Structure-function relationship in Escherichia coli initiation factors. Identification of a lysine residue in the ribosomal binding site of initiation factor by site-specific chemical modification with pyridoxal phosphate.

Incubation of Escherichia coli initiation factor 3 (IF3) with pyridoxal phosphate (PLP) followed by reduction with sodium borohydride resulted in the selective modification and inactivation of this protein. The ribosomal-binding site (RNA-binding site) of IF3 is the target of PLP modification, since (a) the phosphate residue of PLP is required for inactivation; (b) RNA as well as synthetic polynucleotides (especially guanine-containing one) protect IF3 from inactivation; and (c) 30 S, but not 50 S ribosomal subunits, protect IF3 from PLP modification and from inactivation. The incorporation of PLP into IF3 occurred exclusively at lysine residues by reduction of the Schiff bases yielding epsilon-(5'-phosphopyridoxyl)lysine. The PLP-modified lysines were identified by amino acid analysis and sequencing of the PLP-modified peptides. Out of the 20 lysines of the factor, only Lys 2, Lys 5, Lys 99, Lys 112, Lys 166, and an unidentified Lys of the central cluster of the molecule (Lys 86, 87, 90, 91, 96) were found to be modified to varying degrees. The incorporation of 3 to 4 mol of PLP/mol of IF3 is accompanied by a substantial (greater than or equal to 80%) inactivation of this protein; the loss of activity follows apparent first order kinetics, and the inactivation results from the modification of just 1 Lys residue. This essential Lys residue was identified by various criteria to be Lys 112. The identification of an "active region" in the IF3 molecule is emerging from this as well as from other chemical modification studies.

as well as of initiation and elongation factors.2 Among the initiation factors, IF3 interacts with nucleic acids (11)(12)(13)(14)(15)(16) and binds to the 30 S ribosomal subunits via the 16 S rRNA (12,16). In the present paper, we present data on the identification of an active site of IF3 responsible for the interaction of the molecule with the ribosome.

EXPERIMENTAL PROCEDURES
General Preparations-E. coli MRE600 30 S and 50 S ribosomal subunits (17), 16 S and 23 S rRNAs (18), and initiation factors (19) were prepared as described previously. In vitro labeling of initiation factors by reductive methylation with [14C]formaldehyde was performed as described (20). Assay ofIF3 Activity-Unless otherwise specified, IF3 activity was determined by the 30 S poly(U) . N-acetylphenylalanyl-tRNA ternary complex equilibrium perturbation method as described previously (21). Binding of radioactive IF3 to ribosomes was studied by sucrose density gradient centrifugation as described elsewhere (16).
Reaction with Pyridoxal Phosphate-IF3 was incubated with PLP (Serva, Heidelberg) either in 20 mm triethanolamine-HCl, pH 7.8, and 30 mM KCl (TEA buffer) or in 12.5 mM triethanolamine-HCl, pH 7.8, 20 mM KCl, 6 mM Mg acetate, and 4 mM f3-mercaptoethanol (TEA-Mg buffer) as specified in each figure legend. This reaction was stopped by the addition of either unlabeled NaBH4 or NaB[3H]4 (Amersham Buchler) as specified in each figure legend. The reaction mixture was left for at least 10 min at 0 'C to ensure complete reduction by NaBH4 before further processing. To determine the number of PLP molecules incorporated into IF3, bovine serum albumin (1 mg/ml) was added as a carrier to the reaction mixtures containing known amounts of the factor. Acetic acid (0.1 N) was added to decompose excess NaB[3H]4, and after 30 min at 0 'C, the mixture was neutralized by the addition of 0.1 N NaOH. Aliquots of the mixture were then placed on Whatman 3MM filter paper discs and the hot TCA-insoluble radioactivity was determined (22). The number of PLP residues incorporated by each IF3 molecule was calculated from the specific activity of the radioactive borohydride used, assuming 1 tritium atom/PLP incorporated. The stoichiometry of PLP incorporation determined in this way was found to be in good agreement with that determined spectrophotometrically (3).
Protein Chemical Analyses-Protein hydrolysis with Staphylococcus aureus protease (Miles), two-dimensional peptide fingerprinting, amino acid analysis, and manual amino acid sequencing were carried out as described previously (23,24). Active Site of IF3 xolume (CHCUUH and extensive dialvsis against 2% CHtOUH. The product showed a major absorption band at 325 nm, characteristic for the reduced Schiff base (29). After lxophilization, this was dissolved in a small amount of 0.1 N HCI and dried over NaOH. Hvdrolvsis was performed in 5.7 N HCI for 20 h at 110 'C in tacCu. The hvdrolvsate was dried over NaOH, dissolved in H20 and applied as a streak onto a Cel 300 thin layer plate (Macherey-Nagel, Duren).
Electrophoresis was run for 2.5 h at 400 V in the same buffer used for fingerprinting. A single, fluorescent, ninhvdrin-positive band, in addition to the band corresponding to unreacted lxsine. was detected.
The fluorescent band was scraped off and extracted twice with 50%' CH3COOH. The extract was Ivophilized and dissolved in H0. The preparation thus obtained gave onlv one ninhydrin-positive and fluorescent spot after two-dimensional electrophoresis and chromatography.

RESULTS
Mechanism and Specificity of the Reaction The scheme of the reaction of a protein with PLP is shown in Fig. 1. The E-NH-, group of lvsines forms a Schiff base with the aldehyde group of pyridoxal phosphate; the Schiff base is then reduced bv NaBH4 to form E-(5'-phosphopyridoxvl)lysine. This compound can be identified chromatographicallv from its fluorescence or from its radioactivity if tritiated NaBH4 is used to reduce the Schiff base. In this study, we have primarily used the latter method to identify and quarntitate the reaction products of IF3 with PLP.
When E. coli initiation factor IF3 is reacted with PLP, the only product found following acid hvdrolysis of the protein, electrophoresis, chromatography, and autoradiographv is N'pyridoxvl lysine (Fig. 2). When pyridoxal phosphate is substituted by other pyridoxal derivatives, negligible amounts of radioactivity are incorporated by IF3 (not shown), and, unlike with PLP, little or no inactivation of the factor takes place (Table I) tion mixture was extenisively dialyzed againls;t 2( acetic acid and lvophilized. The sample (a< x 10' cpm) was hydrolvzed with 5 7 N HCI for 20 h at 110 -C itn raIeo, dried, and dissoixed in a small amiount of H-0. The hydrolvsate was applied onto a cellulose thin layer sheet (Cel 300) together with a N'^p yridoxxl-lvsxine standard prepared as described under "Experimental Procedures." Electrophoresis was for 2.5 h at 400 V. followed by ascending chromatography in 1-butanol pvridine:acetic acid:water (S:o1 4). The position of the stanldard N`pvridox ll--sine was identified by its fluoresceiice. The plate was then subjected to fluorographv as described (25.). Complete overlap betweeni pyridoxvl lysine fluorescence and radioactixitv was obtained.
TABLF I Effect ofiariouspwidoral derzciotiies on IF) (ctiit V Each sample containing 4.6 jig of IF3 was incubated in TEA buffer (see under "Experimental Procedures") with 2.5 mM of each compound for 10 min at 3 'C (total volume, 30 gl). treated with NaBH1, and assayed for activitv as described under 'Experimental Procedures." Conmpounds 1 and 6 were purchased from Serva. comipounds 2 aind 5 were obtained from Sigma; 4 was from Merck; and 3 was prepared from 1 by reductioni with NaBH1.  1. Scheme of reaction of pyridoxal phosphate with lysine residues in proteins. In the absence of reduction with NaBH4, the formation of the Schiff base is reversible (1). If NaB[H]4 is used for reduction, the resulting E-(5'-phosphopyridoxyl)lvsines will be radioactive. Alternatively, identification of the reaction product can be based on fluorescence properties of pyridoxxl lvsine (3). Sinice trypsin fails to hydrolyze next to pyridoxvl lvsine (2). in the present paper peptide mapping for the identification of the modified peptides was performed following enzvmatic hydrolysis of IF3 with S. aureus protease. Identification of the Residuies Modified by PLP-To determine which Lys residues of IF3 are modified bv the reaction, purified factor was incubated with pyridoxal phosphate, reduced with tritiated NaBH4, dialyzed, hvdrolyzed with S.
aureus protease, and subjected to two-dimensional peptide mapping. A fluorography of such a fingerprint showing four radioactive peptides is presented in Fig. 3A. The identification of these peptides was aided by the use of colored reactions to locate on the plate peptides containing specific amino acids (i.e. Arg, Tyr, His). The unequivocal identification of the leptides, however, was obtained, following their elution from the thin layer plates, bv comparison of their amino acid c(omposition and sequence with the known IF3 sequenice (30).
These results are summarized in Table IL.
The identified peptides (81)1, SP,1, SP7, SP16) correspond to a total of 5 lysine residues (Lys 2, Lys 5, Lys 99, Lvs 112, and Lys 166) out of the 20 lysiies present in the IF3 molecule (30). Of the above lysines, Lvs 166 is modified to a lesser extent, which also varied somewhat from experiment to experiment. The picture shown in Fig. 3A can be regarded as a typical example of this kind of analvsis, and the above-mentioned peptides are always among the main, and often the only targets of PlP miodification. However, depending upon the particular batch of IF3 used, and perhaps upon slight, uncontrollable variations in the reaction conditions, another peptide was sometimes found modified. This peptide was identified to be SP9, which contains 5 lvsines (Lyvs 86, 87, 90, 91, 96) of the central cluster of the IF3 molecule. Although no attempt was made to identify which of the above lvsines are modified, it can be deduced from the stoichiometry of modification that, on average, no more than 1 lvsine is miodified per SP9 peptide. A fluorogram of a two-dimensional fingerprint in which SP9 was modified is shown in Fig. 3B. Inactivation of TFhY3 b PLP Reaction As mentioned above, the PLP modification of IF3 results in the loss of its biological activitv. In the following experiment (Fig. 4A), the rate of inactivation of 1F3 was measured at various PLP concentrations. It can be seen fromi the figure that the loss of IF3 activity proceeds more or less rapidlyv, depending upon the PIP concentration, and that the inactivation follows apparent first order kinetics at any given concentration of PLP, at least during the initial period of incubation. In Fig.   4B we present a log-log plot of the half-times of the inactivation (which are related to the apparent first order rate constants) tver7suis the respective PLP concentrations. This plot yields a straight line with a slope of 0.9, thus indicating that the reaction is first order with respect to the PIP concentra- Fluorogram of a staphylococcal protease peptide map of PLP-modified IF3. In A, IF3 (350 gg) was incuibated in 1 ml of TEA buffer with 0.6 mMi PLI, for 10 min at 37 'C and reduced at 0 'C by 5.5 mM NaB[ H]. The reactionn mixture was exhaustiveix dialvzed against 2`acetic acid, lyophilized. dissolved in o0 mtNx CH COONHI. pH 4, containing 0.1 ImM dithiothreitol, aiid digested with staphxlococcal protease as described uncler 'Experimental Procedures.-The digest was lxophilizeti, dissolved in a small xolume of HW0, and subjected to two dimensional peptide mapping as described under 'Experimental Procedures." Fluorography of the flngerprint was performedt as described (25). B, fluorogram of a hvdolx sate of a different preparation of IF3 modified by PLP as described for A. In this case, SP9 was also among the modified peptides as described in the text. Identification of the peptides nmodified bv PLP To identify the PLP-modified peptides. the radioactive peptides as well as some nonradioactive peptides were scraped from the thin laver plates, extracted with acetic acid, and subjected to amino acid analysis and manual amino acid sequenicing as described under "-Experinmental Procedures." TFhe table lists onlv the results concerninig the radioactive peptides. (1); Ile, 1.6 (1); Lvs, 1.1 (2); SP16 was contaminated with S1P7, so that soome amino acids of the latter (including the NH, terminus) were detected in SIP16.
According to Chang (34), when the NH-terminal amino acid is blocked (in this case, N -methxlmethioniine), the 2nd residue (Lvs) appears as the 1st due to the splitting off of the blocked amino acid before the first cycle. Fit;. 4. PLP concentration dependence of the inactivation of IF3. In A, IF3 was incubated at 10 'C in TEA buffer with the concentration of PLP indicated above each line. At various times, aliquots containing 3.5 pg of IF3 were taken, mixed with 1/15 volunme of 100 mM NaBH4, and chilled at 0 'C. These aliquots were assayed for their biological activitv as described unider -Experimental Procedures.." B. determination of the reactioni order. The rate constants obtained as described for A were plotted tersus the PLP concentration in a log-log plot. The reaction order was obtained from this plot as the slope, following the procedure of Levy et al. (35).
tion and that the inactivation of the factor is due to the modification of at least 1 essential lysine residue. In the following experiment (Fig. 5), the capacity of PLPmodified IF3 to bind to 30 S ribosomal subunits was studied.
It can be seen from the figure that, following the chemical modification, IF3 becomes unable to bind to the 30 S subunits.
Thus, it is likely that the loss of the biological activity of IF3 stemis from its inability to bind to 3(0 S ribosomal subunits. Protection Experiments To find out which of the modified lysine residues is responsible for the inactivation, we performed a number of "substrate" protection experiments which are presented below. The first of these experiments (Fig. 6A) shows that, in the presence of the random polynucleotide poly(AUG), IF3 is strongly protected from PLP reaction. Thus, when IF3 is incubated with 1 mm PLP in the presence of poly(AUG), a negligible amount of PLP is incorporated per IF3 molecule and only a small fraction of the biological activity is lost. In the control experiment, in which IF3 was modified under the same conditions but in the absence of poly(AUG), approximately 3 molecules of PLP were incorporated per IF3 molecule with roughly 80% loss of the biological activity (Fig. 6B).
In addition to reacting with IF3, PLP could theoretically react with poly(AUG). Indeed, some evidence for PLP reaction with RNA has been reported (31). In light of this fact, an experiment was performed to check whether the protection of IF3 by poly(AUG) could stem from the reaction of PLP with the polynucleotide (Fig. 7). In this experiment, IF3 was incubated with PLP and poly(AUG) and reduced with NaB[3H]4 under conditions in which partial protection is obtained. The sample was then divided into two aliquots, one of which was treated with RNases A and Ti. Both samples were then subjected to gel filtration on Sephadex G-25. As seen in the figure, in the sample not treated with RNases, both absorbance at 260 nm and radioactivity appear together in the void volume of the Sephadex column. After enzyme digestion, however, the radioactivity and the absorbance peaks became separated; the radioactive material was still eluted in the void volume of the column, while the A260 absorbing material was retarded. The lack of radioactivity in the latter fractions clearly indicates that the PLP reaction occurred only with IF3 and not with poly(AUG) under experimental conditions where IF3 protection is observed.
To find out whether the protection of IF3 depends upon the type of nucleic acid used, the extent of protection was measured in the presence of various concentrations of different nucleic acids (both synthetic and natural). Fig. 8 summarizes these data. As seen from the figure, there is a substantial protection of IF3 activity in the presence of all nucleic acids used. However, some nucleic acids are more effective than others in protecting IF3. In particular, all guanosine-containing polynucleotides appear to be more effective than comparable amounts (by weight) of polynucleotides containing no guanosine. Among the natural nucleic acids, 16 S rRNA seems to protect IF3 slightly better than does double-stranded DNA. It should be noted, however, that no differences in the capacity to protect IF3 from PLP inactivation were seen between 16 S and 23 S rRNA (see under "Discussion"). It is also noteworthy that, unlike poly(AUG), even high concentrations (0.34 mM) of the initiation triplet, ApUpG, were without effect on the PLP inactivation of IF3; GMP was also without effect (Table  III). These results indicate that the protected site is not merely a phosphate or a nucleotide-binding site, but rather a nucleic acid-binding site.
The lack of specificity between 16 S and 23 S rRNA in protecting IF3 against PLP modification and inactivation is in contrast to the specific protection provided by 30 S ribosomal subunits but not by 50 S subunits. The following experiment presents a comparison of the inactivation of IF3 activity in the presence of either 30 S or 50 S ribosomal subunits as a function of PLP concentration (Fig. 9A) and as a function of the concentration of either ribosomal subunit (Fig. 9B). It can be seen from these figures that 30 S ribosomal subunits are able to provide a substantial protection of IF3 against inacti- vation. Thus, when equimolar amounts of 30 S ribosomal subunits and IF3 are present in the PLP reaction mixture (indicated by the arrow in Fig. 9B), over 60% protection of IF3 activity is obtained, and virtually complete protection is obtained when a 3-fold molar excess of 30 S over IF3 is used. That the protection depends upon the binding of IF3 to 30 S ribosomal subunits can be deduced from the fact that comparable amounts of 50 S ribosomal subunits produce only a minor protection of the IF3 activity and that the capacity of the 30 S subunits to protect is lost in the presence of 0.5 M NaCl, which prevents the ribosomal binding of IF3 (16).
Identification of the Protected Peptides-A comparison of the peptides modified by PLP in the presence or absence of IF3 binding to 30 S ribosomal subunits is presented in Fig. 10 as fluorograms of two staphylococcal protease fimgerprints. Both samples, containing the same amount of IF3, were modified with PLP in the presence of 30 S ribosomal subunits, but only the IF3 of Fig. lOB was actually bound to the ribosomes, while the reaction mixtures of Fig. 1OA contained 0.5 M NaCl, which prevented most of the interaction between the factor and the ribosomes. It is obvious from the figure that when IF3 is ribosome-bound, a much-reduced amount of PLP   ribosomal subunits (0), 50 S ribosomal subunits (0), or without ribosomes (A); PLP was added to the indicated final concentrations and incubation was then continued for 10 min at 37 'C, followed by reduction with NaBH4 (6.25 mM) at 0 'C. The final volume of each reaction mixture was 32 gl. To these, 96 pl of 8 M urea and 4 M LiCI were added. After 15 min incubation at 50 'C, 50 gl of each sample were assayed for IF3 activity. In B, the PLP reaction was essentially identical with that described for A, with the exceptions that the amounts of 30 S (0) or 50 S (0) in the preincubation were varied as indicated, and the PLP concentration in the reaction was 2.5 mm. In addition to the 30 S ribosomal subunit, one sample also contained 0.5 M NaCl (A). After reduction with NaBH4 (see above), the samples were treated with urea/LiCl, incubated, and tested for activity (see above). One hundred per cent activity was calculated by comparison with samples in which comparable amounts of an untreated IF3 were tested in the presence of the equivalent amount of ribosomal subunit.  Fit.. 10. Identification of SP peptides of IF3 protected bv 30 S ribosomal subunits from PLP modifi'cation. In A, IF3 (75 jitg) wras preincubated for 5 min at :37 C with 73 A,,;,) units of ribosonmal subunit in a TEA-Mg buffer containing 0.5 M NaCI, then treated for 10 mnim at 37 'C with ILl (1.25 mm) followed bv reduction with NaB[ Hi (1.25 mM). The total volume of the reaction mixttire was 400 gl. In B, all conditions were identical with those described above.
with the only exceptions being the absence of NaCI and a 2-fold scaling up of the entire reaction mixture. After reduction, the excess NaB[ H], was consumed bv adding I'LlP to a final concentration of 2.5 mm, anid the samples were loaded onto sucrose gradients (i) to 30'', w/v) in TEA-Mg buffer and spun for 3 h at 48000 rpm at 4 'C in an SW 60 Ti rotor. The combined tol)p fractions of the gradients of sample A were lvophilized after extensive dialxsis against 2% acetic acid. The 30 S peak fractions fromi the gradients of sample B were collected bv centrifugation after precipitation by 0.7 volumes of ethanol and suspended in 1 ml of TEA-NIg buffer containing 0.4 M NH1CI to dissociate IFS3 from 30 S. These samples were theni loaded onto a second gradient (10 to 30`(w/v) in a TEA-Mg buffer containing 0.4 M NHFCI) and centrifuged as above. The top fractions from these gradients containing IF3 dissociated from 30 S were combined and Ixophilized after extenrsive dialysis against 2' acetic acid-The IF3 sample thus obtainied was further pulrified from some contamninating ribsoomal proteins bv gel filtration on a Sephadex G-150 superfine column equilibrated with 2%< acetic acid containiing (6 M urea. Purified IF3 was finallv dialyzed against 2%(r acetic acid and lvophilized. Staphvloco c.us protease peptide fingerprinting and fluorographv of these saniples were performed as described under "Experimenital Procedures.is incorporated. In addition, not all peptides are equally protected from PLP modification; thus, while SP1 + SP9, SPY, and SP 16 were strongly protected, the modification of SP1 1 was little affected bv the binding of the factor to the ribosomes. To quantitate the protection of the individual peptides by, 30 S ribosomal subunits, the radioactivitx associated with the various peptides was determined. The results of this experiment are presented in Table IV. An experiment similar to that described in the legend to Fig. 10 and Table IV was also carried out to quantitate the protection of individual SP peptides in the presence of poly(AUG). In this experiment, IF3 activity was protected about 50'% (from 85 to 35%D inactivation) when the modification was carried out in the presence of the polynucleotide. In this case, however, all peptides were found to be substantially protected. The highest protection was found with SP7 (64% ) and SPll (68%), although the latter was not affected by the presence of 30 S ribosomal subunits. SP1 and SP9 were also protected by polv(AUG) (38 and 36%, respectivelv), although proportionately less than by 30 S ribosomal subunits. SPI6 was not heavily modified, even in the absence of polv(AUG), but its protection by the polynucleotide was 47%;. Taken together, one can conclude that the modification of SPlI does not affect the binding site of IF3, since the modification of this peptide, although prevented by the presence of poly(AUG), is hardly affected by the 30 S ribosomal subunits, and IF3 molecules with strongly modified SPII still retain substantial activitv. Furthermore, although we cannot exclude the possibilitv that the modification of peptides SP9 and SP16 leads to the inactivation of the factor, it seems unlikelv that their modification is the major cause of the inactivation. In fact, although these peptides are protected by both 30 S and polv(AUG), SPIl is always found to be modified to a much lower and variable degree compared to the other peptides without any obvious correlatin with the loss of activity, and in many, if not most of the cases, inactivation of IF3 occurs without any modification of SJ9.
T1hus, one can deduce from the protection experiments that the modification of SP7 or SLPl is the most likelv reason for IF3 inactivation; both peptides are always heavily modified in free 1F3 and stronglv protected by pol (AUJG) as well as by 30 S subunits to an extent roughly proportional to the activitv protected.
TABI i IB QiiointCttit'e determ inaition of the extent of m77odifi(,ation of in ditrddcual SP pe)ticledw of IF3 bound an (1 uinbouiind to 30 S ribosomal subunit.s The radioactive spots were cut out and the radioactivitv was determined in a scintillation counter. The radioactivitN remiaining at or near the origin represented 46 and :33'( of the total radioactivitv applied for the unbound and bound IF'3 respecti elx. FPie 11 Comparison of the rate of the inactivation of "long" and "short" IF3 by PLP reaction. A short formr of I1F3 missing the first 6 amino acids was prepared and purified as will be described elsewhere.2 A, SDSpolvacrrlamide (15%) gel electrophoresis of (1) native, (2) mixture of native and short, and (3) purified short IF3. In B, 10 pg of either native (@) or short (0) IF3 were incubated at 10 QC with 2.) mm I'LP in a final volume of 0.15 ml of the TEA buffer. At the indicated times, the samples were reduced with 6.25 mM NaBH, and the actixitv of each sample was determinied as described under 'Experimental Procedures Modification of SP7 Is Responsible for IF3 Inactivation The PLP protection experiments indicated peptides SPi (Lys 2 and 5) or SP7 (Lys 112) as the most likely candidates to play a role in IF3 inactivation. However, a "short" form of IF3 is known in which the first 6 amino acid residues from the NH2 terminus are missing (32,33). Since this shorter form of IF3 is somewhat active, it seems unlikely that Lys 2 and Lys 5 play an essential role in IF3 function, although their modification with the negatively charged PLP could indirectly lead to the inactivation of the factor. Thus, the interpretation of the PLP modification experiments, not only in terms of the location of the active site but, more importantly, in terms of the mechanism leading to the inactivation, would greatly differ depending upon which peptide (SP1 or SP7) was responsible for the inactivation. To solve this problem, a short form of IF3 lacking the first 6 amino acids was produced by mild digestion with proteolytic enzyme according to a procedure which will be described elsewhere. This short form was then purified (see Fig. 1LA) and its PLP-dependent inactivation was determined and compared to that of the native (long) IF3. It can be seen from Fig. liB that the short form of IF3 not only is inactivated by the reaction with PLP, but also that the rate of PLP inactivation of the short IF3 is nearly identical with the rate of inactivation of the native factor.
Thus, even in the absence of Lys 2 and 5, IF3 can be inactivated by PLP reaction. Since fingerprint analysis of the PLP-modified short form of IF3 revealed a pattern of radioactive PLP peptides essentially similar to that of native IF3 (except for the obvious absence of SP1 in the short IF3), we can conclude from the experiment of Fig. 11 that the modification of SP7 (Lys 112) is the major cause of the inactivation of IF3, although we cannot rule out that also the modification of SPI, when present in the intact IF3 molecule, may contribute to the inactivation. DISCUSSION In this paper, we have investigated the reactivity of initiation factor IF3 toward the site-specific lysine reagent pyridoxal phosphate. From the primary structure, it is known that this protein contains 20 lysine residues (30). Among these residues, Lys 2, 5, 99, 112, and 166 are found to be modified by PLP under conditions leading to the nearly complete loss of biological activity. The position of these lysines in the primary sequence of IF3 is schematically presented in Fig. 12, which also shows the position of all other Lys residues. Of the reactive lysines, Lys 166 is normally found to be modified to a lesser and more variable extent than the others. In addition, depending upon the reaction conditions and, more so, upon the IF3 preparation used, another unidentifled lysine belonging to the central cluster of the molecule may become modified.
As a result of the PLP reaction, the IF3 activity is lost. The inactivation reaction is first order with respect to PLP. That 1 or more of the lysines modified by PLP are part of the IF3 active site involved in ribosomal binding can be deduced from various lines of evidence presented under "Results". It was thus shown that the binding of IF3 to the 30 S ribosomal subunits is impaired by PLP reaction, and that IF3 is protected from PLP modification and inactivation in the presence of various nucleic acids, both synthetic and natural, as well as by binding to 30 S ribosomal subunits. On the other hand, no protection was observed in the presence of 50 S ribosomal subunits with which IF3 interacts weakly and nonspecifically, or in the presence of an AUG triplet or GMP. The lack of protection by 50 S ribosomal subunits in contrast to the protection obtained with 23 S rRNA and the equal protection obtained with 16 S and 23 S rRNAs are not surprising, since it had been shown that IF3 can bind almost equally well to both 16 S and 23 S rRNAs and that substantial binding to 50 S ribosomal subunits and spurious binding to 30 S ribosomal subunits can be obtained after stripping some proteins from these particles (12).
As seen under "Results," the phosphate group of PLP is necessary for IF3 inactivation, but the lack of protection by either AUG triplet or GMP, while providing evidence that the protection obtained with nucleic acids is not merely due to the presence of phosphate and/or guanines, suggests that the site protected is a rather complex nucleic acid-binding site. It should be noted, in this connection, that recent data could be interpreted to indicate in average a stretch of 14 nucleotides bound/IF3 molecule (36). In addition, the definite preference of IF3 for guanine-containing nucleotides seen in the polynucleotide protection experiments is consistent with the finding that the binding site of IF3 on the 30 S ribosomal subunit shows some properties of single-stranded, guanosine-containing RNA in being affected by either RNase Ti digestion or kethoxal modification (12,16).
By comparison of the extent of IF3 inactivation and protection with the extent to which the individual lysine residues are modified by PLP and protected in the presence of 30 S ribosomal subunits or poly(AUG), it was concluded that modification of Lys 112 in peptide SP7 is the major cause of IF3 inactivation and that this residue is localized in the ribosomalbinding site of the factor. Concerning the other lysines, it seems unlikely that a role in binding to 30 S ribosomes is played by Lys 99, since this residue is hardly protected by the 30 S ribosomal subunits and its extensive modification does not interfere with IF3 activity. On the other hand, the participation of Lys 2 and 5 in the ribosomal-binding site of IF3 could be compatible with the extent to which these residues are modified by PLP and protected by 30 S and nucleic acids. However, if these 2 lysines play any role in the ribosomal binding of IF3, this must be a marginal one, since a short form of IF3 missing these two residues can still bind, albeit with lesser affinity, to the 30 S ribosomes (32,33). In addition, the fact that the rate of PLP inactivation is nearly identical for the short and the native form of IF3 strongly suggests that the modification of Lys 2 and 5 is not the major cause of inactivation.
Finally, Lys 166 and one of the lysines in the central cluster of the molecule are also modified by PLP-the former, always, but to a variable extent; the latter, only with some IF3 preparations. Due to this irregularity in their behavior, their modifications cannot be regarded as the major cause of the observed IF3 inactivation. This does not exclude, however, H13C-NH> l i their presence in or near the active site of IF3. In particular, the region of the central cluster of lysines, for which an ahelical secondary structure is predicted with 4 lysines oriented on one side of the molecule in a manner suggesting a possible ionic interaction with double-stranded RNA, seems to be a good candidate for an additional site of interaction between the factor and the ribosome.
The presence of Lys 112 in the ribosomal-binding site of IF3 is further supported by the fact that 2 residues away from it, there is a tyrosine residue (Tyr 109) which displays essentially similar properties. Thus, a previous study showed that iodination of Tyr 109 inhibits the binding of IF3 to 30 S ribosomal subunits and that rRNAs and 30 S but not 50 S ribosomal subunits protect this tyrosine from enzymatic iodination (37). Also suggestive is the fact that the rate of chemical modification of Tyr 109 is affected by the presence of IF1 which, in turn, affects the ribosomal binding of IF3. Further evidence for the involvement of at least 1 tyrosine in IF3 binding comes from spectrofluorimetric studies.3 In addition to Tyr 109, other residues in the neighborhood of Lys 112 are likely to be important components of the IF3 active site. Among these, Arg 114 probably constitutes the primary recognition site for the phosphate of PLP (PLP is thought to bridge 2 amino acid residues by interacting via phosphate with one and forming a Schiff base with the amino group of the other). In addition, other possible important residues are Asp 105 and Glu 106 for their potential capacity to recognize guanines in singlestranded regions (38). Finally, since model oligopeptides of the type Lys-Tyr-Lys and Lys-X-Tyr-X-Lys were shown to bind preferentially to single-stranded nucleic acids (39), one can see how the region around peptide SP7 of the IF3 molecule has the potential information to select single-stranded RNA and to recognize guanine residues, two of the established properties of the ribosomal-binding site of IF3.