- Email: [email protected]
Crystal structure of a ribonuclease from the seeds of bitter gourd (Momordica charantia) at 1.75 Å resolution
Biochimica et Biophysica Acta 1433 (1999) 253^260 www.elsevier.com/locate/bba
Crystal structure of a ribonuclease from the seeds of bitter gourd î resolution (Momordica charantia) at 1.75 A Atsushi Nakagawa a , Isao Tanaka a , Ritsu Sakai b , Takashi Nakashima b , Gunki Funatsu b , Makoto Kimura b; * a
Division of Biological Sciences, Graduate School of Science, Hokkaido University, Sapporo 060-0810, Japan b Laboratory of Biochemistry, Faculty of Agriculture, Kyushu University, Fukuoka 812-8581, Japan Received 31 January 1999; received in revised form 4 May 1999; accepted 24 May 1999
Abstract The ribonuclease MC1 (RNase MC1) from seeds of bitter gourd (Momordica charantia) consists of 190 amino acid residues with four disulfide bridges and belongs to the RNase T2 family, including fungal RNases typified by RNase Rh from î Rhizopus niveus and RNase T2 from Aspergillus oryzae. The crystal structure of RNase MC1 has been determined at 1.75 A resolution with an R-factor of 19.7% using the single isomorphous replacement method. RNase MC1 structurally belongs to the (K+L) class of proteins, having ten helices (six K-helices and four 310 -helices) and eight L-strands. When the structures of RNase MC1 and RNase Rh are superposed, the close agreement between the K-carbon positions for the total structure is obvious: the root mean square deviations calculated only for structurally related 151 K-carbon atoms of RNase MC1 and î . Furthermore, the conformation of the catalytic residues His-46, Glu-105, and His-109 in RNase Rh molecules was 1.76 A RNase Rh can be easily superposed with that of the possible catalytic residues His-34, Glu-84, and His-88 in RNase MC1. This observation strongly indicates that RNase MC1 from a plant origin catalyzes RNA degradation in a similar manner as fungal RNases. ß 1999 Elsevier Science B.V. All rights reserved. Keywords: Bitter gourd; Crystal structure; Ribonuclease; (Momordica charantia)
1. Introduction Since McClure et al. discovered that the S-glycoproteins associated with gametophytic self-incompatibility in the Solanaceae has ribonuclease activity [1], there has been growing evidence that RNases in plants, as the cases for RNases in mammals [2], may participate in diverse physiological and developmental processes (for a review see [3]). A large number of sequence information on plant RNases have
* Corresponding author. Fax: +81-92-642-2854; E-mail: [email protected]
become available to date; it has been shown that their amino acid sequences are mutually related. On the basis of sequence homology, the plant RNases have been found to be related to RNases classi¢ed in the RNase T2 family typi¢ed by fungal RNases, such as RNase T2 [4], RNase M [5], and RNase Rh [6]. Furthermore, the plant RNases are subgrouped into two groups S-RNase and S-like RNase; the former includes RNases involved in the gametophytic self-incompatibility and the latter includes RNases not involved in it. However, despite the vast amount of sequence information, very few studies on the tertiary structure of plant RNases have been carried out thus far.
0167-4838 / 99 / $ ^ see front matter ß 1999 Elsevier Science B.V. All rights reserved. PII: S 0 1 6 7 - 4 8 3 8 ( 9 9 ) 0 0 1 2 6 - 0
BBAPRO 35947 2-8-99
254
A. Nakagawa et al. / Biochimica et Biophysica Acta 1433 (1999) 253^260
The 2P,3P-cyclizing RNases have been studied in mammalian as well as microbial cells for many years and classi¢ed into three distinct groups; RNase A, RNase T1 , and RNase T2 families, according to their molecular weights and amino acid sequences. The best characterized protein of the RNase T2 family is RNase Rh from Rhizopus niveus. The three-dimenî resolution sional structure was determined at 2.0 A [7] and the amino acid residues involved in the catalytic reaction and the substrate binding are well de¢ned by site-directed mutagenesis ([8] and references therein). Moreover, on the release of nucleotides from RNA, RNase Rh was identi¢ed to have nonabsolute base speci¢city. However they release the four nucleotides in the order of A s G s U,C [9]. Thus, the fungal RNases grouped in the RNase T2 family can be distinguished from other RNase families by their non-base speci¢city. RNase MC1, isolated from the bitter gourd seeds, consists of a single polypeptide chain of 190 amino acids, having four disul¢de bridges [10]. Sequence comparison of RNase MC1 with other known plant RNases revealed that RNase MC1 has a signi¢cant homology with S-like RNases (42%), and a slightly lower homology with S-RNases (40%). In addition, RNase MC1 shares the two stretches of amino acid sequences around the two active site histidine residues in fungal non-base-speci¢c RNases. However, in spite of sequence homology, RNase MC1 has been revealed to preferentially cleave the phosphodiester bond of the NpU (where N is either A, G, U, or C) [11]. This absolute uracil speci¢city found for RNase MC1 is quite di¡erent from that of fungal RNases, which preferentially cleave RNA at adenylic acid residues, but without absolute base speci¢city. Moreover, the speci¢city of RNase MC1 set by the uracil at the 3P-terminal side of dinucleotide monophosphates is distinct from other RNases, including those in RNase A and RNase T1 families. It is generally known that RNases degrade the phosphodiester bond, primarily recognizing the base located at the 5P-side of the scissile bond, though the dependence of the catalytic rate on the nature of the nucleoside at the 3P-side of the scissile bond is known to be a general feature of RNases. It is thus speculated that RNase MC1 might have diverged from the archetype of fungal non-base-speci¢c RNases, acquiring a subsite structure capable to bind to the uracil
base at the 3P-side of the scissile bond. To obtain a fuller understanding of the enzymatic mechanism of RNase MC1, it was thought to be essential that its tertiary structure be determined at the atomic resolution. De and Funatsu crystallized RNase MC1 for X-ray studies some years ago [12]. In the present study, we determined the crystal structure of RNase MC1 by the single isomorphous replacement method. 2. Materials and methods 2.1. Preparation of the RNase MC1 crystal. RNase MC1 was prepared from the seeds of bitter gourd and crystallized as previously described in [10,12]. The crystals are orthorhombic space group P21 21 21 with cell dimensions a = 67.69, b = 75.42, and î , containing one RNase MC1 molecule c = 38.76 A per asymmetric unit. The heavy atom derivative was prepared by soaking the crystals in solution containing 2 mM of p-hydroxymercuricbenzenesulfonic acid dissolved in the mother liquor. The soaking time was 2 days. 2.2. Data collection and computing. All di¡raction data were collected using CuKK radiation from a MAC science M18X rotation anode î ) with double focusing mirrors. Data (40 kV, 80 mA were collected on a MAC science DIP-2000k di¡ractometer with 1.0³ rotation per frame. All data were integrated using the programs DENZO [13]. Scaling and merging were performed using the program SCALA [14] and other programs in the CCP4 suite [15]. The intensities were reduced to structure factors and scaled relative to each other using the programs in the CCP4 program suite [15]. Results of data reduction are summarized in Table 1. Further calculations for crystallography were mainly performed using the programs in the CCP4 program suite [15]. Two mercury atom sites were found from the anomalous di¡erence Patterson map using SHELXS-97 [16]. Heavy atom parameter re¢nement and phase calculation were carried out using the maximum likelihood approach with the program SHARP [17]. The atomic model was built using the program O [18] and was re¢ned by the program X-PLOR [19].
BBAPRO 35947 2-8-99
A. Nakagawa et al. / Biochimica et Biophysica Acta 1433 (1999) 253^260
255
2.3. Structure determination and re¢nement Re¢nement of heavy atom parameters and phasing î reswas done with the di¡raction data below 2.0 A î olution. The electron density map calculated at 1.9 A resolution obtained after phase extension and phase improvement by the program SOLOMON [20] gave extremely good electron density map (Fig. 1) in spite of poor anomalous signal. The results of phasing calculations are summarized in Table 2. All residues were built into the initial electron density map without ambiguity. Seventy-three water molecules were also identi¢ed in the initial map, and they were included in the initial model for re¢nement. The crystallographic R-factor for the data between 8 to 1.75 î resolution dropped from 0.371 to 0.215 after the A ¢rst ¢ve iterative cycles of positional and B-factor re¢nement. At the ¢nal stage of re¢nement, the model has an R-factor of 19.7% for 95% of the data î and 1.75 A î resolution, including between 10.0 A 190 residues and 25 water molecules, a total of 1145 atoms. Bulk solvent correction was applied. The free R-factor [21] for the remaining 5% of the data within this resolution range is 24.9%. The Ramachandran plot of the model shows that 87.7% of the residues lie within the most favored region and the remaining 12.3% of the residues lie within the additional allowed region. The root mean square deviations from standard values [22] of bond lengths î and 1.246³, respectively. and angles are 0.005 A The atomic coordinates have been deposited in the Protein Data Bank with the accession code 1BK7.
î resolution electron density map. The Fig. 1. Part of the 1.9 A electron density map is made using SHARP [17] and SOLOMON [20] and the re¢ned model of the RNase MC1 coordinates are superimposed. The contour level is 1.3 times the root mean square density of the map.
Table 1 Crystallographic data Data
Native
PHMBS 1
PHMBS 2
î) Resolution (A a Rmerge Rbanom Observed re£ections Independent re£ections I/c(I) Completeness (%) Multiplicity Rciso
1.75 0.055 (0.199)
2.0 0.123 (0.244) 0.063 (0.129) 52 093 11 881 4.4 (2.5) 86.8 (84.2) 4.4 (3.8) 0.289
2.0 0.067 (0.127) 0.046 (0.070) 46 148 12 130 6.1 (5.5) 90.3 (87.0) 3.8 (3.3) 0.252
69 704 20 323 9.1 (3.5) 98.4 (98.2) 3.4 (3.1)
î , PHMBS1: 2.24^2.00 A î , PHMBS2: 2.13^2.00 A î ). Values within parentheses are for the highest resolution shell (Native: 1.95^1.75 A a Rmerge = ggj |GI(h)3I(h)j f|/ggj GI(h)f, where GI(h)f is the mean intensity of symmetry equivalent re£ections. b Ranom = g|GI(+h)f3GI(3h)f|/g(GI(+h)f+GI(3h)f). c Riso = ggj ||FPH |3|FP ||/ggj |FP |, where FPH is the structure factor of the derivatives and FP is the structure factor of the native crystal.
BBAPRO 35947 2-8-99
256
A. Nakagawa et al. / Biochimica et Biophysica Acta 1433 (1999) 253^260
3. Results and discussion 3.1. Structure description Fig. 2 shows a ribbon diagram of the K-carbon tracing of RNase MC1. Dimension of the RNase î , and MC1 molecule is approximately 50U40U35 A structurally belongs to the (K+L) class of proteins, having ten helices (six K-helices and four 310 -helices) and eight L-strands. The amino acid residues (L1: 3^ 9, and L2: 32^40) at the N-terminal region form a two-stranded antiparallel L-sheet, intervening one Khelix (K1; 11^16) and one 310 -helix (K2: 24^27) structures which are held together by the ¢rst disul¢de bridge between residues 15 and 23. The polypeptide chain enters the extended helical structure, having a short L-strand (L3: 43^44), followed by K-helices (K3: 55^61 (310 ), K4: 62^68, K5: 77^87, K6: 89^92 (310 ), K7: 98^111, K8: 114^118, K9: 119^121 (310 ) and K10: 132^143) and one strand L4: 129^131, connected through a series of loops. The amino acid residues at the C-terminal region form an antiparallel pleated L-sheet (L5: 148^153, L6: 160^170, and L7: 176^177), forming a central antiparallel pleated Lsheet with the N-terminal L-strands (L1^L3). The structure is completed by two antiparallel L-strands (L4: 129^131 and L8: 187^189) followed by a stretch of extended polypeptide chain. The amino acid sequence of RNase MC1 was initially determined by the protein chemical methods [10]. Recently, the cDNA encoding RNase MC1
Fig. 2. Ribbon diagram of RNase MC1 molecule. The structure is drawn with Molscript [26] and Raster 3D [27].
was cloned and its nucleotides were sequenced (G. Funatsu, unpublished results). The sequence of RNase MC1 deduced from the cDNA sequence differs in three positions from that reported previously. In the protein analysis, the residues at positions 40 and 165 were identi¢ed as Gly and Gln, whereas they were Gln and Glu by the cDNA sequencing. Further,
Table 2 Phasing statistics Data
Native
Racullis
(centric) (acentric) Phasing powerb (centric) (acentric) Figure of merit (centric) (acentric) (¢nal)c
PHMBS 1
PHMBS 2
Isomorphous
Anomalous
Isomorphous
Anomalous
0.570 0.460
0.690
0.568 0.464
0.858
2.226 3.635
2.399
2.358 3.768
1.953
0.4284 0.4525 0.9247
a
Rcullis = gj |E|/gj ||FPH |3|FP ||, where E is the lack of closure error. Phasing power = G|FH (calc)|/|E|f, where FH (calc) is the calculated heavy atom contribution and E is the lack of closure. c Figure of merit after density modi¢cation by SOLOMON [20,21]. b
BBAPRO 35947 2-8-99
A. Nakagawa et al. / Biochimica et Biophysica Acta 1433 (1999) 253^260
Gln-50 as identi¢ed by protein sequencing was missing in the sequence deduced from the cDNA. The present crystal structure perfectly matches the amino acid sequence deduced from the cDNA sequence. It thus results in that RNase MC1 is composed of 190 amino acid residues with a calculated Mr of 21, 222. The recent analysis by the protein chemical method determined the connection scheme of four disul¢de bridges in RNase MC1 as follows: Cys-15 to Cys-23, Cys-48 to Cys-91, Cys-168 to Cys-179, and Cys-151 to Cys-184 (G. Funatsu, unpublished results), which is in full agreement with that determined by the crystal structure of RNase MC1. 3.2. Structure comparison The three-dimensional structure of RNase MC1 molecule was compared to the RNase Rh molecule, the structure of which has been described in detail [7]. Fig. 3 is the superposition of the backbones of the RNase MC1 and RNase Rh molecules. Two structures are very similar, even with the sequence di¡erence in the residues (26% identity). The root mean square deviation, calculated only for structurally related 151 K-carbon atoms of RNase MC1 and RNase Rh molecules of which K-carbons were closer î , was 1.760 A î as calculated by the 1sqthan 3.8 A improve command in O [18]. The two molecules also possess the same elements of secondary structure, as shown in Fig. 4, which displays the amino acid sequence alignment of the two proteins and the ele-
257
ments of secondary structure observed from the crystallographic models. The di¡erences in the K-carbon tracings mainly occur due to di¡erences in the length of the two polypeptide chains. One insertion found at the Nterminal region in RNase MC1 forms K-helix turn 310 -helix structure possibly stabilized by the disul¢de bridge (Cys-15 to Cys-23) between two-stranded (L1 and L2) structures. Since these two cysteine residues are absent in RNase Rh, but are completely conserved in plant RNases [3], the N-terminal region of plant RNases might fold into a conformation similar to that of RNase MC1. The N-terminal extension, which occurs only in RNase Rh, forms a closed loop that is held together by two disul¢de bridges (Cys-3 to Cys-20 and Cys-10 to Cys-53). This structure is completely missing in the RNase MC1. The other extra residues at the N-terminal (pos. 66^73) and middle region (pos. 118^128) of RNase Rh adopt short and extended loop structures, respectively. Deletion of these residues in RNase MC1 results in the shortened loop structure between two helices K6 and K7, as compared to RNase Rh. Sequence comparison of plant RNases found ¢ve highly conserved regions designated C1^C5 and two hypervariable regions designated HVa and HVb [3]. The conserved regions correspond to Phe-1-Pro-11 (C1), Thr-30-Pro-38 (C2), Lys-87-Ser-92 (C3), Met108-Ile-115 (C4), and Tyr-161-Phe-169 (C5) in RNase MC1. Examining the RNase MC1 structure, the regions C1^C5 are found to adopt secondary
Fig. 3. Comparison of the Ca-trace of the RNase MC1 and RNase Rh. The thick line represents RNase MC1 and the thin line represents RNase Rh molecules. Two molecules are ¢tted with LSQ_IMP command in O [18].
BBAPRO 35947 2-8-99
258
A. Nakagawa et al. / Biochimica et Biophysica Acta 1433 (1999) 253^260
Fig. 4. Sequence alignment of RNase MC1 and RNase Rh. Secondary structure elements indicated are those de¢ned using the program DSSP [28]. K-helices and 310 -helices are indicated by black and gray bars, respectively. L-strands are indicated by arrows. The amino acid residues identical in two proteins are enclosed in boxes. The key residues involved in the catalysis are marked by asterisks and residues involved in the substrate binding are marked by dollar signs.
structures, L1, L2, K5, K7, and L6, respectively, as shown in Fig. 3. This observation suggests that the structure found in this RNase MC1 is principally common to all plant RNases. On the other hand, the hypervariable regions HVa and HVb are located on the extended loop structures between L3 and K3, and K4 and K5, respectively, which are located on the external surface of RNase MC1. Since it is described that the HV regions in S-RNases are involved in pollen recognition [23], the S-RNases may have acquired a variety of amino acids in these regions and display their allele-speci¢c information on the molecular surface. 3.3. Possible active site The active site of RNase Rh has been mapped from the site-directed mutagenesis experiments [8]. The key active site residues include His-46, His-104, Glu-105, Lys-108, and His-109 involved in catalysis, and Trp-49, Asp-51, and Tyr-57 involved in binding the target base. The sequence alignment readily shows their equivalent residues in RNase MC1, as
shown in Fig. 4. All catalytic residues in RNase Rh are completely conserved in the primary structure of RNase MC1 as His-34, His-83, Glu-84, Lys-87, and His-88, while Asp-51 and Tyr-57 in RNase Rh are replaced by Gln-39 and Ser-44 in RNase MC1. Our recent study by the site-directed mutagenesis of RNase MC1 shows that the mutants, in which His34 or His-88 are replaced by the Ala residue, are virtually inactive (Kimura et al., unpublished results). It is therefore likely that His-34 and His-88 in RNase MC1 serve as the catalytic acid and catalytic base, respectively in the transphosphorylation reaction, as the catalytic His-46 and His-109 pair do in RNase Rh [24]. It is known that RNases have two distinct binding sites: the primary site and subsite, for the bases located at 5P- and 3P-terminal ends of the scissile bond, respectively. Base speci¢city is usually attributable to the nature of the interaction of the primary site with the base located at the 5P-terminal end [25]. As for RNase Rh, Trp-49, Asp-51, and Tyr-57 are assigned as building blocks to constitute the primary site. The Trp-49 is conserved in RNase MC1 (Trp-37) and two
BBAPRO 35947 2-8-99
A. Nakagawa et al. / Biochimica et Biophysica Acta 1433 (1999) 253^260
residues appear to adopt an identical conformation, whereas Asp-51 and Tyr-57 are replaced by Gln-39 and Ser-44, respectively, in RNase MC1. In particular, the replacement of Tyr-57 in RNase Rh with Ser44 in RNase MC1 results in the lack of a hydrophobic pocket, which is involved in the purine recognition site (B1 -site) for RNase Rh. As described in Section 1, RNase MC1 is distinct from RNase Rh in that it preferentially cleaves the phosphodiester bond NpU, recognizing the uracil at the 3P-terminal side of the scissile bond. It is thus assumed that the lack of rigid base speci¢city at the 5P-terminal end for RNase MC1 may be due to these substitutions. Obviously, understanding the molecular basis for the uracil speci¢city at the 3P-terminal end for RNase MC1 will require structural studies of the crystal structure of RNase MC1 complexed with suitable nucleic acid fragments. The present crystal structure of RNase MC1 allows us to commence these studies. Acknowledgements We thank Prof. K.T. Nakamura in Showa University for providing us a set of coordinates of RNase Rh and Drs. E. de la Fortelle, J. Irwin, and G. Bricogne in MRC, Cambridge, for their helpful suggestions on using SHARP. A.N. and I.T. are members of the TARA project of Tsukuba University, Japan. References [1] B.A. McClure, V. Haring, P.R. Ebert, M.A. Anderson, R.J. Simpson, F. Sakiyama, A.E. Clarke, Style self-incompatibility gene products of Nicotiana alata are ribonucleases, Nature 342 (1989) 955^958. [2] G. D'Alessio, New and cryptic biological messages from RNases, Trends Cell Biol. 3 (1993) 106^109. [3] P.J. Green, The ribonucleases of higher plants, Annu. Rev. Plant Physiol. Plant Mol. Biol. 45 (1994) 421^445. [4] K. Sato, F. Egami, Studies on ribonucleases in takadiastase, J. Biochem. 44 (1957) 753^767. [5] M. Imazawa, M. Irie, T. Ukita, Substrate speci¢city of ribonuclease from Aspergillus saitoi, J. Biochem. 64 (1968) 597^ 602. [6] M. Tomoyeda, Y. Eto, T. Yoshino, Studies on ribonuclease produced by Rhizopus sp. I. Crystallization and some properties of the ribonuclease, Arch. Biochem. Biophys. 131 (1969) 191^202.
259
[7] H. Kurihara, T. Nonaka, Y. Mitui, K. Ohgi, M. Irie, K.T. Nakamura, The crystal structure of ribonuclease Rh from î resolution, J. Mol. Biol. 255 Rhizopus niveus at 2.0 A (1996) 310^320. [8] M. Irie, K. Ohgi, M. Iwama, M. Koizumi, E. Sasayama, K. Harada, Y. Yano, J. Udagawa, M. Kawasaki, Role of histidine 46 in the hydrolysis and the reverse transphorylation reaction of RNase Rh from Rhizopus niveus, J. Biochem. 121 (1997) 849^853. [9] M. Horiuchi, K. Yanai, M. Takagi, Y. Yano, E. Wakabayashi, A. Sanda, S. Mine, K. Ohgi, M. Irie, Primary structure of a base non-speci¢c ribonuclease from Rhizopus niveus, J. Biochem. 103 (1988) 408^418. [10] H. Ide, M. Kimura, M. Arai, G. Funatsu, The complete amino acid sequence of ribonuclease from the seeds of bitter gourd (Momordica charantia), FEBS Lett. 284 (1991) 161^ 164. [11] M. Irie, H. Watanabe, K. Ohgi, Y. Minami, H. Yamada, G. Funatsu, Base speci¢city of two plant seed ribonucleases from Momoridica charantia and Lu¡a cylindrica, Biosci. Biotech. Biochem. 57 (1993) 497^498. [12] A. De, G. Funatsu, Crystallization and preliminary x-ray di¡raction analysis of a plant ribonuclease from the seeds of the bitter gourd Momordica charantia, J. Mol. Biol. 228 (1992) 1271^1273. [13] Z. Otwinowski, W. Minor, Processing of x-ray di¡raction data collected in oscillation mode, Methods Enzymol. 276 (1997) 307^326. [14] P.R. Evans, Data collection and processing, in: L. Sawyer, N. Isaacs, S. Bailey (Eds.), Proceedings of the CCP4 study weekend, 29^30 January 1993, SERC Daresbury Laboratory, Warrington, UK, 1993, pp. 114^122. [15] Collaborative Computational Project, Number 4, The CCP4 Suite: programs for protein crystallography, Acta Cryst. D50 (1994) 760^763. [16] G. Sheldrick, Patterson superposition and ab initio phasing, Methods Enzymol. 276 (1997) 628^641. [17] E. de la Fortelle, G. Bricogne, Maximum likelihood re¢nement of heavy-atom parameters re¢nement in the MIR and MAD methods, Methods Enzymol. 276 (1997) 472^494. [18] J.P. Abrahams, A.G.W. Leslie, Methods used in the structure determination of bovine mitochondrial F1 ATPase, Acta Cryst. D52 (1996) 30^42. [19] A.T. Brunger, X-PLOR Version 3.1, A system for X-ray Crystallography and NMR, Yale University Press, New Haven, CT, USA, 1993. [20] T.A. Jones, J-Y. Zou, S.W. Cowan, M. Kjeldgaard, Improved methods for building protein models in electron density maps and the location of errors in these models, Acta Cryst. A47 (1991) 110^119. [21] A.T. Brunger, Free R value: a novel statistical quantity for assessing the accuracy of crystal structures, Nature 355 (1992) 472^475. [22] R.A. Engh, R. Huber, Accurate bond and angle parameters for x-ray protein structure re¢nement, Acta Cryst. A47 (1991) 392^400.
BBAPRO 35947 2-8-99
260
A. Nakagawa et al. / Biochimica et Biophysica Acta 1433 (1999) 253^260
[23] D.P. Matton, O. Maes, G. Laublin, Q. Xike, C. Bertrand, D. Morse, M. Cappadocia, Hypervariable domains of self-incompatibility RNases mediate allele-speci¢c pollen recognition, Plant Cell 9 (1997) 1766^1767. [24] K. Ohgi, H. Horiuchi, M. Iwama, H. Watanabe, M. Takagi, M. Irie, Evidence that three histidine residues of a base nonspeci¢c and adenylic acid preferential ribonuclease from Rhizopus niveus are involved in the catalytic function, J. Biochem. 112 (1992) 132^138. [25] J. Steyaert, A decade of protein engineering on ribonuclease T1, atomic dissection of the enzyme-substrate interactions, Eur. J. Biochem. 247 (1997) 1^11.
[26] P.J. Kraulis, MOLSCRIPT: a program to produce both detailed and schematic plots of protein structures, J. Appl. Cryst. 24 (1991) 946^950. [27] E.A. Merrit, M.E.P. Murphy, Raster3D version 2.0, a program for photorealistic molecular graphics, Acta Cryst. D50 (1994) 869^873. [28] W. Kabsch, C. Sander, Dictionary of protein secondary structure pattern recognition of hydrogen-bonded and geometrical features, Biopolymers 22 (1983) 2577^2637.
BBAPRO 35947 2-8-99
Crystal structure of a ribonuclease from the seeds of bitter gourd î resolution (Momordica charantia) at 1.75 A Atsushi Nakagawa a , Isao Tanaka a , Ritsu Sakai b , Takashi Nakashima b , Gunki Funatsu b , Makoto Kimura b; * a
Division of Biological Sciences, Graduate School of Science, Hokkaido University, Sapporo 060-0810, Japan b Laboratory of Biochemistry, Faculty of Agriculture, Kyushu University, Fukuoka 812-8581, Japan Received 31 January 1999; received in revised form 4 May 1999; accepted 24 May 1999
Abstract The ribonuclease MC1 (RNase MC1) from seeds of bitter gourd (Momordica charantia) consists of 190 amino acid residues with four disulfide bridges and belongs to the RNase T2 family, including fungal RNases typified by RNase Rh from î Rhizopus niveus and RNase T2 from Aspergillus oryzae. The crystal structure of RNase MC1 has been determined at 1.75 A resolution with an R-factor of 19.7% using the single isomorphous replacement method. RNase MC1 structurally belongs to the (K+L) class of proteins, having ten helices (six K-helices and four 310 -helices) and eight L-strands. When the structures of RNase MC1 and RNase Rh are superposed, the close agreement between the K-carbon positions for the total structure is obvious: the root mean square deviations calculated only for structurally related 151 K-carbon atoms of RNase MC1 and î . Furthermore, the conformation of the catalytic residues His-46, Glu-105, and His-109 in RNase Rh molecules was 1.76 A RNase Rh can be easily superposed with that of the possible catalytic residues His-34, Glu-84, and His-88 in RNase MC1. This observation strongly indicates that RNase MC1 from a plant origin catalyzes RNA degradation in a similar manner as fungal RNases. ß 1999 Elsevier Science B.V. All rights reserved. Keywords: Bitter gourd; Crystal structure; Ribonuclease; (Momordica charantia)
1. Introduction Since McClure et al. discovered that the S-glycoproteins associated with gametophytic self-incompatibility in the Solanaceae has ribonuclease activity [1], there has been growing evidence that RNases in plants, as the cases for RNases in mammals [2], may participate in diverse physiological and developmental processes (for a review see [3]). A large number of sequence information on plant RNases have
* Corresponding author. Fax: +81-92-642-2854; E-mail: [email protected]
become available to date; it has been shown that their amino acid sequences are mutually related. On the basis of sequence homology, the plant RNases have been found to be related to RNases classi¢ed in the RNase T2 family typi¢ed by fungal RNases, such as RNase T2 [4], RNase M [5], and RNase Rh [6]. Furthermore, the plant RNases are subgrouped into two groups S-RNase and S-like RNase; the former includes RNases involved in the gametophytic self-incompatibility and the latter includes RNases not involved in it. However, despite the vast amount of sequence information, very few studies on the tertiary structure of plant RNases have been carried out thus far.
0167-4838 / 99 / $ ^ see front matter ß 1999 Elsevier Science B.V. All rights reserved. PII: S 0 1 6 7 - 4 8 3 8 ( 9 9 ) 0 0 1 2 6 - 0
BBAPRO 35947 2-8-99
254
A. Nakagawa et al. / Biochimica et Biophysica Acta 1433 (1999) 253^260
The 2P,3P-cyclizing RNases have been studied in mammalian as well as microbial cells for many years and classi¢ed into three distinct groups; RNase A, RNase T1 , and RNase T2 families, according to their molecular weights and amino acid sequences. The best characterized protein of the RNase T2 family is RNase Rh from Rhizopus niveus. The three-dimenî resolution sional structure was determined at 2.0 A [7] and the amino acid residues involved in the catalytic reaction and the substrate binding are well de¢ned by site-directed mutagenesis ([8] and references therein). Moreover, on the release of nucleotides from RNA, RNase Rh was identi¢ed to have nonabsolute base speci¢city. However they release the four nucleotides in the order of A s G s U,C [9]. Thus, the fungal RNases grouped in the RNase T2 family can be distinguished from other RNase families by their non-base speci¢city. RNase MC1, isolated from the bitter gourd seeds, consists of a single polypeptide chain of 190 amino acids, having four disul¢de bridges [10]. Sequence comparison of RNase MC1 with other known plant RNases revealed that RNase MC1 has a signi¢cant homology with S-like RNases (42%), and a slightly lower homology with S-RNases (40%). In addition, RNase MC1 shares the two stretches of amino acid sequences around the two active site histidine residues in fungal non-base-speci¢c RNases. However, in spite of sequence homology, RNase MC1 has been revealed to preferentially cleave the phosphodiester bond of the NpU (where N is either A, G, U, or C) [11]. This absolute uracil speci¢city found for RNase MC1 is quite di¡erent from that of fungal RNases, which preferentially cleave RNA at adenylic acid residues, but without absolute base speci¢city. Moreover, the speci¢city of RNase MC1 set by the uracil at the 3P-terminal side of dinucleotide monophosphates is distinct from other RNases, including those in RNase A and RNase T1 families. It is generally known that RNases degrade the phosphodiester bond, primarily recognizing the base located at the 5P-side of the scissile bond, though the dependence of the catalytic rate on the nature of the nucleoside at the 3P-side of the scissile bond is known to be a general feature of RNases. It is thus speculated that RNase MC1 might have diverged from the archetype of fungal non-base-speci¢c RNases, acquiring a subsite structure capable to bind to the uracil
base at the 3P-side of the scissile bond. To obtain a fuller understanding of the enzymatic mechanism of RNase MC1, it was thought to be essential that its tertiary structure be determined at the atomic resolution. De and Funatsu crystallized RNase MC1 for X-ray studies some years ago [12]. In the present study, we determined the crystal structure of RNase MC1 by the single isomorphous replacement method. 2. Materials and methods 2.1. Preparation of the RNase MC1 crystal. RNase MC1 was prepared from the seeds of bitter gourd and crystallized as previously described in [10,12]. The crystals are orthorhombic space group P21 21 21 with cell dimensions a = 67.69, b = 75.42, and î , containing one RNase MC1 molecule c = 38.76 A per asymmetric unit. The heavy atom derivative was prepared by soaking the crystals in solution containing 2 mM of p-hydroxymercuricbenzenesulfonic acid dissolved in the mother liquor. The soaking time was 2 days. 2.2. Data collection and computing. All di¡raction data were collected using CuKK radiation from a MAC science M18X rotation anode î ) with double focusing mirrors. Data (40 kV, 80 mA were collected on a MAC science DIP-2000k di¡ractometer with 1.0³ rotation per frame. All data were integrated using the programs DENZO [13]. Scaling and merging were performed using the program SCALA [14] and other programs in the CCP4 suite [15]. The intensities were reduced to structure factors and scaled relative to each other using the programs in the CCP4 program suite [15]. Results of data reduction are summarized in Table 1. Further calculations for crystallography were mainly performed using the programs in the CCP4 program suite [15]. Two mercury atom sites were found from the anomalous di¡erence Patterson map using SHELXS-97 [16]. Heavy atom parameter re¢nement and phase calculation were carried out using the maximum likelihood approach with the program SHARP [17]. The atomic model was built using the program O [18] and was re¢ned by the program X-PLOR [19].
BBAPRO 35947 2-8-99
A. Nakagawa et al. / Biochimica et Biophysica Acta 1433 (1999) 253^260
255
2.3. Structure determination and re¢nement Re¢nement of heavy atom parameters and phasing î reswas done with the di¡raction data below 2.0 A î olution. The electron density map calculated at 1.9 A resolution obtained after phase extension and phase improvement by the program SOLOMON [20] gave extremely good electron density map (Fig. 1) in spite of poor anomalous signal. The results of phasing calculations are summarized in Table 2. All residues were built into the initial electron density map without ambiguity. Seventy-three water molecules were also identi¢ed in the initial map, and they were included in the initial model for re¢nement. The crystallographic R-factor for the data between 8 to 1.75 î resolution dropped from 0.371 to 0.215 after the A ¢rst ¢ve iterative cycles of positional and B-factor re¢nement. At the ¢nal stage of re¢nement, the model has an R-factor of 19.7% for 95% of the data î and 1.75 A î resolution, including between 10.0 A 190 residues and 25 water molecules, a total of 1145 atoms. Bulk solvent correction was applied. The free R-factor [21] for the remaining 5% of the data within this resolution range is 24.9%. The Ramachandran plot of the model shows that 87.7% of the residues lie within the most favored region and the remaining 12.3% of the residues lie within the additional allowed region. The root mean square deviations from standard values [22] of bond lengths î and 1.246³, respectively. and angles are 0.005 A The atomic coordinates have been deposited in the Protein Data Bank with the accession code 1BK7.
î resolution electron density map. The Fig. 1. Part of the 1.9 A electron density map is made using SHARP [17] and SOLOMON [20] and the re¢ned model of the RNase MC1 coordinates are superimposed. The contour level is 1.3 times the root mean square density of the map.
Table 1 Crystallographic data Data
Native
PHMBS 1
PHMBS 2
î) Resolution (A a Rmerge Rbanom Observed re£ections Independent re£ections I/c(I) Completeness (%) Multiplicity Rciso
1.75 0.055 (0.199)
2.0 0.123 (0.244) 0.063 (0.129) 52 093 11 881 4.4 (2.5) 86.8 (84.2) 4.4 (3.8) 0.289
2.0 0.067 (0.127) 0.046 (0.070) 46 148 12 130 6.1 (5.5) 90.3 (87.0) 3.8 (3.3) 0.252
69 704 20 323 9.1 (3.5) 98.4 (98.2) 3.4 (3.1)
î , PHMBS1: 2.24^2.00 A î , PHMBS2: 2.13^2.00 A î ). Values within parentheses are for the highest resolution shell (Native: 1.95^1.75 A a Rmerge = ggj |GI(h)3I(h)j f|/ggj GI(h)f, where GI(h)f is the mean intensity of symmetry equivalent re£ections. b Ranom = g|GI(+h)f3GI(3h)f|/g(GI(+h)f+GI(3h)f). c Riso = ggj ||FPH |3|FP ||/ggj |FP |, where FPH is the structure factor of the derivatives and FP is the structure factor of the native crystal.
BBAPRO 35947 2-8-99
256
A. Nakagawa et al. / Biochimica et Biophysica Acta 1433 (1999) 253^260
3. Results and discussion 3.1. Structure description Fig. 2 shows a ribbon diagram of the K-carbon tracing of RNase MC1. Dimension of the RNase î , and MC1 molecule is approximately 50U40U35 A structurally belongs to the (K+L) class of proteins, having ten helices (six K-helices and four 310 -helices) and eight L-strands. The amino acid residues (L1: 3^ 9, and L2: 32^40) at the N-terminal region form a two-stranded antiparallel L-sheet, intervening one Khelix (K1; 11^16) and one 310 -helix (K2: 24^27) structures which are held together by the ¢rst disul¢de bridge between residues 15 and 23. The polypeptide chain enters the extended helical structure, having a short L-strand (L3: 43^44), followed by K-helices (K3: 55^61 (310 ), K4: 62^68, K5: 77^87, K6: 89^92 (310 ), K7: 98^111, K8: 114^118, K9: 119^121 (310 ) and K10: 132^143) and one strand L4: 129^131, connected through a series of loops. The amino acid residues at the C-terminal region form an antiparallel pleated L-sheet (L5: 148^153, L6: 160^170, and L7: 176^177), forming a central antiparallel pleated Lsheet with the N-terminal L-strands (L1^L3). The structure is completed by two antiparallel L-strands (L4: 129^131 and L8: 187^189) followed by a stretch of extended polypeptide chain. The amino acid sequence of RNase MC1 was initially determined by the protein chemical methods [10]. Recently, the cDNA encoding RNase MC1
Fig. 2. Ribbon diagram of RNase MC1 molecule. The structure is drawn with Molscript [26] and Raster 3D [27].
was cloned and its nucleotides were sequenced (G. Funatsu, unpublished results). The sequence of RNase MC1 deduced from the cDNA sequence differs in three positions from that reported previously. In the protein analysis, the residues at positions 40 and 165 were identi¢ed as Gly and Gln, whereas they were Gln and Glu by the cDNA sequencing. Further,
Table 2 Phasing statistics Data
Native
Racullis
(centric) (acentric) Phasing powerb (centric) (acentric) Figure of merit (centric) (acentric) (¢nal)c
PHMBS 1
PHMBS 2
Isomorphous
Anomalous
Isomorphous
Anomalous
0.570 0.460
0.690
0.568 0.464
0.858
2.226 3.635
2.399
2.358 3.768
1.953
0.4284 0.4525 0.9247
a
Rcullis = gj |E|/gj ||FPH |3|FP ||, where E is the lack of closure error. Phasing power = G|FH (calc)|/|E|f, where FH (calc) is the calculated heavy atom contribution and E is the lack of closure. c Figure of merit after density modi¢cation by SOLOMON [20,21]. b
BBAPRO 35947 2-8-99
A. Nakagawa et al. / Biochimica et Biophysica Acta 1433 (1999) 253^260
Gln-50 as identi¢ed by protein sequencing was missing in the sequence deduced from the cDNA. The present crystal structure perfectly matches the amino acid sequence deduced from the cDNA sequence. It thus results in that RNase MC1 is composed of 190 amino acid residues with a calculated Mr of 21, 222. The recent analysis by the protein chemical method determined the connection scheme of four disul¢de bridges in RNase MC1 as follows: Cys-15 to Cys-23, Cys-48 to Cys-91, Cys-168 to Cys-179, and Cys-151 to Cys-184 (G. Funatsu, unpublished results), which is in full agreement with that determined by the crystal structure of RNase MC1. 3.2. Structure comparison The three-dimensional structure of RNase MC1 molecule was compared to the RNase Rh molecule, the structure of which has been described in detail [7]. Fig. 3 is the superposition of the backbones of the RNase MC1 and RNase Rh molecules. Two structures are very similar, even with the sequence di¡erence in the residues (26% identity). The root mean square deviation, calculated only for structurally related 151 K-carbon atoms of RNase MC1 and RNase Rh molecules of which K-carbons were closer î , was 1.760 A î as calculated by the 1sqthan 3.8 A improve command in O [18]. The two molecules also possess the same elements of secondary structure, as shown in Fig. 4, which displays the amino acid sequence alignment of the two proteins and the ele-
257
ments of secondary structure observed from the crystallographic models. The di¡erences in the K-carbon tracings mainly occur due to di¡erences in the length of the two polypeptide chains. One insertion found at the Nterminal region in RNase MC1 forms K-helix turn 310 -helix structure possibly stabilized by the disul¢de bridge (Cys-15 to Cys-23) between two-stranded (L1 and L2) structures. Since these two cysteine residues are absent in RNase Rh, but are completely conserved in plant RNases [3], the N-terminal region of plant RNases might fold into a conformation similar to that of RNase MC1. The N-terminal extension, which occurs only in RNase Rh, forms a closed loop that is held together by two disul¢de bridges (Cys-3 to Cys-20 and Cys-10 to Cys-53). This structure is completely missing in the RNase MC1. The other extra residues at the N-terminal (pos. 66^73) and middle region (pos. 118^128) of RNase Rh adopt short and extended loop structures, respectively. Deletion of these residues in RNase MC1 results in the shortened loop structure between two helices K6 and K7, as compared to RNase Rh. Sequence comparison of plant RNases found ¢ve highly conserved regions designated C1^C5 and two hypervariable regions designated HVa and HVb [3]. The conserved regions correspond to Phe-1-Pro-11 (C1), Thr-30-Pro-38 (C2), Lys-87-Ser-92 (C3), Met108-Ile-115 (C4), and Tyr-161-Phe-169 (C5) in RNase MC1. Examining the RNase MC1 structure, the regions C1^C5 are found to adopt secondary
Fig. 3. Comparison of the Ca-trace of the RNase MC1 and RNase Rh. The thick line represents RNase MC1 and the thin line represents RNase Rh molecules. Two molecules are ¢tted with LSQ_IMP command in O [18].
BBAPRO 35947 2-8-99
258
A. Nakagawa et al. / Biochimica et Biophysica Acta 1433 (1999) 253^260
Fig. 4. Sequence alignment of RNase MC1 and RNase Rh. Secondary structure elements indicated are those de¢ned using the program DSSP [28]. K-helices and 310 -helices are indicated by black and gray bars, respectively. L-strands are indicated by arrows. The amino acid residues identical in two proteins are enclosed in boxes. The key residues involved in the catalysis are marked by asterisks and residues involved in the substrate binding are marked by dollar signs.
structures, L1, L2, K5, K7, and L6, respectively, as shown in Fig. 3. This observation suggests that the structure found in this RNase MC1 is principally common to all plant RNases. On the other hand, the hypervariable regions HVa and HVb are located on the extended loop structures between L3 and K3, and K4 and K5, respectively, which are located on the external surface of RNase MC1. Since it is described that the HV regions in S-RNases are involved in pollen recognition [23], the S-RNases may have acquired a variety of amino acids in these regions and display their allele-speci¢c information on the molecular surface. 3.3. Possible active site The active site of RNase Rh has been mapped from the site-directed mutagenesis experiments [8]. The key active site residues include His-46, His-104, Glu-105, Lys-108, and His-109 involved in catalysis, and Trp-49, Asp-51, and Tyr-57 involved in binding the target base. The sequence alignment readily shows their equivalent residues in RNase MC1, as
shown in Fig. 4. All catalytic residues in RNase Rh are completely conserved in the primary structure of RNase MC1 as His-34, His-83, Glu-84, Lys-87, and His-88, while Asp-51 and Tyr-57 in RNase Rh are replaced by Gln-39 and Ser-44 in RNase MC1. Our recent study by the site-directed mutagenesis of RNase MC1 shows that the mutants, in which His34 or His-88 are replaced by the Ala residue, are virtually inactive (Kimura et al., unpublished results). It is therefore likely that His-34 and His-88 in RNase MC1 serve as the catalytic acid and catalytic base, respectively in the transphosphorylation reaction, as the catalytic His-46 and His-109 pair do in RNase Rh [24]. It is known that RNases have two distinct binding sites: the primary site and subsite, for the bases located at 5P- and 3P-terminal ends of the scissile bond, respectively. Base speci¢city is usually attributable to the nature of the interaction of the primary site with the base located at the 5P-terminal end [25]. As for RNase Rh, Trp-49, Asp-51, and Tyr-57 are assigned as building blocks to constitute the primary site. The Trp-49 is conserved in RNase MC1 (Trp-37) and two
BBAPRO 35947 2-8-99
A. Nakagawa et al. / Biochimica et Biophysica Acta 1433 (1999) 253^260
residues appear to adopt an identical conformation, whereas Asp-51 and Tyr-57 are replaced by Gln-39 and Ser-44, respectively, in RNase MC1. In particular, the replacement of Tyr-57 in RNase Rh with Ser44 in RNase MC1 results in the lack of a hydrophobic pocket, which is involved in the purine recognition site (B1 -site) for RNase Rh. As described in Section 1, RNase MC1 is distinct from RNase Rh in that it preferentially cleaves the phosphodiester bond NpU, recognizing the uracil at the 3P-terminal side of the scissile bond. It is thus assumed that the lack of rigid base speci¢city at the 5P-terminal end for RNase MC1 may be due to these substitutions. Obviously, understanding the molecular basis for the uracil speci¢city at the 3P-terminal end for RNase MC1 will require structural studies of the crystal structure of RNase MC1 complexed with suitable nucleic acid fragments. The present crystal structure of RNase MC1 allows us to commence these studies. Acknowledgements We thank Prof. K.T. Nakamura in Showa University for providing us a set of coordinates of RNase Rh and Drs. E. de la Fortelle, J. Irwin, and G. Bricogne in MRC, Cambridge, for their helpful suggestions on using SHARP. A.N. and I.T. are members of the TARA project of Tsukuba University, Japan. References [1] B.A. McClure, V. Haring, P.R. Ebert, M.A. Anderson, R.J. Simpson, F. Sakiyama, A.E. Clarke, Style self-incompatibility gene products of Nicotiana alata are ribonucleases, Nature 342 (1989) 955^958. [2] G. D'Alessio, New and cryptic biological messages from RNases, Trends Cell Biol. 3 (1993) 106^109. [3] P.J. Green, The ribonucleases of higher plants, Annu. Rev. Plant Physiol. Plant Mol. Biol. 45 (1994) 421^445. [4] K. Sato, F. Egami, Studies on ribonucleases in takadiastase, J. Biochem. 44 (1957) 753^767. [5] M. Imazawa, M. Irie, T. Ukita, Substrate speci¢city of ribonuclease from Aspergillus saitoi, J. Biochem. 64 (1968) 597^ 602. [6] M. Tomoyeda, Y. Eto, T. Yoshino, Studies on ribonuclease produced by Rhizopus sp. I. Crystallization and some properties of the ribonuclease, Arch. Biochem. Biophys. 131 (1969) 191^202.
259
[7] H. Kurihara, T. Nonaka, Y. Mitui, K. Ohgi, M. Irie, K.T. Nakamura, The crystal structure of ribonuclease Rh from î resolution, J. Mol. Biol. 255 Rhizopus niveus at 2.0 A (1996) 310^320. [8] M. Irie, K. Ohgi, M. Iwama, M. Koizumi, E. Sasayama, K. Harada, Y. Yano, J. Udagawa, M. Kawasaki, Role of histidine 46 in the hydrolysis and the reverse transphorylation reaction of RNase Rh from Rhizopus niveus, J. Biochem. 121 (1997) 849^853. [9] M. Horiuchi, K. Yanai, M. Takagi, Y. Yano, E. Wakabayashi, A. Sanda, S. Mine, K. Ohgi, M. Irie, Primary structure of a base non-speci¢c ribonuclease from Rhizopus niveus, J. Biochem. 103 (1988) 408^418. [10] H. Ide, M. Kimura, M. Arai, G. Funatsu, The complete amino acid sequence of ribonuclease from the seeds of bitter gourd (Momordica charantia), FEBS Lett. 284 (1991) 161^ 164. [11] M. Irie, H. Watanabe, K. Ohgi, Y. Minami, H. Yamada, G. Funatsu, Base speci¢city of two plant seed ribonucleases from Momoridica charantia and Lu¡a cylindrica, Biosci. Biotech. Biochem. 57 (1993) 497^498. [12] A. De, G. Funatsu, Crystallization and preliminary x-ray di¡raction analysis of a plant ribonuclease from the seeds of the bitter gourd Momordica charantia, J. Mol. Biol. 228 (1992) 1271^1273. [13] Z. Otwinowski, W. Minor, Processing of x-ray di¡raction data collected in oscillation mode, Methods Enzymol. 276 (1997) 307^326. [14] P.R. Evans, Data collection and processing, in: L. Sawyer, N. Isaacs, S. Bailey (Eds.), Proceedings of the CCP4 study weekend, 29^30 January 1993, SERC Daresbury Laboratory, Warrington, UK, 1993, pp. 114^122. [15] Collaborative Computational Project, Number 4, The CCP4 Suite: programs for protein crystallography, Acta Cryst. D50 (1994) 760^763. [16] G. Sheldrick, Patterson superposition and ab initio phasing, Methods Enzymol. 276 (1997) 628^641. [17] E. de la Fortelle, G. Bricogne, Maximum likelihood re¢nement of heavy-atom parameters re¢nement in the MIR and MAD methods, Methods Enzymol. 276 (1997) 472^494. [18] J.P. Abrahams, A.G.W. Leslie, Methods used in the structure determination of bovine mitochondrial F1 ATPase, Acta Cryst. D52 (1996) 30^42. [19] A.T. Brunger, X-PLOR Version 3.1, A system for X-ray Crystallography and NMR, Yale University Press, New Haven, CT, USA, 1993. [20] T.A. Jones, J-Y. Zou, S.W. Cowan, M. Kjeldgaard, Improved methods for building protein models in electron density maps and the location of errors in these models, Acta Cryst. A47 (1991) 110^119. [21] A.T. Brunger, Free R value: a novel statistical quantity for assessing the accuracy of crystal structures, Nature 355 (1992) 472^475. [22] R.A. Engh, R. Huber, Accurate bond and angle parameters for x-ray protein structure re¢nement, Acta Cryst. A47 (1991) 392^400.
BBAPRO 35947 2-8-99
260
A. Nakagawa et al. / Biochimica et Biophysica Acta 1433 (1999) 253^260
[23] D.P. Matton, O. Maes, G. Laublin, Q. Xike, C. Bertrand, D. Morse, M. Cappadocia, Hypervariable domains of self-incompatibility RNases mediate allele-speci¢c pollen recognition, Plant Cell 9 (1997) 1766^1767. [24] K. Ohgi, H. Horiuchi, M. Iwama, H. Watanabe, M. Takagi, M. Irie, Evidence that three histidine residues of a base nonspeci¢c and adenylic acid preferential ribonuclease from Rhizopus niveus are involved in the catalytic function, J. Biochem. 112 (1992) 132^138. [25] J. Steyaert, A decade of protein engineering on ribonuclease T1, atomic dissection of the enzyme-substrate interactions, Eur. J. Biochem. 247 (1997) 1^11.
[26] P.J. Kraulis, MOLSCRIPT: a program to produce both detailed and schematic plots of protein structures, J. Appl. Cryst. 24 (1991) 946^950. [27] E.A. Merrit, M.E.P. Murphy, Raster3D version 2.0, a program for photorealistic molecular graphics, Acta Cryst. D50 (1994) 869^873. [28] W. Kabsch, C. Sander, Dictionary of protein secondary structure pattern recognition of hydrogen-bonded and geometrical features, Biopolymers 22 (1983) 2577^2637.
BBAPRO 35947 2-8-99