Protein Anatomy: Functional Roles of Barnase Module*

Globular proteins are composed of several modules that are contiguous polypeptide segments of compact conformation. Module boundaries are closely corre- lated with the intron positions of genes that encode proteins. The modules may thus have a one-to-one correspondence with exons in primordial genes. They may also be vestiges of polypeptide segments that ini- tially appeared as primordial proteins in prebiological evolution. Clarification as to whether modules discon- nected from one another have functional potentiality may validate these possibilities. Thus, in this study, each module of a protein was synthesized and assessed for functional potentiality. For this purpose, barnase, a bacterial ribonuclease, was decomposed into six modules (Ml-M6), which were examined to determine whether they have an affinity for RNA and RNase activity. M2, M3, and M6, all of which form a shallow but wide cavity for RNA binding in native barnase, were found to bind to RNA and to possess RNase activ- ity. However, M1 and M5, which support the other modules from the back side, and M4 did not bind to RNA and had no RNase activity. Protein modules with catalytic functions are described in this paper for the first time. That some modules of barnase possess catalytic activity indicates that protein modules may possibly have functioned as primitive catalysts in prebiol- ogical evolution.

Globular conformation is required for proteins to perform specific catalytic functions. A "module" is a compact structural unit in a globular protein or domain (1,2). Module boundaries are closely correlated with the intron positions of genes of globins (3), lysozyme (l), triose-phosphate isomerase (2,4), and other enzymes. An intron was actually shown to be present in a leghemoglobin gene based on examination of identified modules of hemoglobin a-and @-chains (3,5 ) . The correspondence of modules with exons is evidence for the concept "protein in pieces" based on the finding of 'Lgene in pieces" (6).
Polypeptides with M, values of 1000-4000 synthesized from a mixture of amino acid amides under simulated primitive * This work was supported in part by grant-in-aids for scientific research and on priority areas from the Ministry of Education, Science, and Culture of Japan. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisernent" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
I To whom correspondence should be addressed. earth conditions such as a freshwater tide pool were found to have partial secondary structures and to perform catalytic functions (7,8). Modules were observed to be essentially the same size as the polypeptides (2).
Modules (and/or their assembly) may have functioned as primitive proteins in prebiological evolution. For confirmation of this "module hypothesis," the catalytic functions of disconnected modules of barnase were examined.
Barnase is an extracellular nonspecific ribonuclease produced by Bacillus amyloliquefaciens (9) and a monomer with an M, of 12,382 consisting of 110 amino acid residues (10). Its tertiary structure was determined by x-ray crystallography (11). In this study, we used the enzyme to examine the functions of modules. It is one of the smallest globular proteins that, although possessing neither disulfide bonds nor nonpeptide cross-links, has a stable globular conformation.
Cleavage of Escherichia coli 5 S rRNA by modules of barnase was determined as follows. E. coli MREGOO 5 S rRNA (Boehringer Mannheim) was labeled at the 3'-end with [5'-32P]pCp by T4 RNA ligase.
After polyacrylamide gel electrophoresis, the labeled 5 S rRNA band was eluted from the gel by crushing and soaking in 0.5 M sodium acetate, 0.1% SDS, and 0.1 mM EDTA at 15 "C overnight. The eluted 5 S rRNA was precipitated with ethanol and redissolved in 10 pl of cleavage buffer (0.1 M Tris-HCI, pH 7.5) containing a module and incubated at 37 "C for 15 h. After incubation, the reaction was terminated by adding an equal volume of 95% formamide, 10 mM EDTA, 0.025% xylene cyanol, and 0.02% bromphenol blue. Aliquots were electrophoresed on a 12% denaturing polyacrylamide-urea gel T. Noguti, H. Sakakibara, and M. G6, submitted for publication.
The nucleotide sequence was determined by comparison of the degradation patterns of E. coli 5 S rRNA with those of partial alkaline digests on enzymatic sequencing gels. These bands can easily be recognized by the characteristic spacing they produce in the ladders. Electrophoresis of barnase modules was carried out on SDS-polyacrylamide gel (16.5% T, 6% C; 13.5 X 15 X 0.1 cm) according to the method of Schagger and Von Jagow (14), except that yeast RNA (10 mg/15 ml) was added to the separation gel, and the gel was treated with 12.5% glutaraldehyde before staining with Coomassie Brilliant Blue.
Activity staining (15) was performed as follows. To remove SDS, after SDS-polyacrylamide gel electrophoresis, the gel was washed for 30 min at room temperature in 250 ml of 8 mM Tris-HCI, pH 8.0, containing 20% 2-propanol, followed by washing for 15 min in 150 ml of 10 mM Tris-HC1, pH 8.0. The gel was further incubated to cleave RNA for 2 h at 37 "C in 150 ml of 10 mM Tris-HC1, pH 8.0. The gel was washed for 30 min at room temperature in 200 ml of 10 mM Tris-HCI, pH 7.5, containing 25% 2-propanol, stained for RNA for 30 min at room temperature in 10 mM Tris-HCI, pH 7.5, containing 0.2% toluidine blue 0 (Merck), and washed in 10 mM Tris-HC1, pH 7.5, to detect the activity bands.
Module-RNA binding was examined by membrane filter assay. A module (10 p~) was added to 0.1 M HEPES/NaOH, pH 7.5, containing 0-10 A*W units of E. coli MRE6OO 5 S rRNA, incubated for 30 min at 4 "C, and then filtered through a Millipore Ultrafree C3LCCOO membrane filter (exclusion limit, M, 10,000). An unbound module was recovered in the filtrate. No 5 S rRNA was cleaved by modules under the conditions described above. The amount of an unbound module in the filtrate was determined by fluorometric assay of proteins (16). The amount of a bound module to 5 S rRNA was calculated after subtracting that of an unbound module from the total amount of a module.

RESULTS AND DISCUSSION
The RNase activity of modules of barnase toward E. coli 5 S rRNA was assessed by polyacrylamide gel electrophoresis under denaturing conditions. A sample of 5 S rRNA was incubated with 140 FM barnase modules in 0.1 M Tris-HC1, pH 7.5, at 37 "C for 15 h. As shown in Fig. 1 (lanes 2, 3, and 6), 5 S rRNA was very prominent in the controls (lane 7) and was cleaved, and polynucleotide cleavage products were formed. It is thus evident that M2, M3, and M6 catalyze the cleavage of 5 S rRNA, whereas M1, M4, and M5 possess no detectable RNase activity (lanes 1,4, and 5 ) . Native barnase also catalyzed the cleavage of 5 S rRNA. The cleavage patterns of M2, M3, and M6 differed remarkably from that of native barnase. M3 showed the highest RNase activity of the three modules, followed by M2. Although M2, M3, and M6 are 1 2 3 4 5 6 7 8 ribonucleolytic enzymes, their activity toward 5 S rRNA was considerably less than that of native barnase. Decomposition of barnase into modules lowered catalysis by 4 orders of magnitude. The RNase activity of M2, M3, and M6 was higher at 55 "C and pH 9.5 than at 37 "C and pH 7.5 (data not shown). No amino acid mixture corresponding to the amino acid composition of M2, M3, or M6 catalyzed the cleavage of 5 S rRNA under the identical conditions.
To determine whether the RNase activity associated with modules was due to contamination of enzymes with higher RNase activity, high molecular weight (Mr >10,000) contaminants with RNase activity in the synthetic modules were removed by autoclaving and ultrafiltration. M2, M3, and M6 retained significant RNase activity after autoclaving at 120 "C for 10 min or after ultrafiltration with a Millipore Ultrafree C3LCCOO membrane filter (exclusion limit, M, 5000).
To exclude further the possibility of contamination of RNases, RNase activity was examined by cleavage of RNA embedded in SDS-polyacrylamide gels. Synthetic modules of barnase (M2, M3, and M6) were found to show significant RNase activity, whereas none could be detected for M1, M4, or M5 (Fig. 2b). The activity staining bands coincided well with the corresponding protein staining bands (Fig. 2a). This (together with the finding that RNase activity was retained following autoclaving and ultrafiltration) confirms that M2, M3, and M6 actually possess real RNase activity. M2 formed a dimer and a trimer also with RNase activity (Fig. 2, a and  b).
The membrane filter assay for binding of modules of barnase with RNA has indicated that some modules possess RNA binding activity. M2, M3, and M6 bound to E. coli 5 S rRNA; but M1, M4, and M5 lacked RNA binding activity (Fig. 3).  It was quite recently suggested that the positive-charge side chains, Lys-27, Arg-59, and His-102 present on M2, M3, and M6, respectively, are clustered in the active site of barnase and that they are necessary for catalysis and binding of the negatively charged RNA substrate (18).
The use of excess basic amino acid residues (Arg and Lys), rather than acidic ones (Asp and Glu), in the modules may provide some indication of catalytic activity of the modules. M2 and both M3 and M6 have 1 and 2 extra basic amino acid residues, respectively; but M1 and M4 have 2 and 1 extra acidic amino acid residue, and M5 has the same number of acidic and basic residues. Basic amino acid residues have an important role in phosphate binding, and possibly the RNase activity of M2, M3, and M6 may be explained on the basis of the basic amino acid residues.
To estimate further the importance of the net charge on the barnase modules to RNA binding and RNase activities, we chose 22 control peptides possessing net charges of -3 to +5. Table I shows the sequences of the barnase modules and control peptides that are synthesized and that correspond to the short segments (15-26-mer) of different proteins. The control peptides with net charges of more than +3 bound to 5 S rRNA and catalyzed the cleavage of the RNA quite well, whereas this could not be detected for the control peptides possessing negative net charges. These data suggest that the net charges on the peptides are important in RNA binding and RNase activity. In the control peptides possessing net charges of 0 to +2, some had both RNA binding and RNase activities or only RNA binding activity, and others had no significant affinity for RNA and no RNase activity. This finding suggests that the peptides possessing net charges of 0 to +2, at least in the context of these peptides, are not sufficient for RNA binding or RNase activity. Since the net charges on M2, M3, and M6 are within the range of 0 to +2, there may be a different situation in the module itself.
Not determined (because of lack of a primary amino group at the NH2-terminus).
control peptides. It was clear from the spectra that the barnase modules (M2, M3, and M6) as well as the control peptides take predominantly a random coil in aqueous solution (data not shown). These data suggest that the differences in RNA binding and RNase activity for the different peptides are probably not due to differences in secondary structure. Native barnase cleaved a phosphodiester bond of GpN in the loop and hinge regions (Fig. 4a). This result indicates that barnase has a strong preference for guanosine in partial digestion of E. coli 5 S rRNA. Barnase is well known to cleave after only guanosine when catalyzing the hydrolysis of the dinucleotide substrate GpN (33). With longer substrates, however, it preferentially cleaves after guanosine; but it cleaves after other bases, yielding a mixture of mono-and dinucleotides in a total RNA digest (34). Barnase shows the preference order A > G >C = U (35).
On the other hand, barnase module (M2, M3, and M6)induced cleavages in the E. coli 5 S rRNA (Fig. 1) were mostly located in the loop and hinge regions, but their cleavage sites were different from those of native barnase (Fig. 4b).
In the loop regions, one strong cleavage occurred at positions 103 and 104. This result indicates that flexible and dynamic regions appear as preferred target sites for barnase moduleinduced cleavage. The barnase modules preferentially cleaved a phosphodiester bond of YpA. Therefore, the barnase modules are quite different from native barnase, but they resemble pyrimidine-specific bovine pancreatic RNase A (36) in their preference for nucleoside.
The 3"labeled cleavage products ( Fig. 1) from E. coli 5 S rRNA with native barnase and barnase modules showed the same migration as the limited alkaline cleavage product of the same length (data not shown). This indicates that the cleavage products bear a 3"phosphate terminus on the 3'terminal fragment. The presence of a basic amino acid residue in a position favorable for catalysis may be essential for the RNase activity of barnase modules because basic amino acid residues are frequently found in the active site of hydrolytic enzymes. It appears that phosphate binding of native barnase is affected mainly by positively charged residues including His-102, .
M6 with a net charge of +2 showed lower RNA binding and RNase activities than M2 and M3 with net charges of +1 and +2, respectively. Some factors may be responsible for RNA binding, RNase activity, and specificity of barnase modules. Among the factors, imbalance of positively charged amino acid residues and peptide chain length in barnase modules together with their net charge may be much more important. Positively charged residues were delocalized on barnase modules and control peptides possessing RNA binding and RNase activity (Table I). A basic polypeptide such as poly(Leu-Lys) with alternating basic and hydrophobic residues cleaves oligoadenylates more efficiently than poly(Lys) (38). This suggests the importance of imbalance of positively charged residues and hydrophobicity for more efficient cleavage activity. High salt concentration (1 M NaC1) strongly inhibited the RNA binding of M2, M3, and M6 (data not shown). Thus, the interaction between a peptide and RNA appears to be primarily ionic in nature and secondarily nonelectrostatic. M2, M3, and M6 consist of 29,32, and 13 amino acid residues, respectively. The lower RNA binding and RNA catalytic activities of M6 may also be due to the shorter peptide chain length compared with M2 and M3. Barnase modules (M2, M3, and M6) may accelerate the classical base-induced cleavage of a phosphodiester bond of RNA (39). Once the complex is formed through ionic interaction between the phophate groups and protonated basic side chains of barnase modules, the cleavage of the phosphodiester bond occurs via a nucleophilic attack of the free base of the side chain. The base-induced cleavage involves both hydroxyl groups of an RNA-ribose.
The compact conformation of each module and the architecture of barnase as an assembly of modules are shown in Fig. 5 (a and b). The atoms of each module are drawn in different colors. M2, M3, M4, and M6 form a shallow but wide cavity for RNA binding as evident from a front view (Fig. 5a). M1 and M5 appear to support the other four modules from the back side, so that they are in the proper position for producing globular protein (Fig. 5b). Neither module possesses a catalytic site, and both may function only to provide foundation support. The spatial arrangement of Ml"6 is a point of interest.
Barnase is produced by bacteria, and its gene is not split by introns (40), which also is the case for other prokaryotic genes. Thus, assessment of the correspondence of module boundaries of barnase with intron positions in its gene is not possible. However, prokaryotic proteins can be decomposed into modules of essentially the same size as eukaryotic protein modules (2). Module-intron correspondence has been reported for many eukaryotic proteins and their genes. That prokaryotes may have lost introns from their genes (21, 28) appears to be very likely.
It follows from these results that certain modules of barnase function as catalysts in aqueous solution.
It is particularly remarkable that only M2, M3, and M6, which form the active site of native barnase (18), possess RNase activity, while M1 and M5, which support other modules from the back side, have none. This provides an indication as to the manner in which modules with functions evolved and assembled into primitive enzymes. This study strongly supports the hypothesis that certain separate modules possessing weaker catalytic activity may have had functions in early biological evolution. Primordial modules may have been quite common peptides with primitive catalytic functions like control peptides possessing weaker RNase activity, as shown in Table I. They subsequently became connected to each other, forming functional globular proteins through module shuffling. It would appear then that a protein may come to function more efficiently by tuning up catalytic sites in which clusters of likecharged residues or aligned dipoles are used for binding charged species such as substrate, metal ions, cofactors, or other ligands.