Polysaccharide−Polynucleotide Complexes. Part 32. Structural

Department of Chemical Processes and Environments, Faculty of Environmental Engineering, The University of Kitakyushu, 1-1 Hibikino Wakamatsu-ku, Kita...
1 downloads 6 Views 488KB Size
Biomacromolecules 2005, 6, 1540-1546

1540

Polysaccharide-Polynucleotide Complexes. Part 32. Structural Analysis of the Curdlan/Poly(cytidylic acid) Complex with Semiempirical Molecular Orbital Calculations Kentaro Miyoshi,† Kazuya Uezu,*,† Kazuo Sakurai,† and Seiji Shinkai‡ Department of Chemical Processes and Environments, Faculty of Environmental Engineering, The University of Kitakyushu, 1-1 Hibikino Wakamatsu-ku, Kitakyushu, Fukuoka 808-0135, Japan, and Department of Chemistry and Biochemistry, Graduate School of Engineering, Kyushu University, Fukuoka 812-8581, Japan Received December 10, 2004; Revised Manuscript Received February 20, 2005

Natural Curdlan adopts a right-handed 61 triple helix, in which the constituting glucan chains are underpinned with each other by the intermolecular hydrogen bonds. Curdlan can form a stoichiometric complex with polynucleotides [e.g., poly(cytidylic acid), poly(C)]. In this paper, we carried out the MOPAC (semiempirical molecular-orbital package) calculation to examine the molecular structure of the Curdlan/poly(C) complex. The calculation exhibited that two types of hydrogen bonds are formed between the Curdlan and the poly(C); the third nitrogen (N3) in cytosine forms a hydrogen bond with the second OH of one Curdlan chain, and the proton of N4 is interacting with the O2 of another Curdlan chain. In our model, the helix diameter of poly(C) is expanded from 11.0 to 15.3 Å upon complexation. Despite such large conformational changes, the 61 helix structure of poly(C) was maintained even after the complexation. This fact is complementary to the experimental fact that the complexation does not change the band shape of the circular dichroism of poly(C). The chain length dependence of the reaction enthalpy indicated that the complexation becomes thermodynamically more favorable with the chain length increasing. This feature is also consistent with the experimental data. Introduction The β-1,3-D-glucan family is a very abundant cell wall polysaccharide in fungal species.1 Among these polysaccharides, Curdlan is very simple structurally since it only contains linear 1,3-linked repeat units with no 1,6-linked side chain glucosyl units as shown in Figure 1a.1-3 Curdlan takes a right-handed 61 triple helix with a 17.6-Å pitch.4 The triple helix can be dissociated to three single chains by dissolving in dimethyl sulfoxide (DMSO), by increasing the pH above 13, or by heating aqueous Curdlan solutions above 130 °C in an autoclave.1 When water is added to the Curdlan/DMSO solutions or the hot Curdlan aqueous solutions are cooled, gelation can be induced, being similar to other β-1,3-Dglucans such as schizophyllan.5 The gelation mechanism of β-1,3-D-glucans is described as follows: when the single chain is returned to aqueous conditions, partial renaturation of the triple helix can take place between the different chains and the resulting renatured part can play a role at the crosslinking point for the gelation. Brant et al.6,7 and Young et al.8,9 independently showed that when the renaturation process was carried out in considerably dilute solutions (ca. 4-30 µg/mL), the formation of the original triple helix can dominantly occur. In fact, the X-ray crystallography for Curdlan was carried out for the renatured sample from the * To whom correspondence [email protected]. † The University of Kitakyushu. ‡ Kyushu University.

should

be

addressed.

E-mail:

DMSO solution and provided the same results as those of native samples (i.e., triple helix).4 Recently, we found that when the single chain of Curdlan is renatured together with a homo nucleic acid [e.g., poly(cytidylic acid), poly(C); poly(adenylic acid), poly(A); and poly(deoxyadenylic acid), poly(dA)], the Curdlan and polynucleotide chains form a macromolecular complex, instead of forming the Curdlan triple helix.10,11 One of the novel features for this phenomenon is that the complexation occurs in a highly stoichiometric manner and the stoichiometric number indicates that two glucose units in the Curdlan main chain are interacting with one base, as presented in Figure 1b.11 Electron microscopy and X-ray fiber diffraction pattern showed that the complex takes the same rodlike architecture as the original triple helix and the helicity and pitch of the complex coincide with those of the original triple helix.12,13 The structural data suggest that the complex takes a triple helix that can be obtained by replacing one glucan chain in the Curdlan triplex with the polynucleotide chain. Furthermore, our spectroscopic study14 showed that the hydrogen bonding is essential for the complexation. It is believed that one Curdlan strand forms the intermolecular hydrogen bonds with the other two strands in the same x-y planes (z axis is the parallel to the helical axis) along the helical axis, resulting in an inequilateral hexagonal shape perpendicular to the helical axis.15 Recently, Miyoshi et al. re-examined how the hydrogen bonds stabilize the Curdlan triple helix, using a semiempirical molecular-orbital

10.1021/bm049218u CCC: $30.25 © 2005 American Chemical Society Published on Web 04/07/2005

Structural Analysis of the Curdlan/Poly(C) Complex

Biomacromolecules, Vol. 6, No. 3, 2005 1541

Figure 1. Repeating unit of β-1,3-D-glucan and a stoichiometric model for the Curdlan/poly(C) complex. In panel a, the three characteristic angles to determine the glucan conformation are presented, and in panel b, G and B represent the glucose and base moieties, respectively.

Figure 2. Curdlan molecular model proposed by our previous paper,16 from which we constructed a complex model. (a) A side view for the full turn triple helix; here the dotted lines indicate the hydrogen bonds. (b) A cross-sectional view of the helix (perpendicular to the helix). For convenience, in panel b, only two glucoses in each chain are presented. The glucose rings marked with the same number belong to the same chain. The bold lines denote that the atoms are located higher along the helix compared to the atoms indicated by the thin lines. Blue broken wedges indicate a noncoplanar hydrogen bond connected along the helix, traversing three Curdlan chains to make a left-handed helix.

package (MOPAC) and ab initio calculations.16 They proposed other intermolecular hydrogen bonds that are energetically more favorable than the traditional one. Figure 2 shows the new model proposed by Miyoshi et al. In this model, three Curdlan single chains are connected by the intermolecular hydrogen bonds formed between the oxygen atoms (O2 and O2′) in the different chains and on different x-y planes. These intermolecular hydrogen bonds form the consecutive O-H-O array to cross the different chains. In consequence, the helicity of the hydrogen bonding array is left-handed as illustrated by the dotted line in Figure 2. In this configuration, the delocalization energy induced by hydrogen-bond cooperativity17 makes the triplex of the β-1,3D-glucans more stable. Therefore, we decided to use this model to construct a molecular model for the complex. We have examined physicochemical properties for the Curdlan/polynucleotide, and other related complexes and many novel features have been clarified, so far. However,

only a few structural studies are available, because of complexity of this phenomenon. For researching the structural details of the complex, in this study, we explored the molecular structure of the Curdlan/poly(C) complex with modern computational chemistry. The structural flexibility of ribonucleotides is lower than that of deoxyribonucleotides; hence, we employed poly(C) that has an orderly structure and many experimental results of the complex among other poly-ribonucleotides. Computational Methods and Assumptions When we constructed an initial structure of the complex, we started from the Curdlan model of Miyoshi et al.16 Their model used exactly the same atomic coordinates as Deslandes et al.,4 except for the hydrogen-atomic coordinates in the second hydroxyl group. As an initial structure for poly(C), we used the crystallographic data of Arnott et al. for the

1542

Biomacromolecules, Vol. 6, No. 3, 2005

Miyoshi et al.

Figure 3. Calculated molecular model for the Curdlan/poly(C) complex. Curdlan chains are drawn in translucent colors, and poly(C) chain is emphasized. (a) Side view and (b) top view. The arrows indicate the end cytosine moiety which tends to be segregated from the complex.

single-stranded poly(C).18 The pitches of Curdlan and poly(C) are 17.6 and 18.6 Å, respectively. Our previous studies indicated that the Curdlan/poly(C) complex consists of two Curdlan single chains and one poly(C) chain, and the helix parameters (i.e., pitch and number of residues in the pitch) are almost maintained in the complex. To construct such a model, we removed one glucan chain from the Curdlan triple helix and tried to fit one poly(C) chain to the groove occupied by the taken glucan. The reasonable orientation of polynucleotide in the complex was the reducing ends of the glucan strands and 3′ end of the polynucleotide orienting in the same direction as our previous experimental and calculation results of the Curdlan/poly(dA) complex. Hence, we arranged the 3′ end of the poly(C) in the same direction of the reducing ends of the Curdlan strands. Hydrogen bonds could be formed between the cytosine of poly(C) and the second OH in the Curdlan main chain; therefore, we located the poly(C) base close to the second OH to make the base face toward the inside of the helix. This hypothetical triple helix was the initial structure for the calculation. One may suppose that the final structure should depend on the initial one. In fact, when we started the calculation from the structure which was heavily deviated from the above-mentioned one, the pre-optimization (see below) led to dissociation of the complex. Before the calculation for the Curdlan/poly(C) complex, we confirmed the accuracy of the MOPAC calculation for our system by geometry-optimizing the Curdlan triple helix and the poly(C) single chain models from the X-ray crystallographic data.4,18 We carried out pre-optimization for the Curdlan/poly(C) complex model with the molecular mechanics method using MM3 in CAChe, version 5.0 (Fujitsu, Ltd., Japan).19 For the resultant complex, the MOPAC calculation was done with the MOPAC200220 in CAChe, version 5.0. To avoid unfavorable repulsions

between the phosphate anions and/or to simulate electric shielding effects by salts in the experiment, we added the same number of oxonium cations as that of the phosphate anions. All the atoms were allowed to move during the calculation. To simulate the molecules in water, we used the conductor-like screening model (COSMO), setting a dielectric constant of 78.4 at 298 K. In this work, the AM1 method was employed because of accuracy to describe hydrogen bonds.21 For the other parameters, the default values in MOPAC were used. As a result of limitation of our computer power, we could only calculate up to 1600 atoms, corresponding to the 21 residues. According to the previous experimental data, we need a certain length of the chains to stabilize the complex, approximately more than 30 residues. Thus, we evaluated the chain length dependence of the reaction enthalpy, with changing the number of residues from 9 to 21. Results and Discussion Structural Characteristics of the Curdlan/Poly(C) Complex and Hydrogen Bonding. Figure 3 presents the geometry-optimized structure of the Curdlan/poly(C) complex (total of 15 residues, 2.5 pitches). The MOPAC evaluated the pitch to be 18.0 Å, which is 0.4 Å longer than the Curdlan triple helix and 0.6 Å shorter than the poly(C) single helix. The calculation compensates for the difference in the initial pitches; that is, the Curdlan helix extends and the poly(C) helix compresses to fit with each other. Despite these deformations, both chains [Curdlan and poly(C)] maintain their characteristic 61 helix structure. As indicated by the arrows in Figure 3, the terminal cytidine moiety tends to segregate from the helix. This end effect can be ascribed to the fact that there is no pairing segment for the terminal cytidine to push it toward the helix. Calculations for such short chains may not give reliable results on this point. Thus,

Structural Analysis of the Curdlan/Poly(C) Complex

Biomacromolecules, Vol. 6, No. 3, 2005 1543

Figure 5. Top view comparison for the poly(C) conformation between the isolated chain (a) and that in the complex (b).

Figure 4. Hydrogen bonds between Curdlan and poly(C) in the complex. (a) Two types of the hydrogen bond are formed; the third nitrogen (N3) in cytosine forms a hydrogen bond with the second OH of one Curdlan chain, and the proton of N4 is interacting with the O2 of the other Curdlan chain. Panel b illustrates how these hydrogen bonds connect the three chains.

we should have calculated Curdlan/poly(C) models long enough not to be influenced by the end effect. However, this requirement is difficult to attain, owing to the limit of our computer power. The chain length dependence will be discussed in a later section. Polysaccharide conformations are essentially determined by two torsional and one bridge angle around the glycosidic bond, and these key parameters are defined by φ(H1-C1O3′-C3′), ψ(C1-O3′-C3′-H3′), and τ(C1-O3′-C3′) as illustrated in Figure 1a. After MOPAC geometry optimization, these parameters slightly changed from the initial values,4 that is to say, from φ ) 29.1°, ψ ) 9.6°, and τ ) 110.4° to φ ) 27.3°, ψ ) 17.7°, and τ ) 113.4°. The conformational energy map for the β-1,3-D-glucan triple helix was already calculated by Blehm and Sarko.22 When compared with their results and the newly obtained parameter set, our results are within the reasonable range in their map. This coincidence indicates that the Curdlan chain in the complex is not deformed so much. The original Curdlan triple helix has the left-handed intermolecular hydrogen bonds formed between the oxygen atoms on different x-y planes.16 After the complexation, the remaining two Curdlan chains form hydrogen bonds between O2 and O2′ in the same manner as those in the original triple helix. However, the O2-O2′ distance is changed from 3.3 to 2.9 Å. This means that the Curdlan helices come close to each other. The closer distance causes the more stable hydrogen bond; thus, the Curdlan double helix in the complex becomes more stable. As a result of this change, more room is created to accommodate the poly(C) chain between the Curdlan chains. In other words, the complexed poly(C) seems to squeeze the remaining two Curdlan chains to be closer with each other than those in the original Curdlan. Figure 4 presents how the cytosine forms hydrogen bonds with the Curdlan glucoses. The third nitrogen (N3) in

cytosine is an electron-donating group and forms the hydrogen bond with the second OH of Curdlan (Curdlan chain II in the Figure 4a). The proton of N4 is interacting with the O2 of the other Curdlan (Curdlan chain I in Figure 4a). In Figure 4b, comparing this model with the Curdlan triple helix, it is found that the N3 atom exists nearly at the same position as the O2 of the removed Curdlan (Curdlan chain III) glucose. Therefore, the original hydrogen bonding array which had maintained the Curdlan triple helix is replaced by a hybrid array which consists of the glucose versus glucose hydrogen bond and the cytosine versus glucose hydrogen bond. Conformational Changes in Poly(C): Ribose is Deformed upon the Complexation. Figure 5 compares the conformation of the original poly(C) reported by Arnott et al.18 with that of the complex. The figure shows that poly(C) chain deforms in a large extent upon the complexation. The diameter of the helix is 11.0 Å in the original chain, while it is 15.3 Å in the complex. Furthermore, the pitch is decreased by 0.8 Å as described above. According to our previous reports,14,23 the circular dichroism (CD) spectrum is increased upon the complexation. Since there is no absorbance in Curdlan, this CD change is ascribed to the conformational change in poly(C), especially to the exciton coupling between the cytosines involved in helical conformation.10,24 Therefore, the conformational change in poly(C) is directly related to a change in CD upon the complexation. The complexation does not create any new spectrum, it only doubles the positive band around 275 nm.14,23 This means that the spatial arrangement between the adjacent cytosines is not drastically altered upon the complexation. The calculation shows that the complexation does not change the angle between the electric transition moments of the cytosines and the 61 helix structure is maintained. These structural features are consistent with the experimental results obtained from CD measurements. Polynucleotide conformations can be described with combination of the ribose puckering and the backbone torsional angles.11,25 The ribose in poly(C) is inherently nonplanar, and the nonplanarity is termed as puckering. The precise conformation of the ribose can be specified by the pseudorotation (P) defined by eq 1. As expressed by the

1544

Biomacromolecules, Vol. 6, No. 3, 2005

Miyoshi et al.

Figure 6. Comparison of the puckering parameters P between the isolated and complexed poly(C). (a) Definitions of five internal torsional angles in ribose and (b) the pseudorotation wheel. The dilute and dark gray areas indicate the preferred and calculated ranges of P. Typical ribose conformations for each range are illustrated.

Figure 7. Comparison of the backbone torsional angles between the isolated and complexed poly(C) using the conformational wheel. (a) Definitions of six backbone torsional angles in a unit nucleotide and (b) the conformational wheel. The definitions of colors are identical with those in Figure 6.

equation, the five endocyclic torsional angles τ0 - τ4 (see Figure 6a) are related to P. tan P )

(τ4 + τ1) - (τ3 + τ0) 2τ2(sin 36° + sin 72°)

(1)

P is commonly represented by the pseudorotation wheel, which indicates the continuum of ring puckers (Figure 6b). The dilute gray area in Figure 6b indicates the preferred range of P for the ribose conformations for RNAs.25 The preferred range means that most ribose structure determined by crystallography can be represented by these P values. Our calculation for the poly(C) in the complex provides the P values shown by the dark gray area. The dilute and dark gray areas are not overlaid, indicating that the ribose of the poly(C) is distorted upon the complexation. However, comparing HOFs between the ribose in the crystal state and the ribose in the complex, the loss of the thermodynamic

stability by this change of ribose was small. A typical ribose conformation for each case is presented in the figure, and these are called 3′-endo and 2′-exo, respectively. The phosphodiester backbone of a polynucleotide has six torsional angles R, β, γ, δ, , and ζ (Figure 7a) in addition to the five internal sugar torsional angles τ0 - τ4. The backbone cannot rotate freely due to the steric hindrance, and these bond angles are restricted within certain ranges. Figure 7b presents a conformational wheel to show these values. Again, the dilute and dark gray areas mean the reported values and those in the complex. The backbone angles of the poly(C) are approximately consistent with the normally observed range, indicating that the deformation of the backbone was not so large. We evaluated how the HOF of the poly(C) molecule was changed by the conformational change upon the complexation. We took the poly(C) out of the complex without changing the conformation and evaluated the HOF for this

Biomacromolecules, Vol. 6, No. 3, 2005 1545

Structural Analysis of the Curdlan/Poly(C) Complex

that of 12-unit complex since the end effect was significantly observed. Based upon extrapolation of this plot to larger molecules, it appears that the ∆H/unit will go negative between 25 and 30 units and the complex becomes thermodynamically favorable. With increasing chain length and number of hydrogen bond linkages, the structure of the complex is stabilized. We experimentally observed the chain length dependence of the stability for Curdlan/poly(dA) and other related complexes.10,26 When the β-1,3-D-glucan chain is long enough, the critical number of bases to form the complex is around 40 units;26 when the polynucleotide chain is long enough, the critical glucose number is about 50.10 Conclusion and Remarks

Figure 8. Chain length dependence of the standard enthalpy change (∆H) per the structural unit (two glucoses and one cytosine). Here, ∆H/unit is defined by dividing the obtained ∆H by the number of the units (see the text). The illustration schematically demonstrates the thermodynamic parameters and why the coefficients are needed.

structure. As the reference, we constructed a poly(C) chain with 15 residues according to Arnott et al.18 The comparison of these two HOF values showed that the poly(C) with 15 bases in the complex is less stable by 186 kcal/mol than the poly(C) single strand. This means that the energy loss of 12 kcal/mol is increased per one base-unit, which is small enough to be overwhelmed by other favorable interactions such as hydrogen bonds. Chain Length Dependence of Reaction Enthalpy. The standard enthalpy change (∆H) can be defined for the complexation by the following: ∆H ) 3Hcomplex - 3Hpoly(C) + 2HCurdlan3

(2)

where Hcomplex, Hpoly(C), and HCurdlan3 are HOF values for the Curdlan/poly(C) complex, the poly(C) single strand, and the Curdlan triple strands after geometry optimizations, respectively. The coefficients in the equation are necessary to keep stoichiometry. In Figure 8, the illustration schematically demonstrates these thermodynamic parameters and why the coefficients are needed. We can consider a set of two glucoses and one cytosine as an elemental unit. Therefore, we divided the obtained ∆H by the number of the unit and denoted it by ∆H/unit. With this normalization, we can compare ∆H among the different chain lengths. Figure 8 plots ∆H/unit against the number of the units (i.e., chain length). When ∆H/unit is positive, this means that the complexation is thermodynamically unfavorable. The ∆H/ unit decreased with increasing the chain length over 12 units. However, the ∆H/unit of the 9-unit complex is lower than

This paper studied molecular configurations of Curdlan/ poly(C) complex by MOPAC calculation. In the calculated model, the poly(C) chain still took a right-handed 61 helix, similar to that of the original single chain. This fact is consistent with our CD experiments. The complexation expanded the helix diameter from 11.0 to 15.3 Å. This conformational change was related to both changes in the ribose puckering and changes in the backbone conformation. The chain length dependence of the reaction enthalpy suggested that the complexation should become thermodynamically favorable above a certain chain length. We believe that, since the present calculation provides a clear-cut view of the structural aspects for the complex, this model should provide fundamental ideas to understand this novel complexation. Acknowledgment. This work has been financially supported by SORST program of Japan Science and Technology Corporation (JST), and K.S. is thankful to PRESTO “Function and Organization” program. References and Notes (1) Biopolymers, Polysaccharides II: Polysaccharides from Eukaryotes; Wiley-VCH: Weinheim, 2002; Vol. 6. (2) Harada, T.; Misaki, A.; Saito, H. Arch. Biochem. Biophys. 1968, 124 (1), 292-298. (3) Koreeda, A.; Harada, T.; Ogawa, K.; Sato, S.; Kasai, N. Carbohydr. Res. 1974, 33 (2), 396-399. (4) Deslandes, Y.; Marchessault, R. H.; Sarko, A. Macromolecules 1980, 13, 1466-1471. (5) Stokke, B. T.; Elgsaeter, A.; Kitamura, S. Polym. Gels Networks 1994, 2, 173-190. (6) Stokke, B. T.; Elgsaeter, A.; Brant, D. A.; Kuge, T.; Kitamura, S. Biopolymers 1993, 33 (1), 193-198. (7) McIntire, T. M.; Brant, D. A. J. Am. Chem. Soc. 1998, 120 (28), 6909-6919. (8) Young, S. H.; Dong, W. J.; Jacobs, R. R. J. Biol. Chem. 2000, 275 (16), 11874-11879. (9) Young, S. H.; Jacobs, R. R. Carbohydr. Res. 1998, 310, 91-99. (10) Kimura, T.; Koumoto, K.; Sakurai, K.; Shinkai, S. Chem. Lett. 2000, 1242-1243. (11) Koumoto, K.; Kimura, T.; Kobayashi, H.; Sakurai, K.; Shinkai, S. Chem. Lett. 2001, 908-909. (12) Bae, A.; Lee, S.; Ikeda, M.; Sano, M.; Shinkai, S.; Sakurai, K. Carbohydr. Res. 2004, 339 (2), 251-258. (13) Sakurai, K.; Koumoto, K.; Shinkai, S. Macromolecular Nanostructured Materials; Kodansha, Ltd.: Tokyo, 2004. (14) Sakurai, K.; Mizu, M.; Shinkai, S. Biomacromolecules 2001, 2 (3), 641-650. (15) Atkins, E. D. T.; Parker, K. D. Nature 1968, 220, 784-785. (16) Miyoshi, K.; Uezu, K.; Sakurai, K.; Shinkai, S. Chem. BiodiVersity 2004, 1, 916-924.

1546

Biomacromolecules, Vol. 6, No. 3, 2005

(17) Jeffrey, G. A.; Saenger, W. Hydrogen Bonding in Biological Structures; Springer-Verlag: Berlin, 1991. (18) Arnott, S.; Chandrasekaran, R.; Leslie, A. G. W. J. Mol. Biol. 1976, 106, 735-748. (19) CAChe Release 5.0; Serial No. 150270302; Fujitsu, Ltd.: Chiba, Japan, 2002. (20) Stewart, J. J. P. MOPAC Manual, 6th ed.; United States Air Force Academy: Colorad Springs, 1990. (21) Dewar, M.; Xoebisch, E. G.; Healy, E. F.; Stewart, J. J. P. J Am. Chem. Soc. 1985, 107, 3902-3909. (22) Blehm, T.; Sarko, A. Can. J. Chem. 1977, 55, 293-299.

Miyoshi et al. (23) Sakurai, K.; Shinkai, S. J. Am. Chem. Soc. 2000, 122 (18), 45204521. (24) Circular Dichroism: Principles and Applications, 2nd ed.; John Wiley & Sons, Inc.: New York, 2000. (25) Saenger, W. Principles of Nucleic Acid Structure; Springer-Verlag: Tokyo, 1987. (26) Mizu, M.; Kimura, T.; Koumoto, K.; Sakurai, K.; Shinkai, S. Chem. Commun. 2001, 429-430.

BM049218U