Engineering actin-resistant human DNase I for ... - Semantic Scholar

6 downloads 0 Views 2MB Size Report
Goldschmidt-Clermont, P. J., Galbraith, R. M., Emerson, D. L.,. Marsot, F., Nel, A. E. & Arnaud, P. (1985) Biochem. J. 228,. 471-477. 34. McLeod, J. F., Kowalski ...
Proc. Natl. Acad. Sci. USA Vol. 93, pp. 8225-8229, August 1996 Biochemistry

Engineering actin-resistant human DNase I for treatment of cystic fibrosis (DNA enzymology/mutagenesis/protein engineering/actin inhibition/sputum viscoelasticity)

JANA S. ULMER*, ANDREA HERZKA*, KAREN J. Toyt, DANA L. BAKERt, ANTHONY H. DODGEt, DOMINICK SINICROPIt, STEVEN SHAKt, AND ROBERT A. LAzARuS*§ Departments of *Protein Engineering, tBioAnalytical Technology, and *Pulmonary Research, Genentech, Inc., 460 Point San Bruno Boulevard,

South San Francisco, CA 94080

Communicated by James A. Spudich, Stanford University School of Medicine, Stanford, CA, April 10, 1996 (received for review February 7, 1996)

Human deoxyribonuclease I (DNase I), an ABSTRACT enzyme recently approved for treatment of cystic fibrosis (CF), has been engineered to create two classes of mutants: actin-resistant variants, which still catalyze DNA hydrolysis but are no longer inhibited by globular actin (G-actin) and active site variants, which no longer catalyze DNA hydrolysis but still bind G-actin. Actin-resistant variants with the least affinity for actin, as measured by an actin binding ELISA and actin inhibition of [33P] DNA hydrolysis, resulted from the introduction of charged, aliphatic, or aromatic residues at Ala-114 or charged residues on the central hydrophobic actin binding interface at Tyr-65 or Val-67. In CF sputum, the actin-resistant variants D53R, Y65A, Y65R, or V67K were 10to 50-fold more potent than wild type in reducing viscoelasticity as determined in sputum compaction assays. The reduced viscoelasticity correlated with reduced DNA length as measured by pulsed-field gel electrophoresis. In contrast, the active site variants H252A or H134A had no effect on altering either viscoelasticity or DNA length in CF sputum. The data from both the active site and actin-resistant variants demonstrate that the reduction of viscoelasticity by DNase I results from DNA hydrolysis and not from depolymerization of filamentous actin (F-actin). The increased potency of the actin-resistant variants indicates that G-actin is a significant inhibitor of DNase I in CF sputum. These results further suggest that actin-resistant DNase I variants may have improved efficacy in CF patients.

action proposed, wherein DNase I reduces CF sputum viscoelasticity due to the depolymerization of F-actin (7). Purified actin filaments art known to be depolymerized by DNase I (14, 15). In addition, F-actin is in equilibrium with its monomeric globular form (G-actin), which interacts with a large number of actin binding proteins including DNase I (16, 17). G-actin binds DNase I with high affinity and is a potent inhibitor (Ki 1 nM) of DNA hydrolytic activity (18-20), potentially influencing the effectiveness of DNase I in vivo. Thus actin in its different forms may have multiple roles with respect to DNase I efficacy in CF sputum. We sought to elucidate both the basic mechanism of action of human DNase I in CF sputum as well as to assess the significance of G-actin as an inhibitor of DNase I in CF sputum. Our approach was to generate and characterize two classes of mutants: actin-resistant variants that catalyze DNA hydrolysis comparable to wild type but are no longer inhibited by actin, and active site variants which bind actin with affinity comparable to wild type but no longer catalyze DNA hydrolysis. The protein engineering strategy pursued in this paper demonstrates that the reduction of viscoelasticity by DNase I in CF sputum results from DNA hydrolysis and not from depolymerization of F-actin. Furthermore we show that Gactin is a significant inhibitor of DNase I in CF sputum, suggesting that actin-resistant variants may have improved clinical benefit for CF patients.

Respiratory symptoms, recurrent infectious exacerbations, and progressive lung destruction in cystic fibrosis (CF) can be be ascribed to several contributing factors, including a defective CF transmembrane conductance regulator gene, abnormal electrolyte transport, bacterial persistence, and the accumulation of viscous secretions in the airway (1-3). In response to these infections leukocytes infiltrate the airways and, upon lysis, release their contents which include DNA and filamentous actin (F-actin)-two macromolecules that contribute to the viscoelastic nature of CF sputum. DNA comprises about 4-10% of the total dry weight of purulent CF sputum (4, 5); concentrations up to 20 mg/ml have been observed (D.S., unpublished results). F-actin is an abundant protein in leukocytes (6); its concentration in CF sputum is reported to range from 0.1 to 5 mg/ml (7). Recombinant human deoxyribonuclease I (DNase I) reduces the viscoelasticity of CF sputum in vitro and improves lung function and reduces respiratory exacerbations in CF patients (8-12). The ability of DNase I to reduce the viscoelasticity of pulmonary secretions has been attributed to the enzymatic hydrolysis of DNA (8, 13). However, this notion has recently been questioned and an additional mechanism of

MATERIALS AND METHODS Mutagenesis and Expression. Site-directed mutagenesis was carried out by the method of Kunkel et al. (21); mutations were verified by dideoxy sequencing (22). Variant DNA was purified from 500 ml cultures of transformed Escherichia coli XL1 Blue MRF' (Stratagene) using Qiagen tip-500 columns (Qiagen, Chatsworth, CA). Human embryonic kidney 293 cells (ATCC CRL 1573) were grown in serum containing media in 150 mm plastic Petri dishes and log phase cells were transiently cotransfected with 22.5 ,ug DNase variant DNA and 17 jig adenovirus DNA (23). About 16 h after transfection, cells were washed with phosphate-buffered saline (PBS) and the medium was changed to serum free media; cell culture supernatant was harvested after 96 h. Harvests were concentrated 10-fold using Centriprep 10 concentrators and generally contained from 10 to 100 ,g of DNase I variant. DNase Activity Assays. The methyl green assay was used to measure DNA hydrolytic activity of DNase I (24). DNase I concentrations were determined by ELISA by using a goat anti-DNase I polyclonal antibody coat and detecting with a rabbit anti-DNase I polyclonal antibody conjugated to horseradish peroxidase. In both assays, multiple sample dilutions

The publication costs of this article were defrayed in part by page charge payment. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. §1734 solely to indicate this fact.

Abbreviations: CF, cyctic fibrosis; F-actin, filamentous actin; G-actin, globular actin. §To whom reprint requests should be addressed.

-

8225

8226

Biochemistry: Ulmer et al.

were compared with standard curves of purified recombinant human DNase I (Pulmozyme; Genentech) to determine concentrations. The relative specific activity is defined as the DNase I concentration determined in the methyl green assay divided by the DNase I concentration determined in the DNase I ELISA; the data are normalized to wild-type DNase I. Actin Binding ELISA. MaxiSorp plates (Nunc) were coated with 100 ,ul human Gc globulin (Calbiochem) at 10 ,ug/ml in 25 mM Hepes (pH 7.2), 4 mM MgCl2, and 4 mM CaCl2 at 4°C for 16-24 h. After discarding the Gc globulin, excess reactive sites were blocked with 200 ,ul buffer A (25 mM Hepes, pH 7.5/4 mM CaCl2/4 mM MgCl2/0.1% BSA/0.5 mM ATP/ 0.01% thimerosal/0.05% Tween 20). Buffer A was used in all dilution steps unless otherwise noted; incubations were at room temperature for 1 h. The wash buffer was PBS containing 0.05% Tween 20. Rabbit skeletal muscle G-actin [1 mg/ml; prepared by the method of Pardee and Spudich (25) or obtained from Sigma] was dialyzed overnight at 4°C against 5 mM Hepes (pH 7.2), 0.2 mM CaCl2, 0.5 mM ATP, and 0.5 mM 2-mercaptoethanol. After centrifugation at 13,000 x g for 5 min, the amount of G-actin-ATP was quantitated by measuring the absorbance at 290 nm, using 290 = 28.3 mM-1'cm-1 (19). Following the addition of 100 gl of 50 ,ug of G-actin per ml in buffer A, the plates were incubated and washed, and 100 ,lI of cell culture harvest at various dilutions was added. After incubation and washing, 100 pul of anti-human DNase I rabbit polyclonal antibody-horseradish peroxidase conjugate (19 ng/ ml) was added. Following incubation and washing, color development was initiated by adding 100 plI per well of Sigma Fast o-phenylenediamine and urea/H202 reagent (prepared according to Sigma) and stopped by adding 100 ,ul per well 4.5 M H2SO4. The A492 was measured and plotted versus the DNase I concentration. The resultant sigmoidal curves were fit to a four parameter equation by nonlinear regression analysis (26); the EC5o value is the DNase I concentration that produced a half-maximal signal. Actin Inhibition Assays. Varying concentrations of G-actin were preincubated in duplicate for 15 min at room temperature with 0.54 nM DNase I variant in buffer A containing 0.5 mM 2-mercaptoethanol and 150 mM NaCl. Reactions were initiated by the addition of 33P-labeled M13 DNA and salmon testes DNA (Sigma) to a final concentration of 4.1 ,ug/ml, incubated at room temperature for 2 h, and quenched with 25 mM EDTA and cold TCA (6.7% final concentration). After 10 min on ice samples were centrifuged at 9300 x g for 5 min at 4°C and the acid-soluble counts determined. Plots of the fractional activity (cpm inhibited/cpm uninhibited) versus actin concentration were fit by nonlinear regression analysis using KALEIDAGRAPH version 3.0.1 (Synergy Software, Reading, PA) to the following equation to determine the Ki value.

Proc. Natl. Acad. Sci. USA 93 (1996)

RESULTS AND DISCUSSION Protein Engineering. Selected mutations were made based on the x-ray crystal structures of bovine DNase I complexed with either G-actin or DNA (28, 29). A model of a ternary complex of DNase I, actin, and DNA, constructed by superimposing the DNase I structures (28-30), shows that the DNA and actin binding sites are adjacent yet distinct; the inhibition of DNase I by actin due to steric hindrance is apparent (Fig. 1). Bovine pancreatic DNase I shares 78% identity and superimposes with an overall rms deviation for main chain atoms of 0.56 A with the human enzyme, recently solved at 2.2-A resolution (30). The contact residues at the actin binding interface as well as those at active site are conserved. Both human and bovine DNase I have been found to reduce the viscosity of purulent sputum (8, 13). The binding of DNase I to actin involves hydrophobic, electrostatic, and hydrogen bonding interactions over a buried surface area of 1849 A2 (17, 28); the actin binding interface is shown in Fig. 2. Main-chain interactions result from the parallel ,B-strands formed by Tyr-65, Val-66, and Val-67 of DNase I and Gly-42, Val-43, and Met-44 of actin. The DNase I side-chains forming the core of the interface are hydrophobic and include Tyr-65, Val-67, and Ala-114; interactions peripheral to this central hydrophobic region are polar in nature and involve His-44, Asp-53, and Glu-69. Initially, alanine substitutions for these residues were made to assess their relative contribution to actin binding. Further mutagenesis lead to actin-resistant variants that were created by introducing charge, charge reversals, or steric bulk. Active site variants were created by alanine substitution at residues His-134,

Fractional activity =

[EO] + [IO] + Ki -

V([Eo] + [Io] + KJ)2 - (4. [E] * [j) 2 * [EO]]

where [Eo] is the DNase I concentration and [I,] is the total

G-actin concentration. CF Sputum Compaction Assays. Assays using CF sputum collected from four patients and stored at -80°C prior to use were carried out as described (27); sputa were incubated in triplicate with DNase I samples at various concentrations for 20 min at room temperature. Pulsed-Field Gel Electrophoresis. Pulsed-field gel electrophoresis was carried out in 1% agarose gels; after electrophoresis, gels were stained with ethidium bromide and imaged in a Molecular Dynamics Fluorimager.

FIG. 1. Model of a ternary complex of human DNase I with G-actin and a DNA oligonucleotide. Ribbon diagram of the DNase I-actin complex superimposed with the DNase I-d(GGTATACC)2 complex [Brookhaven Protein Data Bank 1ATN (28) and 1DNK (29), respectively] and human DNase I (30). Residues at the active site (His-134, Glu-78, His-252, and Asp-212) are in yellow and those at the actin binding interface (His-44, Asp-53, Tyr-65, Val-67, Glu-69, and Ala114) are in pink. The figure was made by using MIDASPLUS (31).

Proc. Natl. Acad. Sci. USA 93

Biochemistry: Ulmer et al.

FIG. 2. Structure of DNase I complexed with G-actin. The structure (1ATN) was taken from the Brookhaven Protein Data Bank (28); DNase I is represented in grey and actin is depicted in blue. Residues at the actin binding interface that were altered are highlighted in pink. The figure was made by using MIDASPLUS (31).

His-252, Asp-212, and Glu-78; these residues have been im-

in the general acid-base catalysis mechanism that occurs at the scissile phosphate bond (29, 32). In addition, the more conservative substitution of Gln for His at position 134 or 252 was also tested. DNA Hydrolysis and Actin Binding Assays. The relative specific activity of DNase I variants was assessed by comparing the specific DNA catalytic activity of the variant to that of wild type. DNase I concentration was determined by ELISA; reaction with the anti-DNase I antibodies also implied that the mutants were correctly folded. None of the active site variants catalyzed DNA hydrolysis, confirming the critical role of these residues in the enzyme mechanism. In contrast, the actin binding site variants all had similar specific activity compared

plicated

8227

with wild type, demonstrating that changes at the actin binding interface do not affect the active site (Fig. 3 A and C). The binding affinity of DNase I variants for G-actin was assessed by an ELISA where actin was bound to immobilized Gc globulin, a protein whose affinity for actin is unaffected by DNase I (33-35); the EC5o values are plotted in Fig. 3 B and D. The active site variants bound with the same high affinity to G-actin as wild-type DNase I, again confirming the indeof pendent nature of these two sites. However, a wide rangesite reduced affinities for actin was found for the actin binding variants (Fig. 3D). Selected variants were also assayed for actin inhibition of DNase I catalyzed hydrolysis of 33P-labeled DNA in solution. The DNase I variants and their respective Ki values are reported in Table 1. The inhibition of DNasetheenzymatic activity by actin in solution correlated well with relative actin binding affinity determined in the solid-phase actin binding assay (Fig. 3D and Table 1). DNase I Variants at the Actin Binding Interface. Single or aromatic point mutations introducing charged, aliphatic, variants that residues at Ala-114 resulted in actin-resistant were reduced in actin binding by over 10,000-fold relative to wild-type DNase I (Fig. 3D). This large effect is likely due to steric hindrance, since Ala-114 resides on an internal 3-strand and lies at the bottom of a pocket with its side chain in van der Waals contact with that of Val-45 of actin (Fig. 2). The introduction of charge on the central hydrophobic interface at Tyr-65 and Val-67 (Y65R, V67K, V67D) also greatly impaired actin binding (Figs. 2 and 3D). The fact that Y65R and V67K are much more actin resistant than Y65W or V67M implicates charge rather than steric hindrance as the critical factor. Moderate reductions in actin binding were noted upon removal of the aromatic side-chain at Tyr-65 (Y65A) and introduction of a charge reversal at Asp-53 (D53R), which should disrupt the salt bridge formed with Arg-39 and the hydrogen bonds to His-40 of actin (Fig. 3D and Table 1). Mutations at His-44, which interacts with Thr-203 and Glu-207 in actin, and Glu-69, which forms a salt bridge with Lys-61 of

Actin

Active Site Variants

Binding Site Variants

--

A 3

(1996)

oo

1.0_

1.0

I m < a) 0.5

'v' ) wO Z ._ ^

Z

co

1

Q 2] was determined by the actin binding ELISA and is normalized to wild type. The relative [EC5o (variant)/EC5o is(wild type);on mean a logarithmic scale for the actin binding site variants (D); bars labeled with an asterisk (*) represent a lower actin binding affinity plotted limit due to protein expression.

8228 Table 1.

Proc. Natl. Acad. Sci. USA 93 (1996)

Biochemistry: Ulmer et aL Inhibition constants (K1) for G-actin inhibition of

jg/ml

DNase I variant Ki, nM I 1.3 Wild-type DNase 3.9 H44A 24 D53A 47 D53R 36 Y65A >1000 Y65R 161 V67K 3.5 E69R >10,000 A114R The inhibition constants (Ki) were calculated using the known concentrations of DNase I variant and G-actin, the measured fractional activity, and fitting the data by nonlinear regression analysis to the equation for tight binding reversible inhibition as described in Materials and Methods.

actin, had minimal effects (Fig. 3D and Table 1), suggesting that their contributions to the binding energy of the complex are negligible. Activity of DNase I Variants in CF Sputum. Sputum compaction assays were used to measure the relative viscoelasticity of CF sputum after incubation with wild-type and selected DNase I variants (27); the percent change in compaction is shown in Fig. 4. The active site variants (H134A, H252A), which no longer catalyze DNA hydrolysis but still bind G-actin and could therefore depolymerize F-actin, did not reduce sputum viscoelasticity. However, both wild type and the actinresistant variants tested (D53R, Y65A, Y65R, V67K) decreased viscoelasticity in a dose-dependent manner. The data from both sets of mutants demonstrate that the reduction in sputum viscoelasticity by DNase I results from DNA hydrolysis and not actin depolymerization. The hydrolysis of DNA in CF sputum and subsequent reduction in DNA length by wild type and the actin-resistant variant V67K was confirmed using pulsed-field gel electrophoresis (Fig. 5); no reduction in DNA length was seen with the active site variant H134A (D.S., unpublished results). Most importantly, the actin-resistant variants were about 10 to 50-fold more potent in both reducing sputum viscoelasticity and in reducing DNA length compared with wild-type DNase I (Fig. 4 and 5). Based on these results, we conclude that G-actin is a significant inhibitor of DNase I in CF sputum, and that the actin-resistant variants are not subject to this inhibi20

....... fj

..

......I..

f.

10 Qa 0

Cb 2._ -

0

0

-10 CDCD .=

-20

._

Cu> 0

-30 -40

Actin resistant

variants

-50 .I...

0.01

0.1

1

10

DNase Variant Concentration (,ug/mi) FIG. 4. Effect of DNase I variants on CF sputum compaction. The percent compaction of the sputum (mean ± SEM), which correlates with sputum viscoelasticity, is plotted versus the variant concentration; wild type (-); active site variants: H134A (A) and H252A (0); and actin-resistant variants: D53R (-), Y65A (*), Y65R (A), and V67K (v).

V67K lig/mi

wildtype DNase I

33P-labeled DNA-catalyzed hydrolysis by DNase I variants I MW

C

II

I

4.83 1.60 0.53 0.18 0.06 0.02 0.53 0.18 0.06 0.02

FIG. 5. DNA length in CF sputum treated with wild-type DNase and the actin-resistant variant V67K. DNase concentrations and DNA molecular mass markers are indicated; C refers to a control sputum in the absence of DNase. Pulsed-field gel electrophoresis was carried out as described.

tion. The actin-resistant variants tested, which had from about 30 to >1000-fold reduced affinity for actin, were equivalent in the compaction assay, suggesting that they were all uninhibited by actin. Using the equation for tight binding reversible inhibition in the materials and methods as well as assuming that free G-actin is present minimally at its critical concentration of 100 nM (6, 16) and a Ki of 1.3 nM, we calculate that 99, 98, and 37% of wild-type DNase is inhibited at 0.1, 1, and 10 ,ug/ml DNase, respectively, consistent with the data in Fig. 4. Biological Significance and Clinical Implications. The biological significance of DNase I inhibition by actin is not well understood; a potential role during mitosis and apoptosis has been recently proposed (36). DNase I from mammalian sources is highly conserved (37); however, not all mammalian DNases are inhibited by actin (38, 39), suggesting that actin inhibition of DNase I is not- essential for cellular function. In fact, a gene encoding a novel human DNase, termed LSDNase, which is expressed predominantly in the liver and spleen and has 46% amino acid sequence identity to human DNase I, has recently been cloned and expressed (W. F. Baron, C. Q. Pan, R.A.L., and K. P. Baker, unpublished data). Of particular interest is the fact that LS-DNase has DNA hydrolytic activity that is not inhibited by G-actin. The finding that a family of DNase I-like enzymes exists within the human genome raises additional questions as to the cellular biology and biochemistry of DNase as well as the role of actin. The results presented here suggest that actin is a significant inhibitor of DNase I in CF sputum. F-actin is present over a range of concentrations in a limited number of CF sputum studied to date (7); however, the precise amount of actin capable of inhibiting DNase I is unknown. Although the concentration of actin, whether it exists as G-actin, F-actin, or actin in complex with actin binding proteins, in sputum throughout the CF patient population is not well defined, our results suggest that actin-resistant variants may have improved clinical efficacy, especially in patients with high sputum actin levels. We thank C. Eigenbrot and C. Schiffer for helpful discussions

Biochemistry: Ulmer et al. regarding DNase structure; A. Spudich for helpful discussions on actin; Genentech's Assay Services Group for technical assistance; L. O'Connell for tissue culture expertise; M. Vasser, P. Jhurani, and P. Ng for oligonucleotide synthesis; and K. Andow and D. Wood for graphics. 1. Boat, T. F., Welsh, M. J. & Beaudet, A. L. (1989) in Cystic Fibrosis, eds. Scriver, C. L., Beaudet, A. L., Sly, W. S. & Valle, D. (McGraw-Hill, New York), Vol. 2, pp. 2649-2680. 2. Quinton, P. M. (1990) FASEB J. 4, 2709-2717. 3. Collins, F. S. (1992) Science 256, 774-779. 4. Chernick, W. S. & Barbero, G. J. (1959) Pediatrics 24, 739-745. 5. Matthews, L. W., Spector, S., Lemm, J. & Potter, J. L. (1963)Am. Rev. Respir. Dis. 88, 199-204. 6. Stossel, T. P. (1992) in Inflammation: Basic Principles and Clinical Correlates, eds. Gallin, J. I., Goldstein, I. M. & Snyderman, R. (Raven, New York), pp. 459-475. 7. Vasconcellos, C. A., Allen, P. G., Wohl, M. E., Drazen, J. M., Janmey, P. A. & Stossel, T. P. (1994) Science 263, 969-971. 8. Shak, S., Capon, D. J., Hellmiss, R., Marsters, S. A. & Baker, C. L. (1990) Proc. Natl. Acad. Sci. USA 87, 9188-9192. 9. Hubbard, R. C., McElvaney, N. G., Birrer, P., Shak, S., Robinson, W. W., Jolley, C., Wu, M., Chernick, M. S. & Crystal, R. G. (1992) N. Engl. J. Med. 326, 812-815. 10. Ramsey, B. W., Astley, S. J., Aitken, M. L., Burke, W., Colin, A. A., Dorkin, H. L., Eisenberg, J. D., Gibson, R. L., Harwood, I. R., Schidlow, D. V., Wilmott, R. W., Wohl, M. E., Meyerson, L. J., Shak, S., Fuchs, H. & Smith, A. L. (1993) Am. Rev. Respir. Dis. 148, 145-151. 11. Ranasinha, C., Assoufi, B., Shak, S., Christiansen, D., Fuchs, H., Empey, D., Geddes, D. & Hodson, M. (1993) Lancet 342, 199-202. 12. Fuchs, H. J., Borowitz, D. S., Christiansen, D. H., Morris, E. M., Nash, M. L., Ramsey, B. W., Rosenstein, B. J., Smith, A. L. & Wohl, M. E. (1994) N. Engl. J. Med. 331, 637-642. 13. Armstrong, J. B. & White, J. C. (1950) Lancet ii, 739-742. 14. Hitchcock, S. E., Carlsson, L. & Lindberg, U. (1976) Cell 7, 531-542. 15. Weber, A., Pennise, C. R. & Pring, M. (1994) Biochemistry 33,

4780-4786. 16. Sheterline, P. & Sparrow, J. C. (1994) Protein Profile 1, 1-121. 17. Kabsch, W. & Vandekerckhove, J. (1992) Annu. Rev. Biophys. Biomol. Struct. 21, 49-76.

Proc. Natl. Acad. Sci. USA 93 (1996)

8229

18. Lazarides, E. & Lindberg, U. (1974) Proc. Natl. Acad. Sci. USA 71, 4742-4746. 19. Mannherz, H. G., Goody, R. S., Konrad, M. & Nowak, E. (1980) Eur. J. Biochem. 104, 367-379. 20. Pinder, J. C. & Gratzer, W. B. (1982) Biochemistry 21, 48864890. 21. Kunkel, T. A., Roberts, J. D. & Zakour, R. A. (1987) Methods Enzymol. 154, 367-382. 22. Sanger, F., Nicklen, S. & Coulson, A. R. (1977) Proc. Natl. Acad. Sci. USA 74, 5463-5467. 23. Gorman, C., Gies, D. R. & McCray, G. (1990) DNA Protein Eng. Tech. 2, 3-10. 24. Sinicropi, D., Baker, D. L., Prince, W. S., Shiffer, K. & Shak, S. (1994) Anal. Biochem. 222, 351-358. 25. Pardee, J. D. & Spudich, J. A. (1982) Methods Enzymol. 85, 164-181. 26. Marquardt, D. W. (1963) J. Soc. Indust. Appl. Math. 11, 431-441. 27. Daugherty, A. L., Patapoff, T; W., Clark, R. C., Sinicropi, D. V. & Mrsny, R. J. (1995) Biomaterials 16, 553-558. 28. Kabsch, W., Mannherz, H. G., Suck, D., Pai, E. F. & Holmes, K. C. (1990) Nature (London) 347, 37-44. 29. Weston, S. A., Lahm, A. & Suck, D. (1992) J. Mol. Biol. 226, 1237-1256. 30. Wolf, E., Frenz, J. & Suck, D. (1995) Protein Eng. 8 (Suppl.), 79. 31. Ferrin, T. E., Huang, C. L., Jarvis, L. E. & Langridge, R. (1988) J. Mol. Graph. 6, 13-27. 32. Worrall, A. F. & Connolly, B. A. (1990) J. Biol. Chem. 265, 21889-21895. 33. Goldschmidt-Clermont, P. J., Galbraith, R. M., Emerson, D. L., Marsot, F., Nel, A. E. & Arnaud, P. (1985) Biochem. J. 228, 471-477. 34. McLeod, J. F., Kowalski, M. A. & Haddad, J. G., Jr. (1989) J. Biol. Chem. 264, 1260-1267. 35. Houmeida, A., Hanin, V., Constans, J., Benyamin, Y. & Roustan, C. (1992) Eur. J. Biochem. 203, 499-503. 36. Peitsch, M. C., Polzar, B., Stephan, H., Crompton, T., MacDonald, H. R., Mannherz, H. G. & Tschopp, J. (1993) EMBO J. 12, 371-377. 37. Peitsch, M. C., Irmler, M., French, L. E. & Tschopp, J. (1995) Biochem. Biophys. Res. Commun. 207, 62-68. 38. Lacks, S. A. (1981) J. Biol. Chem. 256, 2644-2648. 39. Mannherz, H. G., Kreuder, V., Koch, J., Dieckhoff, J. & Drenckhahn, D. (1982) Biochem. J. 207, 305-313.