Assembly mapping of 30 S ribosomal proteins from Escherichia coli. Further studies.

Abstract Further studies were performed on the sequence of addition of proteins to 16 S RNA during the in vitro reconstitution of 30 S ribosomal subunits from Escherichia coli. Direct binding of protein S17 to 16 S RNA was studied in detail, and the following results were obtained: (a) under reconstitution conditions, a maximum of approximately 1 mole of S17 is bound per mole of 16 S RNA, either alone, or in the presence of all other 30 S proteins; (b) S17 binds only to 16 S RNA and not to 23 S RNA; and (c) radioactive S17–16 S RNA complexes are directly converted (without dissociation) to 30 S subunits by the addition of excess unlabeled total 30 S proteins. From these results, we conclude that the binding of S17 to 16 S RNA is specific. We have also determined the positions of S15, S16, S17, and S12 in the assembly map and have clarified subsequent interactions depending on these proteins. A revised assembly map is presented which incorporates the additional information obtained from these experimental results.

Prom the Institute for Enzyme Irlesearch, Departments of Biochemistry ad Genetics, IJniversity of Wisconsin, Madison, Wisconsin 53706 SUMMARY Further studies were performed on the sequence of addition of proteins to 16 S RNA during the in vitro reconstitution of 30 S ribosomal subunits from Escherichia coli. Direct binding of protein S17 to 16 S RNA was studied in detail, and the following results were obtained: (a) under reconstitution conditions, a maximum of approximately 1 mole of S17 is bound per mole of 16 S RNA, either alone, or in the presence of all other 30 S proteins; (b) S17 binds only to 16 S RNA and not to 23 S RNA; and (c) radioactive S17-16 S RNA complexes are directly converted (without dissociation) to 30 S subunits by the addition of excess unlabeled total 30 S proteins.
From these results, we conclude that the binding of S17 to 16 S RNA is specific.
We have also determined the positions of S15, S16, S17, and S12 in the assembly map and have clarified subsequent interactions depending on these proteins.
A revised assembly map is presented which incorporates the additional information obtained from these experimental results.
Previous studies from this laboratory (I) Iiavc: tl~moiistratecl that i7L vitro reconstitution of 1Lscherichia co& 30 S subunits occurs in a scquentinl, cooperative fashion. It IV515 fountl that untle~ reconstitution conclitions only a few proteins bintl tlilcctly to 16 S RK\'X.
Thohc originally found were S4, S7, M, Sl3, 517 istuclicd as a mixture of SIG and S17), and S2O.l ('crtain other proteins * This work was supported in part by the ('ollcgc of Agriculturc and I,ife Sciences, University of Wisconsin, and by grants from the n'ational Institute of (:encral hIedic:~l Sicicnces ((;Rl-15422) and the IVntional Science Foundation ((+lL3lO%X). This is Paper Ko. 1715 of the Laboratory of (;cnrtics and the 23rd paper in the scrics on "structure and function of hactcrial ribosomes." The preceding paper in this series is llef. 35 In this manner, the sequence of addition of proteins to the 16 8 RN'h was studied, and from the data obtained, an "assembly map" was constructed showing cooperative effects among the various 30 S ribosomal proteins in the asscrnbly reaction (see Fig. 6, il&).
However, our initial studies 011 the assembly mapping of 30 S ribosomal proteins were incomplete (1). First, in these esperimerits the binding of protein Sl5 \vas not investigated. Schaup et (11. (3) subsequently found that protein S15 binds directly to 16 S RN_\. This observation has been confirmed by us and preliminary reports have been made with respect to this protein (4,5). Second, we did not use pure S16 and S17 in construction of the map, but a mixture of thcsc two proteins.
Third, because of technical problems, the position of the protein Sl2 in the assembly map was not cletcrminetl. Finally, although we have clemonstratetl direct incorporation into 30 S subunits of 16 S RNA-protein complexes obtained after the individual binding of S4 and S8 to 16 S RNA, such rigorous proof of specificity of binding was not done with respect to other initial binding proteins.
Subsrqucntly, Schaup et al. (3,6), as well as Garrett et al. (7), studiecl individual binding of 30 S proteins to 16 S RNA and concluded that the binding of S17 to 16  binding experiments were done with a mixture of S16 and S17.
In some experiments, radioactive ribosomal proteins were used We have subsequently found that radioactive S17 ( of the specific activity,2 however, we consider that this value is in groups to the protein amino groups. reasonable agreement with a stoichiometry of 1 mole of S17 per The specific radioactivities of the S17 and 816 used in these experiments were the same as that described in the legend to Table I. mole of 16 S RNA. This conclusion is strengthened by our observation that the saturation value for the binding of S17 to 16 S RNA by itself is close to that obtained in the presence of all other 30 S proteins ( Fig. 1, a and b, and Fig. 2). The same conclusion was obtained when we used [a5S]S17 instead of [i4C]CH3-517 (data not shown).
2. Garrett et al. (7) concluded that the binding of 517 to 16 S RNA is nonspecific because S17 binds equally well to 16 S and 23 S RNA.
We, therefore, examined this possibility.
['%I-CH,-S17 was incubated with an equimolar mixture of 16 S RNA and 23 S RNA under standard reconstitution conditions, and then the binding was examined by sedimentation analysis using sucrose gradients made with either reconstitution buffer or Buffer C. In the former buffer, 23 S RNA tends to aggregate, and therefore, the latter buffer was also used. In both cases azPlabeled 16 S RNA was added to locate the position of 16 S RNA in the gradients.
As shown in Fig. 3, S17 bound exclusively to 16 S RNA.
No binding to 23 S RNA was observed. 3. Finally, we have examined whether the observed 16 S RNA-S17 complex represents a genuine intermediate in the in vitro assembly reaction.
The experiment was similar to the ones reported in our earlier paper (1) with respect to S4 and 58 binding.
[a5S]S17 or [14C]CHa-S17 was bound to 16 S RNA under standard reconstitution conditions. The complex was then mixed with a 3-to 4-fold excess of nonradioactive unfractionated 30 S proteins (TP30), and incubation was continued to assemble 30 S subunits.
The resulting reconstituted 30 S subunits were purified, and the specific activity was compared with that of the original 16 S RNA-S17 complex.
If the 16 S RNA-S17 complex is not an actual intermediate but a nonspecific aggregate, we would expect a large decrease in the specific activity as a result of dissociation of the initial nonspecific aggregate,  I Binding of SlY and S16 to 16 S RNA and 16 S RNA-Protein Complexes The binding of radioactive 517 and S16 was examined as described under "Materials and Methods." The specific radioactivity of [14C]CH~-S17 was calculated to be 2240 cpm per Azao eq and that of the [i4C]CHs-S16, 1450 cpm per AzBO eq. Based on the above specific radioactivity values the maximum amount of S17 bound (in the presence of all 30 S proteins) corresponds to 1.37 moles per mole of 16 S RNA, and for S16, 1.12 moles per mole of 16 S RNA. However, because of possible errors in specific activity determinations,2 we consider these values to be approximate.
It is likely that the maximum binding (lOOg,) observed in both cases corresponds to about 1 mole of radioactive protein per mole of 16 S RNA. Experiment 3 was done using the same proteins and conditions as used in Experiment 2, and the relative amounts were calculated, based on the maximum binding obtained in Exweriment 2. mixing with excess nonradioactive proteins, and subsequent reincorporation into the reconstituted 30 S subunit in the alternative "correct" site. Alternatively, the initial nonspecific aggregate may prevent conversion of 16 S RNA to 30 S subunits. We found that the initial 16 S RNA-S17 complex was converted to 30 S subunits, and no decrease in the specific activity was observed (Table II) .8 Control experiments, in which excess nonradioactive TP30 was added to radioactive S17 before mixing with 16 S RNA, showed an expected large decrease in the specific activity of the reconstituted 30 S subunits.
Thus, we conclude that the 16 S RNA-S17 complex can be converted to 30 S subunits directly without dissociation.
From the above experimental results, we conclude that 517 binds specifically to 16 S RNA in the absence of all other 30 S proteins.
As already mentioned, in order to obtain approximately 1 a A slight but significant increase in the binding of radioactive 517 was noted after the addition of excess TP30 to the preformed 16 S RNA-S17 complex, as shown in Table II. This can be explained by assuming that a small fraction of the S17 bound to 16 S RNA alone may be bound weakly and not recovered by the sucrose gradient centrifugation technique.
However, the binding may be sufficiently strong so that it is not completely diluted by the excess TP30 added but is "chased" into 30 S ribosomes  mole of S17 bound per mole of 16 S RNA, it was necessary to add approximately 3 moles of labeled S17 (Fig. 2). This may reflect the presence of some inactive S17 in the preparation, weak binding of S17 to 16 S RNA, or loss of S17 due to adsorption to glass during the experiments. The results shown in Fig. 2 indicate that at low molar ratios of S17 to 16 S RNA, the amount of S17 bound to 16 S RNA alone is less than that bound when all correct.
As a mixture, both S17 and S16 bind to 16s RNA." This statement was made based on unpublished preliminary experiments. In some earlier experiments, we tested the binding of nonradioactive Sl6 and S17 by analyzing the bound proteins by the twodimensional electrophoresis technique.
With this technique, the spot of S16 is very close to the spot of S17 as well as the spot of the oxidized form of S17 (9). When we examined the binding of Sl6 in the presence of S17, we found two major spots and thought that they were S16 and S17.
In view of the experimental results described in this paper, we now believe that the previous statement "As a mixture, both 517 and S16 bind to 16s RNA" is in- Binding of SIG-As shown in Table I, S16 by itself does not bind to 16 S RNA nor does it bind in the presence of S17 (Table  I and Fig. l~).~ Addition of S4 or S20 gives a small amount of stimulation, but addition of both S4 and S20 results in almost complete binding of SIG, i.e. approximately 1 mole of S16 bound per mole of 16 S RNA (Table I and Fig. Id). In our earlier work, in which a mixture of S16 and S17 was used as 'El," stimulation of the binding of the mixture of S16 and S17 (P9) by S4 (P4a) and S20 (1'14), but not by S8 (1'4b), was observed (cf. Fig. 1 of Ref. 1). Present experiments account for the previous experimental results on the basis of stimulation of S16 binding by S4 and S20 superimposed on the independent binding of S17.
Binding of SS-Our earlier work showed that binding of S5 depends on the presence of S8 and a mixture of S16 and S17 (P9) (1). Table III shows an example of earlier experiments in which a mixture of S16 and S17 was used, and binding of [3H]S5 was examined.
t3H]S5 was bound in the presence of S4, S8, S20, periments was calculated to be 4900 cpm per A260 eq. Based on this value, the amount of S5 bound in the presence of all 30 S proteins corresponds to 0.6 mole per mole of 16 S RNA. In Experiment 2, the amount of S5 bound in the presence of proteins S4, S8, S13, S20, S16, and S17 was somewhat less than in Experiment 1. This is probably due to variability associated with the somewhat weak binding of S5 obtained in the presence of these proteins and the gradual loss of bound S5 during centrifugation (see text for further explanation).
Omission of S8 or both S16 and S17 from the above protein mixture or even from t.he complete protein mixture abolished S5 binding almost completely (Table III).
We have now tested whether S16, S17, or both are required for binding of S5. As shown in Table IV, [r4C]CH3-S5 did not bind to 16 S RNA alone, nor in the presence of S4, S8, S13, S17, and S20 (see also Fig. 4a). Addition of S16 but not S17 to the complex containing S4, S8, S13, and S20 stimulates S5 binding very slightly, and addition of both S16 and S17 produced significant binding of S5 (Table IV and Fig.  45). It should be noted, however, that the extent of S5 binding in the presence of S4, S8, S13, S16, S17, and S20 was still only about 20 to 30% of the maximum binding that can be obtained when all 30 S proteins are present (Table IV and Fig. 4). Thus, there is some quantitative difference between the present and the earlier experiments.
Also, the sucrose gradient pattern in the present experiment suggested that a considerable amount of bound S5 was lost during centrifugation as evidenced by trailing of radioactive material from the ASO peak to the top of the gradient (Fig. 46). It is possible that the labeling of S5 by reductive alkylation weakens 55 binding.
Addition of all of the 30 S proteins appears to be sufficient to stabilize the binding.
Several proteins tested (S7, SQ, and S15) did not significantly improve S5 binding (Table IV). S12 appears to increase binding of S5 slightly and as shown later, S5 appears to assist S12 binding.
We also examined the effect on S5 binding of omitting single 30 S proteins, in the presence of all other proteins (Table IV,  Experiment 2). Omission of S17 or S15 had no effect while omission of S16 or S16 and S17 reduced binding to 60 and 40% of the maximum level, respectively.
Again, some quantitative difference in the results was observed between these and earlier experiments.
This difference may be related to the use of different protein preparations and slightly different reconstitution conditions in the two experiments (cf. legend to Table III).
Effect of S15 on the Binding of S6 and S18--In constructing the assembly map in our earlier work, S15 was not used (1). As mentioned above, we subsequently purified 515 and re-examined its binding, as well as its effect on the binding of other proteins.  with about 0.7 to 0.8 mole of S15 bound per mole of 16 S RNA (data not shown).
In addition, preliminary experiments also showed that a major part of the 16 S RNA-S15 complex can be incorporated directly into 30 S subunits upon mixing with a large excess of TP30, followed by subsequent incubation under the reconstitution conditions. Thus, the binding of S15 to 16 S RNA is specific.
Effects of S15 on the binding of other proteins were then esamined, wherever presence of such effects was suspected. In several cases which were examined, we did not find any effect (e.g. no effect on S5 binding; see Table IV).
We have found, however, a strong effect of S15 on the binding of S18 and S6.
In our earlier work, the binding of S6 could not be observed in the presence of all of the initial binding proteins known at that time (S4, S7, S8, S13, S17 (added together with Sl6), and S20). In addition to these initial binding proteins, the presence of both S9 and S18 was necessary to get relatively good binding of S6. Under such conditions, the dependence of binding of S6 on the presence of S18 was clearly demonstrated (1). In addition, the binding of S18 was also known to be dependent on the presence of S6 (1).
J17e have now found that S15 strongly stimulates the binding of both S6 and S18. As shown in Table V, in the presence of S15 and S18, [3H]CH3-labeled S6 bound to the 16 S RNA-protein complex, and the extent of the binding was about 50% of the maximum binding which could be obtained in the presence of all other proteins.
A similar situation was also observed with respect to the binding of S18. In this case, binding studies were done using unlabeled protein (see "Materials and i\Iethods"). As shown in Table VI, in the presence of S15 and S6, good binding of S18 was observed (about 70% of the maximum binding).
Again, in this system, the binding of S18 is completely dependent on the presence of both S15 and S6. Sll gave a weak additional stimulation, as in the case of S6 binding.
However, in the absence of S15 but in the presence of S4, S8, S13, S16, S17, and S20, Sll strongly stimulates binding of both S18 and S6 (about 5-fold stimulation; data not shown).
Binding of S12-Although S12 is an important protein coded Other proteins added s15, -, --, S6, -S15, S6, -S15, S6, Sll 2Si-SE3 Relative amount of S18 bound % 0 0 67 85 100 for by the str gene (16), and its functional role has been studied extensively (14,16),5 we failed to locate this protein on the assembly map in our earlier work. This was mainly due to difficulty in preparing pure radioactive S12. In the present study, we have used [3H]CH&12 prepared by reductive alkylation of purified S12. As mentioned under "Materials and Methods," the alkylated radioactive S12 retained full functional activity as judged by its ability to reconstitute active 30 S subunits.
Thus, we have now been able to study conditions necessary for binding of s12.
Our earlier work showed that S12 does not bind to 16 S RNA singly, nor in the presence of S4, S8, and S20 (1) (see also Fig. 5).
We have now found that further addition of S16 and S17 to the above mixture caused considerable binding (about 50% of the maximum binding). S20 did not appear to be necessary, but omission of any other component (S4, S8, S16, or S17) abolished S12 binding almost completely (Table VII, Experiment 1) (see also Fig. 5). Addition of S5 so the above protein mixture (S4, S8, S16, S17, S20) resulted in additional stimulation of S12 binding (to about 70% of the maximum binding). S15 might have a slight stimulatory effect. However, the other initial binding proteins, S7 and S13, did not show any significant stimulatory effect.
Omission of single proteins during reconstitution using all the purified proteins indicated that Sl7 has a major role in S12 binding (Table VII, Experiment 2). Omission of Sl6 or S5 had little or no effect, but omission of S16 together with S17 caused a further decrease in the residual binding of S12 observed when S17 alone was omitted.
Thus, we conclude that the major proteins required for S12 binding are S4, S8, S16, and S17. S5 stimulates but is not absolutely required for S12 binding.

u1scuss10u I
The revised assembly map is shown in Fig. 6. (For comparison, the assembly map published in the original paper (1) is also shown (inset). ) We have already discussed the main uncertainties connected with our earlier work.
The revised map retains all of the previous RNA-protein and proteili-protein relationships (shown with arrows), but several additional interactions are incorporated based on the experimental results described in the present paper. In addition, the relationship of S2 to S3 was added. It was previously found that S2 does not bind to 16 S RNA-containing complexes until most of the 30 S proteins are added to incubation mixtures (1). Single component omission experiments showed that omission of S3 as well as many [3H]CH&12 was bound to 16 S RNA under reconstitution conditions in the presence of: (a) S4, S7, S8, S13, S16, and 520; (a) S4, SS, S16, and S17; (c) ZSi-S12.
There arc two major discrepancies on the initial RSh-binding proteins between our conclusion and that obtained by these tvorkcrs. One is concerned with the binding of S17, and the other is concerned with the binding of S13.
Both Schaup et al. (3) and Garrett et al. (7) concluded that the binding of S17 to 16 S RNA observed in our earlier experiments is nonspecific.
In the present paper, wc have presented evidence that the S17 binding is specific, and the 16 S RNA-S17 complex can be converted to 30 S subunit,s without dissociation. The reason for the discrepancy bctwcen the experiments by Schaup et al. (6) and t,hose described here regarding the amount of S17 bound to 16 S RNA is not clear. In their first paper, Schaup et al. (6) observed no sat,uration plateau when they added up to 4 moles of radioactive S17 per mole of 16 S RNA.
At this input protein to RNA ratio, they observed 2 moles of 817 bound per mole of I6 S RNA.
The authors suggested that the nonspecific binding of S17 may be an artifact induced by reduction of the disulfide groups in the protein, implying that they had measured the binding of the reduced form of S17. In the subscquent paper, however, they reportctl that the reduced protein (a mixture of S16 and S17) was unable to bind to 16 S RNA under reconstitution conditions, and stated that the oxidized form, not the reduced form of S17, bouild noilspccifically in their original experiments (3). In our experiments, we have always used the rcduccd form of S17. However, regardless of the oxidation state of S17, Schaup et al. (6) did not observe a clear plateau in the binding of S17. In the data shown for biiltling of S17 by Schaup et al. (6), a Rio-Gel P-150 column was used to separate unbomld protein from 16 S RNA.
Since we have found that S17 sometimes aggregates in the absence of urea, it seems possible that some of the binding observed by these workers represented either binding of aggregated S17 to 16 S RSA or rlution of aggregated protein in the void volume.
It is more difficult to explain the discrepancy hctween our experiments and t,hose reported by Garrett et al. (7) concerning the binding of S17 to 23 S RNA.
We did not observe the binding of S17 to 23 S RNA reported by these workers.
The binding of S17 is completely specific for 16 S RSA\. It is possible that the diffcrcnce between our experimental results and their results could be accounted for by differeilces in the methods used for preparation and storage of purifietl proteins and RNA.
In this regard, it is worthwhile to note that our S17 preparation is pure and not contaminated in any apprrciable amounts by other proteins, including S16 or S15, as judged by NH,-terminal amino acid sequences of these protein preparat,ions, as well as by immunochemical methods.
(The two-dimensional electrophoresis tcchniquc (20) generally used to cheek purity of ribosomal protein preparations is not sensitive enough to determine contamination of S17 by S15 or S16 because thcsc three proteins migrate to a similar region, and, in addition, S17 usually gives two spots, presumably oxidized and reduced forms (9).) Furthermore, all of the proteins used in our experiments are highly active in the reconstitution of functionally active 30 S ribosomal subunits (9). The second discrepancy is related to S13. In our earlier work, we observed weak binding of S13 to 16 S RKA alone. Roth Schaup et al. (3) and Garrett et al. (7) did not observe such binding and speculated that the binding of this protein reported by Mizushima and Nomura (1) was caused by contamination of S13 by S15. The position of elution of S13 in our phosphocellulose chromatography used as a first step of protein purification is entirely different from that of S15 (9). No cross-contamination was observed when the preparations were examined by twodimensional polyacrylamidc electrophorcsis.
In addition, both Zimmermami et nl. (19) and Craven and his co-workers obscrvcd direct binding of S13 urldcr the same conditions. hloreovcr, the latter workers found that Sl3 binds only to 16 S RNA and not to 23 S RNA.G It is also our experience, however, that the degree of binding of S13 to 16 S RNA varies considerably.
As reported earlier (l), S13 binding becomes almost quantitative in the prcscnce of S4, S8, and S20. It is possible t,hat the binding of S13 to 16 8 RNA is weak, and some slight conformational changes which take place during purification or storage of the protein or RNA are sufficient to abolish binding to 16 S RNA alone. Such inactivation may be rcvcrscd, however, by direct or indirect interaction with other proteins such as S4, S8, and S20, resulting in strong binding to 16 S RNA in the presence of these proteins. Although wc have not done further studies on the specificity of t,he weak binding of S13 to 16 S RNA, wc have retained our origina! information in the rcviscd assembly map.
As already emphasized in previous papers (1, al), the arrows connecting proteins to proteins in Fig. 6 represent only the interaction (direct or indirect) that WC dctccted during the experimcnts that were conducted to construct the map. Many other interactions which stabilize ribosomal structure probably occur. In addition, it is conceivable that, proteins other than the initial binding proteins also interact with 16 S RNA in the finished ribosomc structure.
It should also be emphasized that coopcrative interaction between two proteins shown in the assembly map does not necessarily mean direct interaction of the two proteins. For example, protein A may help the binding of protein R by directly interacting with I< resulting in stabilization of the complcx; alternatively, protein A may interact with either 16 S RNA or other proteins and may crcatc a correct, binding site for R indirectly through some conformational alteration. It is probable that the "sequence" described in the assembly map corresponds, at least approsimatcly, to the temporal sequence of assembly.
Other cvidcncc for this assembly sequence comes from the early kiuetic studies of the assembly of 30 S subunits which showed that there is a rate-limiting unimolecular reaction that requires a high activation energy (22). At lower temperatures, a subset of the 30 S ribosomal proteins (RI proteins) interacts with 16  The temperature-dependent conversion of RI to RI* appears to involve a conformatiorial rcarrangemcnt of the RI st,ructure (23). In Fig. 6, those proteins found to be required for this temperature-dcpcndent conversion of RI to RI* as well as three proteins (S20, S13, and S6) not required for the conversion but found in isolated RI particles (23), are shown above the dotted line. Thus, the incorporation of other proteins (under the dotted line) into the reconstituted 30 S particles appears to dcpend on the prior binding of RI proteins and the conformational change discussed above. All of the RI proteins thus identified occupy an earlier part of the assembly map which was constructed using quite different methods.
It is evident from the complcsity of the map, however, as well as the way in lvhich the map was const,ructed, that the precise temporal order during the in vitro assembly would be difficult t'o determine.
In fact, it is probable that there are several differ-ent routes for assembly of 30 S ribosomal subunits, and that differences in free energy of activation among them may be rather small. Some evidence supporting this conclusion has been discussed previously (21,23).
We originally suggested that the assembly map reflects topological relationships between ribosomal proteins in the organized ribosome structure (1). Several investigators isolated ribosome fragments by mild RNase digestion and found that those proteins that cluster together in such subparticles are, in general, the proteins which show inter-relationships in the assembly map. Thk correlation found between the fragment studies and the assembly map has been discussed in detail by these workers in connection with possible three-dimensional ribosomal structures (24-27).
Other investigators have used bifunctional crosslinking reagents to find "neighborhood relationships" among ribosomal proteins in the ribosomes.
So far, the following protein dimers, or trimers, have been identified: S7-S9 (28,29), S&S5 (30, 31),' S13-S19 (29),7 Sl%S21 (30, 32, 33), SWSll-S21 (34), and S2-S3.7 Most of the neighborhood relationships thus identified can be seen in the assembly map except the S13-S19 and S18-S21 pairs. However, it is obvious that the assembly map would not be expected to predict all neighborhood relationships. Nevertheless, positive agreement between many of the cross-linked pairs and the pairs connected in the assembly map support the topological interpretation of the interrelationships between 30 S ribosomal proteins shown in the assembly map. It is hoped that the present assembly map will be useful in designing further experiments to study structure-function relationships as well as the mechanism of assembly of the 30 S ribosomal subunits.