Large-Scale Domain Movements and Hydration Structure Changes in

Here we describe the large-scale domain movements and hydration structure changes in the active-site cleft of unligated glutamate dehydrogenase. Gluta...
0 downloads 0 Views 2MB Size
Biochemistry 2001, 40, 3069-3079

3069

Large-Scale Domain Movements and Hydration Structure Changes in the Active-Site Cleft of Unligated Glutamate Dehydrogenase from Thermococcus profundus Studied by Cryogenic X-ray Crystal Structure Analysis and Small-Angle X-ray Scattering†,‡ Masayoshi Nakasako,*,§,| Testuro Fujisawa,| Shin-ichi Adachi,| Toshiaki Kudo,*,⊥ and Sadaharu Higuchi⊥ PRESTO, Japan Science and Technology Corporation and The Institute of Molecular and Cellular Biosciences, The UniVersity of Tokyo, Yayoi 1-1-1, Bunkyo-ku, Tokyo 113-0032, Japan, and The Harima Institute/SPring-8, RIKEN (The Institute of Physical and Chemical Research), Sayo-gun, Hyogo 679-5143, Japan, and RIKEN (The Institute of Physical and Chemical Research), Hirosawa 2-1, Wako, Saitama 351-0198, Japan ReceiVed October 26, 2000; ReVised Manuscript ReceiVed January 17, 2001

ABSTRACT: Here we describe the large-scale domain movements and hydration structure changes in the active-site cleft of unligated glutamate dehydrogenase. Glutamate dehydrogenase from Thermococcus profundus is composed of six identical subunits of Mr 46K, each with two distinct domains of roughly equal size separated by a large active-site cleft. The enzyme in the unligated state was crystallized so that one hexamer occupied a crystallographic asymmetric unit, and the crystal structure of the hexamer was solved and refined at a resolution of 2.25 Å with a crystallographic R-factor of 0.190. In that structure, the six subunits displayed significant conformational variations with respect to the orientations of the two domains. The variation was most likely explained as a hinge-bending motion caused by small changes in the main chain torsion angle of the residue composing a loop connecting the two domains. Small-angle X-ray scattering profiles both at 293 and 338 K suggested that the apparent molecular size of the hexamer was slightly larger in solution than in the crystalline state. These results led us to the conclusion that (i) the spontaneous domain motion was the property of the enzyme in solution, (ii) the domain motion was trapped in the crystallization process through different modes of crystal contacts, and (iii) the magnitude of the motion in solution was greater than that observed in the crystal structure. The present cryogenic diffraction experiment enabled us to identify 1931 hydration water molecules around the hexamer. The hydration structures around the subunits exhibited significant changes in accord with the degree of the domain movement. In particular, the hydration water molecules in the active-site cleft were rearranged markedly through migrations between specific hydration sites in coupling strongly with the domain movement. We discussed the cooperative dynamics between the domain motion and the hydration structure changes in the active site of the enzyme. The present study provides the first example of a visualized hydration structure varying transiently with the dynamic movements of enzymes and may form a new concept of the dynamics of multidomain enzymes in solution.

Proteins fold and work in water, their natural medium. Because water is a complex fluid with unusual physical properties caused by hydrogen bonds (1), it has great influences on the dynamics of proteins. Therefore, the interface between water and proteins, the hydration structure, † This work was supported by grants-in-aid from the Ministry of Education, Science, Sports and Culture of Japan and the grants for Biodesign and the SR Structural Biology Research Programs from RIKEN. The SAXS experiment was performed under an approval of the organizing committee of SPring-8 (proposal no. 1999B0056-NL-np). ‡ The atomic coordinate is deposited to the Protein Data Bank under the accession code of 1EUZ. * To whom correspondence should be addressed. (M.N.) Phone: (81)-3-5841-8493. Fax: (81)-3-5841-8493. E-mail: nakasako@ iam.u-tokyo.ac.jp. (T.K.) Phone: (81)-48-467-9544. Fax: (81)-48-4624672. E-mail: [email protected]. § Japan Science and Technology Corporation and The University of Tokyo. || The Harima Institute/SPring-8, RIKEN (The Institute of Physical and Chemical Research). ⊥ RIKEN (The Institute of Physical and Chemical Research).

is the subject of much discussion to understand how proteins fold and work in water (2). Hydration structures of proteins have been investigated using various experimental techniques and theoretical simulations, with respect to the amount, the geometry and the influences on the dynamics and stability of proteins (3-8). Recently, cryogenic X-ray crystallography has provided much information on the static pictures of the hydration structures (9, 10). Perhaps, the most interesting result is that proteins are wrapped by hydration shells stabilized through large-scale networks of hydrogen bonds (9). This static picture implies that the dynamic movements in proteins at work require a concerted reorganization of their hydration structures. Although information on the hydration structures is still limited in the static case, snapshots of transiently varying hydration structures around proteins at work are required to describe completely how proteins work in water. For experimentally visualizing transient hydration structures, good targets are multidomain enzymes exhibiting large-

10.1021/bi002482x CCC: $20.00 © 2001 American Chemical Society Published on Web 02/15/2001

3070 Biochemistry, Vol. 40, No. 10, 2001

Nakasako et al.

FIGURE 1: (A) Stereoview of hexameric GluDH from Tp in a crystallographic asymmetric unit. The six subunits are shown as ribbon models and are distinguished by their colors (subunit1, cyan and blue; subunit2, pink; subunit3, green; subunit4, indigo; subunit5, red; subunit6, yellow). The secondary structures in the ribbon model were defined using the DSSP program (47). The colors of C-domain and N-domain are distinguished only in subunit 1. The small spheres indicate the positions of hydration water molecules identified (see also panel B). The bar in the right side corresponds to 50 Å. The same coloring-scheme for the subunits is used for all the figures. (B) A stereoplot showing the distribution of hydration water molecules around subunit 1. The hydration water molecules interacting directly with only subunit 1 are presented as small red spheres. The orange fishnets superimposed on the ribbon model are omit-annealed difference electron density maps of hydration water molecules calculated with the reflections between the Bragg spacings of 8.0 and 2.25 Å and contoured at the 4.0σ level. The residues forming the active site are shown by stick models.

scale and spectacular movements of domains between their unligated and ligated states (11-13). The movements are indispensable to the biological functions, such as substrate binding and catalysis. As observed in T4-lysozyme (1416) and calmodulin (17), multidomain enzymes exhibit spontaneous domain movements in their unligated state, and crystallization of the enzyme sometimes traps the transient state of the movement. Cryogenic X-ray crystal structure analyses for such crystals provide the trapped transient hydration structures as well as the trapped conformational state in the domain movements. In the present study, we selected hexameric glutamate dehydrogenase (GluDH)1 as a research target. GluDH is a multidomain enzyme catalyzing the reaction from glutamate to 2-oxyoglutarate and ammonia in the presence of the

cofactor NAD or NADH (18), and the hexameric form (Figure 1 A) of GluDH is the biologically functional unit in some organisms (19). The subunit of hexameric GluDH consists of two separate domains of Mrs 23K: the nucleotide binding domain (N-domain) and the core domain engaging in the hexamer formation (C-domain) (Figure 1B) (20, 21). Between the two domains, the active-site cleft harbors, and the substrate-binding site is located at the depth of the cleft (Figure 1B). The N-domain is known to exhibit a large-scale movement between the unligated and ligated states (20, 21). In addition, crystal structures of unligated GluDH from 1 Abbreviations: GluDH, glutamate dehydrogenase, rms, root-mean square; Rg, radius of gyration; SAXS, small-angle X-ray scattering; Tp, Thermococcus profundus; Pf, Pyrococcus furiosus; Tm, Thermotoga maritima; Tl, Thermococcus litoralis; Cs, Clostridium symbiosum.

Large-Scale Domain Motion in Glutamate Dehydrogenase various organisms and mutants exhibited variety in the orientations of the two domains (20, 22-29), indicating spontaneous motion of the N-domain in solution like T4lysozyme (16) and calmodulin (17). Mutational studies also supported the spontaneous fluctuation of the N-domain in the unligated state (29). However, no direct structural evidence has been reported for GluDH. In addition, as well as other multidomain enzymes, it is unknown what causes the domain motion and how the motion couples with the hydration water molecules surrounding the enzyme. In the present study, the crystal structure of hexameric GluDH from hyperthermophile Thermococcus profundus (30, 31) was determined at 110 K, including the large number of hydration water molecules. The motions of the N-domains in the six subunits were trapped differently in the crystallization process, and the hydration structures exhibited significant variations, in particular, in the active-site cleft. In addition, small-angle X-ray scattering (SAXS) profiles indicated the motion of the N-domain to be of larger magnitude in solution than in the crystalline state. Here we report the structural changes in both the enzyme and the hydration and discuss the possibility that a few hydration water molecules in the active site directly couple with the domain motion. The present report is the first to reveal that the hydration structures exhibit cooperative variation with the large-scale dynamical motion of enzymes at high resolution. MATERIALS AND METHODS Crystal Structure Determination. The purification and crystallization of recombinant GluDH of Tp were carried out as described previously (31, 32). The crystals obtained belonged to space group P21 with the lattice constants of a ) 112.99, b ) 163.70, c ) 133.07 Å, and β ) 113.46° at 110 K, and one hexamer of GluDH occupied a crystallographic asymmetric unit. The diffraction intensity data were collected by the oscillation method at the BL44B2 beamline (33) of SPring-8 as previously reported (32). The programs DENZO and SCALEPACK (34) were used to process the diffraction intensity data (Table 1). The reflections were collected up to the resolution of 2.0 Å, and those up to 2.25 Å were used in the subsequent analysis, because the RImerge value was higher than 0.30 and the I/σ value lower than 3.0 beyond the Bragg spacing of 2.25 Å (Table 1). The crystal structure of the GluDH hexamer was solved by the molecular replacement method using X-PLOR (35). The search model used for the analysis was constructed from the structural model of GluDH of Pyrococcus furiosus (Pf) (22) [the accession code is 1GTM in the Protein Data Bank (36)]. Through the rotation search followed by the Pattersoncorrelation refinement (37) and the subsequent translation search, one prominent solution was found. However, the structural models of the N-domains in subunits 1, 2, and 5 did not match with the calculated electron density maps, suggesting different quaternary structures in those subunits with the other. A rigid body refinement corrected the position of the N-domains and gave a nice match between the models and the calculated electron density maps. Because of this variation, we did not apply the noncrystallographic 32symmetry constraint to the hexamer in the following structure refinement.

Biochemistry, Vol. 40, No. 10, 2001 3071 Table 1: Statistics of the Diffraction Data and the Refined Structural Model data collection resolution (Å) no. of observed reflections no. of unique reflections completeness (%) I/σ I b Rmerge refinement resolution (Å) unique reflections (F > 2σ) R-factorc R-freed structure model protein (non-hydrogen atoms) sulfate ions hydration water molecules bonds (Å)e angle (deg)e ion pair statisticsf no. of ion pairs per hexamer no. of ion pairs per residue % of charged residues forming ion pairs % of ion pairs formed by Arg/Lys/His % of ion pairs formed by Glu/Asp

100.0-2.25 1 488 939 206 382 99.7 (99.4)a 17.9 (3.4)a 0.063 (0.322)a 8.0-2.25 183 797 0.188 0.250 19 250 16 1931 0.011 1.64 222 0.09 47 74/26/0 48/52

I The highest resolution shell is from 2.33 to 2.25 Å. b Rmerge ) ∑h∑l|Ii(h) - 〈I(h)〉|/∑h∑lIi(h), where Ii(h) is the intensity of ith observation of reflection h. c R ) ∑h|Fobs(h) - Fcalc(h)|/∑hFobs(h), where Fobs(h) and Fcalc(h) are the observed and calculated structure factors of reflection h, respectively. d R-free factor was calculated for 10% of unique reflections, which were not used in the structure refinement throughout (45). e Root-mean-square deviation from ideal stereochemical geometry. f The ion pair statistics were calculated by using the criteria of Barlow and Thornton (46). After the calculation, the geometry of the ion pairs was reexamined manually. a

The structure refinement and model building were carried out with X-PLOR (35) and turbo FRODO (Biographics), respectively. The conventional protocol, including the simulated annealing, was applied under the stereochemical parameters proposed by Engh and Huber (38). The suit of program FESTKOP (9) carried out the picking-up of hydration water molecules from difference Fourier electron density maps. The hydration water molecules included in the model had thermal factors of less than 70 Å2 and exhibited electron density peaks of more than 3.5σ in their omit-annealed difference Fourier electron density maps throughout the subsequent refinement rounds. The final statistics of the refined hexamer model were summarized in Table 1. The program DynDom (12) was used for analyzing the domain movement. The FESTKOP program (9) was used for systematically analyzing the first hydration shell. In the present study, the lower and the upper limits of the hydrogen bond distance were set at 2.4 and 3.4 Å, respectively. Small-Angle X-ray Scattering Experiment. A small-angle X-ray scattering experiment was performed at BL45SX (39) of SPring-8. The X-ray wavelength was tuned to 1.0000 Å and the camera distance was set at 2200 mm. The CCD detector combined with an image intensifier (40) was used for recording the scattering pattern, and the exposure time was 1 s for each measurement. The circular averaging procedure (40) reduced the two-dimensionally recorded SAXS profiles. The scattering experiments were carried out both at 293 and 338 K using a temperature controlled sample cell holder. Scattering profiles of sample solutions and buffers were

3072 Biochemistry, Vol. 40, No. 10, 2001 alternately measured to avoid systematic errors in the data analysis. To correct the concentration effect in the SAXS profile, the concentration of the enzyme solution was varied from 1.0 to 3.0 mg/mL in increments of 0.5 mg/mL. The concentration effect and the radius of gyration (Rg) were analyzed by using the suite of programs, iisgnapr (40). The pair-distribution function P(r) was calculated by using the program GNOM (41). The program DEBYE (Nakasako, unpublished work) was used for the theoretical calculation of the scattering curve, the P(r) function and the Rg value of the crystal structure. RESULTS Variety in the Quaternary Structures between the Six Subunits. Figure 1A shows the overall structure of the GluDH hexamer with 1931 hydration water molecules. The ion-pair statistics of the present structural model were very similar to those for GluDH from Thermotoga maritima (Tm) (24), and the thermostability of this enzyme will be discussed elsewhere. In the crystal lattice, the hexamers interact with adjacent ones through only 14 hydrogen bonds (