Higher plants possess two structurally different poly(ADP-ribose) polymerases


  • Elena Babiychuk,

    1. Laboratorium voor Genetica, Departement Genetica, Vlaams Interuniversitair Instituut voor Biotechnologie (VIB), Universiteit Gent, K.L. Ledeganckstraat 35, B-9000 Gent, Belgium,
    Search for more papers by this author
  • Phillippa B. Cottrill,

    1. Department of Biological Sciences, University of Essex, Essex CO4 3SQ, UK, and
    Search for more papers by this author
  • Sergei Storozhenko,

    1. Laboratorium voor Genetica, Departement Genetica, Vlaams Interuniversitair Instituut voor Biotechnologie (VIB), Universiteit Gent, K.L. Ledeganckstraat 35, B-9000 Gent, Belgium,
    Search for more papers by this author
  • Mayuree Fuangthong,

    1. Laboratorium voor Genetica, Departement Genetica, Vlaams Interuniversitair Instituut voor Biotechnologie (VIB), Universiteit Gent, K.L. Ledeganckstraat 35, B-9000 Gent, Belgium,
    Search for more papers by this author
  • Yimin Chen,

    1. Department of Biological Sciences, University of Essex, Essex CO4 3SQ, UK, and
    Search for more papers by this author
  • Minnie K. O’Farrell,

    1. Department of Biological Sciences, University of Essex, Essex CO4 3SQ, UK, and
    Search for more papers by this author
  • Marc Van Montagu,

    1. Laboratorium voor Genetica, Departement Genetica, Vlaams Interuniversitair Instituut voor Biotechnologie (VIB), Universiteit Gent, K.L. Ledeganckstraat 35, B-9000 Gent, Belgium,
    2. Laboratoire Associé de l’Institut National de la Recherche Agronomique (France), Universiteit Gent, B-9000 Gent, Belgium
    Search for more papers by this author
  • Dirk Inzé,

    1. Laboratorium voor Genetica, Departement Genetica, Vlaams Interuniversitair Instituut voor Biotechnologie (VIB), Universiteit Gent, K.L. Ledeganckstraat 35, B-9000 Gent, Belgium,
    2. Laboratoire Associé de l’Institut National de la Recherche Agronomique (France), Universiteit Gent, B-9000 Gent, Belgium
    Search for more papers by this author
  • Sergei Kushnir

    1. Laboratorium voor Genetica, Departement Genetica, Vlaams Interuniversitair Instituut voor Biotechnologie (VIB), Universiteit Gent, K.L. Ledeganckstraat 35, B-9000 Gent, Belgium,
    Search for more papers by this author

*For correspondence (fax + 32 9 264 5349;
e-mail mamon@gengenp.rug.ac.be).


One of the immediate reactions of the mammalian cell to many environmental stresses is a massive synthesis of poly(ADP-ribose), catalyzed by poly(ADP-ribose) polymerase (PARP). Most of the biological functions attributed to PARP are inferred from experimentation with mammalian cells. In plants, the biology of PARP may be more complicated and diverse than was previously thought. Two poly(ADP-ribose) polymerase homologues were found in plants, the classical Zn-finger-containing polymerase (ZAP) and the structurally non-classical PARP proteins (APP and NAP), which lack the characteristic N-terminal Zn-finger domain. By enzymatic and cytological experiments the recombinant APP protein was shown to be located in the nucleus and to possess DNA-dependent poly(ADP-ribose) polymerase activity in yeast. The nuclear localization was further confirmed by the analysis of transgenic tobacco plants that expressed a translational gene fusion between APP and the bacterial β-glucuronidase. The app promoter was transcriptionally up-regulated in cells pre-determined to die because of deficiency in a DNA ligase I.


Poly(ADP-ribo)sylation is a form of secondary protein modification found in most eukaryotic organisms, but with the surprising exception of yeast (Lindahl et al. 1995;de Murcia & Ménissier de Murcia 1994). This modification is enzymatically catalyzed by poly(ADP-ribose) polymerase (PARP or poly(ADP-ribose) transferase ADPRT;EC2.4.2.30) and involves the attachment of an ADP-ribose moiety derived from NAD+ to glutamic acid residues in the protein, followed by the further transfer of ADP-ribose monomers onto the newly formed adduct. PARP from animals is a nuclear protein of 113–120 kDa which consists of three major functional domains: an amino-terminal DNA-binding domain; a carboxyl-terminal catalytic domain; and an internal automodification domain (Kameshita et al. 1984;Lindahl et al. 1995;de Murcia & Ménissier de Murcia 1994). Through allosteric conformational changes, the enzymatic activity in vitro is greatly increased upon binding to single-strand DNA breaks. The in vivo activity is induced by conditions that eventually result in DNA breaks (Alvarez-Gonzalez & Althaus 1989;Ikejima et al. 1990). This property has led to the hypothesis that PARP is involved at some stage either in the DNA repair process or in the signalling pathways that sense alterations in the integrity of the genome (Lindahl et al. 1995;de Murcia & Ménissier de Murcia 1994). As PARP activation may significantly reduce the cellular NAD+ pool, it has also been suggested that the enzyme may play a critical role in programmed cell death (Heller et al. 1995;Zhang et al. 1994). Despite this apparently important role in the nuclear physiology of animal cells, studies with transgenic mice mutated in the PARP gene (Ménissier de Murcia et al. 1997;Wang et al. 1995, 1997) have shown that PARP is not an essential protein, possibly because of functional redundancy. Cells of PARP-deficient mice are, however, more sensitive to DNA damage (Ménissier de Murcia et al. 1997;Wang et al. 1997) and differ from normal cells of animals in some aspects of induced cell death (Heller et al. 1995).

PARP activity in plant cells was first demonstrated by examining the incorporation of 3H from labelled NAD+ into the nuclei of root tip cells (Payne & Bol 1976;Willmitzer & Wagner 1982). The enzymatic activity was also partially purified from maize seedlings and found to be associated with a protein of an apparent molecular mass of 113 kDa, suggesting that the plant PARP might be similar to that from animals (Chen et al. 1994;O’Farrell 1995). The limited amount of information currently available on the biological function of plant PARP has come from experiments involving PARP inhibitors. These results suggest a possible in vivo role in the prevention of homologous recombination at sites of DNA damage as rates of homologous intrachromosomal recombination in tobacco are increased after application of 3-aminobenzamide (3ABA) (Puchta et al. 1995). Furthermore, application of PARP inhibitors, such as 3ABA, nicotinamide and 6(5H)-phenasthridinone, to differentiating cells of Zinnia or Helianthus tuberosum has been shown to prevent development of tracheary elements (Hawkins & Phillips 1983;Phillips & Hawkins 1985;Shoji et al. 1997;Sugiyama et al. 1995), which is considered to be an example of programmed cell death in plants (Jones & Dangl 1996).

Previously, we have isolated the Arabidopsis app gene, coding for a polypeptide with a calculated molecular mass of 72 kDa (Lepiniec et al. 1995). At its carboxyl terminus, the APP protein is similar to the catalytic domain of PARP from animals, but the N terminus of APP does not show any significant homology to known proteins. APP lacks the Zn-finger DNA-binding domain that is characteristic of the PARP enzymes from animals and is of paramount importance in view of PARP biology as a nick sensor (de Murcia & Ménissier de Murcia 1994). Considering these structural differences and the significant discrepancy in size between APP and maize PARP, it was difficult to conclude that APP was a true poly(ADP-ribose) polymerase based on sequence similarities alone. Monocotyledonous and dicotyledonous plants could possibly have quite different PARPs or a PARP protein with the same domain organization as animal proteins, whereas APP could be either a second PARP or a related protein.

The work presented here was initiated to isolate PARP-encoding cDNA of maize. Simultaneously, we wanted to investigate (i) whether the calculated molecular mass of the encoded protein was close to the biochemically determined value of 113 kDa and (ii) what the differences were between the 113-kDa from maize and the 72-kDa PARP homologue (APP) of Arabidopsis. The molecular cloning data obtained demonstrated that higher plants have two different genes that encode PARP-homologous proteins of 110 kDa and 73 kDa. The maize 110-kDa Zn-finger poly(ADP-ribose) polymerase (ZAP) is highly similar in its sequence and domain organization to the enzyme from animals. Therefore, it seems logical to presume that the PARP activity found previously in maize (Chen et al. 1994) is associated with ZAP. However, the more detailed biochemical characterization of the recombinant ZAP protein will be presented elsewhere. The second 73-kDa PARP-homologous protein from maize, called NAP (non-classical poly(ADP-ribose) polymerase), is a ‘maize counterpart’ of the APP protein from Arabidopsis. Additional biochemical and cytological experiments with APP or NAP proteins were required to state that higher plants have two structurally different PARPs.

Here, we demonstrate that APP, and probably NAP, is a DNA-dependent nuclear poly(ADP-ribose) polymerase. Furthermore, we find that APP might be involved in molecular processes operating during genome destabilization, as has been shown by the increased activity of the app promoter in plant cells deficient in a DNA ligase I.


With the purpose of isolating maize cDNA encoding PARP homologue(s), two approaches were followed. First, we screened a maize cDNA library under low-stringency DNA–DNA hybridization conditions by using a DNA probe prepared from the Arabidopsis app cDNA. Second, PCR amplification of part of the maize PARP was performed by using the first-strand cDNA as a template and two degenerate primers, designed on the basis of the ‘PARP signature’ sequence (de Murcia & Ménissier de Murcia 1994), which is the most conserved amino acid sequence among all known PARP proteins.

The 73-kDa NAP protein from maize is an APP homologue

After the initial cDNA library screening with the app probe and a subsequent 5′ rapid amplification of cDNA ends (RACE) PCR analysis, the nap gene, a maize homologue of the Arabidopsis app, was identified. The NAP protein was 653 amino acids long (molecular mass approximately 73 kDa) and highly similar (61% identity and 69% similarity) to the APP. Most importantly, NAP had an organization of the N-terminus congruent to APP (Fig. 1a), suggesting a rather strict selection pressure on the structure of APP-like proteins in plants. The nap gene was unique in the maize genome (Fig. 2a) and encoded a transcript of 2.4 kb (Fig. 2c).

Figure 1.

The deduced N-terminal amino acid sequences of plant poly(ADP-ribose) polymerases.

(a) Alignment of the sequences upstream of the NAD+–binding domain found in Arabidopsis thaliana APP (A.th. APP; EMBL accession number Z48243) and the maize homolog NAP (Z.m. NAP; EMBL accession number AJ222588). The domain division shown is as previously proposed (Lepiniec et al. 1995). The nuclear localization signal (NLS) located in the B domain is indicated by the bracket. The sequence of the B domain is not very well conserved between dicotyledonous and monocotyledonous plants. The C domain is probably comparable in function to the automodification domain of PARP from animals. The imperfect repeats, A1 and A2, are also present in maize NAP. To illustrate the internally imperfect twofold symmetry within the repeat sequence, the properties of amino acid residues are highlighted below the sequences as follows: •, hydrophobic residue;○, glycine; (+), positively charged residue; (–), negatively charged residue; wavy line, any residue. The axis of symmetry is indicated by the vertical arrowhead and arrowhead lines mark the regions with the inverted repetition of amino acid side chain properties.

(b) Alignment of the DNA-binding and auto-catalytic domains of mouse PARP and maize ZAP. Zn-finger-containing maize ZAP1 and ZAP2 (partial cDNA found by the 5′RACE PCR analysis) are indicated as Z.m. ZAP (EMBL accession number AJ222589) and Z.m. ZAP(race), respectively, and the mouse PARP, M.m. ADPRT (Swissprot accession number P11103). The Zn-fingers and bipartite NLS of the mouse enzyme are indicated by brackets, the Caspase 3 cleavage site by the asterisk, and the putative NLS in the ZAP protein by the bracket in bold below the maize sequence. The amino acid residues that are conserved in all sequences are boxed; amino acid residues with similar physico-chemical properties are shaded with the uppermost sequence as a reference.

Figure 2.

Estimation of the gene copy number and transcript size for the nap and zap genes.

(a) and (b) Maize genomic DNA of variety LG2080 digested with the indicated restriction endonucleases, resolved by agarose gel electrophoresis, blotted, and hybridized with radioactively labelled DNA probes prepared from the 5′ domains of the nap and zap cDNA, which do not encode the NAD+–binding domain. The hybridization pattern obtained with the nap probe (a) is simple and indicates a single nap gene in the maize genome. As can be seen from the hybridization pattern (b), there might be at least two zap genes. To determine the size of the transcripts encoded by the zap and nap genes, approximately 1 μg of poly(A)+ RNA extracted from roots (lane 1) and shoots (lane 2) of 6-day-old seedlings were resolved on an agarose gel after denaturation with glyoxal, blotted, and hybridized with nap (c) and zap (d) 32P-labelled cDNA. 33P 5′ end-labelled BstEII fragments of λDNA were used as a molecular weight markers in both DNA and RNA gel blot experiments; their positions are indicated in kb to the left of each panel.

ZAP is a Zn-finger-containing plant PARP

Using degenerate primers based on very highly conserved regions in the ‘PARP signature’ and the first-strand cDNA as a template, a 310-bp fragment was amplified. The sequence of this fragment showed 55% identity and 64% similarity with human PARP over the same region but was, however, different from the sequence of the nap cDNA. Using the 310-bp fragment as a probe for the library screening, we isolated three cDNAs, the longest of which was sequenced on both strands. This cDNA encoded a PARP-homologous polypeptide of 689 amino acids (molecular mass approximately 109 kDa) and was designated ZAP1 (Fig. 1b). The first Zn-finger of ZAP1 was probably non-functional because it had the sequence CKSCxxxHASV that does not include the third cysteine residue.

We carried out 5′RACE PCR analysis of zap transcripts from the maize line LG2080 (the screened cDNA library was made from the inbred line B734). In all the transcripts from the LG2080 maize plants, there was an insertion of an additional sequence into the coding region, which made the ZAP protein longer by 11 amino acids (980 amino acids, molecular mass approximately 110.4 kDa). The Zn-finger I of ZAP2 was standard and read CKSCxxxHARC (Fig. 1b). The source of this sequence difference is uncertain because it probably did not occur during library construction, but may be due to differences between maize varieties, to the expression of two homologous genes, or to alternative splicing. In fact, maize may have at least two zap genes (Fig. 2b), which encode a transcript of 3.4–3.5 kb (Fig. 2d). The DNA gel blot experiment with a probe prepared from the zap cDNA showed that homologous genes were present in Arabidopsis (data not shown).

Structurally, ZAP was very similar to PARP from animals. It had a well conserved DNA-binding domain composed of two Zn-fingers (36% identity and 45% similarity to the DNA-binding domain of mouse PARP). Even higher homology was shown by comparing only the sequences of the Zn-fingers, Ala1-Phe162 in the mouse enzyme (44% identity and 54% similarity), or a subdomain downstream from the nuclear localization signal (NLS), Leu237-Ser360 in mouse PARP (40% identity and 50% similarity). Whereas the bipartite NLS characteristic of mammalian PARP could not be identified in ZAP, the sequence KRKK fitted a monopartite NLS (Fig. 1b). The putative automodification domain was poorly conserved and was shorter in ZAP than in mouse PARP.

The compilation of the homology of the catalytic domains between plant ZAP, NAP, APP and mouse PARP is shown in Table 1. It should be noted that the NAD+-binding domain of ZAP was more similar to the mammalian enzyme (48% identity) than to that of APP and NAP (40% and 42% identity, respectively), whereas APP and NAP were 68% identical and 76% similar in their catalytic domain.

Table 1.  Pairwise comparison of the NAD+–binding domains1
ProteinsM.m. ADPRTZ.m. ZAPZ.m NAPA.th. APP
  1. 1The sequence similarity over the NAD+-binding domains between PARP from mouse (M.m. ADPRT), maize ZAP (Z.m. ZAP), maize NAP (Z.m. NAP), and Arabidopsis APP (A.th. APP) is presented as a percentage of identical (bold italics) and similar amino acid residues. The ‘Bestfit’ program from the Genetic Computer Group (Madison, WI, USA) was used to obtain the values shown.

  Similarity (%)  
M.m. ADPRT62.155.652.6
Z.m. ZAP48.851.350.6
Z.m. NAP42.440.276.4
A.th. APP41.640.768.5
Identity (%)

APP is a DNA-dependent poly(ADP-ribose) polymerase

Whereas we were convinced that the detected PARP activity in maize (Chen et al. 1994) was associated with ZAP, it was necessary to understand what the NAP/APP protein group represented. We therefore carried out a more detailed study of the APP protein. The choice of yeast as the organism for the expression and enzymatic analysis of the Arabidopsis APP protein was made for a number of reasons. As an eukaryote, Saccharomyces cerevisiae is better suited for the expression of native proteins from other eukaryotic organisms and, unlike most other eukaryotic cells, it does not possess endogenous PARP activity (Lindahl et al. 1995).

The full-length app cDNA was placed in pYeDP1/8–2 under the control of a galactose-inducible yeast promoter. The expression of the APP in yeast was verified by Northern and Western blot analyses (data not shown). To detect the synthesis of poly(ADP-ribose), total proteins were extracted from yeast strains grown under different conditions and incubated in the presence of radioactively labelled NAD+. To prevent the synthesis of poly(ADP-ribose) and the possible automodification of APP in vivo, strains were also grown in the presence of 3ABA, a reversible inhibitor of PARP, which was subsequently removed from the protein extracts during desalting. Figure 3 shows that poly(ADP-ribose) is synthesized by protein extracts of DY(pV8SPA) grown on galactose (Fig. 3a, lanes 1 and 2), but not by a strain containing the empty vector (Fig. 3a, lane 4). It can also be seen that Arabidopsis APP could synthesize polymers up to 40 residues in length (Fig. 3a, lane 1) with the majority of the radioactivity being incorporated into 10–15-mer. This observation is consistent with the polymer sizes detected by other authors (Chen et al. 1994). More radioactivity was incorporated into the polymers when the yeast strain was grown with than without 3ABA (Fig. 3a, lane 1 compared to lane 2). The reason might be that either the APP extracted from inhibited cultures was less automodified (automodification inhibits the activity of PARP) or the labelled NAD+ was used by the enzyme from the uninhibited culture for the extension of existing polymers, resulting in a lower specific activity overall. Under the same reaction conditions, poly(ADP-ribose) synthesized by human PARP, either alone in the reaction buffer or in the presence of a yeast total protein extract from DY(pYeDP1/8–2) (Fig. 3a, lanes 5 and 6, respectively), showed much longer chains, possibly up to 400-mer (de Murcia & Ménissier de Murcia 1994).

Figure 3.

Poly(ADP-ribose) polymerase activity of the APP protein.

(a) The total protein extracts were prepared from DY(pYeDP1/8–2) grown on SDC with 2% galactose (vector GAL) and DY(pV8SPA) grown either on SDC with 2% glucose (app GLU), on SDC with 2% galactose (app GAL), or on SDC with 2% galactose and 3 mm 3ABA (app GAL+3ABA). To detect the synthesis of the poly(ADP-ribose) in these extracts, samples were incubated with 32P-NAD+ for 40 min at room temperature. Two control reactions were performed: 100 ng of the purified human PARP were incubated either in a reaction buffer alone (PARP) (lane 5), or with protein extract made from DY(pYeDP1/8–2) culture grown on glucose (vector GLU+PARP) (lane 6). The autoradiograph obtained after exposure of the dried gel to X-Omat Kodak film is shown. ORi corresponds to the beginning of the sequencing gel.

(b) Stimulation of poly(ADP-ribose) synthesis by DNA in protein extracts from DY(pV8SPA). Amounts of sonicated salmon sperm DNA added to the nucleic acid depleted yeast extracts are indicated in μg ml–1. The synthesis of the poly(ADP-ribose) is blocked by 3ABA, which was added in one of the reactions at a concentration of 3 mm (lane 5). To ensure the maximal recovery of the poly(ADP-ribose), 20 μg of glycogen were included as a carrier during precipitation steps; this, as can be seen, however, resulted in high carry-over of the unincorporated label.

The stimulation of enzymatic activity by nicked DNA is a well known property of PARP from animals (Alvarez-Gonzalez & Althaus 1989). We therefore tested whether the activity of the APP protein was DNA dependent. After removal of yeast nucleic acids (DNA, RNA) and some basic proteins from the galactose-grown DY(pV8SPA) protein extract, the synthesis of poly(ADP-ribose) was analyzed in the presence of increasing concentrations of sonicated salmon sperm DNA. As can be seen in Fig. 3(b), there was a direct correlation between the amount of DNA present in the reaction and the incorporation of 32P-NAD+. Scanning of the phosphor-images indicated that approximately sixfold more radioactivity was incorporated into poly(ADP-ribose) in the reaction mixture containing 40 μg ml–1 of DNA than into that with 2 μg ml–1 of DNA (Fig. 3b, lanes 4 and 2, respectively). The synthesis of the polymer was sensitive to 3ABA in the reaction mix (Fig. 3b, lane 5).

APP is a nuclear protein

In animal cells, PARP activity is localized in the nucleus (Schreiber et al. 1992). The intracellular localization, if nuclear, of APP could provide an important additional indication that APP is a bona fide plant PARP. To this end, we analyzed the localization of the APP polypeptides in yeast cells using anti-APP antisera. Figure 4 shows that the APP polypeptide synthesized in yeast grown on galactose was found mainly in the nucleus (Fig. 4m, q, u). This localization was unaffected by the presence of the PARP inhibitors in the media (Fig. 4q, u).

Figure 4.

In situ localization of APP protein and poly(ADP-ribose) in yeast. DY cells transformed with pYeDP1/8–2 (vector) or pV8SPA (app) were grown to log phase in medium containing glucose (GLU) or galactose (GAL). Cells were fixed, and APP protein was detected with an anti-APP rabbit polyclonal serum (anti-APP) and CY-3 conjugated anti-rabbit IgG sheep antibody. The ADP-ribose polymers were visualized using the monoclonal mouse 10H antibody (H10) and FITC-conjugated anti-mouse IgG sheep antibody. DNA was stained with DAPI.

In addition, we tested whether APP was constitutively active in yeast cells, as has been reported for the human PARP (Collinge & Althaus 1994). Here, fixed yeast spheroplasts were incubated with the monoclonal 10H antibody, which specifically recognizes poly(ADP-ribose) polymers (Kawamitsu et al. 1984). A positive yellowish-green fluorescence signal with the 10H antibody was localized in the nucleus and was observed only in DY(pV8SPA) cells grown on galactose (Fig. 4o). Positive staining was greatly reduced in cells grown in the presence of the PARP inhibitors, 3ABA and nicotinamide (Fig. 4., w). The orange autoimmunofluorescence was due to the filter system used.

To identify the intracellular localization of APP in plant cells, we followed a widely used approach in plant studies, i.e. the examination of the subcellular location of a fusion protein formed between the protein in question and a reporter gene, once the protein fusion was produced in transgenic plants or transfected cells (von Arnim & Deng 1994;Citovsky et al. 1994;Sakamoto & Nagatani 1996;Terzaghi et al. 1997). The part of the APP polypeptide extending from Met1 to Pro407 was fused in frame with GUS and the fusion was expressed in tobacco under the control of the 35S promoter (Fig. 5a). In 2-day-old seedlings from the progeny of four independent lines, most of the GUS activity could be detected in cotyledons and in roots, but not in hypocotyls or root tips (Fig. 5b, panel 1). Because of the transparency of root tissues, it is easier to see that GUS staining was clearly localized in the nuclei of root hairs and epidermal cells (Fig. 5b, panels 2 and 3). Additionally, some diffuse, non-localized staining of other root cells was seen, in particular along the vascular cylinders (Fig. 5b, panel 1).

Figure 5.

Nuclear localization of the APP-GUS translational fusion in transgenic tobacco plants.

(a) Schematic drawing of the expression cassette in the pGCNSPAGUS binary vector. The part of the app cDNA encoding the APP protein from the first methionine (M1) to proline407 (P407) was fused in-frame to the E. coli gene uidA, encoding GUS (Jefferson et al. 1987). The expression of the chimeric cDNA is driven by the 35S promoter from the CaMV (35S). The 5′ non-translated leader of the expected mRNA also contains the Ω leader from the tobacco mosaic virus to increase translational efficiency (data not shown) (Gallie & Walbot 1992). The necessary sequences for polyadenylation are derived from the A. tumefaciens gene encoding nopaline synthase (3′nos).

(b) Histochemical staining for GUS activity in tobacco seedlings transformed with a T-DNA from pGCNSPAGUS (panels 1–3). Two-day-old seedlings were stained for GUS activity. The distribution of activity is shown in a whole seedling (panel 1), its root hairs (panel 2), and root epidermal cells (panel 3). The staining of the root epidermal cells in transgenic plants transformed with uidA under the control of the 35S promoter and used as a control is shown in panel 4. Nuclei are indicated by arrows.

Deficiency in DNA ligase I induces the app gene

PARP in animal cells is one of the most abundant nuclear proteins and its activity is regulated by allosteric changes in the protein upon binding to damaged DNA. We found that the app gene in Arabidopsis had a rather low level of expression, suggesting that transcriptional activation of this gene might be essential for APP function in vivo. To test this hypothesis, the expression of the app gene was studied during in vivo genome destabilization caused by a DNA ligase I deficiency.

Previously, the T-DNA insertion mutant line SK1B2 of A. thaliana Col-5 was isolated (Babiychuk et al. 1997) in the DNA ligase I gene (Taylor et al. 1998). Because female and male haploid cells undergo a number of cell divisions before fertilization, it was expected that the loss-of-function mutation in the DNA ligase I would cause the lethality of gametophytes. For example, the temperature-sensitive allele cdc9 in a DNA ligase of S. cerevisiae, which can be complemented by the Arabidopsis DNA ligase I (Taylor et al. 1998), caused yeast cells to arrest in the S-phase at the non-permissive temperature. Nevertheless, the mutant allele in SK1B2 had normal transmission through both female and male gametes, as judged from the 1:1 segregation of the kanamycin resistance in direct or reciprocal crosses. The mutant allele caused the sporophytic lethality and one-quarter of the seeds were shrunken and dead in mature siliques of self-fertilized, kanamycin-resistant plants of line SK1B2. Hence, either the amount of DNA ligase I synthesized in the diploid, pre-meiotic cells was sufficient to carry out the DNA synthesis in haploid cells, or the allele was leaky, although the T-DNA is integrated into the middle of the gene, or other plant ligases as yet unidentified (Taylor et al. 1998), could partially substitute for the DNA ligase I loss-of-function. We therefore presumed that the lethality in SK1B2 represented the fertilization events of the mutant embryo sacs with mutant pollen and that the formed zygotes or early stage embryos died because of incomplete DNA synthesis during the S-phase of the cell cycle.

We made an app promoter-GUS translational fusion, in which the coding region of GUS was fused in-frame with the first five amino acids of APP and 2 kb of the app 5′ flanking sequences. The app promoter-GUS fusion (Papp uidA) was transformed into Arabidopsis Col-1. The inflorescences of PappuidA plants from a control cross with Col-5 and from a cross with SK1B2 were stained for GUS activity (Fig. 6). The GUS staining pattern that was mostly detected in ageing tissues probably reflected the expression of the app gene, although we have no firm evidence that all of the regulatory sequences were present in the constructs used. This pattern was the same both in the inflorescences of control plants with genotype (Col-1/Col-5; PappuidA/–;LIG1/LIG1) (Fig. 6a) and that of plants with genotype (Col-1/Col-5; PappuidA/–;LIG1/lig1), heterozygous for the mutation in the ligase gene (Fig. 6b). Importantly, some ovules shown in Fig. 6(b) were GUS positive. Out of 551 counted ovules, 107 were GUS positive, which is close to the expected 13:3 segregation. Closer microscopical examination showed that in the GUS-positive ovules only the area occupied by the embryo sac was stained (Fig. 6c). The only difference between the plant shown in Fig. 6(a) and (b) was a mutation in a DNA ligase I gene. We therefore conclude that the app gene was induced because of either the accumulation of DNA breaks or the death of the mutant embryo sacs fertilized with mutant pollen. GUS staining of embryo sacs was found to appear within 24 h after pollination or very soon after fertilization (data not shown).

Figure 6.

Activation of the app promoter-GUS fusion in the embryo sacs of Arabidopsis plants heterozygous for the T-DNA insertion mutation in a DNA ligase I gene, LIG1.

(a) and (b) GUS activity in inflorescences of Arabidopsis plants transformed with the app promoter-gus gene fusion (PappuidA) is revealed by histochemical staining. Inflorescences of F1, glufosinate-resistant plants from a cross Col-1 (PappuidA/–;LIG1/LIG1)×Col-5(LIG1/LIG1) and glufosinate/kanamycin-resistant plants from a cross Col-1 (PappuidA/–;LIG1/LIG1)×Col-5(LIG1/lig1) are shown in (a) and (b), respectively. Note the appearance of the additional GUS staining of some ovules in (b). In these GUS-positive ovules only the area occupied by the embryo sac is stained (c).


This work was initiated to resolve the apparent differences in molecular mass of the poly(ADP-ribose) polymerase enzymes in two plant species. The molecular mass of maize PARP was estimated as 113 kDa by SDS-PAGE and HPLC (Chen et al. 1994), whereas the PARP-homologous Arabidopsis protein APP identified by the molecular cloning of the cDNA had a calculated molecular mass of 72 kDa (Lepiniec et al. 1995). The experiments led to the finding of two types of genes encoding PARP-homologous proteins in plants. We show here that the maize ZAP protein, apparently present in other higher plant species, is similar in sequence and structural organization to PARP from animals. Considering the high evolutionary conservation between proteins of maize and mammals, we would predict that the biology of ZAP will be similar, if not identical, to that of PARP from animals.

The major contribution to the study of poly(ADP-ribos)ylation reactions in cells is the cloning of NAP. This result shows that Arabidopsis APP is not just a mutant derivative of the ZAP, but represents a distinct and novel class of poly(ADP-ribose) polymerase. Not only are NAP and APP very similar in putative domain organization, but APP is a true PARP. This has been shown by (i) the in situ detection of poly(ADP-ribose) polymers in yeast cells expressing app, using the monoclonal anti-poly(ADP-ribose) antibody 10H, and (ii) visualization on sequencing gels of the radioactively labelled polymer synthesized in vitro by yeast protein extracts in the presence of 32P-NAD+.

The PARP activity of APP is DNA dependent; however, presently, it is not known how the DNA dependence of APP is executed in molecular terms, or which particular DNA configurations, if any, are the most effective in stimulating APP activity. Most probably, the APP-specific N-terminal domain binds to DNA. However, our attempts to detect changes in electrophoretic mobilities of various DNA templates after incubation with APP have been unsuccessful.

The APP protein produced in yeast is located largely in the nucleus as revealed with anti-APP antisera. The APP-GUS fusion protein, when synthesized in stably transformed tobacco plants can also be detected in nuclei, this being particularly obvious in root tissues. A stretch of basic amino acids KSKRKR present in a sequence of the B domain of the protein is similar to the SV40 T-antigen-type NLS KR/KXR/K, and may function as an NLS.

The app gene promoter is transcriptionally activated in cells deficient in a DNA ligase I. Because the signal for such induction would most probably be an arrest in S phase and the accumulation of DNA breaks, APP/NAP might thus be involved in the sensing of genome stability, a function believed to be of the foremost importance for PARP from animals. Therefore, ZAP and NAP may represent functional redundancy, the need for which is essential for plant survival in nature. On the other hand, an independent study (A. Levine, personal communication) evidenced that APP might be involved in NAD+ depletion during the H2O2-induced death of soybean cells cultured in vitro. Therefore, APP might be specifically involved in certain steps of plant cell death. We may thus speculate that in ligase-deficient cells there is an initial step that signals genome distress, followed by passive or active degeneration in which APP might be involved. In this respect, it will be interesting to examine the transcriptional activity of the app promoter in other Arabidopsis lines that carry lethal mutations because of defects in other cellular functions, such as DNA synthesis.

Why plants should have two very different PARP proteins and whether other organisms, including animals, also possess various forms of poly(ADP-ribose) polymerase are major questions posed by this study. ZAP is not only strikingly similar to PARP from animals in its predicted domain organization, but also at the sequence level it is more similar to the mouse enzyme, for example, than to APP or NAP. This evolutionary argument suggests that in plants the poly(ADP-ribose) polymerases of the ZAP and NAP groups may have different functions in vivo. ZAP and/or NAP may be more ‘specialized’ for one of the several functions proposed for the mammalian PARP (Lindahl et al. 1995). Alternatively, the NAP group may be involved in other, as yet undiscovered, cellular functions.

Experimental procedures

Yeast and bacterial strains

Saccharomyces cerevisiae strain DY (MATa his3 can1–10 ade2 leu2 trp1 ura3::(3×SV40 AP1-lacZ) (Kuge & Jones 1994) was used for the expression of the APP protein. Yeast transformation was carried out according to Dohmen et al. (1991). Strains were grown on a minimal SDC medium (0.67% yeast nitrogen base, 0.37% casamino acids, 2% glucose, 50 mg l–1 of adenine and 40 mg l–1 of tryptophan). For the induction of the APP expression, glucose in SDC was substituted with 2% galactose.

Escherichia coli strain XL-I (Stratagene, La Jolla, CA, USA) was used for the plasmid manipulations and library screenings which were carried out according to standard procedures (Ausubel et al. 1987;Sambrook et al. 1989). E. coli BL21 (Studier & Moffat 1986) was used for the APP protein expression and Agrobacterium tumefaciens C58C1RifR(pGV2260) (Deblaere et al. 1985) for the stable transformation of plants.

Isolation of maize cDNA

A λZAP (Stratagene) cDNA library from leaves of maize (Zea mays L.), inbred line B734 was kindly provided by Dr Alice Barkan (University of Oregon). Plaques (500 000) were screened according to standard procedures (Sambrook et al. 1989). After screening with the Arabidopsis app probe, one non-full-length cDNA of 1.4 kbp was purified.

Three zap cDNAs were identified after screening with the 310-bp fragment which was obtained by PCR with degenerate primers. These three purified cDNA were all derived from the same transcript because they had identical 3′ non-coding regions; the longest clone (#9) was sequenced on both strands.

PCR analysis

For the PCR with degenerate primers (5′-CCGAATTCGGNTAYATGTTYGGNAA-3′ and 5′-CCGAATTCACNATRTAYTCRTTRTA-3′ with Y = C/T; R = A/G; N = A/G/C/T), the first strand cDNA was used as a template and was synthesized using 5 μg of poly(A)+ RNA from young maize leaves and MuMLV reverse transcriptase. PCR amplifications were performed with Taq DNA polymerase in 100 μl volume using the following conditions: 1 min at 95°C, 2 min at 45°C, 3 min at 72°C, followed by 38 cycles of 1 min at 95°C, 2 min at 45°C, 3 min at 72°C, with a final incubation for 10 min at 72°C.

For the 5′RACE PCR, the template was prepared with the Marathon kit (Clontech, Palo Alto, CA, USA) and 0.5 μg of maize poly(A)+ RNA isolated from inner sheath, outer sheath, and leaves of 1-week-old maize seedlings. The gene-specific, nested primers for PCR amplification were 5′-GGGACCATGTAGTTTATCTTGACCT-3′ and 5′-GACCTCGTACCCCAACTCTTCCCCAT-3′ for nap primers and 5′-AAGTCGACGCGGCCGCCACACCTAGTGCCAGGTCAG-3′ and 5′-ATCTCAATTGTACATTTCTCAGGA-3′ for zap primers.

The amplified PCR products were subcloned and sequenced. A fragment of 800 bp was amplified with nap-specific primers which allowed us to reconstruct the 2295-bp-long sequence of nap cDNA. A 450-bp PCR product was obtained after PCR with zap-specific primers. Eight independent, because of their slight differences in lengths at their 5′ ends, 5′RACE PCR fragments generated with zap-specific primers were sequenced.

Plasmid construction

For the expression of APP in yeast, the full-length app cDNA was excised from pC3 (Lepiniec et al. 1995) as an XhoI-EcoRI fragment. The ends were filled in with the Klenow fragment of DNA polymerase I, and the fragment was subcloned into the SmaI site of the yeast expression vector pYeDP1/8–2 (Cullin & Pompon 1988). The resulting expression vector pV8SPA was transformed into S. cerevisiae strain DY.

For APP expression in E. coli, the complete coding region of the app cDNA was PCR amplified with Pfu DNA polymerase (Stratagene), using the primers 5′-AGGATCCCATGGCGAACAAGCTCAAAGTGAC-3′ and 5′-AGGATCCTTAGTGCTTGTAGTTGAAT-3′, and subcloned as a BamHI fragment into pET19b (Novagene, Madison, WI, USA), resulting in pETSPA. The expression of the full-length APP in E. coli BL21 from pETSPA was very poor. To obtain better expression, pETSPA was digested with NcoI and NdeI or with SmaI, the ends were filled in by the Klenow fragment of DNA polymerase I, and the plasmids were then self-ligated. Of the resulting plasmids pETΔNdeSPA and pETΔSmaSPA, only pETΔNdeSPA gave satisfactory expression of the truncated APP polypeptide (Met310 to His637) in E. coli BL21.

The translational fusion of APP with bacterial GUS was constructed as follows. Plasmid pETSPA was cut with SmaI, treated with alkaline phosphatase, and ligated to a blunted NcoI-XbaI fragment from pGUS1 (Plant Genetic Systems N.V., Gent, Belgium). The ligation mix was transformed into E. coli XL-I and cells were plated onto LB medium supplemented with 0.1 mm isopropyl-β-d-thiogalactopyranoside, 40 μg ml–1 5-bromo-4-chloro-3-indolyl-β-d-glucuronide, and 100 μg ml–1 of ampicillin. In this way, pETSPAGUS was selected as blue colonies. The expression in E. coli of the approximately 110-kDa fusion protein was confirmed by in situ GUS activity gels (Lee et al. 1995). The APP-GUS fusion was placed under the control of the 35S promoter of the CaMV (the Klenow-blunted BamHI fragment from pETSPAGUS was subcloned into SmaI-digested pJD330;Gallie & Walbot 1992) and the resulting expression cassette was subcloned as an XbaI fragment into the XbaI site of the pCGN1547 binary vector (McBride & Summerfelt 1990) to give pGCNSPAGUS. The pGCNSPAGUS was finally introduced into A. tumefaciens C58C1RifR(pGV2260) by the freezing-thawing transformation procedure.

The construction of the app promoter-GUS fusion will be described elsewhere.

Poly(ADP-ribose) polymerase activity assay

Enzymatic activity of the APP was assayed in total protein extracts of yeast strains prepared as follows. DY(pV8SPA) or DY(pYeDP1/8–2) were grown in 50 ml of SDC medium overnight at 30°C on a gyratory shaker at 150 rpm. Yeast cells were harvested by centrifugation at 1000 g, washed three times with 150 ml of 0.1 m potassium phosphate buffer (pH 6.5), and resuspended in 5 ml of sorbitol buffer (1.2 m sorbitol, 0.12 m K2HPO4, 0.033 m citric acid, pH 5.9). Lyticase (Boehringer, Mannheim, Germany) was added to the cell suspension to a final concentration of 30 U ml–1 and cells were incubated at 30°C for 1 h. Yeast spheroplasts were then washed three times with sorbitol buffer and resuspended in 2 ml of ice-cold lysis buffer (100 mm Tris–HCl, pH 7.5, 400 mm NaCl, 1 mm EDTA, 10% glycerol, 1 mm DTT). After sonication, the lysate was centrifuged at 20 000 g for 20 min at 4°C and the supernatant was desalted on a Econo-Pack™ 10 DG column (Bio-Rad, Richmond, CA, USA) equilibrated with reaction buffer (100 mm Tris–HCl, pH 8.0, 10 mm MgCl2, 1 mm DTT). To reduce proteolytic degradation of proteins, the lysis and reaction buffers were supplemented with a protease inhibitor cocktail (Boehringer), one tablet per 50 ml. Nucleic acids were removed from the total extracts by adding NaCl and protamine sulfate to a final concentration of 600 mm and 10 mg ml–1, respectively. After incubation at room temperature for 10 min, the precipitate was removed by centrifugation at 20 000 g for 15 min at 4°C. The buffer of the supernatant was exchanged for the reaction buffer by gel filtration on an Econo-Pack™ 10 DG column.

The assay for the synthesis of poly(ADP-ribose) was adapted from Collinge & Althaus (1994). Approximately 500 μg of total yeast protein were incubated in a reaction buffer supplemented with 30 μCi of 32P-NAD+ (500 Ci mmol–1), unlabelled NAD+ to a final concentration of 60 μm, and 10 μg ml–1 sonicated salmon sperm DNA. After incubation for 40 min at room temperature, 500 μl of the stop buffer (200 mm Tris–HCl, pH 7.6, 0.1 m NaCl, 5 mm EDTA, 1% Na+–N-lauroyl-sarcosine, and 20 μg ml–1 proteinase K) were added and reactions incubated at 37°C overnight. After phenol and phenol/chloroform extractions, polymers were precipitated with 2.5 volumes of ethanol with 0.1 m NaAc (pH 5.2). The pellet was washed with 70% ethanol, dried and dissolved in 70% formamide, 10 mm EDTA, 0.01% bromophenol blue, and 0.01% xylene cyanol. Samples were heated at 80°C for 10 min and then loaded onto a 12% polyacrylamide/6 m urea sequencing gel. Gels were dried on 3 MM paper (Whatman International, Maidstone, UK) and exposed either to Kodak X-Omat X-ray film (Eastman Kodak, Richmond, NY, USA) or scanned using a PhosphorImager- 445SI (Molecular Dynamics, Sunnyvale, CA).

Immunological techniques

A truncated app cDNA encoding an APP polypeptide from amino acids Met310 to His637 was expressed as a translation fusion with six histidine residues at the N terminus after induction of a 500-ml culture of the E. coli BL21(pETΔNdeSPA) with 1 mm isopropyl-β-d-thiogalactopyranoside. The APP polypeptide was purified to near homogeneity by affinity chromatography under denaturing conditions (in the presence of 6 m guanidinium hydrochloride) on a Ni2+–NTA-agarose column, according to the manufacturer’s protocol (Qiagen, Chatsworth, CA, USA). After dialysis against PBS, a mixture of the soluble and insoluble APP polypeptides was used to immunize two New Zealand White rabbits following a standard immunization protocol (Harlow & Lane 1988). For the Western blot analysis, proteins were resolved by denaturing SDS-PAGE (Harlow & Lane 1988;Sambrook et al. 1989) and transferred onto nitrocellulose membranes (Hybond-C; Amersham), using a Semi-Dry Blotter II (Kem-En-Tec, Copenhagen, Denmark).

In situ antigen localization in yeast cells was carried out as described previously (Harlow & Lane 1988). For the localization of the APP protein in yeast spheroplasts, anti-APP serum was diluted 1:3000–1:5000 in Tris-buffered saline-BSA buffer. 10H monoclonal antibody, which specifically recognizes poly(ADP-ribose) polymers (Ikejima et al. 1990) was used in a 1:100 dilution in PBS buffer. The mouse antibodies were detected with the sheep anti-mouse IgG F(ab′)2 fragment conjugated to fluorescein isothiocyanate (FITC) (Sigma) at a dilution of 1:200. Rabbit IgG was detected with CY-3 conjugated sheep anti-rabbit IgG sheep F(ab′)2 fragment (Sigma), at a dilution of 1:200. For the visualization of DNA, slides were incubated for 1 min in PBS with 10 μg ml–1 of 4′,6-diamidino-2-phenylindole (DAPI; Sigma). Fluorescence imaging was performed on an Axioskop epifluorescence microscope (Zeiss, Jena, Germany). For observation of FITC and CY-3 fluorochromes, 23 and 15 filter cubes were used, respectively. Cells were photographed with Fuji Color-100 super plus film.

Plant material and histochemical analysis

Nicotiana tabacum SR1 (Maliga et al. 1975) was used for the generation of stable transformants following the procedure of leaf disc co-cultivation (De Block et al. 1987) with A. tumefaciens C58C1RifR(pGV2260; pGCNSPAGUS). N. tabacum SR1 line transformed with authentic GUS under the control of the 35S CaMV was provided by Dr A. Depicker (Gent, Belgium). Arabidopsis thaliana Col-1 was used for the transformation of the app promoter-GUS fusion following the in situ infiltration procedure. Transformants were selected on the MS medium supplemented with 10 mg l–1 of the ammonium glufosinate (ammonium-dL-homoalanin-4ylmethylphosphinat, PESTANAL®; Reidel-de Haën AG, Seelze, Germany). After two back-crosses to the A. thaliana Col-1, glufosinate-resistant app promoter-GUS transformants were used to pollinate either A. thaliana Col-5 as a control or A. thaliana Col-5 line SK1B2. F1 plants double-resistant to the kanamycin and ammonium glufosinate from a latter cross, and ammonium glufosinate-resistant plants from a former cross were selected and used for the GUS histochemical assays.

For in situ histochemical staining of the GUS activity, plant samples were fixed in ice-cold 90% acetone for 30 min, washed in 0.1 m K2HPO4 (pH 7.8), and then incubated in staining buffer (0.1 m K2HPO4, pH 7.8, 2 mm X-Gluc, 20 mm Fe3+–EDTA) at 37°C. Stained plant tissues were stored in 70% ethanol at 4°C. When necessary, browning of tissues due to phenolic oxidation was reduced by incubation with lactophenol (Beeckman & Engler 1994). The GUS staining was examined under a Jenalumar light microscope (Zeiss). Plant tissues were photographed with Fuji Color-100 super plus film.

Miscellaneous methods

The plasmid construction steps were routinely verified by DNA sequencing carried out according to protocols provided by USB Biochemicals (Cleveland, OH, USA). 32P-labelled DNA probes for nucleic acid hybridization were synthesized by the Ready-Prime DNA labelling kit (Amersham). For DNA and RNA hybridization experiments, the buffer system of Church & Gilbert (1984) was used (0.25 m sodium phosphate, pH 7.2, 7% SDS, 1% BSA, 1 mm EDTA). For Western blot analysis, yeast total proteins were extracted with phenol essentially as described for plant tissues (Hurkman & Tanaka 1986).

For Northern blot analysis, total yeast RNA was extracted with hot phenol as described previously (Ausubel et al. 1987). RNA was resolved on 1.5% agarose gels after denaturation with glyoxal (Sambrook et al. 1989). Hybond-N nylon filters (Amersham) were used for the nucleic acid blotting.


We thank Dr G. de Murcia (Strasbourg, France) for the human PARP protein; Drs M. Miwa and T. Sugimura (Tokyo, Japan) for permission to use 10H monoclonal antibodies; Dr Alice Barkan for the maize cDNA library; Dr A. Levine for sharing results before publication; Dr A. Depicker for seeds of transgenic plants; Dr N. Glab for the yeast expression vector; and are particularly grateful to Dr G. de Murcia for helpful discussions and suggestions; F. De Winter and S. Buyle for technical assistance; N. Barthels and Dr G. Engler for help and suggestions for the GUS analysis of transgenic plants; Dr M. Davey for critical reading of the manuscript; M. De Cock for help in preparing it; and R. Verbanck and K. Spruyt for the artwork. This work was supported by grants from the Vlaams Actieprogramma Biotechnologie (ETC 002) and the International Human Frontier Science Program (RG-434/94 M). P.C. was supported by a postgraduate fellowship from the Biotechnology and Biological Sciences Research Council (BBSRC) and Y.C. by a grant from the Agricultural and Food Research Council (AFRC, now BBSRC). D.I. is a Research Director of the Institut National de la Recherche Agronomique (France).