig. 5b). AtFd1 is marginally more abundant under low nitrate, and AtFd2 does not vary between growth conditions.

Figure 5.

Arabidopsis FNR- and Fd- transcript and protein levels under different nitrogen regimes. (a) Semi quantitative RT-PCR detection of the relative abundance of indicated transcripts in the shoots and roots of Arabidopsis plants grown on ammonium (5 m m), low nitrate (0.1 m m), and high nitrate (6 m m). (b) Immunodetection of relative abundance of indicated proteins in the same tissues was achieved by crude extraction into 1× SDS-PAGE sample buffer and separation of 20 µg protein by SDS-PAGE (10% gel for detection of FNRs, 15% gel for detection of Fds), followed by blotting to PVDF membrane and detection with the indicated antibodies.

Both LFNR gene transcripts are more abundant in low nitrate than ammonium-grown plants. However, at high nitrate, AtLFNR2 expression is greatest while AtLFNR1 transcript levels are low (equivalent to those in ammonium grown plants). Western blots show that total LFNR protein levels were relatively abundant in high nitrate, and scarce in ammonium growth conditions. Our experiments yielded insufficient material for a separation of LFNR isoforms, so it is unclear which LFNR iso-proteins contribute to these changes in protein level.

Like AtFd2, root type AtFd3 expression in shoots is also lowest under high nitrate conditions, and highest in ammonium conditions. In this case the difference is reflected at the protein level. In roots, AtFd3 transcript does not vary greatly between plants grown under different nitrogen regimes, although there is consistently higher expression under high nitrate than any other condition. AtFd3 protein was more abundant in roots of nitrate than ammonium grown plants, and greatest in roots of plants grown at high nitrate.

Expression of both AtRFNR genes in shoots mirrored that of the root type Fd, being highest under ammonium and lowest under high nitrate growth conditions. This difference was most dramatic for AtRFNR1. Reflecting this, two AtRFNR proteins were also most abundant in Arabidopsis shoots grown on ammonium. In roots the expression and protein levels of AtRFNR show the opposite pattern. Both AtRFNR transcripts were most abundant under high nitrate and relatively scarce under ammonium growth conditions. There is an interesting discrepancy between relative expression of AtRFNRs in roots grown on low nitrate. AtRFNR1 transcript is relatively scarce, whereas AtRFNR2 transcripts accumulate equally to those in high nitrate-grown roots. In Fig. 5b, arrows indicate the position of two RFNR bands detected in western blots of root protein samples. The lower RFNR band reflects changes in AtRFNR2 transcript, while the upper band reflects changes in the transcript levels of AtRFNR1.


In addition to the four Fd genes we described previously (Hanke et al. 2004b) we have identified four authentic plant type FNR genes in Arabidopsis. Amino acid sequence and expression data indicate that two of these encode leaf type FNR proteins involved in chloroplast-related functions, while two encode root type enzymes, predominantly involved in non-photosynthetic plastid metabolism. The aim of this work is to gain insight into whether this apparent duplication represents redundancy, or whether there is variation in the metabolic role of the different isoforms, either in terms of gene expression, or protein function.

Evidence for functional differentiation of AtLFNRs

The two LFNR genes identified, AtLFNR1, AtLFNR2 and the two photosynthetic Fds (AtFd1 and AtFd2) were expressed in all green tissues. The only difference in tissue expression of AtLFNR1 and AtLFNR2 was in relative transcript abundance in stems and siliques. It is unclear whether subtle differences between the photosynthetic processes occurring in these two organs equate to variation in the metabolic roles of the two LFNRs. The two AtLFNR isoforms were successfully isolated from a fraction of Fd-binding shoot proteins and chemically identified. Resolution of the N-terminal sequence allowed prediction of the transit peptide cleavage point, revealing a large variation in pI between the two isoforms. Intriguingly, in other species in which two LFNR isoforms have been identified, such as rice, wheat, and maize, the pI values of putative mature proteins also reveal an acidic and a more basic protein is present in each case (see Table 4). Additionally, in all these species, the more basic LFNR iso-protein has a higher identity to AtLFNR2 than AtLFNR1. When these proteins are aligned and used to make a phylogenetic tree (Fig. 6), they clearly indicate the evolution of distinct high pI and low pI subgroups of LFNR. This implies that the presence of a more acidic and a more basic LFNR protein may be a feature common to all higher plant chloroplasts.

Table 4.  pI Comparison in species with two identified LFNRs
Species AccessionpI% ID to AtLFNR1% ID to AtLFNR2
  1. LFNRs were identified by blast on the ncbi website ( using AtLFNR1 and AtLFNR2 as templates. Where N-terminal data was unavailable, mature proteins were predicted by comparison with known N-terminals of LFNRs. pI of mature proteins was predicted at ExPASy (

Arabidopsis1At5g661905.54100 78
2At1g200206.19 78100
Maize1BAA882375.43 86 84
2BAA882366.44 81 88
Wheat1CAD300255.53 81 82
2CAD300247.67 78 82
Rice1XP_4638015.65 83 84
2P413447.67 81 88
Figure 6.

Evolutionary relationship of high pI and low pI LFNR iso-proteins. FNR sequences were identified by blast searches at the National Center for biotechnology Information website ( using AtLFNR1 and AtLFNR2. Where transit peptides were not known they were inferred by analogy with those of known chemically sequenced FNRs. Proteins were aligned using ClustalW (, and the resulting tree was drawn using software at the phylodendron website ( Accession numbers are as shown in Table 4.

LFNR is considered a thylakoid membrane protein, although its gene promoter region differs significantly from those of other thylakoid targeted proteins (Oelmuller et al. 1993), and has been identified as a component of several different membrane complexes. It has also been demonstrated that solubilization of LFNR is part of a response to oxidative stress (Palatnik, Valle & Carrillo. 1997). Our deduced N-terminal sequences of the two AtLFNRs show the transit peptide cleavage point is at an equivalent position, implying entry to the chloroplast and any processing involved in targeting to and association with thylakoid membrane complexes is by a similar mechanism. Although both AtLFNR iso-proteins have been previously localized to thylakoids, there was variation between them in terms of the severity of treatment necessary to extract them from the membrane (Frisco et al. 2004), implying a different mechanism of interaction. In this paper we isolated AtLFNR1 and AtLFNR2 from both chloroplast thylakoid and stromal fractions, but the relative distribution of the two isoforms varied, with AtLFNR1 being the more abundant form in thylakoids, and the ratio of AtLFNR2 being greatly increased in the stromal fraction. This implies that both FNRs are subject to equilibrium between a membrane and a stromal position within the chloroplast. Variable location of the two AtLFNR iso-proteins may be due to association with different complexes, or regulation of this equilibrium. Such a system would offer a mechanism to control the involvement of peripheral membrane proteins in thylakoid metabolism. It is not yet clear whether the difference between pI values of the two isoforms is related to their location within the chloroplast. The evolutionary conservation of separate acidic and basic LFNRs across species boundaries is strong evidence that the two iso-proteins have different functions, and that these are related to their charge.

Although Wang et al. (2003) found no induction of genes corresponding to leaf type Fds or LFNRs on nitrate induction, we have identified large differences between their transcript accumulation in plants grown on different nitrogen growth regimes. AtFd2 transcripts were relatively more abundant in shoots of ammonium- than nitrate-grown plants, but protein levels were equivalent, which may indicate high turnover of transcript. Post-transcriptional regulation has been demonstrated for Chlamydomonas Fd, in which the 5′ untranslated region of Fd mRNA is bound by a redox-oxidation sensor protein that controls the rate of translation (Trebitsh et al. 2000), and pea leaf type Fd, in which the 5′ untranslated region and first third of the coding region contain a light-responsive element that stabilizes the transcript on ribosomes under photosynthetic conditions (Elliott et al. 1989; Hansen et al. 2001). Steady-state transcript levels of AtLFNR2 are most abundant in high nitrate and lowest in ammonium growth conditions (Fig. 5a). This indicates involvement of the gene product in linear electron flow, providing the additional reducing power required in the cell during assimilation of nitrate, rather than ammonium. Total LFNR protein levels are greater in both nitrate conditions than in ammonium-grown shoots, reflecting AtLFNR2 transcripts. By contrast AtLFNR1 transcripts are least abundant in high nitrate. This indicates that, under steady-state conditions, the contribution of AtLFNR1 to reductive nitrogen assimilation is relatively low and that AtLFNR1 and AtLFNR2 have separate functional roles. However, the extent of this functional separation is unclear, and depending on conditions in the chloroplast, activity of one iso-protein may very well supplement the other.

In this study we have demonstrated variation in the relative transcripts of A. thaliana LFNR isoforms between plants grown on different nitrogen regimes. Further evidence of isoform specificity might also be gained from a comparison of diurnal variation in transcripts of these photosynthetic proteins.

Evidence for physiological differences between AtRFNRs

The two AtRFNR genes identified were both expressed in all tissues examined, with the exception of flowers. Relative to other organs, AtRFNR2 expression is much greater than that of AtRFNR1 in roots. Two RFNR proteins detected in Arabidopsis roots were separated by SDS-PAGE, and identified by western blotting. The more abundant, lower band was identified as AtRFNR2 by mass spectrometry, while abundance of the larger protein was consistent with AtRFNR1 expression in shoots and roots under several conditions (Figs 2, 3 & 5). Expression of AtFd3 (a root type Fd) in the same tissues probably reflects a demand for catabolic reduction of Fd, which is most effectively performed by a combination of RFNR and root type Fd (Onda et al. 2000; Hanke et al. 2004b). Leaves contain a significant percentage of non-photosynthetic cells (Bowsher & Tobin 2001) and expression of these genes in seemingly photosynthetic tissues may be due to a demand for reduced Fd in such cells.

A greater abundance of reduced Fd is required for assimilation of nitrate than ammonium and it has already been recorded that AtFd3, AtRFNR1 and AtRFNR2 are transiently induced in Arabidopsis roots by nitrate induction (Wang et al. 2003). Consistent with this, we measured greater transcript and protein levels of AtFd3, AtRFNR1 and AtRFNR2 in roots grown on high nitrate. Intriguingly, there is a difference in regulation of expression of the two AtRFNR genes because AtRFNR2 is also expressed at relatively high levels under low nitrate growth conditions, whereas AtRFNR1 is not. This indicates that AtRFNR2 is specifically involved in root nitrate assimilation regardless of nutrient status, whereas the relatively higher AtRFNR1 expression in roots of plants grown on high nitrate may reflect a more general response to high demands for reduced Fd. The two mature AtRFNR proteins are 89% identical and we consider it unlikely that there is a functional difference, but our analysis of transcript and protein levels suggests that the genes are differentially regulated in response to nitrate and demand for reduced Fd in non-photosynthetic cells.

The role of RFNR in shoots

Further complexity in the regulation of AtRFNR gene expression is revealed in shoots. Interestingly, in a reversal of the pattern in roots, the steady-state transcript levels of both AtRFNR genes and the AtFd3 gene are lowest in plants growing on high nitrate, but high in those grown on ammonium. This suggests simultaneous, tissue-dependent suppression or enhancement of AtRFNR1 and AtRFNR2 expression. Both AtRFNR proteins are also most abundant in shoots grown on ammonium. Assimilation of ammonium rather than nitrate causes acidification of the leaf, whose pH is delicately balanced and poorly buffered during the light period (see Stitt 1999 for a full discussion). We propose that AtRFNR abundance in the shoots of ammonium-grown plants acts to provide reduced Fd to ammonium assimilation in non-photosynthetic shoot cells, thereby relieving such pH stress. Catabolic NADPH supply in green tissues in the light could be through a ubiquitously expressed G6PDH isoform, with reduced sensitivity to redox regulation, as described by Wendt et al. (2000), although analysis of G6PDH isoforms does not clearly identify such an enzyme in A. thaliana (Wakao & Benning 2005). Ammonium arrives in the shoot via the vasculature and Tobin & Yamaya (2001) found that Fd-GOGAT was located in plastids of vascular, as well as photosynthetic, leaf tissues of rice and barley. In addition, Ziegler et al. (2003) showed Arabidopsis Fd-GOGAT promoter-driven expression of GUS in the vascular tissues, as well as photosynthetic tissues, of tobacco leaves. Our detection of relatively higher RFNR transcript accumulation in stems than leaves (Fig. 3) also seems to support this hypothesis. Action of Fd-GOGAT rather than NADH-GOGAT for nonphotosynthetic ammonium assimilation might enable light regulation of the process (pH balance being of less importance in the dark).

Transient induction of RFNR and leaf type Fd genes in shoots on transfer to nitrate has been reported in Arabidopsis (Wang et al. 2003) and maize (Sakakibara 2003). This treatment would probably result in a massive increase in toxic nitrite (Gowri et al. 1992; Crawford 1995; Stitt 1999) and so we hypothesize that such an induction may be due to a rapid and transient increase in demand for reduced Fd by NiR. Our experiments indicate that this does not reflect a specific contribution of RFNR isoforms and root Fd to nitrite reduction in shoots under steady state conditions.

In conclusion, the Arabidopsis genome contains four FNR genes, two of which are leaf isoforms and two root isoforms. These results show that the two LFNR iso-proteins vary in physical properties and subchloroplast location, and indicate that they are functionally distinct. The additional reductant required for assimilation of nitrate, rather than ammonium, is predominantly generated through AtLFNR2. We also present evidence that the two RFNR iso-proteins vary in physiological role, AtRFNR2 being more abundant in roots and more specifically involved in nitrate assimilation. In addition, our results suggest that RFNR proteins play an intriguing role in shoot assimilation of ammonium.


This work was supported by grant in aid 15GSO320 for Creative Scientific Research from the Ministry of Education, Culture, Sports, Science and Technology of Japan and a Joint Research Program between the Japan Society for the Promotion of Science and the Institut National de la Recherché Agronomique France.