Structure and Mechanism of Action of the Cytotoxic Ribonuclease α-Sarcin

4 Structure and Mechanism of Action of the Cytotoxic Ribonuclease c -Sarcin IRA G. WOOL Department of Biochemistry and Molecular Biology The University of Chicago Chicago, Illinois 60637

I. II. III. IV. V. VI. VII. VIII. IX. X. XI. XII.

Discovery of a-Sarcin and Determination of the Basis of Its Cytotoxicity Substrate Specificity of a-Sarcin a-Sarcin Cleavage Site Sequence Structure of a-Sarcin and of Related Aspergillus Toxins Mechanism by Which ot-Sarcin Enters Cells a-Sarcin Recognition Elements in 28S r R N A Conformation of the Sarcin Domain in 28S r R N A Effect of Mutations of the Analog of G4319 in 28S r R N A on Recognition of Oligoribonucleotides by ct-Sarcin Phenotype of Mutations of G2655 in Escherichia coli 23S r R N A Relationship of the ot-Sarcin R N A Identity Element to Selection by the Toxin of the Unique Site of Phosphodiester Bond Hydrolysis Three-Dimensional Structure of Restrictocin, a Homolog of ct-Sarcin. Binding to a-Sarcin Domain R N A and the Catalytic Mechanism Coda References

I. D I S C O V E R Y OF a - S A R C I N A N D D E T E R M I N A T I O N OF T H E BASIS OF ITS C Y T O T O X I C I T Y

In 1956 Birger Olson and colleagues were asked to expand the scope of the antibiotic screening program in the Michigan Department of RIBONUCLEASES: STRUCTURES AND FUNCTIONS

131 Copyright 9 1997by AcademicPress, Inc. All rights of reproduction in any form reserved.

132

Ira G. Wool

Health to include a search for anticancer agents. Soon afterward they obtained a sample of soil from a Michigan farm and isolated from it the mold Aspergillus giganteus (MDH18894) (Olson and Goerner, 1965; Olson et aL, 1965). The mold was found to produce a substance that was marvelously effective in inhibiting the growth of tumors in rats, particularly of sarcoma 180 and carcinoma 755. Birger Olson christened the basic protein, a-sarcin, for antisarcoma, a-Sarcin had a clinical trial, but sad to say it proved to be insufficiently effective and toxic. For more than a decade a-sarcin languished. Its renaissance is an oft-told tale of a potential therapeutic agent whose discovery created great excitement and commensurate expectations, that failed in the clinic, and was buried in disappointment, only to be resurrected by biochemists and molecular biologists because of its utility in analyzing the structure and function of ribosomes and of ribonucleoprotein complexes. Vazquez (1979; Fernandez-Puentes and Vasquez, 1977; Conde et al., 1978) and Cundliffe (Hobden and Cundliffe, 1978), and associates, established that the mechanism of the cytotoxicity of a-sarcin, and of the related Aspergillus toxins restrictocin and mitogillin, was an inhibition of protein synthesis, and that this inhibition was, in turn, the result of an effect on ribosomes. Then in the most important single experiment, Schindler and Davies (1977) showed that a-sarcin inactivated ribosomes by cleaving a fragment, called the a fragment, from the RNA in the large subunit of yeast ribosomes. It was established that only a single fragment was produced by the toxin and that the fragment was probably derived from the 3' end of 28S rRNA (Schindler and Davies, 1977). Restrictocin and mitogillin have the same mechanism of action as asarcin (Fernandez-Luna et al., 1985; Fando et al., 1985). Moreover, the primary structure of restrictocin differs from that of mitogillin by only one residue, and they share 86% amino acid sequence identity with a-sarcin.

II. S U B S T R A T E SPECIFICITY OF a - S A R C I N

When ribosomes are the substrate, the cleavage of RNA by a-sarcin is remarkably specific (Endo and Wool, 1982; Endo et al., 1983). Not only are 5S, 5.8S, and 18S rRNAs spared, but only one phosphodiester bond in 28S rRNA is hydrolyzed (Endo et al., 1983). The specificity cannot be attributed to the unavailability of other phosphodiester bonds

4

Structure and Mechanism of a-Sarcin

133

because, if ribosomes are treated with other ribonucleases, all species of rRNA are progressively digested and no specific oligoribonucleotides are formed (Endo and Wool, 1982). The c~ fragment can be generated from 60S ribosomal subunits as well as from 80S ribosomes but, of course, not from 40S particles. Ribosomes are extremely sensitive to the toxin. A concentration as low as 3 x 10 -8 M will produce the a fragment (Endo et al., 1983). The Vmax for a-sarcin (at 3 x 10 -8 M) with ribosomes is 1.6/xM/min; the Km is 5.3/zM and the kcat is 55 min -1. What is striking about c~-sarcin is that when the substrate is RNA rather than ribosomes, and the concentration of the toxin is higher, its specificity is entirely different, albeit still unusual. If, for example, the substrate is free 28S rRNA, the toxin causes extensive digestion of the nucleic acid (Endo et aL, 1983). It is important that at lower, cytotoxic, concentrations c~-sarcin retains its specificity for the single phosphodiester bond even in free 28S rRNA (Miller and Bodley, 1988). Experiments with homopolymers established that a-sarcin action is preferentially on purines in RNA; experiments with 5S rRNA revealed a most unusual property of the enzymemthat a-sarcin cuts on the 3' side of nearly every adenine and guanine without regard to whether the nucleotide is in a single- or double-stranded region. Hydrolysis by a-sarcin of RNA substrates requires neither monovalent nor divalent cations; divalent cations are inhibitory at any concentration whereas monovalent cations are only inhibitory when the concentration exceeds 0.1 M. Catalysis generates a 3'-phosphate and a 5'-hydroxyl at the site of cleavage (Endo and Wool, 1982). The reaction is not markedly affected by pH but the enzyme is most active at pH 7. a-Sarcin, like many other nucleases, hydrolyzes both RNA and DNA. Digestion of DNA, however, requires about two orders of magnitude more of the toxin for hydrolysis of 50% of the substrate (Endo et al., 1983). Moreover, cleavage of DNA requires magnesium whereas hydrolysis of RNA is inhibited by the cation. Finally, when D N A is the substrate, both purines and pyrimidines are attacked.

III. a-SARCIN CLEAVAGE SITE SEQUENCE

The site of cleavage by c~-sarcin is on the 3' side of the guanosine at position 4325 in 28S rRNA (Fig. 1) (Chan et al., 1983). G4325 is embedded in a purine-rich stretch of 12 nucleotides that is near universal in the

134

Ira G. W o o l

U--A C.-O C--G OAc -'GAu A CGGA / AUCAGCGGGGAAA IIIII .III II oUAOUC AucccA U~

ACUGGcUU cAAGCG GUGGcGGc Uu .III.I C U CGCUGc A

~uuuUU

A U. G u ~-~

G

~\uU

C--a 1 : u6-- 4 3 0 0

G

C

c A uo_

Cr

c

~ ~ A

ucuAA:t~c~~ A44bO CCGcAAG

AGC

,,oc. 9

u

c

~-e~

.G - c

~---'---CAA-- U ~ - - G - - Cu

a ~A<--I

uc"

u

~

--U GU~ -- c

431g ,, ------,,AC

AuC C\v\ _ c\ G c \oA

.u c. \

/

4250

UO '~

.:.ou. ~ u

QAu

4150

.~,:ic~ ~oo~. ~.

G.u

G c uG . , ,

/

A-GG'\'G

Uc

ricin depurination site

-- U u c A U U G U G A A G C I. IIIII

C AA

i

U,. UG_ ,,C ct-sarcin cleavage site" ~"A--U ~'GAA~" ACUUAA

ACu" c C UCGAA

c --G

4200

UUG-Ou 4350 CuuU G U

' G../ U GuA~,;% u' / U _

-C_/ C ~;_/U

u_ ,," G

AGeA

Fig. 1. The a-sarcin domain of 28S rRNA. A portion of rat 28S r R N A with the asarcin domain" the sites of a-sarcin and ricin action are indicated and the stretch of 12 bases that is near universal is underlined.

R N A of cytoplasmic ribosomes. Indeed, this is the longest consecutive sequence of conserved nucleotides in rRNA. It should be noted that c~sarcin has neither antimicrobial nor antifungal activity because it enters neither bacterial nor fungal cells; however, isolated yeast and Escherichia coli ribosomes are sensitive to the toxin. Indeed, all ribosomes that have been tested, including those of the producing organism, are sensitive to the toxin. Obviously, the ribosomes of the producing organism are not inherently resistant to the toxin. How then does the Aspergillus survive the synthesis

4

Structure and Mechanism of a-Sarcin

135

of the toxin? a-Sarcin is synthesized as a protoxin with a 27-aminoacid signal sequence (Lamy and Davies, 1991) that is removed during translation. Prepro-ct-sarcin has a molecular weight of 22,5000 and is processed in the endoplasmic reticulum during translation to the 18,500 molecular weight pro-a-sarcin, which is further processed to the mature form that is packaged into vacuoles for secretion (Endo et al., 1993a,b). Thus protection of the producing organism is the reflection of the biochemistry of the secretory system, which confines the toxin to the endoplasmic reticulum before its synthesis is complete; the endoplasmic reticulum (like the digestive tract) is physiologically outside the cell and the mature toxin cannot regain entry to the cell because the cell membrane of Aspergillus, like that of other fungi, is not permeable to the toxin. In depictions of the secondary structure of the large ribosomal subunit RNA (Wool, 1986; Gutell and Fox, 1988), the a-sarcin cleavage site is in a single-stranded loop (Fig. 1) and the site must, of course, be accessible on the surface of the ribosome. Moreover, the domain was deemed from the start to be important for the function of the ribosomemfirst, because the nucleotide sequence is conserved. The rule for rRNAs is conservation of secondary structure, not primary sequence, so retention of the a-sarcin sequence is impressive. This is not to say that there are no other nucleotide sequences in rRNA that are conserved; there most certainly are, and most of these sequences are important for function. Second, it is surprising that cleavage of a single phosphodiester bond inactivates the ribosome because ribosomes survive mild treatment with nucleases despite many nicks in the rRNA; indeed, intact rRNA is not essential for protein synthesis. Until recently little was known of the function of individual ribosomal components or even of ribosomal domains. As a rule, neither the ribosomal proteins nor the nucleic acids have activity when separated from the particle. In the beginning it was thought that the rRNAs only provided a scaffolding to support the ribosomal proteins that catalyzed the partial reactions of protein synthesis. However, the pendulum has swung in the other direction. Now the rRNAs are envisioned as being responsible for the basic biochemistry of protein synthesis: for the binding of aminoacyl-tRNA, mRNA, and the initiation, elongation, and termination factors; for peptide bond formation; and for translocation. The ribosomal proteins, which are presumed to be a later evolutionary embellishment, are though to facilitate the folding and the maintenance of an optimal configuration of the rRNA, perhaps, in this way conferring on protein synthesis speed and accuracy (Stern et al., 1989). The value that derives from an analysis of the mechanism of action of antibiotics and of toxins that affect ribosomes is in concentrating

136

Ira G. Wool

attention on regions where efforts to comprehend functional correlates of structure are likely to be rewarded. The catastrophic effect of a-sarcin on protein synthesis is ample evidence that the sarcin domain is crucial for ribosome function. This is almost certainly because the domain is involved in EF-1 (or EF-Tu)-dependent binding of aminoacyl-tRNA to ribosomes and EF-2 (or EF-G)-catalyzed GTP hydrolysis and translocation. The conclusion comes from a series of compelling observations: first, that these are the partial reactions of protein synthesis that are most adversely affected by a-sarcin (Vazquez, 1979); and, second, that cleavage at the a-sarcin site in E. coli 23S rRNA interferes solely with the binding of EF-Tu and EF-G (Hausner et al., 1987). Moreover, EFTu and EF-G footprint in the a-sarcin domain (Moazed et al., 1988). In E. coli 23S rRNA EF-Tu protects four nucleotides against chemical modification and these correspond in eukaryotic 28S rRNA to G4319, A4324, G4325, and A4329; G4325 is at the a-sarcin cut site and the other three are in the universal sequence. EF-G also protects four nucleotides: three are the same as the ones protected by EF-Tu, they correspond to G4319, A4324, and G4325.

IV. S T R U C T U R E OF ct-SARCIN A N D OF R E L A T E D Aspergillus TOXINS

The specificity of a-sarcin is likely to be a reflection of the structure of the toxin and of a domain in the ribosome. The protein, whose amino acid sequence has been determined (Sacco et al., 1983), has 150 residues and a molecular weight of 16,987 (Fig. 2). The toxin is very basic, having 20 lysyl, four arginyl, and eight histidyl residues; there are four cysteines and they are linked by disulfide bridges between the residues at positions 6 and 148 and 76 and 132. c~-Sarcin, apart from the disulfide bridges, at least superficially resembles a ribosomal protein; i.e., it is small and basic. Although the amino acid sequences of a-sarcin (Sacco et al., 1983), and of the related ribotoxins mitogillin (Fernandez-Luna et al., 1985) and restrictocin (Lopez-Otin et al., 1984), were first determined directly from the proteins, they have been confirmed from the sequences of nucleotides in cDNAs (Henze et al., 1990; Lamy and Davies, 1991; Lacadena et al., 1994; Wnendt et al., 1993). Comparison of amino acid sequences (Sacco et al., 1983) revealed that a-sarcin is related to certain other nucleases; for example, the toxin has 24% amino acid identity with T1 and 34%

4

Structure and Mechanism of a-Sarcin

137

Alpha-sarcin Restrictocin

1 20 40 60 AVTWTCLNDQKNPKTNKYETKRLLYNQNKAE SNSHHAPLS DGKTGS SYPHWFTNGYDGDG A- T W T C I N Q Q L N P K T N K W E D K R L L Y S Q A K A E S N S H H A P L S D G K T G S SY P H W F T N G Y D G N G

Alpha-sarcin Restrictocin

80 KLPKGRTPI KFGKSDCDRPPKHSKDGNGKTDHYLLEF KL I K G R T P I K F G K A D C D R P P K H S Q N G M G K D D H Y L L E F

Alpha-sarcin Restrictocin

140 RVI Y T Y P N K V F C G I I A H T K E N Q G E L K L C S H RVI Y T Y P N K V F C G I V A H Q R G N Q G D L R L C SH

i00 120 PTF P D G H D Y K F D S K K P K E N P G P A PTF P D G H D Y K F D S K K P K E D P G P A

Fig. 2. An alignment of the amino acid sequences of a-sarcin and restrictocin. The proteins share 86% amino acid identity. Restrictocin and mitogillin differ by only one residuemD25 in mitogillin and $25 in restrictocin.

with U2. Moreover, an alignment of the amino acid sequences had suggested that H49, E95, R120, and H136 in restrictocin might be in the catalytic center; these residues correspond to H50, E96, R121, and H137 in a-sarcin. Mutations, E95G and H136L, were constructed in restrictocin; the former had decreased activity whereas the latter was inactive (Yang and Kenealy, 1992), suggesting that H136 in restrictocin (H137 in c~-sarcin) is involved in catalysis. However, more important than the similarities are the differences between a-sarcin and ribonucleases T1 and U2, and more explicitly the differences in structure that account for the differences in properties, aSarcin is a cytotoxin in part because it has acquired a means of entering cells, no small evolutionary accomplishment for a nuclease. It is a property the related ribonucleases do not share. Moreover, a-sarcin selects and cleaves a single phosphodiester bond in ribosomes whereas ribonuclease U2 causes extensive digestion of the R N A in ribosomes. What one wants to know is the portions of the structure of a-sarcin, absent in related ribonucleases, that account for these properties. The Aspergillus toxins, c~-sarcin (Wawrzynczak et al., 1991), restrictocin (Orlandi et al., 1988), and mitogillin (Better et al., 1992) are candidate "magic bullets" for the treatment of cancer and of viral-infected cells, including those infected with human immunodeficiency virus (HIV), and for use in tissue and organ transplantation. In the most common scenario the toxin is coupled to a specific binding protein or to an antibody directed to an antigen on the surface of cancer cells or cells infected with virus. The intention is to deliver the catalytic toxin exclusively to these cells. Immunotoxins have been constructed and they are effective

138

Ira G. Wool

in vitro against cancer cells; moreover, they do not appreciably harm

nonmalignant cells. However, they are less effective when administered to patients and frequently produce intolerable side effects, although some limited success has been achieved recently. The lack of effectiveness has been attributed to the immunotoxins having poor pharmacodynamicsma short half-life, susceptibility to proteolytic degradation, or capture by the immune system.

V. M E C H A N I S M BY W H I C H a S A R C I N E N T E R S CELLS

a-Sarcin is a cytotoxin and is effective in restricting the growth of tumors, which implies that it is able to enter cells, but how it accomplishes this is a mystery. The toxin appears to have little activity on intact mammalian cells (A. Lin and I. G. Wool, 1983, unpublished data). There is no evidence that o~-sarcin has a specific domain that binds to a particular receptor to effect entry; indeed, a search for receptor-mediated translocation was unsuccessful (A. Lin and I. G. Wool, 1983, unpublished data). a-Sarcin is, on the other hand, an effective inhibitor of protein synthesis in intact virus-infected cells (Munoz et al., 1985; Otero and Carrasco, 1985), in cells of some tumor lines (Turnay et al., 1993), and in cells whose membrane permeability has been modified by phospholipase C (Otero and Carrasco, 1988). What these studies imply is that the entry of a-sarcin into cells requires alteration or damage to the cell membrane by viral infection, by malignant transformation, or by some toxic agent. An extension of this reasoning suggests a physiological function of asarcin: that the toxin acts to limit damage to an organism by mediating the death of cells that have, for example, been infected by virus.

VI. a - S A R C I N R E C O G N I T I O N E L E M E N T S IN 28S r R N A

Early on an effort was made to determine whether the specificity of a-sarcin is dependent on the complex ordered structure of 28S rRNA that derives from the presence of the nucleic acid in a particle containing proteins. A contingent aim was to establish the minimal substrate for c~-sarcin. For this purpose, an oligoribonucleotide (a 35-mer) was synthesized using a synthetic D N A template and the phage T7 RNA polymerase

4

Structure and Mechanism of ~-Sarcin

139

(Endo et al., 1988). This oligomer (Fig. 3A) has the nucleotide sequence and the secondary structure of the domain in eukaryotic 28S r R N A that is attacked by c~-sarcin and is referred to as wild type. Treatment of the synthetic oligoribonucleotide with lower concentrations of a-sarcin, i.e., 2.9 x 10 -8 M, led to the formation of two fragments (Fig. 3B, lane 2) (Endo et al., 1990). About 85% of the substrate is cleaved (Fig. 3C). Cleavage by a-sarcin is at the guanosine at position 21 in the synthetic oligoribonucleotide, which corresponds to G4325 in 28S rRNA and is precisely where the toxin hydrolyzes the nucleic acid in intact ribosomes (Endo et al., 1990). In contrast, a higher concentration of a-sarcin (2 x 10 -6 M) led to cleavage of the substrate at all, or nearly all, of the purines (Fig. 3B, lane 5). This conforms with the observation that higher concentrations of a-sarcin, i.e., higher than is necessary to inactivate ribosomes, lead to the hydrolysis on the 3' side of almost every purine in RNA. To define the features of the structure that prescribe binding to the RNA and that are necessary for the catalysis of hydrolysis, the nucleotides in the a-sarcin domain synthetic oligoribonucleotide (Fig. 4, I) were systematically altered (Endo et al., 1990). Cleavage by a-sarcin of a variant of the wild-type RNA with a transition of the guanosine at position 21 to an A (Fig. 4, II) was decreased to 35% of the control. Transversions of the wild-type G to U (Fig. 4, III) or to C (Fig. 4, IV) reduce hydrolysis to approximately 17 and to 1%, respectively. Thus, there is strong, but by no means absolute, dependence on preservation of the G at the site of covalent modification; the preference is G > A > U ~> C. What is notable is that a-sarcin, which had been presumed to be a purine-specific nuclease, has some activity with pyrimidines. The results prejudice one to consider structure rather than sequence as the more important determinant of specificity. The ct-sarcin domain RNA has a canonical protein-binding structure: a stem, a loop, and a bulged nucleotide (Fig. 4, I). The last occurs in a number of ribosomal protein-binding sites. The suspicion was that the bulged U at position 6 in the substrate (position 4310 in 28S rRNA) is not necessary for a-sarcin action because it does not occur at the comparable site in E. coli 23S rRNA, and the bacterial ribosomes are sensitive to the toxin; just as suspected, a variant that lacked the bulged U in the stem (Fig. 4, V) is as sensitive to the toxin as the wild type (Endo et al., 1990). To test the importance of the stem for a-sarcin action, a linear molecule (a 35-mer) was constructed (Fig. 4, XXIII) that retained the 17 nucleo-

140

Ira G. Wool

Fig. 3. The effect of a-sarcin on a synthetic oligoribonucleotide that mimics the toxin domain in 28S rRNA. The synthetic radioactive a-sarcin domain oligoribonucleotide (35-met) (A) was renatured and then (B) incubated at a concentration of 1.6/zM for 20 min at 20~ without ot-sarcin (lane 1) or with increasing concentrations of the toxin: 2.9 x 10 -8 M (lane 2); 5.9 x 10 8 M (lane 3); 2.9 x 10 7 M (lane 4); 2 x 10 -6 M

4

Structure and Mechanism of a-Sarcin

141

tides in the loop sequence but with the 5' 11 nucleotides and the 3' 7 nucleotides altered so that they would not pair (Endo et al., 1990). a-Sarcin did not cleave the linear substrate specifically. However, there was nonspecific digestion; indeed, the linear RNA appears to be more sensitive to this nonspecific effect of the toxin than the structured RNA. Having established that the stem is required, the number of base pairs in the helix that are needed was determined (Endo et al., 1990). Successive additional base pairs were deleted from the wild-type oligoribonucleotide (Fig. 4, V I - I X ) . A variant having only three pairs in the stem (Fig. 4, IX) was recognized by a-sarcin. It seems likely that the helix is necessary only to tether the ends of the loop so as to allow it to form a specific conformation. In nature the helix may be longer and hence more stable than is required for recognition by a-sarcin; we presume it is neither longer nor more stable than is required for its contribution to the function of this ribosomal domain in protein synthesis. Synthetic oligodeoxynucleotide analogs of E. coli tRNA Ph~ and of tRNA Ly~ are recognized by their cognate aminoacyl-tRNA synthetases. A deoxy analog of the wild-type a-sarcin RNA (Fig. 4, XIV), on the other hand, was not cleaved specifically by a-sarcin (Endo et al., 1990). However, there was nonspecific digestion and, as we had observed before, the hydrolysis of the DNA required magnesium; nonspecific cleavage of RNA by a-sarcin is moderately inhibited by magnesium. The effect of the context of the a-sarcin site guanosine on recognition was evaluated (Endo et al., 1990). It was anticipated that alterations in

(lane 5). A radioautograph of the polyacrylamide-urea gel was used for the analysis. The radioactivity in each of the bands (of 35, 21, and 14 nucleotides) was determined for lanes 2-5, and (C) the percentage of the original substrate that was specifically cleaved was plotted as a function of the concentration of a-sarcin. ( 9 1 6 9 Wild-type 35-mer; cf. A. ( O - - O ) The 34-mer lacking the wild-type bulged nucleotide at position 6. In A, the S and the arrow designate the site of cleavage by a-sarcin. In B, the numbers on the left indicate the number of nucleotides in the RNA. The shadow bands that form a ladder are oligoribonucleotides that most probably were formed by degradation, perhaps due to radiation damage.

S

1 4CG%G4G U G

G

C-G 10-G-C-30 U-4 C-G C.G " 4 ~ ~ t-U0~3' G 5' PPPG

% C"

20

G

C-G 7-G-c-2, U-4 C-G

;.G~~3' G ~'PPPG

G S'PPPG

I G A

cc4 I-6-C-I' C-G U-A y o H 3 ~ . G S'PDPG

4cu

,c4 C-G 9-G.C-2s u-4 C~G C-G 4-u ,P-U0~3' G S*PPPG

io

6,c

A C ~ :16 GA G

4

G ' G ~ G G

AC

S

20

G

G

G

c ~ A

C-G 3-G C - 2 , E-40~3' G ~'DPPG

G

G 4cu C-G

e-6:k-2m C.G C-G

;-uo~~. G S'PPPG

A C G 04 G AG G A

G

cA

ccA

~ - ~ ~ - ~ ~ - uC-G 4cu

A C G 414 G AG

t-U 3' ~~

O OH^'

G S'PPPG

1 G

U

G

$4 C-G 10-G-C-so U-4 C-G uC-G A-u

C-G 80-G-C-ro U-A C ~ G C-G 4.u

~

,!-u0~31 G 5' PPOG

U

cc4

G G

20

G 4cu

Cu cc4

C-G 10-G-C-30 U-A C-G C.6 " 4 ~

G G A

U G

G,

P ~ U $4

or

GAG'& AC o: A~ G

lo-G-C-50 u-4 C-G UC.G A-U

C-G lo-G-C-$a G-U-A-C C-G UC-G A-U 2-U0~3' G 5'PPPG

no-6-C-so

u/:;\n UC-G 4-u

t-U 3' ~ ~

tLU0Ii 3' G

G J'PPPG

5' PPPG

s A C Gro ?G*G G

*c

A

cA

G-!!~.,C/U

10-G-C-30 U-A C.6 UC-G A- U t U 0 H3' G ~'DPPG

G

4

cc4 C-G tO-G-C->o u-4 C.G

ui.6

AC

'4'4 20

G

AC

A

G

G C-G lo-G-C-30 u-A C.G

G?GA 2.

I

G A

C-G 10-G-C-30 U-A C-G

UG ~":!

u

u

U

U

uc"

cuu

C-G lo-G C-30 U-A C-G

u:.

,n-U0"3' G 5,PPPG

$'U0~3' G 5'PPPG

:.O 'H G ~'POOG

3'

5

I u = ? ~ 4 G 20 u U

ucu

G

G

A

C

cuu

G' ,

cC

C-G lo-G-C-30 u-4 C.G UC.G A u t - u 3'~ ~ G 5'pppG

C-GU so-G-C-30 u-4 C-G UC-G 4-u

g . U 3'~ ~

G S'PPPG

C-G to-G-C-30 u-4 C-G

:u:: &-UOH 3' G S'PPPG

Fig*4. The structures (I-XXII) of variants of the a-sarcin domain oligoribonucleotide.

4

Structure and Mechanism of a-Sarcin

143

the 5' adjacent adenosine would not have an appreciable effect because depurination of A4324 by pretreatment with ricin did not affect subsequent cleavage by a-sarcin at G4325 in the same ribosomes. Nonetheless, a series of variants were constructed with alterations of the ricin site A to G, U, or C (Fig. 4, X V - X V I I ) . As expected, all were recognized by a-sarcin with specificity and with normal efficiency. The context was changed in another way: the tetranucleotide GAG(sarcin)A was left intact but the remainder of the universal portion of the loop sequence was engineered so that it was entirely uridines (Fig. 4, XVIII) or, in a second variant, uridines and guanosines (Fig. 4, XIX) (Endo et al., 1990). Neither oligonucleotide is a competent substrate for a-sarcin. Thus context either directly or indirectly influences the recognition of the R N A by c~-sarcin. The specific response to a-sarcin is also lost when the position of the tetranucleotide G A G A in the loop is changed (Fig. 4, X X - X X I I ) (Endo et al., 1990). The results with one of the variants is particularly instructive (Fig. 4, XXI). In this mutant there is still a guanosine at position 21, as in the wild-type substrate; nonetheless, there was no cleavage there. Thus, recognition is not merely of a guanosine at the correct position in the sequence, a-Sarcin appears to appreciate the structure of the loop and this structure is no doubt affected by the nucleotide sequence. The conclusion to this point was that specific recognition of rRNA by a-sarcin requires, in the first instance, a stem and a loop, but that a bulged nucleotide in the stem is not necessary. There is a strong preference for a guanosine at the site of covalent modification and the surounding context, i.e., the 12-base universal nucleotide sequence affects, either directly or indirectly, the binding of a-sarcin and catalysis. The exception is the immediate 5' adjacent adenosine, which has no influence on identification of the RNA. The stem is essential; however, only three of the seven base pairs found in 28S rRNA are needed. The helical stem appears to contribute to recognition only indirectly by tethering the ends of the loop and conferring on the latter a specific conformation. Finally, the position of the tetranucleotide GAG(sarcin)A in the loop conditions recognition, once again, either directly or indirectly by altering the context and hence its conformation. Concurrent experiments with ricin A chain (RA), a cytotoxin that inactivates ribosomes by depurinating the adenosine at position 4324 in 28S rRNA (Fig. 1), the nucleotide 5' adjacent to the a-sarcin cleavage site (Endo and Tsurugi, 1987, 1988), had indicated that RA has an absolute requirement for a G A G A tetraloop; moreover, there was strong

144

Ira G. Wool

evidence that the tetraloop exists, at least transiently, in intact ribosomes (Endo et al., 1991; G10ck et al., 1992). a-Sarcin, on the other hand, does not recognize oligoribonucleotide substrates that have only a G A G A tetraloop. If there is a tetraloop in the sarcin domain in 28S r R N A , it would have to be closed off by a base pair between the cytosine on the 5' side and the guanosine on the 3' side of the G A G A tetranucleotide (Fig. 5 I). A series of mutants was constructed to test this prediction (Gltick et al., 1994). a-Sarcin catalyzes covalent modification of a variant having a reversal of the 5' C and the 3' G (Fig. 5, If). This is a rare change in the G A G A context that is tolerated. In addition, a-sarcin modifies a mutant that has a transversion of the 3' G to a C (Fig. 5, I l l ) , albeit the

R

~ C6

&

U G

G A

A

CUc_GCc

A

ACG A GAG/c

ACG A GAG.,...,C

U G

U G

G A

A

A

CUc_GCC

I0- G-C -30

CUc.GCc

G-C

G-C

U-A C-G C-G UA. U

U-A C-G C-G UA_U

~-UoH 3' G 5' pppG

~-UoH 3' G 5' pppG

~-UoH 3' G 5' pppG

I

It

m

AcG A GAo/U

U G

G A

A

A

U

A-~cGAGAs/A

U G

G A

A

A

A\

Ac G A GA6~,A

U G

G A

A

A

CUc-Gcc

CUc-GCc

CUc-Gcc

G-C

G-C

G-C

U-A

C-G C-G UA_U ~-UoH 3'

Iv

A

U-A C-G C-G UA-u

A\

G 5' pppG

G A

A

U-A

C-G C-G UA_U A-UoH 3' G G 5' pppG

s

U-A

C-G uC-G A-U ~-UoH 3'

G 5' pppG

U

~ACG A GAQ..,,U U G

G A

A

A

CUc-GCc G-C

U-A

C-G C-G UA-U A-UoH 3' G

G 5' pppG

s

Fig. 5. The structures (1-VII) of a set of oligoribonucleotide mutants of the a-sarcin domain RNA in which the 5'-guanosine and the 3'-cytosine that surround the GAGA tetranucleotide have been varied.

4

Structure and Mechanism of a-Sarcin

145

effect is moderately diminished. RA does not recognize this mutant. This is a discontinuity in the identity elements for RA and for a-sarcin. The decrease in cleavage by a-sarcin occasioned by the G-to-C mutation is almost entirely the result of an effect on catalysis. The kcat is decreased almost by an order of magnitude and the kcat/gm by 0.15, but the affinity of the toxin for the substrate is hardly changed. A mutant with simultaneous transversions of the 5' C to A and of the 3' G to U creating a potential A U (Fig. 5, V) pair is affected by asarcin and by RA (Gltick et al., 1994). However, simultaneous changes of the 5' C and the 3' G to A (Fig. 5, VI) or, in a separate mutant, to U (Fig. 5, VII) lead to loss of recognition by RA but not by a-sarcin, although, once again the response to a-sarcin is reduced. Obviously, RA cannot depurinate oligoribonucleotides that lack the capacity to shut off a GAGA tetraloop by forming a closing Watson-Crick pair; it is also clear that the inability to do so modestly impairs, but certainly does not abolish, recognition by a-sarcin.

VII. C O N F O R M A T I O N OF THE SARCIN D O M A I N IN 28S r R N A

From the beginning, experiments have addressed two separate structure-function problems (Wool et al., 1992). The first is the structure of the sarcin domain RNA when resident in the ribosome and the contribution of this structure to the biochemistry of protein synthesis. This is job one. The second problem is the identity elements for the recognition by a-sarcin of a rRNA domain and hence the chemistry of a particular RNA-protein association. The structural requirements for protein synthesis are far more stringent than for toxin recognition (Wool et al., 1992). An observation that may be important for understanding the contribution of the sarcin district of 28S rRNA to the biochemistry of protein synthesis is that the identity elements for RA and for a-sarcin are different (Wool et al., 1992). RA requires a G A G A tetraloop whereas asarcin does not. Moreover, if the capacity to form a G A G A tetraloop is abolished, modification of the R N A by RA is lost, but that by a-sarcin is not. Thus, the identity elements required by RA and by a-sarcin are different, yet the two ribotoxins catalyze covalent modifications of adjacent nucleotides in rRNA. The simplest rationalization of these observations is that the identity elements for a-sarcin lie outside of the G A G A tetraloop.

146

Ira G. Wool

The universal sequence of nucleotides in the sarcin domain did not evolve to maximize the efficiency of recognition by the toxin; the pressure presumably was to facilitate the binding of two related but dissimilar proteins, the elongation factors 1 and 2 (or Tu and G). The elongation factors bind to essentially the same site on the ribosome (Moazed et al., 1988) and a single round of peptide bond formation requires the displacement of the first (EF-1 or EF-Tu) to allow binding of the second (EF-2 or EF-G) and vice versa. Thus, it may be the necessity to bind two different proteins, in an ordered fashion, that explains the extraordinary conservation of the sequence of nucleotides in the loop. These nucleotides may be necessary also for reversible transitions in conformation, which may in turn be needed for protein synthesis. Obviously, what is essential to the resolution of these problems are the details of the structure of the sarcin domain. There is a model of the conformation of the domain derived from nuclear magnetic resonance (NMR) spectroscopy (Scewczak et al., 1993; Szewczak and Moore, 1995). The proposed three-dimensional conformation of the sarcin RNA (Fig. 6) satisfies all of the NMR spectroscopic data, is entirely consistent with the RNA mutants, and perhaps most importantly the hydrogen bonding pattern agrees precisely with chemical modification data obtained independently (H. F. Noller and D. Moazed, 1987, unpublished data; quoted in Szewczak et al., 1993 and Szewczak and Moore, 1995). The structure has, just as was predicted from analysis of mutants, a G A G A tetraloop shut off by a C.G pair (Fig. 6). However, there is a noncanonical, heteropurinic intraloop G - A pair; thus there are only two unpaired nucleotides in the loop. Below the closing C.G pair there is a second heteropurinic A - G pair just as in the tetraloop. This is followed by a reversed Hoogsteen U - A pair, just above which the stacking of the bases in the strands cross over. G10 is far and away the most interesting residue in this RNA. The sequential pattern of nuclear Overhauser enhancement (NOE) connectivities breaks at G10, giving every indication it is a bulged base. Moreover, there is a NOE between A9 and U l l , the only nonsequential aromatic-aromatic N O E in the molecule. The key to the structure, however, is a NOE between the imino proton of G10 and the sugar proton of G19, which implies they are close, and a model has been built in which G10 reaches across the U 1 l - A 2 0 pair toward the ribose of G19 and places its imino proton within hydrogen bonding distance of the phosphate oxygen between G19 and A20. There is next a symmetric homopurinic A - A pair; below this C8 and U7 are juxtaposed to C22 and C23 with no evidence of pairing. Finally, there is a six-base pair

4

Structure and Mechanism of a-Sarcin

147

Fig. 6. The structure of the a-sarcin domain RNA. (Left) Schematic of the secondary structure of the ot-sarcin domain. (Right) Diagram of the hydrogen bonding and base stacking interactions derived from NMR spectroscopy of the same oligoribonucleotide (derived from Szewczak et al., 1993). with permission). The open arrow designates the Nglycosidic bond cleaved by ricin; the filled arrow designates the phosphodiester bond cleaved by ot-sarcin.

148

Ira G. Wool

canonical A-form R N A helix. Thus the R N A , which in secondary structure diagrams is depicted as having a 17-member loop (Fig. 6, on the left), in actuality has a compact conformation that approximates an extended helix (Fig. 7) with few unpaired nucleotides and most strikingly many noncanonical pairs (Fig. 6, on the right). The pattern of chemical modification of nucleotides in the sarcin loop in the free R N A differs from that in the ribosome (H. F. Noller and D. Moazed, 1987, unpublished data). Because the differences include reactivity increases, as well as the expected decreases due to interactions with proteins and neighboring R N A segments, the conformation of the sacrin loop in the ribosome cannot be exactly the same as the one just described. It has been suggested that conformational changes in the sarcin loop trigger translocation (Wool et al., 1992). Because both a-sarcin and R A are able to attack the sarcin loop in the ribosomes, and because a good case can be made that a conformation like the one discussed here is necessary for sensitivity to RA, it seems reasonable to suggest, as Szewczak and Moore (1995) have, that this structure is a conformation that is present in the ribosome at some stage during elongation.

VIII. E F F E C T OF M U T A T I O N S OF T H E A N A L O G OF G4319 IN 28S r R N A O N R E C O G N I T I O N OF O L I G O R I B O N U C L E O T I D E S BY a - S A R C I N

In the three-dimensional conformation of the sarcin domain R N A the guanosine at position 15 (G10 in the oligoribonucleotide used to derive the N M R conformation, G15 in the wild-type oligoribonucleotide described in this section, and G4319 in 28S r R N A ) is bulged (Fig. 6) and produces a prominent kink in the helical structure (Fig. 7) (Szewczak et al., 1993). Although, the R A recognition element can be described with some precision, it is a G A G A tetraloop, that for

Fig. 7. A three-dimensional conformation of the a-sarcin domain RNA. The conformation was obtained from the coordinates deposited in the Brookhaven data base with the program GRASP at a Silicon Graphics workstation. The bulged guanosine (gold) and the sites of action of a-sarcin (blue) and of ricin (purple) are designated. The phosphodiester backbone is represented by a red ribbon. Derived from Szewczak et al. (1993).

4

Structure and Mechanism of a-Sarcin

149

a-sarcin at first could only be d e f i n e d n e g a t i v e l y u m o s t definitively a G A G A t e t r a l o o p is not a c o m p e t e n t substrate. B e c a u s e the c o n f o r m a tion of G15 is such a p r o m i n e n t f e a t u r e of the N M R structure, a t t e n t i o n was d i r e c t e d to it (Gltick a n d W o o l , 1996). F o u r m u t a t i o n s of G15 w e r e constructed: the d e l e t i o n of the n u c l e o t i d e , a t r a n s i t i o n to a d e n o s i n e , a n d t r a n s v e r s i o n s to cytidine a n d to uridine (Fig. 8). T h r e e of the four variants are n o t r e c o g n i z e d by a-sarcin at all; cleavage of the fourth, a G 1 5 A m u t a n t , is b a r e l y d e t e c t a b l e at the highest c o n c e n t r a t i o n of c~-sarcin a n d e v e n this m i n i m a l hydrolysis is n o t consistent. Thus, G4319 is the critical n u c l e o t i d e for c~-sarcin recognition.

IX. P H E N O T Y P E 23S r R N A

O F M U T A T I O N S IN G2655 IN Escherichia coli

O n e way, p e r h a p s the way, to define the function of the sarcin d o m a i n r R N A in p r o t e i n synthesis is by the analysis of the p h e n o t y p e of mutants. M u t a t i o n s have b e e n c o n s t r u c t e d in G2655 in E. coli 23S r R N A (Fig. 9) (M. M a c b e t h and I. G. W o o l , 1995, u n p u b l i s h e d data): the deletion of the n u c l e o t i d e (G2655A), a transition to a d e n o s i n e ( G 2 6 5 5 A ) , a n d t r a n s v e r s i o n s to cytosine ( G 2 6 5 5 C ) and to uridine ( G 2 6 5 5 U ) . G2655 is the b u l g e d g u a n o s i n e that is critical for c~-sarcin r e c o g n i t i o n a n d it is a n u c l e o t i d e that is p r o t e c t e d against chemical

Fig. 12. The structure of the active site in restrictocin. The five-stranded/3 sheet is twisted and forms a cleft; the catalytic residues (in red) are at the open end of the cleft. An inorganic phosphate (in yellow and red) is also at the active site. The two disulfide bonds are in orange. Fig. 13. A model of the complex of restrictocin and a-sarcin domain RNA. The model is from a docking experiment using the coordinates of the NMR conformation of the c~sarcin domain RNA (Szewczak et al., 1993) and those for restrictocin (Yang and Moffat, 1996). On the left, the positions of the bulged guanosine (G10 in Fig. 6) and the guanosine at the cleavage site (G16 in Fig. 6) in the RNA are in orange; the charge distribution on the surface of restrictocin is indicated by red (negative) and blue (positive). On the right, the complex has been rotated 90~to show the complementarity of the surfaces of restrictocin (blue, on top) and of the RNA (red, below).

150

Ira G. Wool R

R

s

R

ACG;o % G

u cc"

C-G =o- G-C-3o U-A C-G C-G UA- U A'UoH3' G G 5' pppG

R

I /s

&,......U

G

cO

C-G

=o- G-C-3o

U-A C-G C-G UA-U A-UOH3' G G 5' pppG

.CG;o % A,.....U

R

~ /s

G

cO

C-G =o- G-C-3o U-A C-G C-G UA- U A'Ur~H G v 3' G 5' pppG

~ /s

.CG;oG% C

u

G

Co cc" C-G

to- G-C-3o

U-A C-G C-G UA- U A-UoH 3' G G 5' pppG

.C ,o % U\ U

G

cc"

C-G

Jo- G-C-3o

U-A C-G C-G UA- U A'UoH 3' G G

5' pppG

Fig. 8. The structures of a set of oligoribonucleotide mutants of the a-sarcin domain RNA in which the guanosine at position 15, the analog of G4329 in 28S rRNA, was varied.

modification by EF-Tu and EF-G. The mutations were constructed in an rrnB gene in a high-copy-number plasmid that has an erythromycinresistance marker. In bacteria in which the plasmid 23S rRNA gene has a chromosometype PiP2 promoter, 70% of the cellular rRNA is derived from the plasmid. The G2655A, G2655C, and G2655U mutants do not grow; they have a dominant lethal phenotype (M. Macbeth and I. G. Wool, 1995, unpublished data). To our surprise and to our mystification, the G2655A mutant grows normally on agar plates and in liquid media. This was a surprise because, to our knowledge, no organism in the biosphere has an adenosine at this position and an adenosine was thought unlikely to reproduce the unusual bulged conformation found in the N M R structure. However, in a competitive growth experiment in which equal numbers of wild-type and G2655A mutant cells were mixed and grown together for approximately 150 generations, there was a definite phenotype: no G2655A mutant cells survived, i.e., they could not compete with the wild type. The plasmid 23S rDNA has silent mutations near the sarcin region (Fig. 9, inset) that allow the source of the R N A in ribosomes (plasmid or chromosomal) to be established; the determination is with specific oligodeoxynucleotide primers and reverse transcriptase. This has made it possible to test the sensitivity of the mutant ribosomes

4 Structure and Mechanism of a-Sarcin

151 6 5 5 - > A. U. C. A

2 U U

uC

,,o,,-*,,.';%

0-

%-%

..,-A S'--

~,cCA. OO U U

C /-^--

I 3'

9

,,~_- ~AO

A

%u

9

. cAO

-O A .U O C_ A A c u ~ ' . . . _ AG G ~ " G _ ,

G AU

.A .c

,,

..C^%% Ou

C

A o o U U ~c-~%

A

uo uC,,

q * S G_ n~'- % O A G~G C / G_GG_U" ~_A C W . 'C . ' U U % ~'C" % GO

III.I A O C A GC A GO U\ U

U tl O--C C--G AC --G

A

Uu2.u" --,~A ~

C -- G I A 9 ~ CGuuGAAG

O U A

Qu*C"*'.

% . . A_

c

O ~~A

AOc uc . . v % GIG kC uAu~," uu A o u , ~ c _ . I A o U C

,,

C_

u_

',_A'" . A

. c-G'~CA "u^," uc

_=U

A./.

o

QoU O,,

Auo%-,,%,.

_A

m

. =7,o

C lu

. . . .

UC:~,'~CA

O AA 285O O--C C --O O--C U--A -A O GC

u ='~

QUt O ^A C C

u

/u CA

/~u

/G U C ~

G

7

C

CA

/ :voo G / u u / G A C

G

Fig. 9. The a-sarcin domain of E. coli 23S rRNA. The mutants of G2655 that were constructed are designated. Inset: Region and identity of silent nearby mutations used to prime specifically for the mutant plasmid-encoded 23S rRNA.

to a-sarcin (M. M a c b e t h and I. G. W o o l , 1995, unpublished data). R i b o s o m e s with plasmid-derived 23S r R N A having a m u t a t i o n at G 2 6 5 5 are not affected by a-sarcin, w h e t h e r the m u t a t i o n is a deletion, a transition to a d e n o s i n e , or a transversion to cytosine or u r i d i n e - - a l l are resistant to the toxin. Thus, the results with r i b o s o m e s c o n f o r m precisely with the results obtained in vitro with o l i g o r i b o n u c l e o t i d e s (see Section VIII).

152

Ira G. Wool

X. R E L A T I O N S H I P OF THE a - S A R C I N R N A I D E N T I T Y E L E M E N T TO SELECTION BY THE T O X I N OF THE U N I Q U E SITE OF PHOSPHODIESTER BOND HYDROLYSIS

How can one explain the relationship between the identity element, G4319, and the cleavage by this small enzyme (recall that c~-sarcin has 150 amino acids) of the phosphodiester bond on the 3' side of the relatively distant G4325? One possibility is that c~-sarcin recognizes or induces a conformation of the sarcin domain that is different than that derived from NMR spectroscopy; the assumption is that in this putative alternate conformation the recognition nucleotide and the hydrolytic site are closer. This should not be dismissed out of hand because, as has been pointed out, the sarcin loop R N A is not very stable thermodynamically; moreover, the interpretation would accord with the speculation that conformational changes in the domain drive translocation. Unfortunately the possibility is not easily tested. The bulged G10 (the G4319 analog) is about 15 A away but on the same side of the sarcin loop as the phosphate group of the diester that is cleaved. This suggests a second possibility: that c~-sarcin recognizes G4319 and then cuts at a fixed distance. The virtue of this proposal is that it can be tested. The test requires the construction of a suitable substrate: the basic operation is to insert two pairs of nucleotides between G14.A17 and C13.G18 so as to change the distance between G15 (G4319) and the wild-type hydrolytic site (Gltick and Wool, 1996). In preliminary experiments, promising results were obtained with this construct (Fig. 10A): c~-sarcin-catalyzed cleavage occurred at a new, equidistant site just as predicted. This was confirmed in separate experiments by identification of the source of the cleavage fragments with 5' and 3' end-labeled oligoribonucleotides. Nonetheless, the results were not decisive; it was clear that distance was not the only determinant. The nature of the inserted nucleotides was important, at least in a negative sense; they could not form Watson-Crick pairs. In the initial experiment, the inserted 3' side nucleotides were guanosine (17a) and adenosine (17b) to preserve the sequence of the wild-type cut site; the 5' side nucleotides, 13a and 13b, were randomized (Fig. 10A). When a construct was synthesized with uridine (13a) and cytosine (13b) inserted on the 5' side to complement the guanosine (17a) and adenosine (17b) inserted on the 3' side, thereby permitting the formation of two presumably stable pairs in the loop (i.e., C13b 9 G17a and U13a 9 A17b), no hydrolysis by

4

Structure and Mechanism of ~-Sarcin

153

c~-sarcin occurred at either the new or old sites. However, if the 5' insertion was either adenosine (13a) and guanosine (13b) or cytosine (13a) and adenosine (13b) (Fig. 10B), which are unlikely to form stable pairs with the nucleotides inserted on the 3' side, i.e., guanosine (17a) and adenosine (17b), there was cleavage by the toxin at the new site. This is a strong indication that c~-sarcin requires an open, or at least an unstable, structure in the sarcin loop; a requirement that was implied by earlier experiments. But even with these constructs there was a small amount of residual cleavage at the old, or wild-type, site. For this reason the wild-type G - A cleavage sequence was changed to C - C (Fig. 10C) (Gltick and Wool, 1996). Hydrolysis with this substrate was exclusively at the new site. It needs to be noted that hydrolysis by c~-sarcin at the new site in the variant oligoribonucleotide is not as efficient as at the old site in the wild-type RNA. An assessment of the extent of the decrease in efficiency as a function of c~-sarcin concentration indicates that cleavage at the new site is only about 25% of that at the old site. What is particularly important, however, is that cleavage at the new site retains its dependence on the G4319 analog; just as with the wild-type substrate, deletion of that guanosine or its transversion to a cytidine abolishes hydrolysis. The results suggest that G4319 is the identity element for c~-sarcin recognition, that the toxin binds to G4319 or at least in close proximity to the nucleotide, and that binding allows c~-sarcin to orient itself and to catalyze hydrolysis of a phosphodiester bond at a fixed distance and at a fixed position in space relative to G4319. Manipulation of the structure of the R N A substrate has uncovered this aspect of how recognition is related to catalysis. That cleavage at the new site in the variant R N A is never as efficient as at the old site in the wild-type R N A indicates that not all of the structural features of the relationship have as yet been defined.

XI. T H R E E - D I M E N S I O N A L S T R U C T U R E OF R E S T R I C T O C I N , A H O M O L O G OF a~-SARCIN. B I N D I N G TO a - S A R C I N D O M A I N RNA AND THE CATALYTIC MECHANISM

The three-dimensional structure of restrictocin, refined to a resolution of 1.7 .A,, has been determined by X-ray diffraction of single crystals

154

Ira G. Wool

Fig. 10. Analysis of the relationship of the distance between the a-sarcin identity guanosine, the analog of G4319 in 28S rRNA, and the site of cleavage. The structure of the wild-type oligoribonucleotide is in Fig. 6. (A) A mixture of 16 mutant oligomers in which the nucleotides at positions 13a and 13b were randomized (they are designated N).

( Y a n g and Moffat, 1996). Recall that restrictocin shares 86% a m i n o acid identity with a - s a r c i n and has the s a m e m e c h a n i s m of action. Restrictocin has two a helices and two a n t i p a r a l l e l / ~ sheets (Fig. 11). T h e structural core is c o m p o s e d of a p e r p e n d i c u l a r ct helix of t h r e e turns p a c k e d against a five-stranded antiparallel fl s h e e t that is stabilized by

4

Structure and Mechanism of a-Sarcin

155

Fig. 10. (Continued). (B) The residues at 13a and 13b are cytosine and adenosine.

hydrophobic residues. The NH2 terminus has a long two-stranded antiparallel/3 sheet. A particularly distinctive feature of restrictocin is the large connecting loops between/3 strands. The structural core forms a cleft that has the putative catalytic residues and an inorganic phosphate believed to be derived from the crystallization buffer (Fig. 12). These five residues in restrictocin (Y47, H49, E95, R120, and H136) exactly reproduce five in the catalytic center of RNase T1. Although restrictocin and T1 share 24% amino acid identity, the latter has only 104 residues whereas restrictocin has 149. That the structure

156

Ira G. Wool

Fig. 10. (Continued). (C) the residues at positions 16 and 17 are changed to cytosine. The open arrow designates the site of cleavage in the wild-type substrate; the filled arrow designates the new site.

of the core in restrictocin closely resembles that in T1 strongly suggests that the mechanism of phosphodiester bond cleavage is the same. Studies of the T1 catalytic mechanism had suggested a two-step reaction (cf. Yang and Moffat, 1996): the first is a phosphoryl transfer in which the base-acid couple H49 and E95 (the residues and positions are in restrictocin but identified by analogy to T1) act as the general base to abstract, a proton from the 2'-hydroxyl of a ribose and H136 serves as

4

Structure and Mechanism of ~-Sarcin

157

Fig. 11. The secondary structure elements in restrictocin. A ribbon diagram of the secondary structure of restrictocin in which the elements are designated. The core of the enzyme consists of a five-stranded antiparallel/3 sheet (strands B3, B4, B5, B6, and B7) stabilized by an c~ helix (HI) having three turns that is perpendicular to the/3 sheet. The NH2 terminus has a second antiparallel/3 sheet with two strands (BI and B2) linked by loop L1, which has six residues that are not resolved. There is a helix (H2) with a single turn in the loop (L3) that links strands B3 and B4.

a general acid to protonate the 5'-oxygen of the leaving nucleotide of the RNA (which now has a 5'-hydroxyl) yielding a 2',3'-cyclic phosphate intermediate. The inorganic phosphate in the restrictocin structure may occupy the site used by the phosphorus of the 2',3'-cyclic intermediate. In the hydrolysis reaction, the roles of the catalytic residues are reversed; H136 serves as the general base and activates a water molecule; H49 and E95 donate a proton to the 2'-oxygen. The activated water attacks the phosphorus in the cyclic intermediate and forms the second product, an RNA with a 3'-phosphate. The role of R120 is not clear; it may electrostatically stabilize the phosphate group or it may support the

158

Ira G. Wool

geometry of the active site through a network of hydrogen bonds. It needs to be noted that there is some disagreement about the T~ catalytic mechanism, particularly about the assignment of the residues that serve as general base and general acid. Thus, the analysis of site-directed mutants of restrictocin designed to probe the catalytic mechanism will be helpful. Restrictocin has secondary structural elements very much like those in T~ and some other ribonucleases; what sets restrictocin apart from T~ and the other ribonucleases is the extent and the complexity of its loops (Figs. 11 and 12). Presumably it is amino acid sequences and tertiary structures in one or more of the loops that endow restrictocin with its specificity for a single nucleotide in 28S rRNA and that, perhaps, mediate entry of the toxin into the cell. In this interpretation it is the loops that give restrictocin its toxic phenotype. Loop 4, which spans residues 99 to 117, is exposed to solvent, is mobile and rich in lysine residues, and is critically positioned to play a role in the specificity of substrate binding. Restrictocin has a dome and a planar side with the catalytic residues at one end of a shallow cleft that is extended by a platform formed by loops L2 and L4; the concave surface of the platform is a possible domain for the binding of RNA (Fig. 13). Using the coordinates for the NMR conformation of the sarcin domain RNA, a docking experiment was done with restrictocin (Yang and Moffat, 1996). The program GRASP was used and advantage was taken of the fit of the concave surface of restrictocin and the crown (the G A G A tetraloop) of the RNA (Fig. 13). The lysine-rich loop L4 was found to be positioned in the major groove of the helical stem. This fit has the phosphorus of the phosjghodiester bond between G16 and A17 at the cleavage site only 0.93 A from the phosphate in the restrictocin crystal structure. In the model derived from the docking experiment, a platform formed by loops L2 and L4 provides most of the interface with the RNA: K42 in loop L2 and K106, Kll0, K l l l , and Kl13 in loop L4 could form salt bridges with phosphates in the RNA. The side chain of D108 or of S109 might form hydrogen bonds with bases in the RNA and facilitate discrimination. The positively charged ridge formed by K60, K63, K80, K88, and R65 in loop L3 is close to the sugar-phosphate backbone of G13-G16 in the R N A to further facilitate the interaction. Perhaps the most important feature of the model is that the bulged G10, the critical c~-sarcin recognition element in the RNA, is close to the lysine-rich loop L4. Thus, the structural elements in restrictocin that account for the specificity of RNA recognition are the platform formed by loops L2 and L4, and the positively charged ridge in loop L3.

4

Structure and Mechanism of a-Sarcin

159

XII. C O D A

Willie Sutton, when asked why he robbed banks, replied in what has become a cliche, "because that is where the money is." We have studied the mechanism of action of c~-sarcin because we want information on the molecular basis of the function of ribosomes in protein synthesis, on the chemistry of a very specific p r o t e i n - R N A interaction, and on the mechanism of catalysis by a special ribonucleasemand the c~-sarcin domain r R N A is where that information is. The studies of the mechanism of action of c~-sarcin have been fruitful and there is every indication they will continue to be.

REFERENCES

Better, M., Bernhard, S., Lei, S. P., Fishwild, D., and Carroll, S. (1992). Activity of recombinant mitogillin and mitogillin immunoconjugates. J. Biol. Chem. 267, 1671216718. Chan, Y. L., Endo, Y., and Wool, I. G. (1983). The sequence of the nucleotides at the c~sarcin cleavage site in rat 28S ribosomal ribonucleic acid. J. Biol. Chem. 258, 1276812770. Conde, F. P., Fernandez-Puentes, C., Montero, M. T. V., and Vazquez, D. (1978). Protein toxins that catalytically inactivate ribosomes from eukaryotic microorganisms. Studies of the mode of action of alpha sarcin, mitogillen and restrictocin: Response to alpha sarcin antibodies. F E M S Microbiol. Lett. 4, 349-355. Endo, Y., and Tsurugi, K. (1987). The mechanism of action of ricin and related toxic lectins on eukaryotic ribosomes. The site and characteristics of the modification in 28S ribosomal RNA caused by the toxins. J. Biol. Chem. 262, 5908-5912. Endo, Y., and Tsurugi, K. (1988). The RNA N-glycosidase activity of ricin A-chain. The characteristics of the enzymatic activity of ricin A-chain with ribosomes and with rRNA. J. Biol. Chem. 263, 8735-8739. Endo, Y., and Wool, I. G. (1982). The site of action of c~-sarcin on eukaryotic ribosomes. The sequence at the c~-sarcin cleavage site in 28S ribosomal ribonucleic acid. J. Biol. Chem. 257, 9054-9060. Endo, Y., Huber, P. W., and Wool, I. G. (1983). The ribonuclease activity of the cytotoxin c~-sarcin. The characteristics of the enzymatic activity of c~-sarcin with ribosomes and ribonucleic acids as substrates. J. Biol. Chem. 258, 2662-2667. Endo, Y., Chan, Y. L., Lin, A., Tsurugi, K., and Wool, I. G. (1988). The cytotoxins c~sarcin and ricin retain their specificity when tested on a synthetic oligoribonucleotide (35-mer) that mimics a region of 28S ribosomal ribonucleic acid. J. Biol. Chem. 263, 7917-7920.

160

Ira G. Wool

Endo, Y., Gltick, A., Chan, Y. L., Tsurugi, K., and Wool, I. G. (1990). RNA-protein interaction. An analysis with RNA oligonucleotides of the recognition by c~-sarcin of a ribosomal domain critical for function. J. Biol. Chem. 265, 2216-2222. Endo, Y., Gl0ck, A., and Wool, I. G. (1991). Ribosomal RNA identity elements for ricin A-chain recognition and catalysis. J. Mol. Biol. 221, 193-207. Endo, Y., Oka, T., Tsurugi, K., and Natori, Y. (1993a). The biosynthesis of a cytotoxic protein, c~-sarcin, in a mold of Aspergillus giganteus I. Synthesis of prepro- and proot-sarin in vitro. Tokushima J. Exp. Med. 40, 1-6. Endo, Y., Oka, T., Tsurugi, K., and Natori, Y. (1993b). The biosynthesis of a cytotoxic protein, ot-sarcin, in a mold of Aspergillus giganteus. II. Maturation of precursor form of ot-sarcin in vitro. Tokushima J. Exp. Med. 40, 7-12. Fando, J. L., Alaba, I., Escarmis, C., Fernandez-Luna, J. L., Mendez, E., and Salinas, M. (1985). The mode of action of restrictocin and mitogillin on eukaryotic ribosomes. Eur. J. Biochem. 149, 29-34. Fernandez-Luna, J. L., Lopez-Otin, C., Soriano, F., and Mendez, E. (1985). Complete amino acid sequence of the AspergiHus cytotoxin mitogillin. Biochemistry 24, 861-867. Fernandez-Puentes, C., and Vazquez, D. (1977). Effects of some proteins that inactivate the eukaryotic ribosome. FEBS Lett. 78, 143-146. Gltick, A., and Wool, I. G. (1996). Determination of the 28S ribosomal RNA identity element (G4319) for alpha-sarcin and the relationship of recognition to the selection of the catalytic site. J. Mol. Biol. 256, 838-848. Gl0ck, A., Endo, Y., and Wool, I. G. (1992). Ribosomal RNA identity elements for ricin A-chain recognition and catalysis. Analysis with tetraloop mutants. J. Mol. Biol. 226, 411-424. Grtick, A., Endo, Y., and Wool, I. G. (1994). The ribosomal RNA identity elements for ricin and for c~-sarcin, mutations in the putative CG pair that closes a G A G A tctraloop. Nucleic Acids Res. 22, 321-324. Gutcll, R. R., and Fox, G. E. (1988). A compilation of large subunit RNA sequences presented in a structural format. Nucleic Acids Res. 16, r175-r269. Hausncr, T. P., Atmadja, J., and Nicrhaus, K. H. (1987). Evidence that the G2661 region of 23S rRNA is located at the ribosomal binding sites of both elongation factors. Biochimie 69, 911-923. Hcnzc, P. P., Hahn, U., Erdmann, V. A., and Ulbrich, N. (1990). Expression of the chemically synthesized coding region for the cytotoxin ot-sarcin in Escherichia coli using a secretion cloning vector. Eur. J. Biochem. 192, 127-131. Hobden, A. N., and Cundliffe, E. (1978). The mode of action of alpha sarcin and a novel assay of the puromycin reaction. Biochem. J. 170, 57-61. Lacadena, J., Pozo, A. M., Barbero, J., Mancheno, J. M., Gasset, M., Ofiaderra, M., LopezOtin, C., Ortega, S., Garcia, J., and Gavilanes, J. G. (1994). Overproduction and purification of biological active native fungal c~-sarcin in Escherichia coli. Gene 142, 147-151. Lamy, B., and Davies, J. (1991). Isolation and nucleotide sequence of the Aspergillus restrictus gene coding for the ribonucleolytic toxin restrictocin and its expression in Aspergillus nidulans: The leader sequence protects producing strains from suicide. Nucleic Acids Res. 19, 1001-1006. Lopez-Otin, C., Barber, D., Fernandez-Luna, J. L., Soriano, F., and Mendez, E. (1984). The primary structurer of the cytotoxin restrictocin. Eur. J. Biochem. 143, 621-634.

4

Structure and Mechanism of a-Sarcin

161

Miller, S. P., and Bodley, J. W. (1988). ot-Sarcin cleaves ribosomal RNA at the ot-sarcin site in the absence of ribosomal proteins. Biochem. Biophys. Res. Commun. 154, 404-410. Moazed, D., Robertson, J. M., and Noller, H. F. (1988). Interaction of elongation factors EF-G and EF-Tu with a conserved loop in 23S RNA. Nature (London) 334, 362-364. Munoz, A., Castrillo, J. L., and Carrasco, L. (1985). Modification of membrane permeability during Semliki Forest Virus infection. Virology 146, 203-212. Olson, B. H., and Goerner, G. L. (1965). Alpha sarcin, a new antitumor agent. I. Isolation, purification, chemical composition, and the identity of a new amino acid. Appl. Microbiol. 13, 314-321. Olson, B. H., Jennings, J. C., Roga, V., Junek, A. J., and Schuurmans, D. M. (1965). Alpha sarcin, a new antitumor agent. II. Fermentation and antitumor spectrum. Appl. Microbiol. 13, 332-326. Orlandi, R., Canevari, S., Conde, F. P., Leoni, F., Mezzanzaniza, D., Ripamonti, M., and Colnaghi, M. I. (1988). Immunconjugate generation between the ribosome inactiving protein restrictocin and anti-human breast carcinoma MAB. Cancer Immunol. Immunother. 26, 114-120. Otero, M. J., and Carrasco, L. (1985). Proteins are cointernalized with virion particles during early infection. Virology 160, 75-80. Otero, M. J., and Carrasco, L. (1988). Exogenous phospholipase C permeabilizes mammalian cells to proteins. Exp. Cell Res. 177, 154-161. Sacco, G., Drickamer, K., and Wool, I. G. (1983). The primary structure of the cytotoxin ot-sarcin. J. Biol. Chem. 258, 5811-5818. Schindler, D. G., and Davies, J. E. (1977). Specific cleavage of ribosomal RNA caused by alpha sarcin. Nucleic Acids Res. 4, 1097-1110. Stern, S., Powers, T., Changchien, L. M., and Noller, H. F. (1989). RNA-protein interactions in 30S ribosomal subunits: Folding and function of 16S rRNA. Science 244, 783-790. Szewczak, A. A., and Moore, P. B. (1995). The sarcin/ricin loop, a modular RNA. J. Mol. Biol. 247, 81-98. Szewczak, A. A., Moore, P. B., Chan, Y. L., and Wool, I. G. (1993). The conformation of the sarcin/ricin loop from 28S ribosomal RNA. Proc. Natl. Acad. Sci. U.S.A. 90, 9581-9585. Turnay, J., Olmo, N., Jiminez, A., Lizarbe, M., and Gavilanes, J. (1993). Kinetic study of the cytotoxic effect of cz-sarcin, a ribosome inactivating protein from Aspergillus giganteus, on tumor cell lines: Protein biosynthesis inhibition and cell binding. Mol. Cell. Biochem. 122, 39-47. Vazquez, D. (1979). "Inhibitors of Protein Biosynthesis," pp. 80-81. Springer-Verlag, Berlin. Wawrzynczak, E., Henry, R. V., Cumber, A., Parnell, G., Derbyshire, E., and Ulbrich, N. (1991). Biochemical, cytotoxic and pharmacokinetic properties of an immunotoxin composed of a mouse monoclonal antibody Fib75 and the ribosome-inactivating protein c~-sarcin from Aspergillus giganteus. Eur. J. Biochem. 196, 203-209. Wnendt, S., Felske-Zech, H., Henze, P., Ulbrich, N., and Stahl, U. (1993). Characterization of the gene encoding o~-sarcin, a ribosomal-inactivating protein secreted by Aspergillus giganteus. Gene 124, 239-244. Wool, I. G. (1986). Studies of the structure of eukaryotic (mammalian) ribosomes. In "Structure, Function, and Genetics of Ribosomes" (B. Hardesty and G. Kramer, eds.), pp. 391-411. Springer-Verlag, New York.

162

Ira G. Wool

Wool, I. G., Gltick, A., and Endo, Y. (1992). Ribotoxin recognition of ribosomal RNA and a proposal for the mechanism of translocation. Trends Biochem. Sci. 17, 266-269. Yang, R., and Kenealy, W. (1992). Regulation of restrictocin production in Aspergillus restrictus. J. Gen. Microbiol. 138, 1421-1427. Yang, X., and Moffat, K. (1996). Crystal structure of the highly specific Aspergillus ribotoxin, restrictocin. Structure 4, 837-852.

Fig. 7.

Fig. 12.

Fig. 13.