Cloning and characterization of the P subunit of glycine decarboxylase from pea (Pisum sativum).

A pea leaf cDNA library constructed in lambda gt11 was screened with an antibody raised to the P subunit of glycine decarboxylase. One of the positive clones isolated was sequenced and shown to contain an open reading frame, which encoded the entire P subunit polypeptide. Aligning the deduced amino acid sequence with the amino acid sequence determined directly from the NH2 terminus of the mature P subunit shows the presence of a putative 86 amino acid leader sequence, presumably required for import into the mitochondria, and gives a Mr of the mature protein of 105,000. Comparison of this deduced amino acid sequence with the sequence of a pyridoxal phosphate-containing peptide isolated from the P subunit of chicken liver glycine decarboxylase shows remarkable conservation. The P subunit, however, shows little sequence homology with other published amino acid decarboxylases. Expression of the P subunit mRNA shows a pattern very similar to that of the corresponding polypeptide: it is strongly light induced and is expressed at a much higher level in leaves than in other tissues. Southern blot analysis suggests that the P subunit is encoded by a small multigene family.

In both animal and plant mitochondria glycine decarboxylase (GDC)' catalyzes the oxidation of glycine to form COz, NH3, NADH, and 5,10-methylenetetrahydropteroyl-~-glutamic acid (Motokawa and Kikuchi, 1974;Clandinin and Cossins, 1975). The enzyme is composed of four subunits: P (containing pyridoxal phosphate), H (a carrier protein containing lipoamide), T (a transferase responsible for producing 5,10-methylenetetrahydopteroyl-~-glutamic acid), and L (lipoamide dehydrogenase required to complete the cycle). In plants GDC is responsible for the decarboxylation step of the photorespiratory pathway in which COz is lost from the plant as a result of the oxygenase reaction of ribulose bisphosphate carboxylase/oxygenase. Consequently, GDC is present in very large amounts in the mitochondria of light-grown pea leaves, where it may constitute 30-50% of the matrix protein (Bour-* This work was funded by the Cambridge Laboratory. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. The nucleotide sequence(s) reported in this paper hos been submitted x59773.
to the GenBankTM/EMBL Data Bank with accession number(s) §To whom correspondence and requests for reprints should be addressed.
The abbreviations used are: GDC, glycine decarboxylase complex; SDS, sodium dodecyl sulfate; HEPES, 4-(2-hydroxyethyl)-l-piperazineethanesulfonic acid. guignon et Walker and Oliver, 1986a). The presence of GDC appears to be light dependent since the amounts of the polypeptides and the activity of the enzyme increase dramatically upon exposure of etiolated seedlings to light (Day et al., 1985;Walker and Oliver, 198613).
All four subunits of GDC have been individually isolated from pea (Walker and Oliver, 1986a;Bourguignon et al., 1988) and all four are required for maximum enzyme activity (Walker and Oliver, 1986a). Although the subunits are normally purified independently of one another, certain subunits, such as P and H, demonstrate an association under some conditions (Hiraga and Kikuchi, 1980a;Oliver et al., 1990). In addition, Neuberger et al. (1986) have shown that at high concentrations of mitochondrial matrix protein all four subunits may be retained behind an XM 300 ultrafiltration membrane whereas the subunits applied individually are not . This suggests that under appropriate conditions the enzyme exists as a complex. Oliver et al. (1990) have recently shown that this weakly bound complex has a stable stoichiometry of 1 L subunit dimer/2 P subunit dimers/ 27 H subunit monomers/9 T subunit monomers.
The initial step in the glycine cleavage reaction is the formation of a Schiff base between the carbonyl group of P subunit pyridoxal phosphate and glycine with the release of COz. Although this step only occurs at very low rates in the absence of the H subunit (Hiraga and Kikuchi, 1980b;Walker and Oliver, 1986a), the P subunit is considered the true glycine decarboxylase (Hiraga and Kikuchi, 1980b). Much information is now available on the primary structure of the H subunit from chicken and bovine liver (Fujiwara et al., 1986(Fujiwara et al., , 1990 and from pea (Kim and Oliver, 1990;Macherel et al., 1990).
In addition, the sequence of the T subunit from bovine liver has recently been published (Okamura-Ikeda et al., 1991). Studies on the structure of the P subunit and its interaction with the other subunits of the enzyme have been hampered by the availability of only the amino acid sequence of a small pyridoxal phosphate-binding peptide from the protein purified from chicken liver (Fujiwara et al., 1987). However, Kume et al. (1991) recently reported the cloning of cDNAs encoding the P subunit from chicken and human, and here we report for the first time the cloning and characterization of a fulllength cDNA clone for the P subunit of GDC from plants and the deduced amino acid sequence of the complete polypeptide.

EXPERIMENTAL PROCEDURES
Plant Material-Pisum sativum cultivar "Birte" was grown under natural day length in a greenhouse with a 20 'C day and 16 "C night temperature in seedtrays containing vermiculite after surface sterilizing the seeds in 10% (v/v) commercial bleach solution. Plants were harvested within 2 weeks except for those required for mRNA extraction from mature green leaves and embryos, which were grown in pots containing a mixture of John Innes No. 1 compost and grit.

5355
Etiolated seedlings were grown for 6 days in the dark at 20 "C before exposure to continuous light at 20 "C for various periods of time (100 pmol quanta of photosynthetically active radiation r n -' s-l). Plants for dark treatment were grown in the glasshouse for 3 weeks prior to the 60-h continuous dark period.
Protein Purification and Antibody Production-The P subunit was purified from an acetone precipitate (Walker and Oliver, 1986a) of pea leaf mitochondria prepared by the method of Day et al. (1985) by running a 5040% saturated ammonium sulfate fraction of the resuspended matrix proteins on a 12% SDS-polyacrylamide gel (Laemmli, 1970) and electroeluting the very prominent band of M, 97,000 which has previously been identified as the P subunit (Walker and Oliver, 1986a). H protein was gel-purified in the same way from the 75-100% ammonium sulfate fraction (Walker and Oliver, 1986a). Electroeluted proteins were freeze-dried and 300 pg of P and 100 pg of H were injected subcutaneously into separate rabbits after taking samples of preimmune serum. The proteins were equally divided between three injections made at 10-day intervals, the first using Freund's complete adjuvant and the others incomplete adjuvant. Serum was collected at 10-day intervals after the third injection, and IgG fractions were purified using a protein A-Sepharose column (Ey et al., 1978).
Protein Sequence Analysis-The NH2-terminal protein sequence was determined using an Applied Biosytems (Warrington, Cheshire, United Kingdom (U. K.) 470A gas-phase protein sequencer equipped with a 120A on-line phenylthiohydantoin analyzer. The P protein bands cut from Immobilon blots of low pH SDS-polyacrylamide gels (Moos et al., 1988) containing the 50-60% ammonium sulfate fraction (see above) and stained with Brilliant Blue R were placed directly onto the sequencer reaction cartridge and held in place with trifluoroacetic acid-etched glass fiber discs (Hunkapillar et al., 1983).
Antibody Characterization and GDC Assays-SDS-polyacrylamide gels were electroblotted onto nitrocellulose and development of Western blots was essentially as described by Blake (1984) but including a 1 M salt wash after incubation with the primary and secondary antibodies. The ability of the anti-P IgG to inhibit the activity of glycine decarboxylase was assessed by immunoprecipitation of the P subunit and assaying the supernatant for the ability to decarboxylate [ l-14C]glycine. Immunoprecipitations were carried out using matrix proteins prepared according to Neuberger et al. (1986). Incubations contained 20 mM HEPES, pH 7.5, 1 mM dithiothreitol, 100 p M pyridoxal phosphate, 3.5 mg of protein A-Sepharose, 4.3 pg of mitochondrial matrix protein, and 270 pg of bovine serum albumin or preimmune and immune IgG in various ratios in a total volume of 275 pl. Samples were incubated at room temperature for 45 min, before adding protein A-Sepharose. They were then mixed end over end for a further 30 min, spun at 10,000 X g for 2 min, and 230 p1 of the supernatant assayed for enzyme activity (Walker and Oliver, 1986a).
Nucleic Acid Extraction and Analysis-RNA was purified by the method of De Vries et al. (1988) and DNA purified from the same preparation by ethanolic precipitation from the supernatant remaining after LiCl precipitation of the RNA and subsequent banding on a caesium chloride/ethidium bromide density gradient. Poly(A)-containing RNA was purified on oligo(dT)-cellulose. RNA was size fractionated on agarose gels containing formaldehyde according to standard techniques. Hybond N+ was used with NaOH for Southern and Northern blots according to the manufacturer's instructions (Amersham International, Amersham, U. K.). Filters for RNA dotblot analysis were prepared using the method of White and Bancroft (1982), each sample was split between two replica dots, and each preparation was applied at 2.5 and 5 pgldot. Hybridization of cDNA clones to dot-blots was standardized by expressing the results relative to hybridization to the rDNA. Hybridization to the rDNA was measured by probing dot-blots of the same samples (four replicas) with the pea ribosomal DNA cosmid clone cDB107, the standard deviation was less than 18% of the mean, and no significant difference was observed between means. Hybridizations were performed according to Domoney and Casey (1983) and the filters washed in 0.2 X SSC, 0.1% SDS and exposed to Kodak XAR5 x-ray film. Hybridization was quantified by scanning the autoradiogragh using a densitometer. Results are presented as the average of readings made at the two different loadings of RNA.
Restriction enzyme digestions of pea genomic DNA were performed according the instructions of the enzyme manufacturers, and the digestions were judged to be greater than 98% complete by observing the pattern given with hybridization to the rDNA.
cDNA Cloning and Analysis-cDNA was synthesized using a cDNA synthesis kit (Amersham International, Amersham, U. K.) according to the manufacturer's instructions from 5 pg of poly(A)-containing RNA purified from mature leaves of varying size. Following screening with the P subunit antibody, positive clones (15) were subcloned into the KpnI site of Bluescript. Sequencing was performed on doublestranded plasmid DNA (Murphy and Kavanagh, 1988) following sequential exonuclease III/mung bean nuclease digestion of the insert (Fig. 1). The entire clone was sequenced separately on each strand. The H subunit clone (pGDH1) was isolated in a similar manner to the P subunit clone and was positively identified on the basis that the sequence of 200 nucleotides at the 5' end were identical to those described by Kim and Oliver (1990).

RESULTS
Characterization of the P Subunit Antibody-The antibody was judged to be monospecific on the basis that it gave only one band of M, 97,000 when used to probe Western blots of SDS-polyacrylamide of total pea leaf protein extract (see Fig.  4). The antibody was tested for its ability to immunoprecipitate the P subunit by measuring the inhibition of the GDC activity of a mitochondrial extract following immunoprecipitation of the P subunit. The results (Fig. 2) show that over 90% of GDC activity could be inhibited by the antibody, consistent with the antibody recognizing the P subunit of GDC.
Identification of a P Subunit cDNA Clone-The P protein antibody was used to screen lo5 plaques from a pea leaf cDNA library constructed in Xgtll. Positive clones (15) were isolated and one of these clones, which contained an insert of 3.4 kilobases, was subcloned into a plasmid vector for sequencing and named pGDP1. The sequence (Fig. 3) contained an open reading frame of 3171 nucleotides, and the amino acid sequence deduced from this included a sequence matching exactly the amino acid sequence that we had determined directly from the NH2 terminus of the mature protein. upstream methionine residue, prior to an in frame stop codon, lies 86 amino acids upstream of the mature NH, terminus and is assumed to be the translation start site. The 86 amino acid leader sequence shows many characteristics of a mitochondrial targeting sequence: rich in serine (20%) and arginine (15%) (von Heinje, 1985), with the regions of hydroxylated amino acids (including one run of 8 consecutive hydroxylated residues) interspersed with positively charged amino acids (Rosie et al., 1986) and an arginine residue at -2 relative to the cleavage site (Hendrick et al., 1989).
The DNA sequence shows the presence of a 3"untranslated region of 140 nucleotides downstream of two consecutive translation stop codons. Like many plant genes, there does not appear to be an AATAAA polyadenylation signal anywhere in this sequence. The most likely signal is the GATAAT motif located 13 nucleotides upstream from the poly(A) tail (Joshi, 1987).

Expression and Organization of the P Protein
Genes-In order to examine not only the level of P protein mRNA expression but also that of other GDC subunits, all measurements of P subunit mRNA are accompanied by measurements of H protein mRNA levels on the same samples. In addition, to correlate mRNA levels with protein levels, Western blots of protein extracts made from the same plant material as used for the mRNA extractions were probed with the P and H subunit antibodies. Glycine oxidation occurs at a much faster rate in mitochondria isolated from leaves compared to those from other tissues (Gardestrom et al., 1980). We therefore examined the tissue-specific pattern of the GDC mRNA expression by probing Northern blots containing poly(A)+ mRNA from pea leaves, roots, and embryos with pGDP1 and pGDH1. The highest levels of mRNA expression are seen in the leaves, although for the P subunit in particular, a significant amount of mRNA can be found in other tissues (Fig. 4). In addition, for both P and H there is a strong correlation between the level of the mRNA and the level of the corresponding protein (Fig. 4).
Since it is known that the level of GDC in the mitochondria of etiolated seedlings increases dramatically on exposure to light (Day et al., 1985;Walker and Oliver, 1986b) and H subunit mRNA has previously been shown to be light inducible (Macherel et al., 1990;Kim and Oliver, 1990), the effect of light on the expression of the P subunit mRNA was examined. In order to standardize the results, mRNA levels were examined in total RNA and since the P subunit mRNA (3.7 kilobases) is masked on Northern blots by the very large Results are normalized by expressing the hybridization relative to the hybridization to the ribosomal RNA. The corresponding protein levels are shown as Western blots (loaded with 15 pg of total protein/track) probed with the P and H subunit antibodies. amount of ribosomal RNA, measurements of RNA levels were performed using dot-blots. In the first experiment (Fig. 5), the exposure of etiolated seedlings to light produced a large increase in the level of P subunit mRNA and protein. The level of P subunit mRNA increased to 90% of its maximum level after exposure to light for 6 h. This corresponded to an increase of greater than 4-fold compared to dark grown levels. This result has been confirmed by Kim et al. (1991) who have found a similar result. After the same time in the light, H subunit mRNA had reached less than 10% of its maximum level. Similar to the results seen for tissue specificity, the level of the P and H subunits appeared to closely follow the level of the corresponding mRNA (Fig. 5). Over the time course described above, light not only causes the induction of proteins, but also a large amount of leaf development, which may also contribute to the changes in gene expression observed. To try and separate these two factors, mature green plants were put in the dark for 60 h then re-exposed to light. After 60 h in the dark, the mRNA for both the P and H subunits had declined to a fraction (<2%) of their before-darkenening levels (Fig. 6). After returning the plants to the light for 24 h, however, the mRNA level had returned to, or in the case of P subunit mRNA far exceeded, those of the level prior to dark treatment. Over the time course of the experiment, the Western blots indicated that there was very little change in the P or H subunit levels ( Fig. 6).
In order to determine the copy number of the genes encoding the P subunit mRNA, Southern blots of pea genomic DNA as well as plasmid DNA markers were probed with pGDP1.
The P subunit appears to be encoded by a small multigene family (Fig. 7) with approximately two genes for the P subunit/haploid genome in pea.

DISCUSSION
The clone (pGDP1) isolated contains an open reading frame which encodes the polypeptide sequence of the entire P sub-  I I I I I I I I I I I I I I I I I I I I I I I I I  unit of GDC. The deduced amino acid sequence together with our NHz-terminal amino acid data suggests that the mature protein contains 971 amino acids with a M, of 105,000. This is larger than previous estimates of 97,000-98,000 (Bourguignon et al., 1988;Walker and Oliver, 1986a) made from gel electrophoresis. The difference may be accounted for by the overall negative charge of the P subunit causing it to run anomalously on SDS-polyacrylamide gels. In addition, there is an unusually long 86 amino acid leader sequence, which is presumably for targeting the protein to the mitochondria. Long mitochondrial targeting sequences are generally associated with proteins which require further intramitochondrial sorting (see Schatz, 1987, for review). Available evidence, however, suggests that the P subunit is located in the mitochondrial matrix (Neuburger et al., 1986). The significance of the large targeting sequence and any special function it may perform awaits further functional analysis.
During the induction of the P subunit following the exposure of etiolated seedlings to light, the level of the protein closely follows that of the corresponding mRNA: the same result is seen for the H subunit (see also Macherel et al., 1990;Kim and Oliver, 1990), suggesting that under these conditions the protein level is controlled primarily by the level of corresponding mRNA. Following a period of 60 h of continuous darkness, the level of P subunit mRNA in the leaves of mature plants dropped sharply. The corresponding protein, however, showed relatively little change, implying that the P subunit is a stable protein, and little of it is turned over during this period. In comparison to the expression of the P subunit during light induction, the expression of the H subunit appears to show a lag and, although reasonable quantities of the P subunit mRNA are present in tissues other than leaves (roots or embryos), H subunit mRNA only appears at comparatively low levels. In the mitochondria of mature leaves, GDC exists as a stoichiometric complex and this stoichiometry presumably reflects that of the active enzyme. The reason why in other tissues, or at other stages of development/ greening, GDC subunits appear to accumulate with different stoichiometry is unclear. A relationship between the levels of P and H subunit is implied from the work of Blackwell et al. (1990), who have characterized a GDC mutant from barley which shows reduced levels of the P and H subunits but apparently normal levels of the T and L subunits. Given that the P subunit is apparently stable in the absence of normal levels of the H protein, it appears that either the level of the H subunit is somehow governed by the level of the P subunit or that though the P and H genes show subtly different patterns of expression, they share a common regulatory factor. Comparison of the deduced amino acid sequence shown here with the amino acid sequence of a pyridoxal phosphatecontaining peptide from the GDC P subunit of chicken liver (Fujiwara et al., 1987) shows remarkable similarity (Fig. 8). Some 46 of the 54 amino acids are identical including a run of 25 consecutive amino acids, which includes a region rich in glycine just to the carboxyl side of the site of attachment of the pyridoxal phosphate. Given the high degree of sequence conservation it is probable that Lys7" in our sequence corresponds to the site of attachment of the pyridoxal phosphate. In common with all other characterized amino acid decarboxylases, this lysine is preceded by a histidine (Vaaler and Snell, 1989). The hydrophobic glycine-rich region is conserved in several amino acid decarboxylases and has been suggested to be involved in substrate binding (Moore and Boyle, 1990). In addition, Fujiwara et al. (1987) suggest that the large number of glycine residues are required to give the steric freedom required for access of the lipoyl moiety of the H subunit to the active site. Alignment of the complete amino acid sequences for the mature proteins of human and chicken (Kume et al., 1991) with that of pea shows that the similarity between them extends throughout the protein sequence. The amino acid sequence of the mature Pisum P subunit is 56% identical and 73% similar to the mature sequences of the human and chicken proteins which are more closely related (91% similar and 83% identical; Kume et al., 1991). Two notable regions of identity exist between all three proteins. The first is from Gln405 to Met433, and the second is from His787 to Pros1' in the Pisum protein. The second domain includes the lysine residue which binds the pyridoxal phosphate and the glycine-rich region which have been identified in the human and chicken proteins (Kume et al., 1991). Such strong conservation between unrelated species clearly suggests involvement of these domains in either the catalytic activity or in the interaction between subunits of the glycine decarboxylase complex. Apart from those mentioned above, the P subunit of GDC shows no obvious sequence similarity to any other characterized amino acid decarboxylases.
The P subunit of GDC shows several unique catalytic properties, such as its ability to catalyze the exchange of the glycine carboxyl carbon with CO, (Hiraga and Kikuchi, 1980b). It is also characterized by its large size (Mr 105,000) when compared to other amino acid decarboxylases (e.g. M, of 56,000 for tryptophan decarboxylase from Catharanthus roseus (De Luca et al., 1989)). Interestingly, bacterial GDC in which the polypeptide size and composition of the P subunit have been determined exists as a a& tetramer containing polypeptides of M, 54,000-63,000 (Freudenberg and Andreesen, 1989;Gariboldi and Drake, 1984). Comparison of the sequence of a bacterial enzyme with that of the P subunit from Pisum may shed some light on the evolution of the eukaryotic GDC P subunit.