J. Mol. Biol. (1991) 218, 695-698

Purification,

Crystallization and Preliminary Crystallographic Analysis of Porcine Aldose Reductase

Ossama El-Kabbanil”f, Sthanam V. L. Narayanal, Y. Sudhakar Babu2 Karen M. MooreI, T. G. Flynn3, J. M. Petrash4, Edwin M. Westbrook Lawrence J. DeLucas’ and Charles E. Buggl ‘The University of Alabama at Birmingham Center for Macromolecular Crystallography P.O. Box 7%THT, UAB Station Birmingham, AL 35294, U.S.A. ‘BioCryst, Inc. 1075 13th Street South, Suite 310 Birmingham, AL 35205, U.S.A. 3Department of Biochemistry Queen’s University Kingston, Ontario Canada K7L 3N6 4 Washington University School of Medicine Dept. of Ophthalmology & Visual Sciences P.O. Box 8096, 660 S. Euclid Avenue St Louis MO 63110, U.S.A. Biological,

5Argonne National Laboratory Environmental & Medical Research Argonne, IL 60439, U.S.A.

(Received 12 October 1990; accepted 11 January

1991)

Large crystals of porcine aldose reductase have been grown from polyethylene glycol solutions. The crystals are triclinic, space-group Pl, with a=81.3 A, b= 859 8, c=56.6 a, a= 102.3”, fl= 103.3” and y = 790”. The crystals grow within ten days to dimensions of 0.6 mm x 04 mm x 0.2 mm and diffract to at least 2.5 A. There are four molecules in the unit cell related by a set of three mutually perpendicular non-crystallographic 2-fold axes.

disease Diabetes mellitus is a debilitating affecting millions of people worldwide, and is the cause of significant morbidity and mortality due to progressive impairment of the visual, renal, nervous and vascular systems. Damage to these tissues results from biochemical and metabolic alterations occurring in response to chronic hyperglycemia. Recent evidence suggests that the metabolic products of aldose reductase (ALR-2$), the first and rate-limiting enzyme of the polyol pathway of

glucose metabolism, may be a biochemical link between hyperglycemia and diabetic complications in the eye and other major organ systems (Kinoshita & Nishimura, 1988). Inhibitors of aldose reductase, when administered to experimentally induced diabetic rats, essentially prevent the onset of diabetic complications such as cataract, retinopathy and neuropathy (Kador, 1988). While some aldose reductase inhibitors have undergone limited clinical testing, few have demonstrated substantial efficacy as judged by most objective parameters and many have been associated with undesirable sideeffects (Stribling, 1990). Thus, development of more potent and selective aldose reductase inhibitors will

7 Author to whom all correspondence should be addressed. $ Abbreviation used: ALR-2, aldose reductase. 695 002%2836/91/080695-04

$03.00/O

0

1991 Academic

Press Limited

696

0. El-Kabbani

clearly be required for effect’ive long-term prophylactic treatment of diabetic patients. All of the ALR-2 inhibitors developed to date are non-competitive or uncompetit,ive inhibitors of the enzyme and bear no structural resemblance to any ALR-2 substrates. Rational design of aldose reductase inhibitors is ideally based on the enzyme’s molecular structure as determined by X-ray crystallography. To this end, we have purified porcine muscle ALR-2 to apparent homogeneity, and have grown crystals of the enzyme by the vapor diffusion technique using polyethylene glycol as a precipitant. The crystals diffract to at least 2.5 A. Fresh pork muscle was obtained and used immediately or stored at -60°C. NADPH, enzyme substrates and buffer salts were purchased from Sigma Chemical Company. Sephadex G-100 and Blue SepharoseCL-6B were obtained from Pharmacia, and DEAE-Sephacel from Sigma. The Bio-Sil SEC 125 column was purchased from The electrophoresis supplies were Bio-Rad. purchased from Pharmacia, and ultrafiltration membranes were obtained from Amicon. ALR-2 activity was determined at 23 “C by monitoring the rate of NADPH oxidation with a Beckman DU-50 spectrophotometer as described (Hayman & Kinoshita, 1965). The sample cuvette contained 200 ~1 of sample, 100 ~1 of 3 mMNADPH, and 27 ml of buffer solution containing 50 rnM-sodium phosphate, 0.4 M-lithium sulfate, 5 mM-o,L-glyceraldehyde, 5 mM-2-mercaptoethanol (pH 6.2). The isolation and purification of aldose reductase was carried out at 4°C using a modification of published procedures (Cromlish 8: Flynn, 1983a, 6). Approximately 500 g of pork muscle was cut into small pieces and homogenized in one liter of buffer A 5 mM-2-mercapto(10 mM-sodium phosphate, ethanol, 5 mM-EDTA (pH 7)). The homogenate was centrifuged for 30 minutes at 9000 revs/min, after which the supernatant was filtered through glass wool and solid ammonium sulfate added to 37.5% saturation. The solution was stirred for one hour, 30 minutes at and then centrifuged for 9000 revs/min. Ammonium sulfate was added to the resulting supernatant to bring it to 66 o/o saturation. The solution was stirred for one hour and centriat 9000 revs/min. The fuged for 30 minutes resulting pellet was dissolved in 75 ml of buffer A and applied to a G-100 column (5 cm x 100 cm) at the rate of 80 ml/h. Fractions were assayed for aldose reduct’ase activity, and the active fractions pooled and concentrated to approximately 75 ml by ultrafiltration using an Amicon YMlO membrane. The sample was dialyzed exhaustively against buffer A and applied to a Blue Sepharose CL-6B column (25 cm x 30 cm) at 60 ml/h. The column was washed with 500 ml of buffer A and the enzyme eluted using a 09 to 95 M-NE&l gradient in buffer A. Fractions with ALR-2 activity were pooled, concentra.ted to approximately 50 ml, and dialyzed against buffer B (5 mM-sodium phosphate, 5 mM2-mercaptoethanvl (pH 7.4)). The sample was

et al.

applied to a DEAF,-Sephacel column at 60 ml/h, and the column was washed with 500 ml vf buffer B. Enzyme was eluted with a gradient of 0.0 to 0.2 M-sodium phvsphat’e in buffer B. All fractions were collected and assayed for ALR-2 activity. The active fractions were pooled and concentrated to I.0 ml. The concentrated sample was then injected on a high-pressure liquid chromatography TSK-125 size exclusion column (2.15 cm x 60 cm) and eluted with a buffer containing 906 M-sodium phosphate, 9.15 M-sodium chloride, 5 rnM-2-mercaptoethanol (pH 6.5). Active fractions were pooled and assayed for protein concentration by the Bradford (1976) method using bovine serum albumin as the standard. The sample volume was adjusted to obtain a final concentration of 8 to 12 mg protein/ml (final protein recovery of 6 mg). The protein was judged to be pure and homogeneous by observation of a single band on both SDS/polyacrylamide gel electrophvresis and ana.lytical isoelectric focusing gels. The enzyme was dialyzed against a buffer containing 25 mM-&k%, 5 rnul-2-mercaptvethanol (pH 6.2) and stored at 4°C until used for cryst,allization trials. Crystals of ALR-2 were grown at 4°C by vapor diffusion methods (hanging-drop technique; McPherson, 1985). Five-p1 droplets of protein previously dialyzed against, 25 mM-Mes buffer with 5 mM-2-mereaptoethanvl (pH 6.2) were mixed with 5 ~1 of a solution of 7.5% (w/v) polyethylene glycvl 3000, 25 mM-Mes, 5 mi%-2-mercaptvethanvl (pH 62). The droplets were vapor equilibrated against a similar buffer system containing higher percentages (typically 20%) of polyethylene glyeol. Within seven to ten days, showers of small triclinic crystals appear. By using the small crystals as seeds in additional crystallization experiments, crystals as large as 0.6 mm x 0.4 mm x 02 mm in size can be obtained within ten days (Fig. 1). TV accomplish this, 25 2-1.11vapor diffusion droplet experiments were set up and allowed to equilibrate for three days. After this time the 25 droplets were coalesced into one drop on a cover slide and a small (0.05 mm) seed crystal was inserted into the droplet using a 62 mm capillary syringe system atta.ched to a micromanipulator stage. The large droplet,

Figure E. &y&al vf’ porcine aidose reductaae. Clrystal dimensions: I.0 mm x 0.25 mm x 0.05 mm.

Communications -

697

Figure 2. Oscillation photograph of triclinic crystal of porcine aldose reductase, taken at the synchrotron radiation source at Daresbury, England. The edge of the circle represents 2.1 A data.

containing the seed crystal was then inverted and sealed over the same reservoir solution used for the initial crystallization experiments, and incubated at 4°C. Within three or four days the seed crystal grew to its final size. The crystals diffract strongly to 2.5 A resolution and are stable for more than 60 hours in CuKa-radiation (40 kV, 100 mA) generated by a fine-focus rotating anode generator (RU-200, Rigaku). Using synchrotron radiation, a diffraction limit of 2.1 A (1 A=O.l nm) was observed (Fig. 2). Three mutually consistent axial zones from precession photographs were indexed, indicating that the crystal system is triclinic, space group Pl, with unit cell dimensions a = 81.3 A, b = 85.9 8, c = 56.6 8, a= 102*3”, /I= 103.3” and y=79.0”. The crystals are isomorphous with those previously reported from pig lens tissue (Rondeau et al., 1987). Using an estimated molecular weight of 38,000 and assuming that there are three monomers present per unit cell, a T/‘, value of 3.26 A/D a and 62% solvent content are calculated; assuming four molecules per unit cell, a 8, value of 2.45 A/Da and 50% solvent content are calculated. Both sets of values are within the range found for protein crystals examined by Matthews (1968). X-ray diffraction data using three native aldose reductase crystals were recorded on a Nicolet multi-wire area detector and processed by the Xengen program package (Howard et al., 1987). The final merged native data had 46,515 out of 53,813 possible unique reflections, with 37,610 reflections measured more than once within the 2.5 A resolution sphere. The value of Rsym is 9.2% and the overall mean ratio Z/o(Z) = 17.6 for the 2.5 A resolution data set. Rsym is defined as:

Figure 3. Self-rotation function (K= 180” section) for porcine aldose reductase: data between 12.0 A and 50 a resolution were included with the radius of integration at 28 A. The crystallographic a-axis is along X-axis, c* along 2, and b* in the YZ plane, where XYZ are the righthanded orthogonal references axes in MERLOT (Fitzgerald, 1988). ‘I-’ is positive from 2 to Y and CDis positive from Y to -X in the X Y plane. (0) and (A) represent the axis directed away from the viewer and closed symbols are either in the plane or coming towards the viewer. The numbering of density peaks is as described in Table 1 and contours are drawn at 1 0 intervals for peaks above the 2 CJlevel.

resolution with an integration radius of 28 A, the calculated rotation function displayed six significant peaks ranging from five to seven times background levels, with the next highest peak being two times background. These six peaks represent two sets of three mutually perpendicular 2-fold axes and each set is related to the other by a center of inversion. Table 1 illustrates the peak heights, $ and 5~ values for each S-fold axis (i.e. IC= 180”). This observation confirms the presence of four molecules in the unit cell related by 222 symmetry. Extensive screening to determine an adequate mother liquor to stabilize aldose reductase crystals

Table Rotation Peak no.

where 1 is the mean intensity of the N reflections with intensities Zi and common indices h, k, 1. A self-rotation function (Fig. 3) was calculated MERLOT program package (Fitzgerald, using 1988). Using diffraction data between 12 A and 5 A

1 2 3 4 5 6

1

function

results

Peak height

*

rp

K

12 64 51 68 52 64

37.5 70.0 61.2 143.5 119.0 1090

135.5 16.3 275.3 314.4 96.2 197.3

180.0 180.0 1800 180.0 180.0 180.0

698

0. El-Kabbani

has been unsuccessful. As a result, highly concentrated solutions of heavy-atom derivatives were made and added directly to the protein droplet containing ALR-2 crystals. Thus far, one derivative, potassium platinum nitrite, was found to be isomorphous and suitable for X-ray analysis. The concentration of the heavy-atom in the drop was approximat’ely 15 m&r and the soaking time ten days. Final derivative merged data (resolution 3.19 A) had 19,758 out of 26,292 possible unique reflections with 16,663 reflections measured more than once. The value for Rsym is 9.5% and the average ratio I/o(l) = 10.8 at this resolution. The mean fractional difference between the native and derivative data sets equals 13.8% for data between 15 A and 6 A resolution. Three platinum sites were located from _ a three-dimensional isomorphous difference Patterson map using 2377 reflections (1>30 (I)). The positional parameters, isotropic temperature factors and occupancies for the platinum sites were refined by lack-of-closure-error minimization (Blundell & Johnson, 1976). A minor platinum site was located from a difference-difference Fourier map. The platinum positions are rela,ted by non-crystallographic 222 symmetry. Phases calculated for these positions yielded an overall figure of merit of 0.40, using data to 4-O A resolution. The electron density map calculated using the observed native amplitudes and phases from the platinum sites was subjected to solvent flattening (Wang, 1985). The resultant map revealed a clear demarkation of the solvent boundary and the molecular center of the tetramer in the unit cell. Currently, a search for additional heavy-atom derivatives is underway, and 4-fold symmetry averaging of the electron density map calculated using phases obt,ained from the platinum derivate is in progress. The primary structure of ALR-2 has been determined in a number of species by peptide sequencing (Schade et al., 1990) and complemenbary DNA sequencing (Carper et al., 1987; Petrash & Favello, 1989; Nishimura et al., 1990), revealing a high degree ( > 80 ye) of interspecies sequence homology. Thus, the structure of porcine ALR#-2 is expected to resemble closely that of ALR,-2 from other species. The three-dimensional structure of porcine ALR-2 will be used to design active-site directed inhibitors Edited

et al.

that may be more clinically efficacious than the noncompetitive inhibitors that are now available. This work was supported by research grants: from the Juvenile Diabetes Foundation, Lions International Eye Foundation, BioCryst; Inc., XASA grant XAGW-813. BRSG RR05807, CORE P30-EY03039; (to JXP.) from the P;ational Eye Institute (EY058.56); and in part by grants to the Department of Ophthalmology (Washington University) and t’o J.M.P. (Robert E. McCormick Scholar Award) from Research to Prevent Blindness, Inc. We also thank Mrs Holly Gettinger for technical assistance.

References Blundell, T. L. $ Johnson, L. P;. (1976). Protein Crystallography pp. 333-336; Academic Press, New York. Bradford, M. M. (1976). ilnal. Biochem,. 72; 248-254. Carper, D., h’ishimura, C., Shinohara, T., Dietzchold, B., Wistow, G., Craft, C., Kador, P. & Kinoshita, J. H. (1987). FEBX Letters, 220, 209-213. Cromlish, J. A. & Flynn, T. G. (1983a). J. Biol. Chem. 258, 3416-3424. Cromlish, 3. A. $ Flynn, T. G. (1983b). J. Riot. C&n?. 258, 358333586. Fitzgerald, M. D. P. (1988). J. AppZ. CrystalEogr. 21; 273-278. Rayman, S. & Kinoshita, J. II. (1965). J. Biol. Chem. 240, 877-882. Howard, A. J.; Gilliand, G. L., Finzel, B. C., Poulos, ‘I. L.* Ohlendorf, D. H. & Salemme, F. 12. (1987). J. Appl. Crystallogr. 20; 382-387. Kador, P. F. (1988). DiabetesllVetabal. Rev. 8, 325-352. Kinoshita, J. H. & Nishimura, C. (1988). Diabetes/ Metabol. Rev. 4, 323-337. Matthews, B. W. (1968). J. MOE. Biol. 33, 491-497. McPherson A. (1985). Methods Enzymol. 114, 112-120. Xshimura. C., Matsuura, Y.; Kokai. Y., Akera, T., Carper, D., Morjana, IV., Lyons, C. & Flynn, T. G. (1990). J. Biol. Chem. 265, 9788-9792. Petrash, J. M. & Favello, A. D. (1989). Curr. Eye Res. 8, 1021-1027. Rondeau, J. M.; Samama, J. P., Samama, B.; Barth. P.. Noras, D. & Biellman, J. F. (1987). J. Afol. Biol. 195, 945-948. Schade: S. Z., Early, S. L.; Williams, T. R.; Kezdy, F. J., Heinrikson, R. IL., Grimshaw, C. E. & Doughty, C. 6. (1990). J. Biol. Chem. 265, 628-635. Stribling, D. (1990). Exp. Eye Res. 50, 621-624. Wang, B. C. (1985). Acta Crystallogr. 12, 813-815.

by W. Hendrickson