Ribosome-inactivating proteins (RIP) are a family of plant enzymes for which a unique activity was determined: rRNA N -glycosidase at a specific universally conserved position, A 4324 in the case of rat ribosomes. Recently we have shown that the RIP from Saponaria officinalis have a much wider substrate specificity: they are actually polynucleotide:adenosine glycosidases. Here we extend studies on substrate specificity to most known RIP: 52 purified proteins, both type 1 (single-chain) and type 2 (two chain, an enzymatic chain and a lectin chain) were examined for adenine release on various substrates including RNAs from different sources, DNA, and poly(A). All RIP depurinated extensively DNA and some released adenine from all adenine-containing polynucleotides tested. From experimental evidence the entire class of plant proteins, up to now called ribosome-inactivating proteins, may be classified as polynucleotide:adenosine glycosidases. The newly identified substrates may be implicated in the biological role(s) of RIP.
The class of plant proteins called ribosome-inactivating protein (RIP) are enzymes which catalyse an irreversible damage to ribosomes hydrolising the glycosidic bond of a unique, highly conserved adenosine residue on the major rRNA: A 4324 in the case of rat liver rRNA (reviewed in ref. 1 ). Until recently this was the only known substrate for RIP.
This homogeneity in substrate recognition by the various RIP has been challenged by several observations: (i) some saporins, RIP from Saponaria officinalis , and, to a lesser extent, other type 1 RIP, release more than one molecule of adenine from each ribosome ( 2 , 3 ); (ii) PAP (a RIP from the leaves of Phytolacca americana ) is capable of depurinating artificial polynucleotidic loops which are not substrate for ricin ( 4 ); (iii) saporin-L1 is capable of depurinating a variety of polynucleotides ( 5 ) and (iv) some other RIP, namely Hura crepitans RIP and some isoforms of PAP, depurinate both ribosomal and non-ribosomal substrates ( 6 ). The substrate specificity of saporin-L1 has been studied in detail, together with the optimal conditions for the enzymatic activity ( 7 ). We decided to extend the study on substrate specificity to as many RIP as available, taking into account that optimal conditions for enzymatic activity as determined for saporin-L1 appear to be quite different from those used so far for the determination of enzymatic activity of RIP on ribosomes. The results presented here show that all RIP tested do act on DNA and many on different polynucleotidic substrates, releasing adenine from the sugar phosphate backbone of poly- and polydeoxy-nucleotides. The plant proteins formerly called RIP may thus be classified as polynucleotide:adenosine glycosidases.
Materials and Methods
Sources of polynucleotide:adenosine glycosidases (i.e. RIP) and botanical classification of RIP producing plants can be found in ( 1 ). Ricin, Ricinus agglutinin (RCA120), volkensin and saporins were purified as described in ( 3 , 8 , 9 ), respectively. Pokeweed antiviral protein (PAP) has been fractionated by chromatography onto Blue-Sepharose as practised for PAP-C ( 10 ) into two isoforms: isoform 1 (lower affinity for the dye) and isoform 2 (higher affinity for the dye). Recombinant saporin-S6 [saporin-S6r: SAP-1 and SAP-3 ( 11 ), SAP-C ( 12 )], trichosanthin and recombinant ricin A chain were generous gifts, respectively from Professor M. Soria, Milan, Italy, from Dr H.W. Yeung, Hong Kong, and from Professor J.M. Lord, Coventry, UK. Other polynucleotide:adenosine glycosidases were prepared as described in the appropriate references cited in the review by Barbieri et al. ( 1 ).
Poly(A), rRNA from Escherichia coli , genomic RNA from tobacco mosaic virus (TMV) and MS 2 were from Boehringer GmbH, Mannheim, Germany. Genomic RNA (m ssRNA positive + one small satellite) from artichoke mottled crinkle virus (AMCV), a generous gift from Dr E. Benvenuto, Rome, Italy, was prepared by phenol extraction and ethanol precipitation from purified virus isolates. DNA from herring sperm (hsDNA) from Sigma (St Louis, MO, USA) was mechanically sheared and made RNA-free by treatment with DNase-free RNase A (Boehringer) for 2.5 h at 37°C. DNA was then repeatedly precipitated in ethanol to remove the enzyme and, when indicated, was melted by heating at 90°C for 5 min, followed by rapid cooling in ice. Adenine used as standard was from Sigma. Materials and equipment for low pressure chromatography were from Pharmacia LKB (Uppsala, Sweden). All other reagents were of analytical or molecular biology grade and, when possible, RNase-free. Water was Milli-Q (Waters-Millipore). Chloroacetaldehyde was prepared according to ( 13 ).
Determination of polynucleotide:adenosine glycosidase activity
Polynucleotide:adenosine glycosidase activity was determined by measuring adenine released from the substrate by HPLC after derivatisation to its fluorescent derivative ethenoadenine essentially as described in ( 7 ). Reaction conditions, previously determined as optimal for saporin-L1 ( 7 ), are reported in the legends to the pertinent Tables and Figure 1 .
Effect of ribosome-inactivating proteins on genomic viral ribonucleic acids
Fifty-two purified proteins, representing the vast majority of known RIP, were tested for depurinating activity on MS 2 genomic RNA ( Table 1 ). Reaction conditions chosen were those optimal for the determination of RIP activity on rat liver ribosomes [inhibition of poly(U)-directed phenylalanine polymerisation] as determined in previous experiments (results not shown). These screening experiments were performed with a high concentration of RIP to detect activity under conditions which could be non-optimal, and in order to overcome the possible lack of cofactors ( 14 ). These enzyme concentrations are anyhow lower than those naturally occurring in subcellular compartments where RIP are localised [e.g. Phytolacca americana and Saponaria officinalis tissues ( 15 , 16 )] and lower than those necessary to show activity of ricin and other RIP on naked ribosomal RNAs [for a review see ( 1 )].
Only some saporins, RIP from Saponaria officinalis , released significant amounts of adenine from MS 2 RNA, in excess of one mol per mol of RNA ( Table 1 ). All other RIP showed low or no detectable activity on this substrate in present experimental conditions. A variability in the activity of different batches of natural saporin-S6 was found, which may be due to the fact that preparations of natural saporin-S6 are a variable mixture of different isoforms ( 17 ). Thus we tested three different recombinant saporin-S6 gene products, which were all inactive on viral RNAs. Saporin-L1, the RIP with the highest depurination rate on MS 2 RNA, was also active on plant virus RNAs, from tobacco mosaic virus (TMV) and artichoke mottled crinkle virus (AMCV), although with lower rates.
Activity of ribosome-inactivating proteins on herring sperm DNA, rRNA and poly(A)
All RIP of either type tested released adenine from hsDNA, although with different activity: RIP from Cucurbitaceae (bryodins, luffin a, momordin I, trichosanthin and trichokirin), barley RIP, mapalmin and all type 2 RIP were substantially less active than other RIP. Surprisingly, ricin and related two-chain RIP, which must be reduced to act on ribosomes ( 18 ), were active on DNA even before reduction. Denaturation of DNA prior to assay did not increase significantly depurination rates, except in the case of momordin I (results not shown).
Activity on purified E.coli rRNA in present conditions was highly variable. To be noted however that all type 2 RIP, either reduced or not, depurinated rRNA only marginally.
Only saporins and RIP from Phytolacca dioica (PD-S) depurinated extensively poly(A) at pH 6.0; saporin-L1, -L2 and -R2 being the most active ( Table 2 ); marginal depurination of poly(A) was observed with Hura crepitans RIP and bryodin-R, whilst no activity on this substrate was detected with the other RIP tested, including recombinant saporin-S6r and those (PAP-R and trichosanthin) which were marginally active on viral genomic RNA.
The dependance of the reaction on enzyme concentration was determined for PD-S2 (a RIP from the seeds of Phytolacca dioica ) ( Fig. 1 ). From linear regression analysis a release of 856 (from hsDNA) and 657 [from poly(A)] mol of adenine per mol of enzyme was calculated.
The experiments described above show that all tested RIP remove more than one adenine residue from DNA, and some also from rRNA and other polyribonucleotides. The possibility of a contamination of RIP preparations by other enzymes seems to be excluded by the most careful purification procedures ( 3 ), by the low amount of enzyme necessary for activity ( Fig. 1 ) and by the fact that RIP from all sources, including recombinant ones, were active on hsDNA. To our knowledge there are no enzymes with the activity described in this work. The nearest are probably DNA glycosylases [reviewed in ( 19 )] whose similarities with RIP have been discussed in ( 7 ).
Within the limit of the small number of observations, there is consistency between results obtained using bacteriophage RNA (from MS 2) and plant viral RNAs, possible natural substrates.
Type 2 RIP do not require reduction for activity on DNA, whereas they must be reduced to depurinate ribosomes. The different steric hindrance of ribosomes and hsDNA may account for this difference. The recent report that a mutant non-reducible form of ricin is still cytotoxic ( 20 ) suggests that depurination of non ribosomal substrate(s) may contribute to the toxicity of ricin and related toxins.
Depurination of poly(A), RNA and DNA proceeds in the absence of any cofactor, although an influence of cofactors, as those characterised by Carnicelli et al. ( 14 ) cannot be excluded. DNA, RNA, and poly(A) all undergo multiple deadenylation by some RIP indicating that a specific nucleotide sequence in the substrate is not required for these enzymes to act.
Several molecular structures of RIP have been elucidated in detail and only one possible enzymatic site has been identified indicating that the activity on ribosomes and on other substrates is due to the same active site.
Interactions between RIP and DNA have been reported previously. Relevant to present work are the observations that mice poisoned with ricin presented single-strand breaks in DNA ( 21 ) and some type 1 RIP and ricin ( 22–26 ) introduced nicks into supercoiled double-stranded DNA. Interestingly, both ricin and camphorin, another type 2 RIP, acted unreduced ( 23 ), consistently with what observed in present work with hsDNA ( Table 2 ). It was found also that gelonin (referred to as GAP31) transforms supercoiled DNA into relaxed and linear forms, and that gelonin-derived peptides bind to DNA and RNA ( 27 ). It is possible that these RIP actually removed adenine from DNA, which became more fragile and then broke spontaneously or under the action of contaminating nucleases.
DNA and ribosomes, that are depurinated consistently by all RIP, possibly are the best candidates for the natural substrate of these enzymes. Action on non-ribosomal RNA and poly(A) may be either accidental or expression of a possibly additional biological role. Thus the subdivision of RIP based on substrate specificity may be correlated to their natural activity.
RIP with high activity on DNA come from plant species belonging to the order of the Caryophyllales, whilst RIP from Cucurbitaceae have a lower activity, thus suggesting that there may be a evolutionary-related difference either in the substrate fine specificity or in the requirements for maximal activity. Moreover, it appears that the different forms of RIP produced by the same plant ( 1 ) often are functionally different, all being active on mammalian ribosomes but having very different activity on nucleic acids. This suggests that RIP with distinct substrate specificity produced by a single plant species may have a different functional role. No correlation between the tissue origin (seed, leaf, etc.) and the spectrum of activity on nucleic acids was observed.
Present results may help to understand the function of RIP in Nature. Several roles have been proposed for RIP, each one with an associated substrate for the enzymatic action [reviewed in ( 1 )]. Some of these roles may be readdressed on the basis of the newly identified substrate specificities of RIP, if the latter apply also in vivo :
a defensive role against plant viruses ( 1 , 28 , 29 ). In the widely accepted mechanism of antiviral action the target of RIP are the autologous ribosomes: viral infection alters the cell structure, RIP gain access to ribosomes and inactivate them, with arrest of protein synthesis, death of the cells, and prevention of viral replication and spread. Although a direct action on the entire virus has been excluded since first experiments, present results suggest that the RIP deadenylating different forms of RNA could act directly on viral RNA. Localisation of PAP in high concentrations between cell wall and plasma membrane in pokeweed leaves ( 15 ) should allow for RIP-viral nucleic acid contact. Since a DNA template is present at some stage of all viral infections, depurination of virus-derived DNA may also give protection against all viruses. Present work indicates that retroviral DNA could be damaged as suggested for HIV ( 24 ), and, although definitely not physiological, anti-HIV activity of RIP is suggestive of mechanisms which may work also in plant systems;
a role in the regulation of cell metabolism. In this hypothesis the targets are autologous ribosomes, which are inactivated to stop protein synthesis by senescent and/or altered ribosomes, or in senescent tissues. RIP activity appears, or increases when seeds mature ( 3 ) and in senescent or stressed leaves ( 6 , 29 , 30 ), in correlation to events leading to arrest of the metabolism of plant cells. The activity on DNA may well contribute to this regulation, since the removal of few or even one adenine residue may be sufficient to disrupt transcription. An effect on poly(A) tails of mRNAs and on other RNAs may also be of some importance in the regulation of cell metabolism. Depurination of autologous DNA may be a cause of programmed cell death in the senescence of leaves. Also, DNA damage may have a role in the pathogenesis of the lesions induced by RIP, and particularly in the DNA fragmentation typical of apoptosis observed in cultured cells exposed to RIP of either type ( 31–37 ) and in tissues of rats poisoned with ricin and abrin ( 38 ). It should be recalled, however, that inhibition of protein synthesis is per se sufficient to induce apoptosis both in rat liver ( 39 ) and in cultured cells ( 31 , 32 , 40 ), and thus the apoptosis induced by RIP could be accounted for by the ribosomal inactivation brought about by these proteins.
The identification of new in vitro substrates may allow for the identification of yet undiscovered plant and/or animal poly-nucleotide:adenosine glycosidases, which might not be active on mammalian ribosomes, the substrate now used to detect RIP.
This study was supported by the Università di Bologna, Funds for selected Topics , by grants from the Ministero dell'Università e della Ricerca Scientifica, from the Consiglio Nazionale delle Ricerche, special projects Biotecnologie e Biostrumentazione and ACRO , from the Associazione Nazionale per la Ricerca sul Cancro, from the Regione Emilia Romagna, and by the Pallotti's Legacy for Cancer Research.