Cytokinin oxidase from Zea mays: purification, cDNA cloning and expression in moss protoplasts

Authors

  • Nicole Houba-Hérin,

    1. 1 Laboratoire de Biologie Cellulaire INRA, Route de S t-Cyr, 78026 Versailles Cedex, France, and2Laboratoire de Microséquençage des Protéines, Institut Pasteur, 25, rue du Dr Roux, 75724 Paris Cedex 15, France
    Search for more papers by this author
  • 1 , Claude Pethe,

    1. 1 Laboratoire de Biologie Cellulaire INRA, Route de S t-Cyr, 78026 Versailles Cedex, France, and2Laboratoire de Microséquençage des Protéines, Institut Pasteur, 25, rue du Dr Roux, 75724 Paris Cedex 15, France
    Search for more papers by this author
  • 1 , Jacques D’Alayer,

    1. 1 Laboratoire de Biologie Cellulaire INRA, Route de S t-Cyr, 78026 Versailles Cedex, France, and2Laboratoire de Microséquençage des Protéines, Institut Pasteur, 25, rue du Dr Roux, 75724 Paris Cedex 15, France
    Search for more papers by this author
  • and 2 Michel Laloue 1 ,

    1. 1 Laboratoire de Biologie Cellulaire INRA, Route de S t-Cyr, 78026 Versailles Cedex, France, and2Laboratoire de Microséquençage des Protéines, Institut Pasteur, 25, rue du Dr Roux, 75724 Paris Cedex 15, France
    Search for more papers by this author

  • EMBL Data Library accession number Y18377.

*For correspondence (fax +33 130833099; e-mail laloue@versailles.inra.fr ).
†These authors contributed equally to the work presented here.

Summary

Cytokinins are degraded by cytokinin oxidases (CKOs) which catalyse cleavage of theN6-(isopent-2-enyl)-side chain resulting in formation of adenine-type compounds. CKO activity has been recorded in many plants and is thought to play a key role in controlling cytokinin levels in plants. Several partially purified CKOs have been characterised but no genes have been isolated yet. CKO activity is known to be inhibited by phenylureas, cytokinin agonists. We used 1-(2-azido-6-chloropyrid-4-yl)-3-(4-[3H])phenylurea ([3H]-azidoCPPU) to photolabel a glycosylated CKO from maize kernels. This enabled us to purify the enzyme. Peptide sequences were determined and the corresponding cDNA was cloned. The deduced amino acid sequence shares homology domains with FAD-dependent oxidases. An original assay based on transient expression of the enzyme in moss protoplasts allowed the functionality of the recombinant enzyme to be demonstrated.

Introduction

Cytokinins are hormones that control plant growth and development including growth of lateral buds, leaf expansion and delay of leaf senescence. Together with auxins, cytokinins are required for cell division and shoot formation in tissue culture ( Miller et al. 1956; Skoog & Miller 1957). Naturally occurring cytokinins are N6-substituted adenine derivatives and their metabolism in plants is well established ( McGaw & Burch 1995; for a review). However, the initial step(s) of their biosynthesis in plants are still unknown. A gene coding for the enzyme that adds an isoprenyl chain on the amino group of AMP leading to a cytokinin molecule has only been found so far in some phytopathogenic bacteria, like Agrobacterium and Pseudomonas ( Morris 1995).

In tobacco callus and cell suspensions, the active cytokinins are thought to be the free bases ( Hecht et al. 1975 and Laloue 1980, respectively). The base, riboside and nucleotide forms of cytokinins can be interconverted by purine metabolism enzymes ( Chen 1981). Some of the corresponding genes recently have been cloned and sequenced from Arabidopsis thaliana and the moss Physcomitrella patens ( Moffatt et al. 1992; Schnorr et al. 1996; von Schwartzenberg et al. 1998).

Irreversible or transient cytokinin inactivation can proceed through conjugate formation. N-glucosylation and alanine conjugation lead to biologically inactive and very stable cytokinins ( Mok & Martin 1994). Martin et al. (1997) recently cloned from Phaseolus vulgaris seeds the gene coding for the O-xylosyltransferase, an enzyme which forms O-xylosylzeatin from zeatin. The cytokinin-O-glycoside conjugates can be hydrolysed into more active cytokinins and are assumed to serve as storage products. Brzobohaty et al. (1993) demonstrated the release of free cytokinin from cytokinin-O-glucoside conjugates by a β-glucosidase encoded by a cloned Zea mays cDNA.

In addition, cytokinins can be irreversibly inactivated by cytokinin oxidase (CKO) which oxidatively cleaves the N6-side chain of cytokinin bases and ribosides to yield adenine and adenosine, respectively ( Whitty & Hall 1974). Cytokinin degradation indeed was observed to be a significant component of cytokinin metabolism in tissues of many plants supplied with radiolabelled cytokinins. In tobacco cells, exogeneously supplied N6-(Δ2-isopentenyl)adenosine (iPA) is readily catabolized with a half-life of 3 h ( Terrine & Laloue 1980). Therefore, changing the expression of CKO should be a very powerful tool for modulating cytokinin levels in plants. CKOs have been partially purified from three main plant sources: maize kernels ( Burch & Horgan 1989; Meilan & Morris 1994) or seedlings ( Burch & Horgan 1992; Schreiber et al. 1995), wheat germ ( Laloue & Fox 1989) and Phaseolus callus tissue ( Chatfield & Armstrong 1988). CKOs are generally described as glycoproteins. Highly diverse molecular weights were reported for CKOs in different plants (i.e. from 25.1 to 94 kDa) or even in the same plant ( Armstrong 1994; for a review). However, as yet, no genes have been isolated.

Some diphenylureas which are very potent cytokinin agonists have been shown to inhibit CKO activity ( Laloue & Fox 1985, 1989). This property has been largely confirmed and extended to other urea derivatives like thidiazuron ( Chatfield & Armstrong 1986). This inhibitory effect has been shown to be non-competitive ( Burch & Horgan 1989; Wang & Letham 1995).

We developed an azido-derivative of N-(2-chloro-4-pyridyl)-N′-phenylurea (CPPU) ( Dias et al. 1995) and showed that this probe can be used to photolabel and purify cytokinin-binding proteins ( Gonneau et al. 1998; Nogué et al. 1996). As expected, tritiated azidoCPPU was shown to photolabel CKO which was then purified by 2D-gel electrophoresis. We report here the cloning of a cDNA for maize CKO. Amino acid sequencing of internal peptides made it possible to design degenerate oligonucleotides. The cDNA was recovered and sequenced. The activity of the encoded protein was demonstrated using a transient expression assay in moss protoplasts.

Results

[3H]-azidoCPPU labelling of cytokinin oxidase (CKO)

Partial purification of a maize CKO.

CKO enrichment starting from 400 g of maize kernels is presented in Table 1. The initial purification procedure was adapted from Chatfield & Armstrong (1988) and Burch & Horgan (1989). After anion exchange chromatography the yield in active CKO was 2.3% with a purification factor of at least 160. However, no discrete peak corresponding to the activity was obtained indicating that the enzyme was not purified to homogeneity. Analysis by SDS–PAGE also confirmed that the active fraction contained a protein mixture (data not shown).

Table 1.  Purification of maize cytokinin oxidase
Steps
Activity
(nmol min–1) a
Volume
(ml)
Proteins
(mg)
Specific activity
(pmol mg–1 min–1)
Yield
(%)
Purification
fold
  1. a nmol of iPA degraded per min at 30°C (pH 7.5). Assays performed with [3H]-iPA 2 μm.

Crude extract126.08602236.056.3100.0
30–50% ammonium sulfate fraction56.0651105.050.744.0
ConA-Sepharose chromatography18.13815.71153.014.020.5
ResourceQ FPLC2.9280.39206.02.3163.5

Photolabelling of maize CKO.

The partially purified postconA Sepharose fraction of maize CKO was exposed to [3H]-azidoCPPU and irradiated for promoting covalent binding of the probe to affine proteins. SDS–PAGE analysis and fluorography then showed a radiolabelled band corresponding to a molecular weight of 63 kDa (data not shown). Competition experiments using substituted ureas like CPPU and purine-type cytokinins which are substrates of the enzyme led to a displacement of the photolabel (data not shown). The post-conA labelled fraction was analysed further by FPLC anion exchange and gel permeation chromatography, respectively ( Fig. 1a,b). In both cases, we noticed that the CKO activity peak was exactly superimposed on the radioactive peak. These cochromatography results indicated that the labelled protein most likely corresponded to CKO.

Figure 1.

CKO activity coincides with [3H]-azidoCPPU photolabelled proteins after FPLC anion exchange separation and gel permeation.

An aliquot of post-conA maize CKO fraction (activity 0.7 nmol min–1 ml–1) was photolabelled.

(a) A 2-ml aliquot was applied to a ResourceQ column which was eluted with a NaCl gradient. Fractions of 2 ml were collected.

(b) A 100-μl aliquot of the same fraction was applied to a TSKSW gel permeation column eluted at a flow rate of 1 ml min–1 and fractions of 250 μl were collected. In both cases, fractions were assayed for CKO activity and radioactivity.

Functional evidence that the labelled protein is CKO using a suicide substrate.

Suicide substrates are mechanism-based enzyme inactivators and acetylenic and/or allenic groups are classical enzymatically activable functional groups (for details, see Walsh 1982). HA8, an N6-substituted adenine with a C4 allenic chain (– CH2-CH = C = CH2), is the most active of a series of such suicide substrates which inhibit CKO activity in a time and concentration dependent fashion (C. Pethe, unpublished results). The design of these inhibitors was reasoned on the basis of the formation, in the active site, of an enimine in the isoprenic chain, as demonstrated with wheat CKO ( Laloue & Fox 1985). Thus oxidation of HA8 by the enzyme should lead to an alleneimine intermediate of greater nucleophilicity expected to allow the addition of nucleophilic groups of the enzymatic site.

As illustrated in Fig. 2, inactivation (87% after 2 h incubation) of maize CKO in the presence of HA8 resulted in a 77% reduction of the photolabelling (lane 2) as compared to control samples which had been incubated either without any substrate (lanes 1 and 4) or with N6-(Δ2-isopentenyl)adenine (iP) (lane 3). Thus, inhibition of the labelling is strictly related to the inactivation of the enzyme since HA8, without preincubation and at the final concentration of 0.1 μm, had a limited effect on CKO photolabelling (lane 4), as previously established in competition experiments (results not shown). iP was without any significant effect (lane 3), because it was entirely degraded by CKO during the course of the incubation.

Figure 2.

CKO photolabelling with [3H]-azidoCPPU is reduced after specific inactivation of the enzyme with the allenic suicide substrate HA8.

Four aliquots (0.4 ml) of a post-conA fraction of maize CKO were incubated at 30°C for 2 h: samples 2 and 3, in the presence of HA8 and iP 0.4 μm, respectively; samples 1 and 4, without any substrate. Then all samples were diluted fourfold with buffer. Each sample was then photolabelled with [3H]-azidoCPPU 0.2 μm after 5 min of further incubation at 4°C in the presence of the probe alone (samples 1–2–3) and the probe plus HA8 0.1 μm (sample 4). All samples were then analysed by SDS–PAGE and fluorography. Cytokinin oxidase activity of each diluted sample was measured in parallel. Upper and lower signals on the left lane correspond to the 69 and 46 kDa [14C]-labelled molecular weights. ND = not done.

Hence, these results show that the labelling of the protein with the probe [3H]-azidoCPPU is significantly reduced when the enzyme has previously been inactivated through a mechanism-based reaction with a specific suicide substrate. It can be concluded that the labelled protein indeed is CKO. This probe was then used to trace CKO through the last steps of its purification.

[3H]-azidoCPPU-mediated purification of maize CKO

Maize CKO photolabelled with [3H]-azidoCPPU after conA-Sepharose chromatography was then purified on a ResourceQ column as illustrated in Fig. 1(a). The labelled fractions were further characterised by 2D-gel electrophoresis. Fluorography revealed the presence of one major labelled spot of approximately 63 kDa ( Fig. 3b). Silver staining showed a major protein spot and minor spots having the same molecular weight of about 63 kDa but differing in pI ( Fig. 3c). The labelled spot corresponded to one of the minor spots and was weakly stained by Coomassie blue ( Fig. 3a) suggesting that CKO was still a minor component of the extract.

Figure 3.

Localization of maize photolabelled CKO by 2-D gel electrophoresis and fluorography.

Half of an 8-ml aliquot of the post-conA fraction was photolabelled with [3H]-azidoCPPU 0.5 μm and purified by anion exchange chromatography (as in Fig. 1a) and analysed by 2D-gel electrophoresis. The equivalent of 3.7 ml of the post-conA fraction was applied to the gel stained with Coomassie blue (panel a) and analysed by fluorography (panel b). The equivalent of 0.3 ml was also analysed and the gel silver stained (panel c). The remaining half of the photolabelled post-conA fraction purified by anion exchange chromatography was further treated with cold azidoCPPU 5 μm (3 rounds) and a 1/10th aliquot was analysed by 2D-gel electrophoresis and silver staining of the gel (panel d). Arrows indicate the position of the azidoCPPU-modified CKO.

At this stage, we considered the possibility that the electrophoretic behaviour of the azidoCPPU-modified enzyme could differ from that of the unmodified enzyme. This was indeed the case, as shown after protein staining of the 2D-electrophoresis gel ( Fig. 3d) of the post-ResourceQ fraction treated with 5 μm cold azidoCPPU (3 rounds). The spot corresponding to the radiolabelled protein and indicated by an arrow on the figure had indeed increased in intensity ( Fig. 3d) as compared to the untreated sample ( Fig. 3c).

We exploited this modification in the electrophoretic behaviour of the azidoCPPU-treated CKO to purify the enzyme. A total of seven preparative 2D-gels were run and stained with amido-black. The spots of interest were recovered, pooled and submitted to amino acid sequencing after endoproteolytic cleavage since the N-terminus of the protein was blocked. The digested sample amounted to about 6 μg protein. Four internal peptide sequences were obtained ( Fig. 4): WNRFVEMK (peptide 1); RLLSPGQDIFN (peptide 2); TYLARHTDRSDWVRHFGAAK (peptide 4) and GILQGTDIVGPLIVYPLNK (peptide 5).

Figure 4.

Nucleotide and deduced amino acid sequences of the CKO cDNA.

The nucleotide sequence is presented in the top line and the deduced one letter amino acid sequence is shown below. The sequenced peptides are boxed. ORF-flanking oligonucleotides are indicated by arrows. The initial 297 bp sequence corresponds to nucleotides 1289–1585. Amino acid (a.a) and nucleotide (nt) numbering is given at the left and the right side of the figure, respectively. (EMBL accession number is Y18377.)

Isolation of a CKO cDNA

Poly A+ RNA was prepared from maize cobs harvested about two weeks after anthesis. RT–PCR was performed by using degenerate oligonucleotides designed in both orientations from the four peptide sequences. Twelve oligonucleotide combinations were tested in ‘touchdown’ PCR to uncover all possible orders of the peptides. One combination resulted in an amplification product, i.e. a PCR reaction with primers designed from the sequences of peptides 5 and 1 (Co51 and Coa1) resulted in an approximately 300 bp amplified fragment by using as a template an aliquot of an initial PCR reaction performed with primers designed from the sequences of peptides 5 and 2 (Co51 and Coa23). This DNA fragment was cloned and four clones were sequenced. The 297 bp long sequences were all identical. The deduced amino acid sequence is bordered by the parts of peptides 5 and 1 which were used to design the oligonucleotides. This sequence also contains the C-terminus of peptide 5 and the whole of peptide 4 ( Fig. 4).

By using specific primers designed within this 297 bp long sequence, RACE PCR was performed on poly A+ RNA so that both 3′- and 5′-RACE reaction products would overlap over a 81 bp stretch. A CKs7-AP2 (approximately 400 bp) 3′-RACE reaction product and a CKa7-AP1 (approximately 1600 bp) 5′-RACE reaction product were recovered (AP1 and 2 are the Clontech anchored adaptors for RACE PCR). They were cloned and sequenced. At least two different clones and the two complementary strands contributed to the sequence determination for each RACE reaction product. The reconstructed nucleotide sequence is given in Fig. 4. Its linearity was confirmed by sequencing the independently amplified PCR product corresponding to the ORF region (see below). It has an open reading frame of 1602 bp that codes for a protein of 534 amino acids with a calculated molecular mass of 57.2 kDa which comprises all four peptide sequences. Since there is a stop codon upstream of the first ATG in all three reading frames, it is very likely that the ORF does not extend 5′ to the longest 5′-RACE clone found.

The deduced CKO amino acid sequence shows sequence similarity with a FAD-binding domain (a.a. approximately 170–240) found in several oxidases, suggesting that CKO is a FAD-dependent oxidase ( Fig. 5). A GHS motif is also found in the CKO sequence (a.a. 104–106). This motif is typical for those enzymes that covalently bind to FAD through a histidine residue. A high sequence identity (30%) was observed with a hypothetical 47.9 kDa oxidoreductase encoded by the FAS operon of Rhodococcus fascians and the best match (42% identity) was with a hypothetical protein of Arabidopsis thaliana (AC002510).

Figure 5.

Alignment of the CKO-deduced amino acid sequence with several FAD-dependent oxidases.

Shaded regions indicate identical (black) or similar (grey) amino acids. AC002510: hypothetical protein in Arabidopsis thaliana; FAS 5: hypothetical 47.9 oxidoreductase in Rhodococcus fascians (P46377); HDNO: Arthrobacter oxidans 6-hydroxy- d-nicotine oxidase (X05999); OX/RED: FAD-dependent oxidoreductase in Streptomyces coelicolor (AL009204); OXRTG: Rattus norvegicusl-gulonolactone oxidase (P10867)

Southern and Northern hybridisation analyses

A 0.6 kb long cDNA fragment (a 3′ segment downstream of the SacI site at position 1082 in Fig. 4) was used as a probe for Southern and Northern hybridisations. Under low and high stringency washing conditions, the probe hybridises to one major fragment of genomic DNA for all the enzymes tested ( Fig. 6a,b). It was checked by restriction analysis of PCR-amplified genomic DNA that these enzymes do not cut the gene, HindIII excluded, at least in the region extending between the translation start and stop codons (data not shown). Some weakly hybridising bands were observed in all lanes even after high stringency washing indicating the presence of related genes in the maize genome. This was also observed when using an internal 1.1 kb long NotI-BamHI cDNA fragment (bp 340 to bp 1464 in Fig. 4) (data not shown).

Figure 6.

Gene copy number and transcript size for the CKO gene. Maize var. Nobilis genomic DNA (8 μg) was digested with the indicated restriction endonucleases.

The Southern blot was hybridized with a radiolabelled 0.6 kb probe (bp 1082–1675 in Fig. 4). The membrane was washed either at low stringency (a) or at high stringency (b). RNA was purified from maize var. Nobilis cobs harvested 2 weeks after anthesis. Total RNA (10 μg, lane 1) and poly A+ RNA (1 μg, lane 2) were resolved by gel electrophoresis and the Northern blot (c) was hybridised with the 0.6 kb probe. DNA and RNA length markers are given at the left of each panel in kbp.

Total RNA and mRNA were purified from 2-week-old cobs. They were submitted to Northern analysis using the same probes. Figure 6(c) shows that the CKO mRNA is about 1.8 kb long, a size which is compatible with the longest reconstructed cDNA (1776 bp). The transcript is not abundant and is only clearly detected in a poly A+ RNA population.

Transient expression of the cloned CKO in moss protoplasts

As an expression system that should be appropriate for the production of a correctly folded glycosylated plant protein, we used a transient expression assay in Physcomitrella patens protoplasts. Moss protoplasts were chosen because they do not require exogenous auxins and cytokinins for growth and division in contrast to higher plant protoplasts. The ORF-corresponding cDNA region was amplified from poly A+ RNA with primers spanning the start and stop codons (CKs8 and CKa8) and cloned. One clone, whose sequence was identical to the sequence presented above, was subcloned into the expression vector pLBR19. Protoplast aliquots were transformed with pLBR19 carrying the CKO- or the β-glucuronidase-ORF (GUS-ORF), respectively. The GUS-ORF-transformed cells were used as a control for the CKO-ORF transformed cells and vice versa. Two days after transformation, CKO activity was measured and detected in the culture medium of the CKO-ORF transformed cells ( Table 2). Only a weak activity was recovered in extracts made from these CKO-ORF transformed cells (data not shown). Incubation of [3H]-iPA in the presence of the culture medium from the control samples for up to 2 h did not lead to any adenosine production. GUS activity was assayed in the cells as a measure for the transformation efficiency of the protoplast batches. These results show that a catalytically active maize CKO is expressed in the transformed moss protoplasts.

Table 2.  Transient expression of maize cytokinin oxidase in the culture medium of moss protoplasts
CKO activity (%) bGUS activity (AU min–1) c
Protoplast batch aGUS-ORFCKO-ORFGUS-ORFCKO-ORF
  • a

    Batches 1,2 and 3 consisted of one sample per construct; batch 4 consisted of three samples per construct.

  • b Assays performed with [3H]-iPA 0.05 μm for 30 min at 30°C. Results are given as the percentage of iPA conversion into adenosine.

  • c GUS activity is expressed in arbitrary units per min using 4-methylumbelliferyl-β- d-glucuronide as a substrate. Results are given for 3.6 × 105initially treated protoplasts.

  • AU = arbitrary unit; ND = not done.

10100ND0
2010033380
305810750
407513650

Discussion

Cytokinin oxidase is a key and specific enzyme of cytokinin metabolism. Evidence presented in this paper indicates unambiguously that we have purified and cloned a maize cytokinin oxidase. Purification of the enzyme was assisted by the use of tritiated azidoCPPU which was shown to photolabel the active enzyme and to displace its electrophoretic migration towards a more basic pH. The probe-modified CKO could then be isolated from a set of proteins having the same molecular weight. Degenerate oligonucleotides related to the amino acid sequences of internal peptides allowed us to amplify a 297 bp cDNA fragment which was cloned and sequenced. A complete cDNA sequence could then be obtained after 5′- and 3′-RACE PCR using specific oligonucleotides. Finally, the cytokinin oxidase ORF was cloned in a plant expression vector and its activity was clearly demonstrated in a transient expression assay using Physcomitrella patens protoplasts.

Photolabelling of the glycosylated CKO was very instrumental in identifying the protein after SDS–PAGE. Under such conditions, the protein has a molecular mass of 63 kDa which differs significatively from the previously reported 78 kDa for maize kernels CKO ( Burch & Horgan 1989) and 70 kDa for maize seedlings CKO ( Schreiber et al. 1995 ). These determinations are quite different from the 44 kDa molecular mass of the native enzyme measured by gel filtration using FPLC columns ( Fig. 1b; Burch & Horgan 1989). Glycosylated peptides often exhibit higher apparent sizes on SDS–PAGE gels because they bind less SDS molecules. Variations in the CKO glycosylation pattern could explain the discrepancy between the previously reported sizes of maize CKO observed on SDS–PAGE gels. In these studies, band identification as CKO was further supported by the observation that polyclonal antibodies raised against CKO purified fractions cross-reacted with CKO activity ( Burch & Horgan 1989; Schreiber et al. 1995 ). However, it is also possible that monitoring the last steps of CKO purification by SDS–PAGE was misleading and that the observed interaction between the antibodies and CKO activity was due to the highly antigenic glycosidic residues present on both CKO and the 70 or 78 kDa protein, respectively.

Preliminary attempts to express recombinant CKO in Escherichia coli cells were not successful and we therefore chose to analyse the functionality of the enzyme by using a plant system that should process the protein correctly. Moss protoplasts have the advantage of not requiring auxins and cytokinins for growth and division, although cytokinins are important in controlling moss development ( Ashton et al. 1979 ). Activity was mainly recovered in the culture medium. Either the enzyme is liberated in the medium by cell lysis or it is excreted. Therefore, functionality of the expressed recombinant enzyme is the final proof that we purified and cloned a maize CKO.

The reaction catalysed by CKO requires molecular oxygen and there is evidence that the reaction involves an iminopurine intermediate ( Laloue & Fox 1985). CKO is generally believed to be a copper-containing oxidase ( Armstrong 1994 for a review). However, no conclusive experimental data support this assumption. The deduced amino acid sequence of maize CKO cDNA has a domain (from position ≈ 170 to position ≈ 240) which is approximately 50% similar to a domain found in FAD-dependent oxidases. Furthermore, it has a conserved GHS motif (position 104/106) involved in the binding of FAD to a histidine residue in these flavoproteins ( Fig. 5). The highest homology is observed with the translated product of an Arabidopsis gene. This gene product is currently being characterised (F. Nogué, personal communication). FasV is part of the fas operon of Rhodococcus fascians encoding genes required for fasciation in plants ( Crespi et al. 1994 ). One of the genes belonging to this operon codes for cytokinin synthase ( Crespi et al. 1992 ). The FasV protein has been tentatively described as an electron transporter by Goethals et al. (1995 ). FasV homology with FAD-dependent oxidases was already pointed out by Mushegian & Koonin (1995). This homology indicates that maize CKO is most likely a flavoprotein. This conclusion is further supported by our observation (results not shown) that maize CKO is strongly inhibited by diphenyliodonium (DPI), an inhibitor of flavoprotein oxidoreductases ( O’Donnell et al. 1993 ).

By analysing the activity of glycosylated and unglycosylated CKOs in Phaseolus callus tissues, Kaminek & Armstrong (1990) suggested that a compartmentation could exist in the cells that keep the glycosylated form in the cell wall or plasmalemma and the unglycosylated form in an internal compartment. Is glycosylation important for functionality and/or localization? This question requires further investigation. Because of our purification scheme, we isolated a glycosylated protein, and Southern analysis indicates that the gene is likely to be present in one copy. We cannot rule out, however, the possibility that some related genes exist. Moreover, at least for the four sequenced peptides, no difference was observed between the commercial sweet corn variety and the hard corn field variety Nobilis. We did not isolate any cDNA other than the sequence reported here. However, CKO may belong to a family of genes with different expression patterns as suggested by Jones & Schreiber (1997). According to Northern analysis, the CKO gene seems to be weakly expressed in kernels harvested about two weeks after anthesis, a time that has been reported to correspond with a peak of CKO activity ( Dietrich et al. 1995 ).

We now possess a tool for analysing the role of cytokinin oxidase in plant development in general and in endosperm/embryo development in particular. It should be interesting to overexpress CKO in plants and to identify if, and how, this overexpression affects their phenotypes. Transgenic plants overproducing cytokinins have been already produced by transformation with the Agrobacterium tumefaciens IPT gene ( Faiss et al. 1997 ; Smigocki & Owens 1989). Interestingly, increased cytokinin metabolism/inactivation seemed to occur in these plants. Zhang et al. (1995 ) observed an NAA-mediated oxidative breakdown of both exogenous radiolabeled zeatin-type cytokinins and endogenous cytokinins in tobacco tissue transformed with the IPT gene. These observations were supported by Motyka et al. (1996 ) who showed an increase in cytokinin oxidase activity in both leaves and roots of transgenic tobacco plants expressing the IPT gene. It will be of special interest to confirm if and how cytokinin homeostasy is affected in plants overexpressing CKO or in mutants of Arabidopsis thaliana having a disrupted CKO gene.

Experimental procedures

Plant material

Commercial sweet corn was used for protein purification. For molecular biology experiments, kernels originated from field hard corn variety Nobilis, a mid-early hybrid, Pau Semences, 64230 Lescar, France. Cobs were cleaned, frozen in liquid nitrogen and kept at –70°C before use. The Gransden wild-type strain of the moss Physcomitrella patens ( Ashton & Cove 1977) was used for transient expression assays.

Chemicals

[3H]-iPA (17.8 mCi μmol–1) was prepared according to Laloue & Fox (1989). HA8 was a gift from R. Mornet (Angers University, France). iP, iPA, methyl-α- d-mannopyranoside, 2,5-diphenyloxazole, trypsic inhibitor, leupeptin, DTT, bovine serum albumin, Chaps, and the DOC–TCA kit were purchased from Sigma, St Louis, MO, USA. Pefabloc was from Interchim, Montluçon, France; concanavalinA-Sepharose was from Pharmacia, Uppsala, Sweden; polyclar AT was from Boehringer, Ingelheim, Heidelberg, Germany.

CKO purification

Extraction. Eight aliquots of 50 g frozen kernels were ground to a fine powder in a coffee grinder with four pulses of 10 sec each. The resulting powder was homogenized in a cold mortar with 800 ml of cold extraction buffer (0.1 m phosphate buffer pH 6.8 adjusted prior to use at 25 ng l–1 trypsic inhibitor, 1 mg l–1 leupeptin, 1 m m Pefabloc, 2 m m DTT, 5 m m EDTA, 33 g l–1 Polyclar AT). Homogenates were pooled, left at 4°C for 30 min with intermittent stirring and centrifuged at 20 000 g for 90 min. Supernatants were pooled and filtrated through two layers of Miracloth (Calbiochem, La Jolla, CA, USA).

Ammonium sulfate fractionation. Solid ammonium sulfate was added to the extract to 30% saturation over a period of 20 min at 4°C. After 1 h, the precipitate was eliminated by centrifugation (20 000 g for 90 min). Ammonium sulfate was again added in the same way to the collected supernatant to 50% saturation. After centrifugation, the pellet was then solubilized in 2 m m phosphate buffer pH 6.8 containing 2 m m DTT and the solution was dialysed overnight against three changes of EB buffer (20 m m Tris–HCl buffer pH 7.4, 0.5 m NaCl and 2 m m DTT). The dialysed extract was clarified by centrifugation at 40 000 g for 1 h, aliquoted and stored at –20°C.

ConcanavalinA-Sepharose chromatography. The extract was applied on a concanavalinA-Sepharose column (2.5 cm inner diameter, 6 cm heigth) equilibrated with 100 ml EB buffer. The sample (65 ml for 400 g kernels) was applied to the column which was washed in EB buffer without NaCl until the return of OD to the baseline and eluted in the same buffer containing 0.2 m methyl-α- d-mannopyranoside. Four ml fractions were collected and assayed for CKO activity. The appropriate fractions were pooled, aliquoted and stored at –20°C.

Anion exchange chromatography. An FPLC apparatus (Pharmacia) equipped with an anion exchange column (ResourceQ, Pharmacia, 6 ml) was used. Equilibration was performed in T buffer (20 m m Tris–HCl buffer pH 8.0).Two ml sample aliquots were applied to the column. The column was then washed in T buffer and eluted in a linear gradient of 1 m NaCl in the same buffer (increments were 2.5 m m NaCl min–1 for 1 h). Two ml fractions were collected and assayed for CKO activity and/or radioactivity as required. The appropriate fractions were pooled, aliquoted and stored at –20°C.

Gel permeation chromatography

A TSK3000SW gel permeation column (Tosohaas, Montgomery Ville, PA, USA) (7.5 mm I.D. × 300 mm height) was equilibrated with Tris–HCl 20 m m pH 7.8, Na2SO4 0.1 m, Na2EDTA 5 m m buffer. Elution was performed in the same buffer. The gel filtration standards were from BioRad Laboratories, Hercules, CA, USA. Fractions of 250 μl were collected and assayed for CKO activity and radioactivity.

Assay for CKO activity

Unless otherwise stated, assays were performed according to Laloue & Fox (1989), using [3H]-iPA 2 μm. The radioactive adenosine formed was separated from the substrate by HPLC on a C8 Lichrospher 60, RPselect B, 5 μm column (125 × 4 mm) from Merck. The HPLC system was a Waters gradient instrument 600E equipped with a programmable multiwavelength detector 490E. Radioactivity was measured with an on line Flo-one β instrument from Radiomatic.

Photoaffinity labelling

One ml protein aliquots obtained after conA-chromatography were incubated with 0.5 μm[3H]-azidoCPPU (333 Gbq mmol–1) at 4°C for 30 min (unless otherwise stated) on a shaker in the dark. After transfer in quartz cuvettes, samples were placed at a distance of 25 cm from a 254 nm, 6 W lamp (Bioblock Scientific, Illkirch, France) and irradiated for 5 min (energy = 0.3 mW cm–2). Photoincorporation of cold azidoCPPU was performed five times in the presence of 10 μm azidoCPPU, prior to preparative 2D-gel electrophoresis.

Before anionic exchange and gel permeation chromatography, excess of radioactive probe was removed by using desalting columns (NAP columns from Pharmacia) equilibrated with T buffer.

Polyacrylamide gel electrophoresis

Sample preparation. After anionic exchange chromatography, proteins were precipitated with DOC-TCA, pelleted by centrifugation at 12 000 g for 15 min, washed three times in methanol containing 0.1 m ammonium acetate and once in methanol, vacuum-dried and then solubilized in Laemmli (1970) or IEF buffer (8 m urea, 2% Pharmalytes 3–10, 32 m m DTT, 8 m m Chaps, 10 μM leupeptin, 1 m m Pefabloc and a few bromophenol blue cristals) for 1 and 2D-gel electrophoresis, respectively.

SDS–PAGE and fluorography. SDS–PAGE was performed on 0.75 mm thick 12% polyacrylamide running gel with 5% polyacrylamide stacking gel. The peptide pattern was visualized by Coomassie blue or silver staining. [14C]-labelled molecular weight markers were from Amersham, Cleveland, OH, USA.

For fluorographic analysis, the Coomassie blue stained gels were soaked in glacial acetic acid containing 20% (w/v) 2,5-diphenyloxazole for 60 min according to Skinner & Griswold (1983). Radiolabelling was visualized after exposure at –70°C of the dried gels to preflashed Kodak X-O-Mat films. The fluorographic signals were quantitated after digitalization with an OmniMedia 6CX-XRS scanner and analysis of the data on a Bioimage 1D software (Millipore Co, Bedford, MA, USA). They were corrected to take into account differences in the protein contents of the lanes which were measured on the Coomassie blue stained gel by the same technique.

2D-gel electrophoresis. Samples were submitted to 2D-gel electrophoresis on a Multiphor II apparatus from Pharmacia. IEF was performed on 11 cm or 18 cm Immobiline Dry strips pH 4–7 for 16 h at 33 or 44–45 kVh, respectively, and, if necessary, stored at –70°C before use. For the second dimension, the IEF gel strips were soaked at room temperature in 50 m m Tris–HCl buffer pH 6.8, 6 m urea, 30% glycerol, 1% SDS (w/v) for 15 min in the presence of 16 m m DTT and for 15 additional min in the presence of 240 m m iodoacetamide. The strips were then transferred onto the surface of a horizontal ExcelGelTM (Pharmacia, 8–18% gradient SDS-polyacrylamide) according to the procedure described by Pharmacia. Electrophoresis was carried out at 20 mA in the stacking gel and 40 mA in the running gel. Peptides were visualised by silver staining for analytical electrophoresis, amido-black for preparative gels and Coomassie blue for fluorography. In that case, gels were removed from their gel-bond support and treated as SDS–PAGE gels.

Protein determination

All protein determinations were done according to the Lowry’s procedure with the Protein Assay kit from Sigma. Bovine serum albumin was used as a standard.

Determination of amino acid sequences

Spots were cut out of the amido-black stained 2D-gels and partially dehydrated in a Speed-Vac. Gel pieces were rehydrated in 150–200 μl of 0.1 m Tris–HCl, pH 8.6, 0.01% Tween 20 (Pierce, Rockford, IL, USA) and digested with endoproteinase Lys-C from Lysobacter enzymogenes (Boehringer, Mannheim, Germany) at a final concentration of 2 μg ml–1 for 18 h at 35°C.

The supernatant was recovered and the pellet was rinsed with 60% acetonitrile. The acetonitrile rinse was added to the supernatant and acetonitrile was removed in a Speed-Vac. Sample was injected onto a DEAE-HPLC column linked to a C18 reverse phase HPLC column eluted with a 0–45% acetonitrile, 0.1% TFA gradient ( Kawasaki & Suzuki 1990). Peaks recorded at 210 nm were collected manually and frozen (–20°C) until sequencing. Sequencing was performed on Applied Biosystems 473 and 494 sequencers.

Nucleic acid isolation

Plasmid DNA was purified on QIAGEN-tips (QIAGEN GmbH, Hilden, Germany). Maize DNA was prepared from kernels according to Dellaporta et al. (1983 ) and purified by CsCl gradient. Total RNA extraction from kernels was adapted from Hall et al. (1978 ). Poly A+ RNA was purified by two rounds of spun-column chromatography by using the mRNA purification kit of Pharmacia.

DNA cloning

DNA amplification fragments were excised from electrophoresis gels and recovered by using a Geneclean kit (Bio 101 Inc, Vista, CA, USA). They were cloned in a pBluescript (KS) T-vector. This vector was prepared by digesting the plasmid with EcoRV restriction enzyme and adding thymidine 3′-overhangs according to Marchuk et al. (1991 ).

The plasmids were used to transform XL1-Blue (Stratagene, La Jolla, Ca, USA) or DH10B (Gibco BRL, Paisley, Scotland) Escherichia coli strains. Oligonucleotides spanning the start and the stop codon DNA region (CKs8, CKa8) were used to amplify the ORF-cDNA. Amplification was performed with the Advantage Klen Taq Polymerase (Clontech Laboratories, Palo Alto, CA, USA) in the presence of DMSO 4%. One pBluescript clone in which the CKO-ORF was in opposite orientation compared to the β-gal promoter (clone 16) was completely sequenced. The CKO-ORF fragment was directionally cloned in the plant vector pLBR19, a derivative of pJIT60 ( Guerineau et al. 1992 ). This vector carries a duplicated CaMV 35S promoter and a CaMV terminator. A SalI blunt-ended EcoRI fragment was cloned in pLBR19 that had been digested with HindIII, made blunt and further digested with SalI. The junction between the promoter and the ORF was checked by sequencing. This construct was named pLBR1916. pLBR19 with GUS-ORF was a gift of P. Mourrain (INRA Versailles, France).

DNA sequencing

DNA was sequenced by Sanger’s method. Cloned fragments were either sequenced in a cycle sequencing procedure with the Dye primer kit (PRISMTM, Perkin Elmer, New Jersey, USA) or with the Dye terminator kit (ABI PRISMTM, Perkin Elmer) and specific internal primers (CKs10, CKs11, CKs12, CKs13, CKs16, CKa10, CKa11, CKa12, CKa13). Electroeluted PCR fragments were sequenced with the Dye terminator kit (ABI PRISMTM, Perkin Elmer). Analysis was performed on an Applied Biosystems 370A DNA sequencer in the laboratory or at Cybergène (Saint-Malo, France).

Southern analysis

Genomic DNA (8 μg) was fractionated on a 0.8% agarose gel and blotted onto HybondTM N+ membranes. The blot was fixed by UV crosslinking in an UV Stratalinker (Stratagene). Membrane was probed with gel-purified restriction fragments that were radiolabelled with the hexamer kit of Pharmacia. Probes were a 1.1 kb NotI-BamHI internal restriction fragment of a 5′-RACE clone and a 0.6 kb SacI-HindIII restriction fragment of clone 16, respectively. Pre-hybridisation and hybridisation steps were performed at 65°C in formamide-free Church’s buffer (0.25 m Na2HPO4 pH 7.4, 2 m m EDTA, 7% SDS, 0.2 mg ml–1 heparin and 0.1 mg ml–1 denatured salmon sperm DNA). Washing was done for 15 min in 2 × SSC, 0.1% SDS at 50°C (i.e. low stringency conditions) and the blot was submitted to autoradiography. Membrane dehybridisation was performed according to the manufacturer’s instructions and total removal of the probe was checked by autoradiography. Pre-hybridisation and hybridisation steps were performed as described above. Washing was done for 30 min in 2 × SSC, 0.1% SDS at 55°C followed by 30 min washing in 0.2 × SSC, 0.1% SDS at 65°C (i.e. high stringency conditions).

Northern analysis

Total RNA (10 μg) and poly A+ RNA (1 μg) were fractionated on a 1.2% agarose formaldehyde gel and blotted onto a GeneScreen membrane (DuPont-NEN Research Products, Boston, MA, USA). RNA immobilisation and hybridisation were identical to Southern analysis. Washing was done for 20 min in 2 × SCC, 0.1% SDS at 55°C. Molecular weight markers (0.24–9.5 kb) were from Gibco BRL.

Oligonucleotide sequences

(a) Degenerate oligonucleotides

Coa1: CAT(TC)TCIAC(AG)AAIC(TG)(GA)TTCCA = an antisense primer related to peptide 1 (WNRFVEM);

Coa23: TT(AG)AAIAT(AG)TC(TC)TG(ACGT)CC(ACGT)GG = an antisense primer related to the C-terminus (PGQDIFN) of peptide 2;

Co51: ATI(CT)TICA(AG)GGIACIGA(CT)AT(ACT)GT(ACGT)GG = a sense primer related to the N-terminus (ILQGTDIVG) of peptide 5;

Co52: GTIGGICCI(CT)TIAT(ACT)GTITA(CT)CC = a nested sense primer related to peptide 5 (VGPLIVYP).

(b) Specific oligonucleotides

CKs2: TGGCGCCCAACGACCTG; CKs3: TGCAGGAGCAGAACAGG;

CKs6: ACAAATCCATGTGGGACGACGGCA;

CKs7: GGATCCTGCGCTTCTGCGACCTC;

CKs8: GCTAATGGCGGTGGTTTATTACCTGCTG;

CKs10: CCCTACACCATCGCGTTCC; CKs11: TTCCTTGACCGGGTGCACG;

CKs12: TCCTGTACCCGTCGTCCAC; CKs13: ATGGGGAGATGGTGACGTG;

CKs16: CGACCGACCTGGCGAACAC; CKa5: GCCGCGCCAGGTAGGTCTTG;

CKa7: GGACCCAGTCACTGCGGTCCGTGT;

CKa8: CCAAGAAACTAAAACATGCATGGGCTAT;

CKa10: CTTGTTGAGCGCCACCTCC; CKa11: CTGGCCCATGAGGGAGTGG;

CKa12: CACGTAGCTCATCGGGCC; CKa13: GATCCACACCTGCTCGCC.

Oligonucleotides were purchased from Genset, Paris, France and Eurogentec, Liège, Belgium.

Rt–pcr

The cDNA first strand was reverse transcribed from 2 μg poly A+ RNA by using the Pharmacia Ready to Go kit and random hexamers as primers (Boehringer). Six independent reactions were performed. Each reaction product was split and used for two RT–PCR reactions. A touch-down procedure ( Don et al. 1991 ) was used, consisting of 11 cycles with an annealing temperature scaled down from 1°C each cycle, starting from 62°C to 52°C (45 sec at 94°C, 30 sec at the annealing temperature and 2 min at 72°C) followed by 35 cycles (45 sec at 94°C, 30 sec at 52°C and 1 min 30 sec at 72°C). The PCR reaction mixture was as advised by the company. Forty picomoles of the degenerate oligonucleotides (Coa1, Coa23, Co51) were used in a 50 μl reaction. Bacterial transformants were screened by PCR with Co52 and Coa1.

Race pcr

Poly A+ RNA was treated with the Marathon TM cDNA amplification kit from Clontech. One μg was used in each reaction. The procedure was carried out exactly as advised by the company. Specific primers were CKs6, CKs7 and CKa7.

The final 3′-RACE reaction product was obtained after two rounds of amplification with the ExpandTM Long template PCR system (Boehringer) and primers CKs6-AP1 followed by a nested PCR with primers CKs7-AP2 and the Taq polymerase (Perkin Elmer), respectively. AP1 and AP2 are the anchored adaptators included in the MarathonTM kit. A 5′-RACE reaction product was visible after one round of amplification with the Clontech Advantage Klen Taq Polymerase in the presence of DMSO 4%. It was reamplified by gel puncturing by using the same primers. Bacterial transformants were screened by PCR with CKs2, CKs3, CKa5 and the universal primers.

Transient expression in moss protoplasts

Moss cultures and protoplast preparations were performed according to Schaefer & Zry—d (1997). Improved protoplast preparations were obtained by overnight digestion at 25°C in the dark with 0.05% driselase (Fluka Chemie AG, Switzerland) and 0.02% macerozyme R10 and 0.1% cellulase Onozuka R10 (Yakult Biochemicals, Nishinomiya, Japan) according to Bourgin et al. (1979 ). Briefly, after several washes in mannitol 8.5%, the protoplasts were counted and resuspended in mannitol 8.5%, MgCl2 15 m m, MES 0.1%, pH 5.6. PEG-mediated direct DNA transfer into protoplasts was performed as described by Schaefer et al. (1994 ). The samples were kept for 20 h in the dark and transferred for 24 h in the light. The culture medium was collected after centrifugation. Aliquots were analysed for CKO activity as described, except that [3H]-iPA was 0.05 μm. Cells were washed twice in new culture medium and disrupted by three cycles of freezing and thawing. The resulting crude extracts were assayed for CKO and GUS activities. GUS assay was performed according to Jefferson et al. (1987 ).

Acknowledgements

We thank Fabien Nogué for his help in the completion of the manuscript. We also thank Herman Höfte and Ian Small for critically reading the manuscript. The technical help of Amel Majira is acknowledged. We thank Daniel Duval for providing the maize line, Jean-Pascal Meunier and Jean-Marie Pollien for growing the plants and Kirk Schnorr for his past contribution to this project. This work is dedicated to the memory of Jean-Pierre Bourgin on the occasion of the 4th anniversary of his death.

Note Added in Proof

During the completion of this paper, the sequence of the maize ckx1 gene (AF044603) from R.O. Morris and J.G. Laskey (University of Missouri, Columbia, USA) was made available in the databank. Our cDNA sequence differs at seven positions in the ORF leading to three amino acid changes.

Ancillary