In bacteria, two release factors have been characterized. RF-1' recognizes ... were first dialyzed into a thiol-free buffer, 50 mM Hepes-KOH, pH. 7.6, 50 mM KCl, ...
Val. 261, No. 5, Issue of February 15, pp. 2289-2293,1986 Printed in U.S.A.
T H EJOURNAL OF BIOLOGICAL CHEMISTRY 0 1986 by The American Society of Biological Chemista. Inc.
The Ribosomal Binding Domain of the Escherichia coli Release Factors MODIFICATION OF TYROSINE IN THE N-TERMINAL RELEASEFACTORS 1 AND 2DIFFERENTIALLY*
DOMAIN OF RIBOSOMAL PROTEIN L11AFFECTS
(Received for publication, June 20, 1985)
Warren P. TateSQ,Kim K. McCaughanS, Christina D. Ward$, Vicki G. SumpterS, Clive N. A. Trotman$, Marina Stoffler-Meilicke, PeterMalyll, andRichard Brimacombe From the $Department of Biochemistry, University of Otago, Dunedin, New Zealand and the Max-Planck-Institut fur Molekulare Genetik, Abteilung Wittman, D 1000, Berlin-33, Germany
Ribosomal protein L11 is one of only two ribosomal proteins significantly iodinated when Escherichia coli 50 S subunits are modified by immobilized lactoperoxidase, and the major target has been shown previously to be tyrosine at position 7 in the N-terminal domain. This modification reduces in vitro termination activity with release factor (RF)-1 by 70-90%, but RF2 activity is less affected (30-50%). The loss of activity parallels incorporation of iodine into the subunit. The 50 S subunits from L11-lacking strains of bacteria have highly elevated activity with RF-2 and low activity with RF-1. The iodination does not affect RF-2 activity but reduces the RF-1 activity further. Ribosomal proteins, L2, L6, and L25, are significantly labeled in Lll-lacking ribosomes in contrast to the control 50 S subunits. L11 has been modified in isolation and incorporated back efficiently into L11-lacking ribosomes. This L1 1, iodinated also predominantly at Tyr 7, is unable to restore RF-1 activityto Lll-lacking ribosomes in contrast to mock-iodinated protein. These results suggest the involvement of the N terminus of L11 in the binding domain of the bacterial release factors and indicate that there are subtle differences in how the two factorsinteract with the ribosome.
the factors into a functional complex with the termination codon. The N-terminal partof L11 seems critical since antibodies against a 64-amino acid fragment from this region strongly inhibitedin vitro termination in contrast to antibodies against other parts of the molecule. This part of L11 has been localized in the region of the 50 S subunit where the L7/ L12 stalk originates ( 7 ) . When the 50 S subunit of E. coli is iodinated under mild conditions with immobilized lactoperoxidase only two proteins, L5 and L11, are highly labeled; of the 1-2 mol of iodine incorporated per molof subunit about half is found in L11 (9). L11 contains two tyrosine residues, at positions 7 and 61 (10). Most of the iodine is found on the tyrosine 7 (9). We have utilized this finding to define further the binding domains of the two release factors by modifying L11 in situ and also in isolation prior to reconstituting intoribosomes lacking L11. MATERIALS AND METHODS
E. coli tRNAN“ was from Boehringer Mannheim, Enzymobead
lactoperoxidase/glucose peroxidase from Bio-Rad, and NalZ5Iand L[3H]methionine (10 Ci/mmol) werefrom Amersham Corp. E. coli MRE 600 was from the Public Health Laboratory Service, Porton, United Kingdom, and E. coli strains AM68 and AM76 were from Dr. E. Dabbs (11).Ribosomes were isolated from each strain as for tight couples but omitting the final gradient centrifugation step (12). In bacteria, two release factors have been characterized. Subunits were prepared from the 70 S ribosomes by gradient centrifin sucrose for 17 h a t 23,000 rpm in a Beckman SW 27 rotor RF-1’ recognizes UAA and UAG and RF-2, UAA and UGA ugation in either of two buffers: 0-33% (w/v) sucrose, 20 mM Tris-HCI, pH (1).While earlier studies with antibiotics (2, 3), antibodies 7.6, 50 mM NH40Ac, 0.5 mM Mg(OAc),; or 10-30%(w/v) sucrose, 50 against ribosomal proteins (4),and ribosomal core particles mM Tris-HC1, pH 7.6, 200 mM NH,Cl, 3 mM Mg(OAc)Z, 10 mM 0( 5 ) suggested that the two factors had very similar binding mercaptoethanol. E. coli tRNANe‘ was aminoacylated with [3H]medomains, recently it was found that theEscherichia coli ribo- thionine and formylated as described (13). Release factors, RF-1 and somal protein L11 suppresses RF-2 activitybut promotes the RF-2, were prepared from extracts of E. coli MRE6OO according to activities of RF-1 (6, 7 ) . This effect is secondary to a require- the procedure of Caskey et al. (13). Iodination of Ribosomal Subunits and Protein LI I-The subunits ment for the stalk proteins of the 50 S subunit L7/L12 (8). were first dialyzed into a thiol-free buffer, 50 mM Hepes-KOH, pH The modulating effect of L11 on the release factor activities 7.6, 50 mM KCl, 5 mM Mg(0Ac)Z. The iodination reaction was has been demonstratedboth when L11 is omitted from a performed on 2 A2wnmsubunits or an amount of protein L11 equivaribosomal subunit reconstituted from cores and split proteins lent to that found in 10 AZmnm 50 S subunits generally in 50 pl in the (6) and with ribosomes lacking L11 isolated from mutant presence of20 p M NaI and, where appropriate, 1 pCiof Nalz5I, strains of E. coli (7). L11 exerts its effect on the binding of together with 5 rl of Enzymobead lactoperoxidase/glucose oxidase reagent (1 vial in 500 pl of water). The reaction was initiated with * The costs of publication of this article were defrayed in part by the addition of fl-D-glucoseto 0.2%. After incubation at 25 “C for 30 the payment of page charges. This article must therefore be hereby min (subunits) or 15 min (L11) the reaction was terminated with pmarked “aduertisement” in accordance with 18 U.S.C. Section 1734 mercaptoethanol (1%) and theEnzymobeads were removed by centrifugation. The iodinated subunits were precipitated with ethanol solely to indicate this fact. forgel analysis or used without further treatment for functional § Recipient of a fellowship from the Humboldt Foundation and a studies. With iodinated L11 the sample was dialyzed into a reconstigrant from the Medical Research Council of New Zealand. tution buffer, 20 mM Tris-HC1, pH 7.6,400 mM NH4C1, 0.2 mM To whom reprint requests should be addressed. ll Present address: Biochemisches Institut der Universitat Zurich, EDTA, 20 mM Mg(OAc)t, and4 mM P-mercaptoethanol, or used Winterthurerstr. 190, CH-8057, Zurich. directly in a reconstitution protocol with ribosomes lacking L11. The ‘The abbreviations used are: RF, release factor; Hepes, 4-(2- L l 1 was added at a slight molar excess (1.25) over the 70 S ribosomes hydroxyethy1)-1-piperazineethanesulfonic acid. from either AM68 or AM76 and incubated at 4 “C for 15 min in a
2289
Ribosomal Domain of Bacterial Factors Release
2290
buffer, 20 mM Tris-HC1, pH 8.0, 50 mMNHdC1, 10 mM Mg(OAc)s, 4 mM 0-mercaptoethanol. I n Vitro Termination Reactions-For preparation of ribosomal complexes containing [3H]fMet-tRNAat theP site 50 pl of a mixture containing 20 mM Tris-HC1, pH 7.5, 150 mM NH4C1, 10 mMMgC12, 50 pmol of 70 S ribosomes or 50 S and 30 S subunits, 0.2 A 2 W n m unit of AUG and 10-15 pmol of [3H]fMet-tRNA (4000 cpm/pmol) were incubated for 30 min a t 30 “C. For in vitro termination 5 pl of this complex was added to 45 pl of a solution to give 50 mM Tris-acetate, pH 7.2, 30 mM Mg(OAc)z, 75 mM ammonium acetate, and 0.08 AZ6,, units of UAA and 0.5-5 pg of RF-1 or RF-2 (depending on the purity). Analysis of Iodinated Proteins-The 50 S subunits were treated with ribonucleases A and T1 in sodium dodecyl sulfate buffer, and the proteins were precipitated with ethanol (16). The proteins were then separated by two-dimensional gel electrophoresis, the radiolabeled proteins were located by autoradiography, and theappropriate spots cut out andcounted (17). RESULTS
The extent of the incorporation of iodine into protein L11 when the 50 S subunit was modified, together with the presence of almost all of the label in one of the two tyrosines, tyrosine 7 of the 141 amino acids, implied that most of the ribosomal particles would carry the modification in this position (9). Since this was within the N-terminal region of L11 which had been shown to be important for the activities of the two release factors (7) the method allowed an analysis of the importance of tyrosine 7 in the binding domains of the factors. Initially it was checked that the ribosomal subunits used in thefunctional studiesindeed had thesame iodination pattern since they were prepared by a different procedure; L5 and L11 were the proteins significantly labeled while other proteins were labeled to a small extent (see Fig. 4, panel B ) . This isconsistent with the original studies. When iodinated 50 S subunits were constituted into 70 S substrates for i n vitro termination the resulting ribosomal complexes were as active as their unmodified counterparts in their peptidyltransferase activity, essential for the peptidyltRNA hydrolysis step of the termination reaction. The release of the model peptide, m e t , from the complex in response to the release factors and termination codon in contrast was significantly affected by the iodination as shown in Fig. 1. Mock-iodinated suhunits where glucose wasomitted from the immobilized lactoperoxidase system (open bars)showed activity undiminished from normal subunits, whereas the iodinated subunits (closed bars) consistently had only 10-30% of the activity of the controls with RF-1 and 50-70% with RF-2. These data suggested that thetyrosine residue or its environment was important to the domain of activity of the release
factors although RF-1 activity wasmore sensitive to the modification. The relationship between the loss of RF-mediated termination activity observed in Fig. 1 and the incorporation of radiolabeled iodine into the 50 S particles is shown in Fig. 2. Since the rate of incorporation of iodine and the loss of activity are rapid, this result hasbeen presented as afunction of the concentration of glucose which initiates the labeling. At concentrations below 0.01% glucosethere was little incorporation or loss of activity but at 0.01% glucose there was about 50% incorporation (open squares) and also about 50% of the final loss of activity observed for each release factor (RF-1, closed circles; RF-2, open circles). Moreover, no further modification of the subunit occurred when the loss of activity of the subunits for termination in vitro had reached a plateau (0.05-0.1% glucose). Nor could further loss of activity or more extensive labeling be obtained by adding a further aliquot of iodinating reagent. However, in this lattercase the proportion of iodine in L11 compared with the other minor labeled proteins decreased slightly. Apparently the less exposed tyrosines of other ribosomal proteins continued to be labeled while no further labeling of L11 was proceeding. The extent of labeling from these experiments suggested that 0.5-0.8 mol of iodine was incorporated per mol of L11 although the nm. amount of L11 hasbeen estimated only from AZ3,, The- other ribosomal protein extensively labeled by the immobilized lactoperoxidase is L5. This protein is not required, however, for in vitro termination since it can be omitted from reconstituted ribosomes without loss of release factor-dependent functions (6). It is possible that the loss of activity observed above after iodination of the 50 S subunit results partially from the modification of other ribosomal proteins. We have, therefore, modified 50 S subunits isolated
RF-2
Ir
FIG. 1. Effect of iodination of 50 S subunits on in vitro termination with RF-1and RF-2. The 50 S subuits were iodinated as described under “Materials and Methods” and reconstituted with unmodified 30 S subunits into substrate for in vitro termination. Control subunits were treated identically except glucose to initiate the iodination reaction was omitted. Backgrounds without added release factor (400 cpm) were subtracted in each case.Openbars, mock-iodinated subunits; closed bars, iodinated subunits.
FIG. 2. Relationship between incorporation of lZ5I into 50 S subunits and loss of in vitrotermination activity. The extent of incorporation of ‘‘’1 into 50 S subunits was monitored. Each reaction contained 21,900 cpm of Nal”I in addition to the 20 p M nonradioactive NaI and various concentrations of glucose to initiate the reaction. Subunits from parallel reactions not containing the radioactive iodine were used to make substrate for in vitro termination reactions. The activity of each was tested inRF-1- and RF-2-mediated reactions. Backgrounds without added release factor (400 cpm) were subtracted ineach case, and thevalues of the mock-iodinated subunits (100% values) were 1,625 cpm for RF-1 and 2,160 cpm for RF-2. A background of 750 cpm was subtracted for the incorporation of iodine from the mock-iodinated reaction (no glucose). 0, incorporation of [lZ5I];0, RF-1-mediated reaction; 0,RF-2-mediated reaction.
Ribosomal Domain
of Bacterial Factors Release
from two L11-lacking strains of E. coli, AM68 and AM76 (11). AM68 has very low activity with RF-1 unless a 10-fold higher concentration of factor is added whereas AM76 has some activity with RF-1 at normal concentrations. In Fig. 3 the activities of the mock-iodinated and iodinated L11+ and L11ribosomes are shown as a function of the concentrations of the release factors added. The L11+ ribosomes after iodination showed 10-20% activity with RF-1, and activity was not restored by excess factor (panel A ) . Similarly the RF-2 activity was reduced by 50% and was not restored with excess factor (panel C). The activity of RF-1 on L11- AM76 ribosomes was decreased after modification of the subunit (panel B ) suggesting part of the loss observed with L11+ ribosomes might be independent of L11; in contrast activity with RF-2 was unaffected (panel D) suggesting all of the loss of activity (panel C) was dependent upon L11. Ribosomes lacking L11 proved to be a suitable control to test the specificity of how iodinating L11 affects the binding domain of the two release factors. However, the absence of L11 may mean tyrosines on other proteins are now exposed and, therefore, available for labeling. We have, therefore, compared the labeled ribosomal proteins from the Lll-deficient strains AM68 and AM76 withthose from acontrol strain (Fig. 4). Panel A shows a tracing of a stained gel; the other panels are autoradiographs. Panel B shows a control 50 S subunit; panels C and D show the L11-deficient subunits from AM68 and from AM76. Proteins L5 and L11 were strongly labeled in thecontrol, and the radioactivity in these two spots accountsfor over half the total. In contrast thetwo L11-lacking subunits were strongly labeled in L5, showed no L11, but L2, L6, L10, and L25 were now significantly labeled. When L11 was inserted back into the L11-deficient subunit significant labeling in this group of proteins (L2, L6, and L25) was not found just as in the control. These dataindicate that while the L11-lacking ribosomes have proved useful to assess whether a low level of labeling of other ribosomal proteins
A
C
2291 B
D
2 l*
FIG. 4. Two-dimensional polyacrylamide gel profiles of iodinated 50 S ribosomal proteins from control and Lll-lacking subunits. Subunits were iodinated and the proteins isolated as described under “Materials and Methods.” Gels were stained and autoradiographed. The gelswere run from left to right in the first dimension and top to bottom in the second dimension. Pawl A, 50 S proteins stained with Coomassie Blue; panel B, lZ5I-labeledproteins from L11+ subunits;puwl C, lZ5I-labeledproteins from L11- subunit proteins from L11- subunit (AM76). (AM68); panel D, 1Z51-labeled Note L11 appears as a double or triple spot inthis gel system (9).
has aneffect on the terminationreaction in vitro they arenot the ideal control. The loss of activity of RF-1-mediated reactions afteriodination (Fig. 3) probably reflects the iodination of other proteins at thebinding domain. As an alternative approach to labeling L11 in situ we have RF1 ILll* ribosomes) A RF1 lLl1ribosomes) B iodinated it separately and then restored it to the ribosomal subunit, leaving the ribosomal background unmodified. The labeling pattern of the two tyrosines in L11 may differ when the protein is modified in situ in the 50 S particle or in isolation. We have used cleavage with cyanogen bromide after iodination of L11 in thesetwo situations to test this. Tyrosine 7 is in a small fragment of 16 amino acids while tyrosine 61 is in a large fragment of 81 amino acids. Tyrosine 7 has 77% of the incorporated label when L11 is modified in isolation and 85% when it is modified in situ. In both cases, therefore, tyrosine 7 is the prime target for the modification. The iodination of the tyrosines did not hinder the incorporation of L11 intothe reconstituted particle. Only the subunit reconstituted with the iodinated L11 rather than the mock-iodinated L11 had radioactivity associated with it (Fig. 5). About 70% of the labeled L11 was reconstituted when added at a molar equivalence to the subunit. Thus the effect of iodinating L11 aloneon termination in vitro could be measured. Ribosomes lacking L11 have low activity with RF-1 which release factor I PI) FIG. 3. In vitro termination activity of 50 S subunits from can be restored by reconstitution of L11 whereas the elevated be depressed to about 50%.L11 lacking control and AM76 (L11 deficient)strainsof E. coli after activity with RF-2 can modification with iodine. After reconstitution of the iodinated 70 S ribosomes were reconstituted with mock-iodinated L11 subunits into substrates for in vitro termination their activity in RF- or iodinated L11 and theeffects on RF-1-and RF-2-mediated 1- and RF-2-mediated reactions was measured. Background in the absence of release factors (900-1200 cpm for control 50 S and 500- activities determined. Mock-iodinated L11 restored the activ700 cpm for L11- 50 S)were subtracted in each case. For RF-1 1 pl ity of AM68 ribosomes with RF-1 whereas the iodinated contained 50 pg of protein and for RF-2 1-2 pg of protein. 0, mock- protein did not (Fig. 6). A similar resultoccurred with AM76 iodinated 0, iodinated. ribosomes, but therewas significant activity in theabsence of
I /
Ribosomal Domain of Bacterial Release Factors
2292
approach is independent of any differences in the labeling pattern of L11 in situ or in isolation but still allows the L11 to be restored to anunmodified background. DISCUSSION
L11 is the 50 S ribosomal protein most highly labeled with immobilizedglucose peroxidase/lactoperoxidase (9), which iodinates only the exposed tyrosines. Most of the label is found intyrosine 7 with a small proportion in tyrosine 61 (9). Since tyrosine 7 is highly exposed it is agood candidate to be in thedomain of interaction of the release factors. The activity of RF-1 is very strongly inhibited by this modification whereas that of RF-2 is partially blocked. Protein L5 is significantly iodinated by the procedure, and others are to minor a degree. RF-2FIG. 5. Reconstitution of iodinated L11 into particles lack- Particles lacking L5 are active for both RF-1- and ing L11. L11 was mock iodinated or iodinated and, after removal of mediated reactions (6), but each iodinated particle with a the immobilized glucose oxidase/lactoperoxidase, it was reconstituted modified L11 may have one or more other modifications. The into 70 s ribosomes from the L11-lacking strain, AM68. Samples were use of L11-deficient particles provided one method of deterapplied to sucrose gradients and sedimented under conditions pro- mining that the loss of RF functions resulted from the iodimoting dissociation of the ribosomes into subunits. Sedimentation is from right to left. Panel A, mock-iodinated subunits; Panel B, iodi- nation of L11. RF-2 function was unaffected on iodinated particles lacking L11, but therewas some lossof activity with -, absorbance at 254 nm. nated subunits. 0, lZ5Iand lZ5I-Lll; RF-1. When L11 is missing, however, tyrosines on four proteins, L2, L6, L10, and L25, are significantly more exposed. Protein L6 andL10 has been mapped on the 50 S subunit by B RF-1 A RF-2 immunoelectron microscopy to be closeto L11 below the base of the L7/L12 stalk (14) and so are in the vicinity of the domain of interaction of the release factors. Indeed, recently an antibody preparation against L6 has been found to inhibit RF-1-dependent reactions.*L25 has been mapped on the stalk E n side of the central protuberance, and some antibody-binding 4 sites extend toward the L11 domain (15). The position of L2 0 on the subunit has not yet been elucidated. Our data suggest I1 a# that even if these two proteins prove not to be in the region c of L11 an absence of this protein must perturb the confora II mation of the subunit to alter the exposure of residues on E other proteins. An alternative more direct approach was to iodinate L11 0 AM76 AM68 AM76 before restoration into the L11-lacking particle. The conforFIG. 6. Activity in RF-I- and RF-2-mediated reactions of mation of L11 in solution and the relative exposure of the 70 S ribosomes reconstituted with iodinated L11. L11 was tyrosines to thereagent are likely to be different from that of iodinated and restored to L11-deficient 70 S ribosomes from the the protein when on the subunit. The second tyrosine at strains AM68 and AM76. The activity of these ribosomes in RF-1- position 61 is still not significantly labeled, however, when and RF-2- mediated reactions was compared with those containing the modification is carried out on the protein in isolation. A no L11 or mock-iodinated L11. Background values in the absence of release factors (1000 cpm, AM68, 1300 cpm, AM76) were subtracted. third approach involved iodination of L11 in situ, isolation of the protein within a TP50, and reconstitution of the modified L11 from this fraction into an L11-lacking particle. QualitaL11 which was doubled by the mock-iodinated protein but tively all approaches to iodinate L11 gave similar effects on not by the iodinated L11. The results with the RF-2-mediated the release factor mediated reactions; RF-1 activity was very reactions were less clear reflecting the smaller effect the sensitive to the modification whereas RF-2 activity was only iodination of L11 has on RF-2 functions and the relatively partially affected. modest depression of RF-2 activity by L11. In this case the Collectively the dataimply that thetwo release factors have mock-iodinated protein depressed the activity by half with subtle differences in their domains of interaction involving AM68 ribosomes and somewhat less with AM76 ribosomes, L11. In thepresence of L11 both factorshave the same specific and themodified protein consistently showed some modulat- activity, but in its absence there is a 50-fold difference (6). ing effect on RF-2. The extent of this compared with the Subfractions of an IgG against L11 have previously been unmodified protein has been variable as is reflected in Fig. shown to effect RF-1- and RF-2-mediated reactions in a 6B. These data support the findings from the iodination of reciprocal manner implying that theinteraction of the factors L11 in situ that RF-1 is sensitive to modification of tyrosine at the L11 domain must be different (7). The N-terminal 7 whereas RF-2 functions are less affected. region ofL11 has otherunique residues amenable to chemical As an alternative protocol a protein fraction -(TP50) con- modification, an arginine at position 64 and a cysteine at taining L11 has been isolated from iodinated and mock- position 38. Modification of these residues will provide a more iodinated 50 S subunits and reconstituted into L11 lacking 50 detailed profile of the domain of interaction of the release S particles. The TP50 had no effect on normal 50 S subunits factors. We have modified the free lysine residues within the but that from the mock-modified subunits restored RF-1 activity to L11-lacking subunits. The TP50 from modified W. P. Tate, M. Stoffler-Meilicke, and G. Stoffler, unpublished subunits was unable to restore RF-1-dependent activity. This data.
L
I
‘c
AM68
Ribosomal Domain of Bacterial Release Factors protein by reductive methylation without loss of activity of the reconstituted ribosomes for release factor-mediated reactions. Acknowledgments-Dr. Eric Dabbs kindly supplied the L11-E. coli strains, AM68 and AM76. We thank Dr. H. G. Wittmann, Dr. G. Stoffler for constant interest and support, and Dr. A. Carne for helpful discussions.
1.
2. 3.
4. 5.
REFERENCES Caskey, C. T. (1980) Trends Biochem. Sci. 5,234-237 Brot, N., Tate, W. P., Caskey, C. T. & Weissbach, H. (1974) Proc. Natl. Acad. Sci. U. S. A. 7 1 , 81-92 Caskey, C. T. & Beaudet, A. L. (1971) in Molecular Mechanisms of Antibiotic Actions on Protein Biosynthesis and Membranes (Munoz, E., Garcia-Fernandez, F. & Vazquez, D., eds) pp. 326336, Proceedings of the Symposium at Granada, Spain, Elsevier, Amsterdam Tate, W. P., Caskey, C. T. & Stoffler, G . (1975) J . Mol. Biol. 93, 375-389 Armstrong, I. L. & Tate, W. P. (1980) FEBS Lett. 109, 228-232
2293
6. Tate, W. P., Schulze, H. & Nierhaus, K. H. (1983) J. Biol. Chem. 258,12si~-1282o 7. Tate. W. P.. Doenin. M. J.. Noah. M.. Stoffler-Meilicke. M.. and Stoffler, G. (1584)’J.Bioi. Chem. 2 5 9 , 7317-7324 8. McCaughan, K. K., Ward, C. D., Trotman, C. N. A. & Tate, W. P. (1984) FEBS Lett. 175,90-94 9. Maly, P., Wower, J., Zobawa,M. & Brimacombe, R. (1983) Biochemistry 22,3157-3162 10. Dognin, M. J. & Wittmann-Liebold, B. (1977) FEBS Lett. 8 4 , 342-346 11. Dabbs, E. R. (1977) Mol. Gen. Genet. 151,261-267 12. Noll, M., Hapke, B., Schreier, M. H., and Noll, H. (1973) J. Mol. Biol. 75,281-294 13. Caskey, C. T., Scolnick, E., Tompkins, R., Milman, G . & Goldstein, J. (1971) Methods Enzymol. 20, 367-375 14. Stoffier-Meilicke, M., Noah, M. & Stoffler, G . (1983) Proc. Natl. Acad. Sci. U. S. A. 80,6780-6784 15. Noah, M. (1982) Ph.D. thesis, Techische Universitat, Berlin 16. Ulmer, E., Meinke, M., Ross, A., Fink, G . & Brimacombe, R. (1978) Mol. Gen. Genet. 160,183-193 17. Wower, J., Maly, P., Zobawa,M. & Brimacombe, R. (1983) Biochmistry 22,2339-2346 I
,
