The RNA N-glycosidase activity of ricin A-chain. The characteristics of the enzymatic activity of ricin A-chain with ribosomes and with rRNA.

Ricin A-chain cleaves the N-glycosidic bond at A-4324 in 28 S rRNA when intact rat ribosomes are the substrate. Cleavage occurs at a concentration of the toxin of 1 X 10(-10) M, and specificity for this single residue is retained when the concentration is as high as 3 X 10(-7) M. The apparent Michaelis constant (Km) for the reaction is 2.6 microM, and the turnover number (Kcat) is 1777 min-1. The same N-glycosidic bond is cleaved by ricin A-chain in naked 28 S rRNA, but at a greatly reduced rate. The Km value for this reaction is 5.8 microM. The results suggest that the A-chain has a similar affinity for 28 S rRNA in ribosomes and in the absence of ribosomal proteins. Ricin A-chain has no effect on 23 S rRNA in Escherichia coli ribosomes, however, the N-glycosidic bond at A-2600 in naked 23 S rRNA is cleaved by the toxin; this corresponds to the ricin site in eukaryotic 28 S rRNA. Since the Km value (3.3 microM) for the reaction with E. coli 23 S rRNA approximates that obtained with rat liver ribosomes, it is possible that E. coli ribosomal protein(s) protect this site against ricin attack in intact ribosomes. Ricin A-chain also acted on naked 16 S rRNA cleaving the N-glycosidic bond of adenine at position 1014. The results suggest that ricin A-chain recognizes a specific structure in rRNA, perhaps a loop and stem having the sequence GAGA in the loop.

Ricin A-chain cleaves the N-glycosidic bond at A-4324 in 28 S rRNA when intact rat ribosomes are the substrate. Cleavage occurs at a concentration of the toxin of 1 X 10"' M, and specificity for this single residue is retained when the concentration is as high as 3 X 10" M. The apparent Michaelis constant (K,) for the reaction is 2.6 PM, and the turnover number (Kcat) is 1777 min". The same N-glycosidic bond is cleaved by ricin A-chain in naked 28 S rRNA, but at a greatly reduced rate. The K , value for this reaction is 5.8 PM. The results suggest that the A-chain has a similar affinity for 28 S rRNA in ribosomes and in the absence of ribosomal proteins. Ricin A-chain has no effect on 23 S rRNA in Escherichia coli ribosomes, however, the N-glycosidic bond at A-2600 in naked 23 S rRNA is cleaved by the toxin; this corresponds to the ricin site in eukaryotic 28 S rRNA. Since the K,,, value (3.3 PM) for the reaction with E. coli 23 S rRNA approximates that obtained with rat liver ribosomes, it is possible that E. coli ribosomal protein(s) protect this site against ricin attack in intact ribosomes. Ricin Achain also acted on naked 16  Ricin is a cytotoxic protein derived from castor beans (1) that inhibits protein synthesis by inactivating ribosomes (1).
The mechanisms of the inactivation is the hydrolysis of a Nglycosidic bond adjacent to the a-sarcin site in 28 S rRNA (2)(3)(4). The specificity of the effect of ricin A-chain, the catalytic subunit of the toxin, on ribosomes is remarkable. Only one N-glycosidic bond in 28 S rRNA is cleaved, and the other rRNAs are unaffected. The site of cleavage in rat 28 S rRNA is the adenosine at position 4324; that nucleotide occurs in a sequence, AGUACGAGAGGAAC, that is highly conserved ( 5 ) . The A-chain also acts on deproteinized 28 S rRNA causing a cleavage, probably at the same site as that in the ribosomes, however, at a reduced rate (3). Ricin A-chain does not inactivate prokaryotic ribosomes (5), and no N-glycosidic bond is cleaved in the rRNA of the particle even with a high concentration of the toxin (3). Why ricin A-chain is without activity on Escherichia coli ribosomes is not obvious since the nucleo-. 1 tide sequence and the secondary structure around the ricin site are highly conserved (5, 6). We have now examined the characteristics of the enzymatic activity of ricin A-chain with ribosomes and with naked rRNA from rat and from E. coli.

EXPERIMENTAL PROCEDURES
General-The following procedures were either described or cited previously (2,3): the preparation of rat liver polysomes, the treatment of rat liver and hepatoma ribosomes with puromycin, the preparation of ribosomes from E. coli, the incubation of the ribosomes with ricin A-chain, the extraction of rRNA with phenol and sodium dodecyl sulfate, aniline-induced chain scission at the ricin site in the RNA, the analysis of the nucleic acids by polyacrylamide gel electrophoresis, the preparation of the R-fragment' and other oligonucleotide fragments, the methods used for sequencing of RNA, and the source of the materials.
Preparation of 3 2 P -~l e d Ribosomes from Hepatoma CeuS and from E. coli-Uniformly labeled rRNA was prepared from ribosomes obtained from rat hepatoma cells, AH-130, or E. coli strain K-A19 that were grown in the presence of [32P]orthophosphate (7,8).
Calculation of the Extent of Cleavage of the N-Glycosidic Bond in the Large rRNAs by the Action of Ricin A-chain-Ribosomes or naked rRNA were incubated with ricin A-chain, and the RNA was extracted from the reaction mixture. The sample was treated with aniline to induce scission of the phosphoribose backbone on the 3'-side of the ricin-sensitive nucleotide, and the oligonucleotides were separated by polyacrylamide gel electrophoresis. The bands containing the Rfragment, 5.8 S, or 5 S rRNA were identified after exposure of the xray film and they were excised, and the radioactivity in each band was measured in a liquid scintillation counter. We assume that there is not selective loss of any rRNA species during preparation of the nucleic acids. Thus, the molar ratios of the R-fragment to 5.8 S rRNA in rat ribosomes or to 5 S rRNA in E. coli ribosomes, determined by measurement of the radioactivity, should reflect their ratio in ribosomes. The details of the method for the calculation of the concentration of R-fragment were described earlier (9).

Specificity of the Effect of Ricin A-chain on Rat Ribosomes
and on Naked rRNA-Ricin treatment of rat 80 S ribosomal particles, and subsequent scission of the rRNA with aniline, generated the R-fragment (Fig. lA), confirming an earlier observation (3). A concentration of the toxin of 1 X 10"' M (an enzyme to substrate ratio of 3333) was effective in cleaving the N-glycosidic bond at A-4324 in 28 S rRNA, and, even if the concentration was increased by 3 orders of magnitude (to 3 X M), only the R-fragment was produced (Fig. 1A). Production of the R-fragment was complete in 10 min ( Fig.   ' The abbreviations used are: R-fragment, the oligonucleotide that is cleaved from the 3' end of the large rRNA after treatment with ricin A-chain followed by exposure to aniline-the rat R-fragment contains 449 nucleotides and the E. coli R-fragment has 244 nucleotides; 553-fragment, the oligomer of 553 nucleotides that is cleaved from the 3' end of rat 28 S rRNA presumably by the action of a contaminant ribonuclease during the preparation of ribosomes (3).

8735
A B mM NaCI, 10 mM MgC12) were incubated at 37 "C with ricin A-chain in the concentrations indicated. After 10 min for ribosomes or 60 min for naked rRNA, the nucleic acids were extracted from the reaction mixture with phenol and sodium dodecyl sulfate. The RNA (5 pg) was treated with aniline at acidic pH at 60 "C for 10 min to induce scission of the phosphoribose chain at an apurinic site. The samples were analyzed by electrophoresis in 2.5% polyacrylamide, 0.5% agarose composite gels. The RNA was stained with ethidium bromide. The arrowheads designate the R-fragment and the arrows the Sfragment.
2C) when the concentration of ricin A-chain was 1 X lo-' M ( Fig. 2A). Treatment of protein-free rat rRNA with ricin A-chain caused cleavage of a N-glycosidic bond probably at the same site in 28 S rRNA as in ribosomal particles since the mobility of the fragment formed is similar (Fig. 1B). The naked rRNA, however, required 4 orders of magnitude more of the A-chain and longer incubation with the toxin as well (Fig. 2

.
Whether a smaller RNA fragment that contains the ricin site is also a substrate for the A-chain was tested. For this purpose, the 3"terminal fragment of 28 S rRNA that has 553 nucleotides was isolated from untreated ribosomes, and its 5' terminus was made radioactive (3). The fragment was incubated with ricin A-chain, and the RNA was analyzed by gel electrophoresis after treatment with aniline. The A-chain acts on this smaller fragment as is evidenced by the production of a 5"terminal oligonucleotide having a size similar to that of the R-fragment marker (Fig. 3A, lanes 1-5). During the course of these experiments, we found that only magnesium ion was essential for the activity of ricin. The optimum concentration of magnesium is about 10 mM (Fig. 3A), and, under the condition employed here, 64% of the molecules were affected. Monovalent cations such as ammonium, sodium, and potassium at concentrations exceeding 0.2 M inhibited the activity of the enzyme when assessed in the presence of 10 mM MgCl (data not shown). We noted that the molar ratio of enzyme to substrate in these experiments with the 553-fragment was about 100 times higher than with intact 28 S rRNA (Fig. 2C).
The 5'-32P-labeled 553-fragment that had been treated with ricin A-chain was subjected to RNA sequencing reactions, and a pattern of bands (Fig. 3B, lanes 3 and 6 ) was observed that is similar to those obtained with reference rRNA prepared from ricin-treated ribosomes (Fig. 3B, lanes 4 and 7). The extent of the cleavage of the band a t each time interval was determined.
Treatment of the 553-fragment with A-chain made the phosphodiester bonds at G-4323 and at A-4324 resistant to the action of ribonucleases (cf. arrowheads in Fig. 3B). (The faint bands seen at these two sites are likely to derive from hydrolyses in unaffected fragments.) Thus, the A-chain acts on the 553-fragment to cleave the N-glycosidic bond at the same site as in ribosomes. These results suggest that the A-chain recognizes a specific structure in the RNA. In addition, it is apparent that removal of ribosomal proteins decreases the sensitivity of this bond to the toxin. Thus, r-protein(s) modulate the response to the toxin.
Preparations of 32P-labeled 80 S ribosomes or of naked rRNA, at concentrations of 0.5 to 2.7 PM, were treated with ricin A-chain, followed by aniline, and the radioactivity in the R-fragment was determined (cf. "Experimental Procedures"). These steady state kinetic experiments were performed so that it was the initial velocity of the reaction that was assessed. From a double reciprocal plot of the data (Fig. 4) (lanes 3 and 6). For comparison, similar experiments were carried out on the 553-fragments from nontreated ribosomes (lanes 2 and 5) and from ricin-treated ribosomes (lanes 4 and 7). The alkaline digests (lane 1 ) were prepared with the control 553-nucleotide fragment. The arrow denotes the ricin site, A-4324, in 28 S rRNA. is 1777 min" ( Table I). These values are in reasonable agreement with those reported by Olsnes et al. (11) who used an indirect method. A similar experiment was carried out with naked rRNA and appreciably the same K,,, value was obtained; however, the Lt was much lower ( Fig.   4 and Table I).
Assuming that Kl and K2 are > > K3 (in the reaction illustrated in the legend to Table I), the finding that there is no significant difference in the K,,, values with the two substrates suggests that ricin A-chain recognizes a site in naked 28 S rRNA and binds there with the same affinity as in ribosomes. This interpretation conforms to the observation of Hedblom et al. (12) who estimated the dissociation constant (Kd) for the A-chain on rat liver ribosomes, employing Scatchard plot analysis, to be 3 PM which is close to our K,,, value. The value for Kmt with naked 28 S rRNA, it should be noted, is only an approximation since these experiments were done in noncatalytic conditions.

Effect of Ricin A-chin on Naked rRNAs from E. coli R h -
somes-Ricin A-chain does not inactivate prokaryotic ribosomes (1) nor does it cleave an N-glycosidic bond in the RNA of these ribosomes (3). However, the reason the A-chain is without effect on E. coli ribosomes is not apparent inasmuch as 23 S rRNA contains a sequence homologous to the ricin site of rat 28 S rRNA. In an attempt to solve this problem, total rRNA was isolated from E. coli ribosomes and was incubated with varying amounts of ricin A-chain (Fig. 5). An analysis of the RNA by gel electrophoresis, after the treatment of the sample with aniline, gave an unexpected result. Ricin treatment of the naked rRNA led to the cleavage of Nglycosidic bonds at more than one site as is apparent from the production of four fragments rather than two (Fig. 5, lanes  3-8, bands designated a, b, c, d). These bonds are not affected

TABLE I Michuelis constant and turnover number of ricin A-chain
The initial reaction velocity was measured using 32P-labeled ribosomes or ribosomal RNA. The parameters were calculated from the reciprocal plot of the data. Ricin A-chain catalyzed reaction.
where S is substrate, R is ricin A-chain, E-S is binary complex as an intermediate of the reaction, P is product, and K is velocity constant. by the A-chain when the rRNAs are in ribosomes (Fig. 5, lune 1) which suggests that r-protein(s) act, in some way, to protect the rRNA from ricin attack. The A-chain at a concentration of 2 X lo6 M was effective in producing fragments a and d, and, even if the concentration was increased 75-fold, no more than four fragments were formed. The formation of the fragments from 23 S rRNA was dependent on the concentration of the A-chain (Fig. 2B) and on the incubation time (Fig. 2C), and the kinetics were similar to those observed with 28 S rRNA. The amounts of fragments b and c appeared to be less than that of fragments a and d (Fig. 5). Fragments a and d were derived from 23 S rRNA and fragment b and c from 16 S rRNA; formation of the fragments was dependent on treatment with aniline (Fig. 5, lanes 10-13).

Kat
The sites of cleavage of 23 S and 16 S rRNAs by ricin A- in 25 pl of Buffer B at 37 "C for 60 min with concentrations of the A-chain indicated. Ribosomal particles were incubated in the same way but in Buffer A. The RNA was analyzed as described in the legend to Fig. 1. In lanes 3-8 the substrate was naked rRNA. In lane 1, the substrate was ribosomes. In lanes 10 and 11, preparations of 23 S rRNA, and, in lanes 12 and 13, 16 S rRNA, were analyzed after treatment with ricin A-chain, but either before (lanes 10 and 12) or after (lanes 11 and 13) exposure to aniline. In lanes 2 and 9, rRNA from a-sarcin-treated ribosomes was analyzed; the a-fragment serves as a marker.  E. coli rRNA (cf. Fig. 5) were made radioactive at the 5' termini with [yS2P]ATP and T4 kinase. The fragments were repurified by electrophoresis in 3.5% polyacrylamide gels and were partially digested with ribonuclease T1 (G), ribonuclease U2 (A), or ribonuclease PhyM (A/U). Alkaline digests of the oligonucleotides are designated OH. The digests were separated by electrophoresis in 20% polyacrylamide gels. In A , fragment a and the a-fragment were analyzed; in B, fragment b was analyzed. Nucleotides are numbered from their 5' terminus (20). chain were determined. The fragments a and b were isolated by sucrose gradient centrifugation, and each 5' terminus was labeled with [y3*P]ATP after treatment with phosphatase. Each of the radioactive fragments was repurified by gel elec-  16 S rRNA (18). and E. coli 5 S rRNA (19) are depicted. The ricin site in yeast 25 S rRNA was deduced from the size of the oligonucleotide and from the sequence. Because of the similarity in the ricin-insensitive structures of yeast and rat, we show only the latter. The site of ricin action in the conserved tetranucleotide sequence, GAGA (boxed), is designated with an arrow. trophoresis, and the 5"terminal nucleotide sequences were determined by an enzymatic method. The 5'-end of fragment a is (3-2661 which is adjacent to the a-sarcin site (Fig. 6A).
a-Sarcin cleaves the phosphodiester bond between G-2661 and A-2662 of E. coli 23 S rRNA in ribosomes. This is the first direct determination of the a-sarcin cleavage site in E. coli 23 S rRNA. The results indicates that ricin A-chain cleaves the N-glycosidic bond at A-2660 of 23 S rRNA since the /3-elimination reaction catalyzed by aniline at acidic pH resulted in chain scission on the 3' side of A-2660 giving G-2661 as the 5'-end of fragment a. This residue (A-2660) corresponds to A-4324 in rat 28 S rRNA. In a similar manner, the site of action of ricin A-chain on 16 S rRNA was identified as A-1014 (Fig. 623). Fragment c is derived from the 5'-end of 16 S rRNA (data not shown).
The K,,, value for the action of ricin A-chain on E. coli 23 S rRNA is 3.3 p~, which is close to the value obtained with rat ribosomes (Table I). The K,, for the reaction with 23 S rRNA is close to that observed with naked 28 S rRNA as the substrate. The results indicate that 23 S and 28 S rRNAs contain an equivalent structure for ricin action.
The finding that ricin A-chain cleaved the N-glycosidic bond of A-1014 in 16 S rRNA was unexpected. This, however, provided the opportunity to derive a consensus structure for the site of ricin attack on rRNA. The secondary structures proposed for the ricin-sensitive sites in rat 28 S, yeast 26 S, and E. coli 16 S and 23 S rRNAs were examined (Fig. 7A). It appears that the ricin A-chain acts on an adenosine residue in the sequences GAGA present in a loop with a stem of 6 or 7 base pairs in length (Fig. 7A). There are a number of such loop and stem structures in rRNA. These structures are presented in Fig. 7B. From the comparison of the structures in A and B, we propose that ricin A-chain acts on the first adenosine residue in the sequence of GAGA in a loop with a stem of 6 or more base pairs. In addition, the location of that adenosine residue in the loop seems to be another requirement for the A-chain action.

DISCUSSION
Direct measurement of the K,,, and Kcat for ricin A-chain yielded values of 2.6 K M and 1777 min, respectively. These values are in agreement with those reported by Olsnes et al. (11) who used an indirect method. The K,,, we obtained accounts for the extreme toxicity of ricin since the intracellular concentration of ribosomes is of the order of 1 PM (20).
It is remarkable that the K,,, for ricin A-chain with naked 28 S rRNA is the same as that for ribosomal particles. This suggests that the A-chain is able to bind to the same site on naked 28 S rRNA as in ribosomes and with a similar affinity.
The interpretation is in fact substantiated by the direct measurement of the dissociation constant (Kd). Hedblom et al. (12) have shown that 1 molecule of ricin binds to one rat liver ribosome with a Kd of 3 p~. Thus, ricin A-chain appears to recognize a specific structure in rRNA and the recognition does not seem to require protein. Moreover, intact 28 S rRNA is not essential since the 553-nucleotide 3'-fragment that contains the ricin site can serve as a substrate for the toxin. Eukaryotic ribosomal protein(s) may condition ricin action at a step after binding since their removal results in a large reduction in the Kcat.
Ricin A-chain cleaves an N-glycosidic bond in E. coli 23 S rRNA only in the absence of r-proteins. The site of action of the A-chain is A-2600, adjacent to the a-sarcin site. The K,,, and the Kcat for the reaction with E. coli 23 S rRNA were almost the same as for rat 28 S rRNA supporting the assumption that the A-chain recognizes a specific structure in the RNA. It is important that the A-chain does not act on 23 S rRNA in ribosomal particles. This suggest that prokaryotic rprotein(s) in some way interfere with the binding of the Achain. Indeed, Hedblom et al. (12) have shown directly that ricin A-chain does not bind to E. coli ribosomes.
An unexpected finding was that ricin A-chain also cleaves an N-glycosidic bond at position A-1014 in 16 S rRNA. This leads to the proposal that it is the tetranucleotide sequence, GAGA, in a loop with a stem of 6 or 7 base pairs that is recognized by ricin. Recently, Montfort et al. (21) have presented a three-dimensional structure for ricin A-chain derived from crystallographic analysis. They have found a cleft in the molecule and have suggested that it is the active center of the protein. This cleft may accommodate the recognition structure we have proposed during the cleavage reaction.