Ribosome-inactivating proteins (RIPs) are N-glycosylases that remove a specific adenine from the sarcin/ricin loop of the large rRNA in a manner analogous to N-glycosylases that are involved in DNA repair. Some RIPs have been reported to remove adenines from single-stranded DNA and cleave double-stranded supercoiled DNA. The molecular basis for the activity of RIPs on double-stranded DNA is not known. Pokeweed antiviral protein (PAP), a single-chain RIP from Phytolacca americana, cleaves supercoiled DNA into relaxed and linear forms. Double-stranded DNA treated with PAP contains apurinic/apyrimidinic (AP) sites due to the removal of adenine. Using an active-site mutant of PAP (PAPx) which does not depurinate rRNA, we present evidence that double-stranded DNA treated with PAPx does not contain AP sites and is not cleaved. These results demonstrate for the first time that PAP cleaves supercoiled double-stranded DNA using the same active site that is required for depurination of rRNA.
Ribosome-inactivating proteins (RIPs) produced by plants, fungi and bacteria exhibit strong inhibitory activity on protein synthesis by removing a specific adenine (A4324) from 28S rRNA of rat ribosomes or its equivalent residue A2660 from 23S rRNA of Escherichia coli ribosomes (1,2). A4324 is located at a highly conserved and surface-exposed stem-loop region (sarcin/ricin loop) of the rat 28S rRNA. Depurination of the sarcin/ricin loop interferes with both the EF-Tu (or EF-1)-dependent binding of aminoacyl-tRNA and the GTP-dependent binding of EF-G (or EF-2) to ribosomes, arresting protein synthesis at the translocation step (3).
RIPs can be divided into two classes: (i) type I RIPs, such as pokeweed antiviral protein (PAP), which consist of a single polypeptide chain possessing RNA N-glycosylase activity; and (ii) type II RIPs, such as ricin, which have a type I-like catalytic subunit (A chain) linked to a cell binding subunit (B chain) (3). PAP and several other RIPs exhibit strong antiviral activity against both plant and animal viruses, including influenza virus (4), poliovirus (5), herpes simplex virus (6) and HIV (7). Positive correlations were reported between RIP-catalyzed depurination of tobacco ribosomes and antiviral activity of exogenously applied RIPs (8). However, several lines of evidence suggested that antiviral effects of PAP and other RIPs might not be only due to inhibition of host protein synthesis. Studies with vero cells infected with herpes simplex virus indicated that PAP inhibited viral DNA synthesis by 90%, while only 30% of host protein synthesis was inhibited (9). Similar studies with PAP against HIV-1 showed that virus replication could be significantly inhibited at PAP concentrations which are not toxic to the host cell (7). We have shown that a non-toxic C-terminal deletion mutant of PAP inhibited viral infection, but did not show detectable depurination of tobacco ribosomes in vivo (10). These results suggested that PAP and other RIPs might interfere with virus replication by a mechanism other than host ribosome inactivation. One possible mechanism is that RIPs might directly depurinate the nucleic acid of the invading pathogens. If this hypothesis is true, then the sarcin/ricin loop would not be the only target of RIPs. Li et al. showed that a single chain RIP, trichosanthin from Trichosanthes kirilowii, cleaved supercoiled double-stranded DNA into relaxed and linear forms in vitro (11). Similar activity of other RIPs on supercoiled double-stranded DNA template was observed with dianthin, gelonin, luffin, cinnamomin and saporin (12,15). Gelonin inhibited growth of the malaria parasite, Plasmodium falciparum, in erythrocytes by eliminating the 6 kb mitochondrial DNA of the parasite, suggesting that biological activity on DNA may contribute to the cytotoxicity of gelonin in vivo (16).
Recent results indicated that gelonin, PAP and ricin damage single-stranded DNA by removing adenines and cleaving at the resulting abasic sites (17). Although it has been shown that RIPs from 50 different species are able to depurinate synthetic nucleic acids, including herring sperm DNA and poly(A) (18), previous reports describing the activity of RIPs in cleaving and linearizing supercoiled DNA templates in vitro did not rule out the possibility of nuclease contamination in the purified protein preparations (19). In this report, using a highly purified preparation of PAP and its enzymatically inactive form, PAPx, which cannot depurinate rRNA, we demonstrate for the first time that PAP cleaves double-stranded supercoiled DNA using the same active site required to depurinate rRNA.
Materials and Methods
Purified PAP was purchased from Calbiochem (USA), chloroacetaldehyde from Fluka (Switzerland), aniline from Fisher (Pittsburgh, PA) and radioisotopes from NEN.
Purification of PAPx
A mutant PAP (PAPx), which contained the E176V mutation at its active-site, was previously described (20). The full-length cDNA encoding PAPx was transformed into tobacco cv Samsun (nn) by Agrobacterium-mediated transformation (10). Purification of PAPx was performed as described by Irvin (21) with the following modifications. Mature leaves (100 g) of transgenic tobacco line NT144 (10), expressing high levels of PAPx, were ground in 400 ml of pre-chilled buffer containing 50 mM Tris-HCl (pH 7.5), 0.2 mM EDTA, 5 mM 2-mercaptoethanol and 1% polyvinylpyrrolidone. The extract was filtered through four layers of cheesecloth and two layers of Miracloth. The filtrate was precipitated by slowly adding ammonium sulfate to 60% saturation and stirring for 2 h at 4°C. After centrifugation at 11 000 g for 20 min, ammonium sulfate was added to the supernatant to 95% saturation and the mixture was stirred overnight at 4°C. The pellet was collected by centrifugation at 11 000 g for 30 min and dissolved in 20 ml of Q buffer (20 mM Tris-HCl, pH 7.5, 0.2 mM EDTA and 1 mM 2-mercaptoethanol). The sample was dialyzed in four changes of Q buffer. Following dialysis, the protein sample was centrifuged again at 11 000 g for 10 min, and the supernatant was loaded onto Q buffer-equilibrated Q15 anion exchange filters (Sartorius) connected to a Bio-Rad FPLC Biologic System. The flow-through from the filters was then loaded onto a Bio-Rad S2 cation exchanger column. After washing with 3 vol of Q buffer, the proteins were eluted in a 0–30% NaCl gradient in Q buffer. The fractions containing PAPx were assayed by SDS-PAGE followed by immunoblot analysis.
SDS-PAGE and silver-staining
PAP and purified PAPx (50 ng) samples were separated through a 12% SDS-polyacrylamide gel. Following electrophoresis, the gel was fixed in a solution containing 50% methanol, 10% acetic acid, 0.05% formaldehyde at room temperature for 1 h. The gel was washed with 50% ethanol three times each for 20 min, and then treated with 0.02% Na2S2O3·5H2O for 1 min. After rinsing three times with water, the gel was incubated with 0.2% silver nitrate solution containing 0.03% formaldehyde for 20 min. After washing with water, the gel was transferred to 6% sodium carbonate solution containing 0.0004% Na2S2O3·5H2O and 0.05% formaldehyde until the appearance of bands. The gel was then fixed in a solution containing 50% methanol and 10% acetic acid and air-dried in gel drying film.
DNase activity assay of PAP and PAPx
One microgram of pG-1 DNA (22) was incubated with different concentrations of PAP or PAPx in 10 µl of RIP buffer (50 mM Tris-HCl, pH 7.5, 50 mM KCl, 10 mM MgCl2) at 37°C for various times as indicated in each figure. The samples were analyzed in a 1% agarose gel and visualized by staining with 0.05% ethidium bromide.
Aniline treatment of plasmid DNA was carried out using the method previously described for rRNA (10). After the DNA samples were treated with PAP or PAPx, 50 µl of 0.1 M aniline in 3% acetic acid was added to the samples. The samples were incubated on ice for 10 min and precipitated with ethanol.
Labeling of DNA treated with PAP
pG-1 plasmid (100 ng) was first treated with 100 ng of PAP or PAPx in 10 µl RIP buffer at 37°C for 30 min. The samples were then extracted twice with phenol-chloroform (1:1), followed by ethanol precipitation and washing. The dried pellets were dissolved in 5 µl of sterile water. The samples were then labeled with E.coli DNA polymerase I Klenow fragment and [α-32P]dCTP at room temperature for 15 min. Unincorporated nucleotides were removed by passing the sample through a G-50 column. The final reaction products were separated through a 1% agarose gel, and autoradiographed.
Adenine assay by a fluorimetric method
Adenine was quantified using the method described by Zamboni et al. (23). pG-1 DNA (6 pmol) was incubated with 60 pmol of PAP in a 50 µl volume containing 50 mM Tris-HCl pH 7.5, 50 mM KCl and 10 mM MgCl2 at 37°C for 1 h. After incubation, plasmid DNA was precipitated with 2 vol of ethanol. The ethanol-soluble fractions were diluted to 1 ml with water and 0.4 ml of 0.14 M chloroacetaldehyde containing 0.1 M sodium acetate (pH 5) was added to each sample. The samples were incubated at 85°C for 1 h. After cooling to room temperature, fluorescence was measured in an Amico SPF-125 spectrophotofluorometer at an excitation wavelength of 310 nm and an emission wavelength of 410 nm. Purified adenine at concentrations ranging from 1 to 1200 pmol (Sigma) were used as standards for calculation of adenine in the samples.
PAP cleaves supercoiled DNA into relaxed and linear forms
We used a highly purified PAP preparation to determine if PAP can act on supercoiled DNA. Sequence analysis of this protein sample revealed only a single N-terminal sequence corresponding to PAP, suggesting that purified PAP is homogenous. To determine if PAP is active on supercoiled DNA templates, pG-1 plasmid (9 kb) was incubated with increasing concentrations of PAP. As shown in Figure 1, upon incubation with PAP, the supercoiled form of pG-1 was converted to the relaxed and linear forms in a concentration-dependent manner. DNA treated with 1 µg PAP was completely converted to the relaxed and linear forms (Fig. 1). Similar results were obtained with other supercoiled plasmids, such as pGEM, pUC18 and pBR322 (data not shown). RNA N-glycosylase activity of PAP on rRNA requires Mg2+ for optimal activity (24). As expected, the activity of PAP on double-stranded DNA was greatly stimulated in the presence of 10 mM divalent cations, Mg2+, Ca2+ or Mn2+ (data not shown).
PAPx does not cleave supercoiled double-stranded DNA
DNA glycosylases catalyze cleavage of the N-glycosylic bond linking a damaged base to the DNA sugar-phosphate backbone. Some of these enzymes exhibit an abasic (apyrimidinic/apurinic) lyase (AP lyase) activity (25). To determine if depurination is required for cleavage of double-stranded DNA by PAP, we used a previously isolated PAP mutant, PAPx, which had a point mutation at the active site (E176V). We previously showed that E176V mutation abolished rRNA glycosylase activity in yeast and transgenic plants (10,20). We purified PAPx from the transgenic tobacco line NT144-12, which expressed the highest levels of PAPx (10), using a protocol similar to that used for purification of PAP from pokeweed (21). A major peak containing a 29 kDa protein eluted at the 10% salt concentration. We confirmed that this protein peak corresponds to PAPx by immunoblot analysis. PAPx purified by this method is homogenous as judged by SDS-PAGE and silver-staining (Fig. 2).
Purified PAPx did not depurinate ribosomes in vivo (10) or inhibit translation in rabbit reticulocyte lysate (20). The activity of purified PAPx on supercoiled DNA was assayed as previously described for PAP. As shown in Figure 3, purified PAPx did not affect supercoiled DNA, while similar amounts of purified PAP produced linear and relaxed forms of the supercoiled DNA template. These results indicate that Glu176, required for rRNA depurination, is also critical for PAP to relax and cleave supercoiled double-stranded DNA.
Plasmid DNA treated with PAP contains AP sites
RIP-mediated depurination of the large ribosomal subunit RNA results in susceptibility of the RNA sugar-phosphate backbone to hydrolysis at the depurination site. Hence, when depurinated rRNA is treated with aniline, a small fragment is released, which is 367 nt in the yeast 26S rRNA (3). To determine if there are AP sites on double-stranded supercoiled DNA treated with PAP, we incubated PAP treated plasmid DNA with aniline. As shown in Figure 4, when plasmid DNA was treated with PAP and subsequently with aniline, both relaxed and supercoiled forms of DNA were hydrolyzed (Fig. 4, PAP+). When DNA was treated with PAPx followed by aniline, neither form was significantly affected (Fig. 4, PAPx+). Similarly, aniline treatment in the absence of PAP did not have a significant effect on the supercoiled or relaxed forms of plasmid DNA (Fig. 4, Control). Some DNA was lost due to precipitation, which was evident after aniline treatment in PAPx-treated plasmid DNA (PAPx-versus PAPx+). However, there was a significant difference in the amount of intact DNA remaining when plasmid DNA incubated with PAP or PAPx was treated with aniline (PAP+ versus PAPx+). Both supercoiled and relaxed forms of DNA treated with PAP were hydrolyzed after aniline treatment, suggesting that PAP creates multiple AP sites that are susceptible to cleavage by aniline. In contrast, only a slight decrease in the amount of DNA was observed when plasmid DNA was incubated with PAPx followed by aniline. Similar results were obtained when treatment with 0.1 M NaOH was used to detect AP sites. Plasmid DNA treated with PAP was hydrolyzed after NaOH treatment, while plasmid DNA treated with NaOH without prior treatment with PAP was not affected (data not shown).
PAP depurinates double-stranded supercoiled DNA
To obtain direct evidence for depurination of supercoiled DNA, we used a fluorimetric assay to determine if adenine is released from supercoiled plasmid DNA treated with PAP (23). Adenine can be converted to a fluorescent product, etheno-adenine, upon reaction with chloroacetylaldehyde. Fluorescence of the adenine derivative at 410 nm is proportional to the amount of adenine present in the samples (23). As shown in Table 1, there is an ∼10-fold increase in adenine levels after plasmid DNA is treated with PAP, compared to the control DNA without PAP treatment. These results indicate that PAP removes adenine from supercoiled plasmid DNA.
Nicks generated by PAP have free 3′ hydroxyl termini
Many DNA glycosylases, which remove damaged bases, also carry AP lyase activity. AP lyases break the DNA backbone by catalyzing β-elimination, producing a trans α,β unsaturated aldehyde which cannot be extended by DNA polymerases (26). PAP may cleave double-stranded supercoiled DNA by a similar mechanism; alternatively, it may nick the DNA by an endonuclease mechanism. We used the E.coli polymerase I Klenow fragment to determine if PAP-treated DNA contains a free 3′-OH terminus that could be labeled with [α-32P]dCTP by the exchange reaction. In the presence of a single nucleotide, the 3′-5′ exonuclease activity of the Klenow fragment will degrade double-stranded DNA from the 34-hydroxyl terminus, while its polymerase activity will fill in the single-stranded region, incorporating [α-32P]dCTP. As shown in Figure 5, plasmid DNA digested with BamHI was labeled. Plasmid DNA treated with PAP was also labeled and the position of the labeled bands corresponded to both the relaxed and linear forms. The relaxed and supercoiled forms of the untreated plasmid DNA (pG-1) or plasmid DNA treated with PAPx were not labeled, indicating that they are not nicked. The nicked DNA generated by PAP could be labeled by the Klenow fragment of DNA polymerase, suggesting that phosphodiester bond cleavage is not through β-elimination catalyzed by DNA glycosylase/AP lyases. These results are in agreement with the earlier observation that the nicks generated by treatment of supercoiled DNA with cinnamomin and camphorin have free 5′-phosphate termini, which could be labeled with [γ-32P]ATP after treatment with alkaline phosphatase (14). These results indicate that PAP cleaves double-stranded DNA by an endonuclease mechanism and Glu176 at the active site is required for cleavage of plasmid DNA.
There is controversy regarding the activity of RIPs on substrates other than the sarcin/ricin loop of the rRNA and about the mechanism of double-stranded DNA degradation. Ling et al. (14) reported that the nicked and linear forms generated by the action of RIPs on supercoiled DNA were not labeled by [3H]sodium borohydride, leading them to conclude that RIPs do not act by a DNA glycosylase mechanism. Furthermore, since boiling the ricin A chain destroyed its depurination activity on 28S rRNA, but only reduced its ability to cleave DNA, the two activities were described as independent (13). More recent results showed that the N-glycosylase and DNase activities of PAP were separately and specifically inactivated by chemical modification and heat, and recombinant PAP purified from E.coli did not have DNase activity, leading to the conclusion that the DNase activity of PAP is due to the presence of contaminating nucleases (19). Barbieri et al. suggested that the DNase activity associated with RIPs could be due to non-specific depurination of DNA (18). Recently, Nicolas et al. (17) have shown that gelonin, PAP and ricin damage single-stranded DNA by removing a specific set of adenines and cleavage at the resulting abasic sites. Using an oligonucleotide as a substrate, they demonstrated that these enzymes are monofunctional glycosylases which form a Schiff base intermediate which is more characteristic of bifunctional glycosylase/lyase enzymes (17).
Different RIPs have conserved residues that are clustered near the catalytic site, suggesting that any other activity might be the product of the same active site. If the same active site is responsible for both activities, elimination of one activity might have a similar effect on the other. Alternatively, if the DNase activity is due to a contamination, elimination of the RNA N-glycosylase activity might have no effect on the DNase activity. In this report, we demonstrate that PAP can convert supercoiled double-stranded DNA into relaxed and linear forms using the same active site required for depurination of rRNA. Glu176, which is conserved among all RIPs sequenced to date and has been shown to be critical for depurination of rRNA (10), is also critical for cleavage of double-stranded DNA, since mutation of Glu176 to Val abolishes the activity of PAP on both nucleic acid substrates.
During cleavage of supercoiled duplex DNA molecules by PAP, supercoiled molecules are rapidly converted to nicked relaxed molecules, possibly as a result of nicking at transient single-stranded regions in the supercoiled DNA. At high enzyme concentrations, the nicked form is converted to linear molecules. Partial single-stranded A-T-rich regions have been reported in supercoiled DNA (27), which are susceptible to attack by a single-stranded DNA specific endonuclease (28). These singlestranded regions may mimic the sarcin/ricin loop of ribosomal RNA, and may be the potential targets for PAP. A-T-rich regions of pBR322 are the preferential targets for cleavage by gelonin (15). Using atomic force microscopy, Wu et al. (29) showed that several RIPs bind to single-stranded regions of supercoiled DNA. Thus, the unique topology of supercoiled DNA seems to be required for PAP recognition since the linearized double-stranded DNA is not a substrate for single chain RIPs like cinnamomin and camphorin (14). It has been reported that single-stranded DNA is the preferred substrate for the nuclease activity of gelonin (30), supporting the hypothesis that PAP and other RIPs may preferentially target the single-stranded regions of double-stranded supercoiled DNA. Similarly, single-stranded DNA endonucleases are able to nick supercoiled DNA because of the partial single-stranded character of such molecules (31). Following rapid nicking of supercoiled molecules, PAP may cleave single-stranded regions opposite nicks at a reduced rate, linearizing the supercoiled plasmid DNA.
The stoichiometry of the reaction, which represents an excess of the enzyme is consistent with previous observations on the activity of RIPs on naked RNA. Endo and Tsurugi (32) demonstrated that ricin A-chain depurinated rat rRNA at A4324 in naked 28 S rRNA much less efficiently than in intact ribosomes, suggesting that ribosomal proteins may play a role in modulating the sensitivity of ribosomes to RIPs. We recently demonstrated that PAP binds to ribosomal protein L3 to access yeast ribosomes (33), suggesting that the differences in sensitivity of ribosomes and naked RNA may be due to interactions of RIPs with ribosomal proteins. A similar situation may exist with DNA, where proteins or other factors that interact with DNA may play a role in modulating the sensitivity of DNA to PAP in vivo.
DNA glycosylases remove damaged bases from DNA by cleavage of the N-glycosylic bond. The simple glycosylases are limited to the N-glycosylic bond cleavage, the product of which is an abasic site (25). The glycosylase/AP lyases carry out the glycosylase reaction and the subsequent AP lyase reaction, using an amino group as a nucleophile, resulting in an imino enzyme intermediate. This intermediate may be hydrolyzed, forming an abasic site or may undergo a β-elimination reaction resulting in lyase scission of the sugar-phosphate backbone (25). We present evidence that the DNA nicked by PAP contains free 3′-hydroxyl termini that can be labeled, indicating that PAP does not cleave DNA by the β-elimination reaction characteristic of DNA glycosylase/AP lyases. Instead, it removes adenines from DNA as a monofunctional glycosylase and cleaves supercoiled duplex DNA molecules as an endonuclease. Structure-function analyses of several DNA repair glycosylases identified a critical aspartic or glutamic acid residue that if mutated abolishes the glycosylase activity (34). Since a point mutation at a critical glutamic acid residue abolished the activity of PAP on double-stranded DNA, the results presented in this report are consistent with those obtained with DNA glycosylases.
In Saccharomyces cerevisiae, the RAD1 and RAD10 genes are involved in DNA nucleotide excision repair and in a pathway of mitotic recombination that occurs between direct repeat DNA sequences (35). The RAD1 and RAD10 complex, which degrades single-stranded DNA endonucleolytically, also cleaves supercoiled duplex DNA molecules (36). The N-terminal sequence of PAP shows 40% amino acid sequence homology to the yeast RAD10 gene product. The homology between PAP and enzymes involved in DNA excision repair suggest that PAP and other RIPs may access DNA by mimicking enzymes involved in DNA repair. However, in contrast to DNA glycosylases, which remove damaged, mismatched bases, PAP can remove unmodified adenines from double-stranded supercoiled DNA, suggesting that it may protect plants or animals from infection by damaging the DNA of a pathogen.
In vitro depurination activity on templates such as poly(A), DNA and RNA has been described for 50 RIPs (18). Using DNA labeled in the purine ring of adenine as a substrate, Brigotti et al. recently showed that PAP-S, ricin and shiga-like toxin I release labeled adenine from double-stranded DNA (37). We show here that the active site required for depurination of rRNA is also required for depurination and cleavage of supercoiled, doublestranded DNA. Although the same active site is required for depurination of rRNA and DNA, the sequences required for recognition of DNA substrates may differ from those required for recognition of rRNA. We have previously shown that an intact active site is required for antiviral activity of PAP (10). If depurination of viral RNA or DNA is the basis for antiviral activity, PAP mutants which do not depurinate host rRNA may still inhibit viral infection. In support of this, a non-toxic C-terminal deletion mutant which did not show detectable depurination of host ribosomes in vivo inhibited viral infection, suggesting that antiviral activity can be dissociated from depurination of host rRNA (10). These results suggest that rRNA depurination may not be the only mechanism for antiviral activity; DNA-damaging activity of PAP may also contribute to its antiviral action.
We thank Drs Peter Day, Jonathan Dinman and Katalin Hudak for critical reading of the manuscript, and Dr Maureen Bonness for the purified PAPx. This work was supported by the National Science Foundation grant MCB96-31308 to N.E.T. and the New Jersey Commission on Science and Technology.