If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
* This work was supported by Deutsche Forschungsgemeinschaft Grant KO2065/3 (Research Group FOR 383 “Sulfur Metabolism in Plants: Junction of Basic Metabolic Pathways and Molecular Mechanisms of Stress Resistance”). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Sulfate assimilation provides reduced sulfur for synthesis of the amino acids cysteine and methionine and for a range of other metabolites. Sulfate has to be activated prior to reduction by adenylation to adenosine 5′-phosphosulfate (APS). In plants, algae, and many bacteria, this compound is reduced to sulfite by APS reductase (APR); in fungi and some cyanobacteria and γ-proteobacteria, a second activation step, phosphorylation to 3′-phosphoadenosine 5′-phosphosulfate (PAPS), is necessary before reduction to sulfite by PAPS reductase (PAPR). We found previously that the moss Physcomitrella patens is unique among these organisms in possessing orthologs of both APR and PAPR genes (Koprivova, A., Meyer, A. J., Schween, G., Herschbach, C., Reski, R., and Kopriva, S. (2002) J. Biol. Chem. 277, 32195-32201). To assess the function of the two enzymes, we compared their biochemical properties by analysis of purified recombinant proteins. APR from Physcomitrella is very similar to the well characterized APRs from seed plants. On the other hand, we found that the putative PAPR preferentially reduces APS. Sequence analysis, analysis of UV-visible spectra, and determination of iron revealed that this new APR, named PpAPR-B, does not contain the FeS cluster, which was previously believed to determine the substrate specificity of the otherwise relatively similar enzymes. The lack of the FeS cluster in PpAPR-B catalysis is connected with a lower turnover rate but higher stability of the protein. These findings show that APS reduction without the FeS cluster is possible and that plant sulfate assimilation is predominantly dependent on reduction of APS.
Sulfur is essential for life. It is present in the amino acids cysteine and methionine as part of proteins and peptides; in many coenzymes and prosthetic groups, such as FeS centers, thiamine, lipoic acid, etc.; and in a variety of secondary metabolites, e.g. glucosinolates and alliins in plants. In these compounds, sulfur is present in the reduced form of organic thiols; however, the major form of sulfur available in nature is sulfate. This has to be reduced and incorporated into organic compounds in the sulfate assimilation pathway, which is present in photosynthetic organisms, fungi, and many bacteria, but not in metazoans and bacteria adapted to a parasitic lifestyle. In the sulfate assimilation pathway, sulfate is taken up into cells by sulfate transporters. Before reduction, it is activated by adenylation to adenosine 5′-phosphosulfate (APS),
The abbreviations used are: APS, adenosine 5′-phosphosulfate; APR, adenosine-5′-phosphosulfate reductase; PAPS, 3′-phosphoadenosine 5′-phosphosulfate; PAPR, 3′-phosphoadenosine-5′-phosphosulfate reductase; PpPAPR and PpAPR-B, P. patens 3′-phosphoadenosine-5′-phosphosulfate reductase and homolog, respectively; PpAPR, P. patens adenosine-5′-phosphosulfate reductase; GFP, green fluorescent protein; AtAPR, A. thaliana adenosine-5′-phosphosulfate reductase; EcPAPR, E. coli 3′-phosphoadenosine-5′-phosphosulfate reductase; PaAPR, P. aeruginosa adenosine-5′-phosphosulfate reductase.
2The abbreviations used are: APS, adenosine 5′-phosphosulfate; APR, adenosine-5′-phosphosulfate reductase; PAPS, 3′-phosphoadenosine 5′-phosphosulfate; PAPR, 3′-phosphoadenosine-5′-phosphosulfate reductase; PpPAPR and PpAPR-B, P. patens 3′-phosphoadenosine-5′-phosphosulfate reductase and homolog, respectively; PpAPR, P. patens adenosine-5′-phosphosulfate reductase; GFP, green fluorescent protein; AtAPR, A. thaliana adenosine-5′-phosphosulfate reductase; EcPAPR, E. coli 3′-phosphoadenosine-5′-phosphosulfate reductase; PaAPR, P. aeruginosa adenosine-5′-phosphosulfate reductase.
which is catalyzed by ATP sulfurylase (Fig. 1). APS can be reduced directly to sulfite by APS reductase (APR) or undergo a second activation step, phosphorylation by APS kinase to 3′-phosphoadenosine 5′-phosphosulfate (PAPS). Subsequently, PAPS is reduced by thioredoxin-dependent PAPS reductase (PAPR) to sulfite. Sulfite is further reduced to sulfide by sulfite reductase and incorporated into the amino acid acceptor of O-acetylserine (reviewed in Refs.
APS and PAPS differ only by a single phospho group, similar to NADH and NADPH. Accordingly, APR and PAPR are homologous proteins sharing ∼20% identical amino acid residues, including two highly conserved domains, a sulfonucleotide-binding domain, and a catalytic domain containing the active-site cysteine (
). The reaction catalyzed by APR can be divided into three steps. In the first step, APS binds to the protein, and a reductive transfer results in sulfite bound to the active-site cysteine in a stable reaction intermediate (
). In the second step, free sulfite is released by the action of the C-terminal domain in plants or free thioredoxin in bacteria. In the third step, the glutaredoxin active site or thioredoxin is regenerated by reduction with GSH or thioredoxin reductase (
). Not only do all confirmed APRs, including dissimilatory APR from sulfate-reducing bacteria, possess the FeS cluster, but when the bifunctional APR/PAPR from Bacillus subtilis is deprived of its cofactor, reduction of APS (but not PAPS) is abolished (
). The deduced amino acid sequence of the putative P. patens PAPR (PpPAPR) is more similar to the sequences of PAPRs from Escherichia coli and fungi than to the sequence of APR from P. patens (PpAPR) (Fig. 2). PpPAPR lacks the two cysteine pairs necessary for binding the FeS cluster and does not contain the thioredoxin-like C-terminal extension. However, neither the enzyme activity of the gene product nor its function in P. patens sulfate assimilation has been confirmed yet (
). To gain insight into the biological function of the two proteins, we performed biochemical analysis of the recombinant proteins. Here, we provide evidence that, although the protein encoded by the novel cDNA is capable of PAPS reduction, it is much more active with APS as a substrate. Thus, the APS reduction catalyzed by PpPAPR is not dependent on the FeS cluster, and as a consequence, this enzyme should be renamed PpAPR-B.
Plant Material—P. patens (Hedw.) B.S.G. was cultured in liquid or solid Knop medium as described (
Localization of PpAPR and PpPAPR—The intracellular localization of PpAPR and PpPAPR was addressed by transient expression of vectors encoding C-terminal green fluorescent protein (GFP) fusion proteins. The complete open reading frames of PpAPR and PpPAPR were amplified from P. patens RNA by reverse transcription-PCR with primers APRlokfw (TCAGATCTATGGCGTTGAAGGTTGCATCAC) and APRlokrev (CTCCCGGGTTGCAATGCTTGCACAAACCCAAG), introducing BglII and SmaI restriction sites, and primers PAPRlokfw (TCGGATCCATGGCGATGGCTATGGCGGAT) and PAPRlokrev (CTCCCGGGTGTTCTACCACCAACGTGAATTCC), adding BamHI and SmaI sites, respectively. The PCR products were cleaved with the corresponding restriction enzymes and ligated into the BglII-SmaI-digested GFP reporter plasmid mAV4 (
), was used. The identity of plastids is evident from the red chlorophyll autofluorescence.
The plasmid DNA for transfection was prepared with the QIAfilter plasmid maxi kit (Qiagen Inc.). Protoplasts were transiently transfected with 50 μg of circular plasmid DNA. Isolation and regeneration of protoplasts in the absence of any antibiotic drug were carried out as described previously (
). Six days after transfection, the localization of GFP fluorescence was analyzed in live moss protoplasts by confocal laser scanning microscopy (Leica TCS 4D) using 488 nm excitation and two-channel measurement of emission from 510 to 580 nm (green/GFP) as well as 590 nm (red/chlorophyll).
Overexpression of PpAPR and PpPAPR in E. coli—PpAPR and PpPAPR (
) were produced in E. coli using the pET14b expression system and purified to homogeneity by SDS-PAGE with the His·Tag® system (Novagen).
Enzyme Assays—APR and PAPR activities were measured as the production of [35S]sulfite, assayed as acid-volatile radioactivity. The reaction mixture contained 5 μg of recombinant PpPAPR (0.1 μg for measurements with APS) or 4 ng of recombinant PpAPR, 50 mm Tris-HCl (pH 9), 500 mm MgSO4, 5 mm dithioerythritol, 1-200 μm [35S]PAPS (specific activity of 1.67 kBq/nmol; PerkinElmer Life Sciences) or 1-330 μm [35S]APS (synthesized according to Ref.
). To analyze the pH dependence of the reaction, a series of Tris-HCl buffers with varying pH values was used. For calculation of kinetic parameters, the data were linearized according to the Lineweaver-Burk method. Protein concentrations were routinely estimated according to Bradford (
Electronic Spectra—UV-visible spectra were recorded on a Lambda 16 instrument (PerkinElmer Life Sciences) equipped with a temperature-controlled cell compartment.
Bioinformatics—Sequence analysis was performed with the web-based platform Biology WorkBench (workbench.sdsc.edu/). Analysis of the P. patens genomic sequence was performed on the COSMOSS homepage (www.cosmoss.org). The subcellular localization of APR and PAPR was analyzed by programs available on-line, viz. SherLoc (
), resulting in the same tree topology. The protein sequences used for analysis were obtained from the GenBank™ Data Bank as follows: Arabidopsis thaliana APR (AtAPR)-1 (NP_192370), AtAPR2 (NP_176409), and AtAPR3 (NP_193930); PpAPR (CAD22096) and PpPAPR (CAD32963); Adiantum capillus-veneris (BP918674); Ceratopteris richardii (AAT09441); Enteromorpha intestinalis (AAC26855); Chlamydomonas reinhardtii (AAM18118); Selaginella moellendorffii APR and PAPR (obtained from analysis of genomic sequence at moss.nibb.ac.jp/); Pseudomonas aeruginosa (NP_250447); Burkholderia cepacia (AAD50979); Sinorhizobium meliloti (AAD55759); B. subtilis (P94498); Mycobacterium tuberculosis (P65668); E. coli (P17854); Salmonella typhimurium (P17853); Synechococcus (Q55309); Synechocystis (P72794); Saccharomyces cerevisiae (P18408); Penicillium chrysogenum (AAG24520); Schizosaccharomyces pombe (Q10270); Selaginella lepidophylla (AAT09442); and Marchantia polymorpha (combined from expressed sequence tags BJ859590 and BJ853552). Homology modeling was performed with MODELLER Version 9 and was based on the structures of E. coli PAPR (EcPAPR; Protein Data Bank code 1SUR) and P. aeruginosa APR (PaAPR; code 2GOY, chain B) (
Subcellular Localization of PpAPR and PpPAPR—Physcomitrella is unique in possessing genes homologous to plant APR as well as to bacterial and fungal PAPRs. To obtain a first insight into the biological function of the two proteins, we analyzed their intracellular localization. Plant sulfate assimilation occurs in plastids. When compared with their bacterial counterparts, both PpAPR and PpPAPR contain N-terminal extensions carrying all signs of targeting peptides. Several computer prediction programs uniformly indicated localization of both proteins in plastids with high probability (Table 1). To corroborate the prediction, we created constructs coding for C-terminal fusions of PpAPR and PpPAPR with GFP and transfected Physcomitrella protoplasts. Plastid localization of both fusion proteins was confirmed by confocal laser scanning microscopy displaying co-localization of the GFP fluorescence signals with red autofluorescence of chlorophyll (Fig. 3). The same signal distribution was obtained with a control construct encoding GFP fused to the chloroplast-targeting peptide of FtsZ2-1 (
). Interestingly, the distribution of the two enzymes within the moss chloroplasts differed significantly. Whereas the PpPAPR-GFP signal was uniformly distributed within the plastids, PpAPR-GFP fluorescence displayed a spotty pattern (Fig. 3) similar to that described previously for Physcomitrella hydroperoxide lyase (
Biochemical Properties of PpAPR and PpPAPR—To confirm the sulfonucleotide specificities of the two enzymes, we expressed them in E. coli and added an N-terminal His tag for easy subsequent purification. As expected, purified PpAPR was dark yellow in color, similar to recombinant APR from A. thaliana or Lemna minor (
), whereas PpPAPR was colorless. The UV-visible spectrum of PpAPR was very similar to that of AtAPR2, indicating the presence of an FeS cluster as a cofactor, whereas no indication for the same cofactor was detected in the spectrum of PpPAPR (Fig. 4). These findings were confirmed by iron measurements. Although recombinant PpAPR and AtAPR2 contained 3.2 nmol of iron/nmol of protein, no iron was associated with recombinant PpPAPR (data not shown). On an SDS-polyacrylamide gel, recombinant PpAPR was detectable as a single band of 50 kDa, whereas the molecular mass of PpPAPR was 32 kDa (data not shown).
Substrate Specificity of PpAPR and PpPAPR—PpAPR catalyzes the reduction of APS dependent on dithioerythritol or GSH. The properties of the enzyme are similar to those of the well characterized enzyme from A. thaliana (
). However, in contrast to AtAPR2, PpAPR was inhibited by APS concentrations >20 μm (Fig. 5A). The affinity of both enzymes for APS was very high, with an estimated Km of 6 μm (Table 2). From this, we calculated a Kcat of 37.5 s-1 and a Kcat/Km of 6.25 × 106 liter m-1s-1. PpPAPR was able to reduce PAPS in a concentration-dependent manner. However, the activity was ∼50,000-fold lower than that of PpAPR with APS as a substrate (Fig. 5B). The Km for PAPS was estimated to be 20 μm, which is comparable with that of EcPAPR known from literature (22.5 μm) (
) or determined using bacterial extracts (18 μm) (Fig. 5B). Accordingly, the Kcat and Kcat/Km values were very low: 7.5 × 10-4 s-1 and 37.3 liter m-1s-1, respectively. PpAPR did not show any significant activity with PAPS as a substrate (Fig. 5B). Surprisingly, however, PpPAPR was much more active with APS as a substrate than with PAPS (Fig. 5C). The APS reduction rate increased almost linearly at an APS concentration of up to 100 μm and decreased afterward to ∼50% of the maximal value. Compared with PpAPR, the APR activity of PpPAPR was several thousandfold lower at low APS concentrations, but at APS concentrations >50 μm, the difference was only ∼200-fold. The Kcat of PpPAPR for APS reduction was 0.176 s-1, leading to a Kcat/Km of 3520 liter m-1s-1. EcPAPR was not able to reduce APS at all (Fig. 5C). To confirm the substrate specificity of the PpPAPR protein, competition assays with unlabeled APS or PAPS added to the APR or PAPR assays were performed. Whereas the addition of PAPS did not affect APR activity (Fig. 5D), the addition of APS inhibited sulfite production from PAPS (Fig. 5E). Thus, the putative PAPR from P. patens, despite a high similarity to PAPRs from fungi and E. coli, catalyzes the reduction of APS. Therefore, this enzyme has been renamed PpAPR-B.
TABLE 2Kinetic constants of PpAPR, PpAPR-B, and AtAPR2
Similar to other plant APRs, PpAPR can use both dithioerythritol and GSH as electron donors. The reaction catalyzed by PpAPR and AtAPR2 was saturated at 4 mm dithioerythritol (Fig. 6A). It needed, however, a substantially higher GSH concentration. AtAPR2 activity reached its maximum at 30 mm GSH and was inhibited by higher GSH concentrations (Fig. 6B). On the other hand, PpAPR was capable of maximal activity at GSH concentrations of 50 and 100 mm. The maximal reaction velocity of AtAPR2 was ∼50% higher with dithioerythritol than with GSH, whereas the activity of PpAPR was identical with both reductants. Because the GSH concentrations necessary for full activity were much higher than the physiological concentrations of 1-5 mm, in vivo, APR seems not to function at its full potential.
PpAPR-B (but not PpAPR) was dependent on thioredoxin (Fig. 6C). The addition of thioredoxin increased PpAPR-B activity with both substrates, APS and PAPS. The APR activity of PpAPR-B was saturated at 2.5 μm thioredoxin, but PAPS-dependent sulfite production was still increasing. On the other hand, PpAPR activity was not affected by thioredoxin. This was in line with our expectations, as the protein already contains the thioredoxin-like domain necessary for electron transfer (
Additional Characteristics of PpAPR and PpAPR-B—The pH dependence of APS reduction by PpAPR was very similar to that of AtAPR2, with low activity at pH ≤ 7 and higher activity at pH ≥ 8.5 (Fig. 7A). PpAPR-B activity in terms of APS reduction increased gradually and less markedly with increasing pH. On the other hand, the PAPS-dependent activity of PpAPR-B showed a clear optimum at pH 8 (Fig. 7B). To test whether PpAPR or AtAPR might be capable of PAPS reduction at a different pH than in the standard assay, we also measured this reaction at varying pH values. However, the two enzymes did not show any indication of PAPR activity. As a control for PAPR, an extract of E. coli was analyzed in the same way and showed a pH optimum between pH 8 and 8.5 with a slower decrease of activity at more alkaline pH (Fig. 7B).
Plant APRs are very unstable enzymes because of the rapid loss of their iron-sulfur clusters (
). Indeed, PpAPR and AtAPR2 were completely inactivated after 1 day at -20 °C. In contrast, PpAPR-B retained 30% of its activity after the same storage time (data not shown). When kept at 4 °C, AtAPR2 started to lose activity after 24 h and was completely inactivated after 4 days. PpAPR did not lose any activity for 48 h, but the enzyme was inactivated afterward and retained only ∼15% of the initial activity at days 4 and 5. On the other hand, PpAPR-B was stable over 5 days (data not shown). The two APR isoforms of P. patens differed also in their temperature dependence. Whereas 37 °C was the optimal temperature for both of them, PpAPR was slightly more active at lower temperature, but was sensitive to high temperature, losing 70% of its activity at 45 °C (data not shown). PpAPR-B and AtAPR2 retained full enzyme activity at 45 °C.
APR activity has been shown previously to be stimulated by sulfate and other salts (
). Congruently, the activities of AtAPR2 and PpAPR were increased by 18- and 24-fold in the presence of 500 mm MgSO4 compared with measurements in 50 mm Tris-HCl, dithioerythritol, and APS without any other salt addition (Fig. 8). PpAPR (but not AtAPR2) was also highly stimulated by (NH4)2SO4. The addition of 500 mm MgCl2 and KNO3 resulted in lower APR activity than without any salt. PpAPR-B was stimulated by all four salts used in the experiment, although the level of induction by sulfate was not as high as for PpAPR (Fig. 8). In contrast, none of the salts at 0.5 m stimulated the activity of EcPAPR.
Modeling of the APS/PAPS-binding Pocket—APS and PAPS differ only by the presence of a phosphate group. According to sequence analysis, PpAPR-B was predicted to catalyze PAPS reduction. However, the enzyme preferentially reacted with APS. Because the crystal structures of EcPAPR and PaAPR are known (
), we aimed to identify which amino acid changes led to the different substrate specificity. Fig. 9 shows the comparison of the sulfonucleotide-binding pockets of PpAPR-B, PaAPR, and EcPAPR with bound substrates. Evidently, the overall structures of the binding sites in all three proteins are very similar. The model predicts a close proximity of Asp85 and Arg187 of PpAPR-B to the 3′-hydroxyl of ribose carrying the phospho group in PAPS (Fig. 9A). A similar configuration with two large and charged amino acid residues (Asp and Gln) is present in the APS-binding pocket of PaAPR (Fig. 9B). On the other hand, in EcPAPR, the phosphate seems to be accommodated because of substitutions of Asp85 by Gly and Arg187 by Leu (Fig. 9C).
PpAPR-B Homologs in Other Plant Species—Because the PpAPR-B gene has higher sequence similarity to the genes from fungi and bacteria than to those from plants, it was possible to speculate that its presence in Physcomitrella is a result of a recent horizontal gene transfer (
). However, expressed sequence tags from S. lepidophylla (lycophyte) and M. polymorpha (liverwort) encoding PpAPR-B orthologs were also found in data bases. More important, the genome of another lycophyte species, S. moellendorffii, is being sequenced, and the trace files are available in the public data base. Sequence analysis revealed that S. moellendorffii possesses a single copy ortholog of PpAPR-B and a single gene for the plant-like APR. The four lower plant PpAPR-B homologs are highly conserved at the amino acid level, and in phylogenetic analysis, they cluster together (Fig. 2), revealing that they have a common evolutionary origin. Indeed, when expressed in E. coli, S. lepidophylla APR-B was active as APR, similar to PpAPR-B (data not shown). Therefore, it seems that, sometime during evolution, the substrate specificity of the original PAPR was altered to APS. Because lycophytes are already vascular plants, the common ancestor of bryophytes, liverworts, and vascular plants had to possess both genes. Interestingly, however, APR from S. moellendorffii does not possess the invariant cysteine residue in the active center, which is conserved among all APRs and PAPRs. Because this cysteine is essential for catalysis, the gene product will most probably not be functional. This and other nucleotide substitutions result in branching of the S. moellendorffii APR before the node separating plants and green algae (Fig. 2).
It has been generally adopted that PAPRs exist in fungi and bacteria only, whereas plants reduce APS directly (
). The putative PAPR from P. patens (PpAPR-B) is in many respects similar to bacterial and fungal PAPRs. Its amino acid sequence is 26.6 and 25.7% identical to PAPRs from E. coli and S. cerevisiae, but only 20.6 and 20.3% identical to APRs from A. thaliana and P. patens. In the phylogenetic analysis, it clearly clusters with other PAPRs from bacteria and fungi (Fig. 2). More important, it does not possess the two Cys pairs that bind the iron-sulfur cluster, which have been considered as markers for APS-dependent reduction (
). In addition, the enzyme does not possess the thioredoxin-like domain found in plant APRs. Therefore, the enzyme was clearly expected to catalyze thioredoxin-dependent sulfite production from PAPS. Indeed, sulfite production from PAPS dependent on the substrate concentration was measured (Fig. 5B), and the reaction was strongly stimulated by thioredoxin (Fig. 6C). The affinity for PAPS (Km = 20 μm) (Table 1) was highly comparable with that of the E. coli and yeast enzymes, with Km values of 22.5 and 19 μm, respectively (
). The latter value is even lower than the Vmax of PpAPR-B in the PAPR reaction (1.4 milliunits/mg), which was also determined in a heterologous system using E. coli thioredoxin. Therefore, it can be expected that the reaction velocity with homologous thioredoxin will be ∼10-fold higher, in the range of 14 milliunits/mg. In contrast, the Vmax of EcPAPR is ∼500-fold higher (5.1-6.6 units/mg). PpAPR-B thus indeed seems to act as a PAPR.
However, testing the ability of PpAPR-B to catalyze APS reduction revealed that, with this substrate, the enzyme is much more active. Sulfite production from APS was ∼50,000-fold higher than that from PAPS and was also dependent on thioredoxin. In competition assays in which unlabeled APS or PAPS was added to radioactive PAPS or APS, respectively, PAPS was not able to compete with [35S]APS and to reduce the APS reduction rate. On the other hand, unlabeled APS reduced the production of [35S]sulfite from [35S]PAPS, showing that the enzyme acts primarily as an APR. Among the plant APRs, PpAPR-B is the only bifunctional one because other tested enzymes (PpAPR and AtAPR2) were not capable of PAPS reduction. Dual substrate specificity for both APS and PAPS was described, however, for the CysH protein from B. subtilis (
). In contrast to PpAPR-B, which catalyzes APS reduction ∼100 times more efficiently (cf. Kcat/Km for APS and PAPS reduction), only an ∼5-fold difference in Kcat/Km between the two substrates was found for the B. subtilis enzyme. The Kcat was comparable between APS and PAPS reduction, and so the difference in efficiency was driven by a higher affinity of the protein for PAPS (
). The PAPR activity of PpAPR-B thus seems to be too low compared with the rate of APS reduction to merit the enzyme to be called bifunctional. The enzyme was thus named PpAPR-B to avoid confusion with the numbering of APR isoforms of higher plants, e.g. APR1, APR2, and APR3 of A. thaliana.
These findings thus contradict the accepted view that APR activity is linked to the presence of an iron-sulfur cluster (
). Indeed, all sulfonucleotide reductases using APS analyzed to date possess the cofactor, including dissimilatory APR, plant APR with the thioredoxin-like domain, and bacterial APR. In contrast, no FeS clusters were found in PAPRs from fungi, yeast, and enteric bacteria, except for the bifunctional enzyme from B. subtilis (
). PpAPR-B does not possess the FeS cluster, as demonstrated by UV-visible spectroscopy, and consequently, no iron bound to the protein could be measured. The enzyme thus represents a new form of APR independent of the FeS chemistry. The cost of the lack of cofactor is substantial, as on the basis of the Kcat/Km, PpAPR-B is 1770-fold less efficient than PpAPR. At physiological APS concentrations <10 μm, the reaction velocity of PpAPR is 2500 times higher than that of PpAPR-B. Why there is such a difference in activity between the two proteins is not obvious. The low reaction velocity of PpAPR-B explains, however, why we were not able to detect APR or PAPR activity in crude extracts of moss knock-outs in PpAPR (
). In addition, the two isoforms differ in their optimal reaction conditions and in that PpAPR-B requires the addition of thioredoxin. If both enzymes are present at the same level in moss, the in vitro activity of PpAPR-B contributes <0.1% of the total activity, which is under the detection limit of the assay. Despite the much lower reaction velocity, however, PpAPR-B is clearly able to compensate for the lack of APR in ΔAPR (PpAPR knock-out) plants in vivo (
What could be the evolutionary advantages of this novel APR isoform? An additional PAPR could be able to salvage sulfur from PAPS otherwise lost in secondary metabolism. As PpAPR-B is also capable of this reaction, this could be one of the potential benefits to retain an alternative enzyme to PpAPR. The major advantage of PpAPR-B compared with PpAPR is, however, its high stability. Plant APRs are rapidly inactivated and have a quick turnover. Possessing a second, stable isoform of the enzyme lowers the need for a constant replacement of the enzyme. In addition, the new isoform of APR does not need the FeS cluster and thus is advantageous under iron-limiting conditions, e.g. in aquatic habitats of the mosses. This could also be achieved with PAPR, but the consequence of the change in the active center is saving one ATP/molecule of APS reduced by the new enzyme and thus a substantial saving of energy. Consequently, PpAPR-B contributes to the huge pool of metabolic moss genes that are absent in seed plants (
). However, it seems that APS reduction is the only route of sulfate assimilation in both seed and basal plants.
As PpAPR-B evidently evolved from a PAPR by changing the substrate-binding site, it was interesting to find out which amino acid changes led to the alteration of substrate specificity. The two sulfonucleotides APS and PAPS differ only by the presence of a phosphate group modifying the 3′-residue of ribose. A similar modification distinguishes NAD+ and NADP+ dinucleotide cofactors. A great effort was put into understanding the differences in the binding of the two cofactors and engineering the corresponding binding pockets (
). These residues are invariant in >30 plant APRs and are present in all confirmed bacterial APRs. The bifunctional APR/PAPR from B. subtilis has Gly in place of Asp66, probably to enable binding of both substrates. Some other bacterial enzymes that possess the two Cys pairs and were therefore assumed to be APRs possess Ser instead of Asp66 or Asp instead of Glu65. EcPAPR possesses two small amino acids in place of Glu65 and Asp66, i.e. Gln and Ala, respectively. The sequence of PpAPR-B and the orthologs in Selaginella and Marchantia is Gly84-Asp85. Therefore, these enzymes are more similar to the plant/bacterial APRs than to PAPRs. Homology modeling confirmed the higher similarity of the binding pocket of PpAPR-B to that of PaAPR compared with EcPAPR. It also revealed a close proximity of Asp85 to the 3′-hydroxyl of ribose, which may be responsible for the alteration of the binding pocket to accommodate APS instead of PAPS. However, a second residue, Arg187, also seems to interact with the 3′-OH group. A small uncharged amino acid (Leu) is at this position in most bacterial PAPRs, so it seems that also this residue is involved in the determination of substrate specificity. Site-directed mutagenesis experiments will be necessary to show whether these amino acid residues are indeed implicated in the distinction between APR and PAPR. The change in substrate specificity resulted, however, in a reduction in catalytic efficiency compared with EcPAPR, which has a 50-fold higher Kcat/Km compared with PpAPR-B. Our current findings thus show that APS reduction without the FeS cluster is possible and that the usefulness of the two cysteine pairs as markers to distinguish APS- and PAPS-dependent sulfate reduction has to be re-appraised.
We thank Andreas Zimmer for introduction to programs on the prediction of subcellular localization of proteins, Louis Gremillon for help with confocal microscopy, Michael McMa-nus for critical reading of the manuscript, and Anne Katrin Prowse for help with editing the manuscript.