Proc. Natl. Acad. Sci. USAVol. 86, pp. 3639-3643, May 1989Biophysics

Structure of activated aconitase: Formation of the [4Fe-4S] clusterin the crystal

(Fe-S enzyme/x-ray diffraction)

A. H. ROBBINS* AND C. D. STOUTtDepartment of Molecular Biology, Research Institute of Scripps Clinic, 10666 North Torrey Pines Road, La Jolla, CA 92037

Communicated by Helmut Beinert, January 24, 1989 (received for review October 12, 1988)

ABSTRACT The structure of activated pig heart aconitase[citrate(isocitrate) hydro-lyase, EC 4.2.1.3] containing a [4Fe-4S] cluster has been refined at 2.5-A resolution to a crystallo-graphic residual of 18.2%. Comparison of this structure to therecently determined 2.1-A resolution structure of the inactiveenzyme containing a [3Fe-4S] cluster, by difference Fourieranalysis, shows that upon activation iron is inserted into thestructure isomorphously. The common atoms of the [3Fe-4S]and [4Fe-4S] cores agree within 0.1 A; the three commoncysteinyl S., ligand atoms agree within 0.25 A. The fourthligand of the Fe inserted into the [3Fe-4S] cluster is a water orhydroxyl from solvent, consistent with the absence of a freecysteine ligand in the enzyme active site cleft and the isomor-phism of the two structures. A water molecule occupies asimilar site in the crystal structure of the inactive enzyme.

The stereospecific dehydration/rehydration reaction cata-lyzed by aconitase [citrate(isocitrate) hydro-lyase, EC4.2.1.3] requires iron (1-3), which is present in the enzyme asan iron-sulfur cluster (4, 5). In the inactive, aerobicallyisolated beef heart enzyme this cluster is [3Fe-4S]; the clusterhas cubane-like geometry (6-8). Upon activation of theenzyme with Fe2+ under reducing conditions, a [4Fe-4S]cluster is formed (6-12). The fourth Fe added to form the[4Fe-4S] cluster is directly involved in coordinating to sub-strates (13-16). Mossbauer (13, 14) and electron nucleardouble resonance spectroscopy in conjunction with 170 and13C labeling experiments (15, 16) show that the fourth Fe (Feasite, refs. 13 and 14) coordinates one carboxyl of substrate(citrate, cis-aconitate, or isocitrate) (16). When nitroisocit-rate is complexed to reduced activated aconitase both thehydroxyl group and H70 (x = 1 or 2) are bound simultane-ously (15). Kinetic experiments ofaconitase turnover in 3H20suggest that the enzyme traps protons or water from thesolvent (17); the proton abstracted from substrate is con-served in product by the enzyme (2). Literature leading to thecurrent mechanistic understanding of the aconitase reactionhas been reviewed (18).

Interconversion of a [3Fe-4S] to [4Fe-4S] cluster has beenobserved in Desulfovibrio gigas ferredoxin (19). In addition,the [3Fe-4S] cluster in this protein can incorporate Co2+ orZn2+ into the fourth site of the cluster (20, 21). The geometryof the [3Fe-4S] cluster in the 1.9-A resolution structure ofAzotobacter vinelandii 7Fe ferredoxin is very similar to thatof [4Fe-4S] cubanes (22-24), consistent with extended x-rayabsorption fine structure results for aconitase (6) and D.gigas ferredoxin (25).

Recently, the structure of inactive [3Fe-4S] aconitase frompig heart has been solved and refined at 2.1-A resolution(unpublished data). In this paper we report the refined struc-ture of the enzyme at 2.5-A resolution following activation in

the crystal. Comparison ofthe [3Fe-4S] and [4Fe-4S] aconitasestructures, which are isomorphous, provides crystallographicevidence for a Fe-S cluster interconversion. Previously, it wasshown from analysis of the anomalous difference Pattersonmap for inactive aconitase that the Fe-Fe distances are 3.0 times or(I) (Table 2).

*Present address: Miles Research Division, West Haven, CT 06516.tTo whom reprint requests should be addressed.

3639

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

Dow

nloa

ded

by g

uest

on

June

2, 2

021

3640 Biophysics: Robbins and Stout

Table 1. Data collection and scalingNo. of No. of

No. of total independentData set crystals reflections reflections Resolution, A Rsymm(l)* Rdiff(F)tInactive 3 350,360 47,583 2.1 0.106Activated 2 283,187 28,269 2.5 0.134 0.124

*R = Zh>li~i - II/hZhaI, where IT is the mean intensity of the i observations of reflection h.tR = 1hj1Fph1 - 1FpII/jIFp1, where lFphl and IFpI are the activated and inactive structure amplitudes, respectively.

Refinement calculations were done using the programX-PLOR (28). The structure of inactive aconitase was takenfrom the 2.1-A refinement (R factor 0.209, 43,288 indepen-dent reflections 2 O0O.F in the resolution range 5.0-2.1 A,6171 atoms including 322 water molecules with B c 60 A2,and rms deviation from ideality of 0.026 A for bonds). Themodel was modified with the program Frodo (29). The aminoacid sequence of 755 residues of the pig heart enzyme usedin refinement was derived from the DNA (H. Zalkin, personalcommunication) and the protein (W. E. Brown, personalcommunication). It should be noted that theN terminus oftheprotein has not been identified from the DNA or proteinsequence or in the electron density. Residue 1 in both crystalstructures is the first residue for which there is identifiabledensity. Coordinates of both the activated and inactivestructures have been deposited with the Protein Data BankA complete description of the inactive structure determina-tion will be presented elsewhere.

RESULTS

A difference Fourier map calculated at 2.5-A resolution withcoefficients [IFI(activated) - IFI(inactive)] and phases from therefined, inactive aconitase structure revealed a single, posi-tive electron density peak. This peak is shown in Fig. la inrelation to the electron density and model for the [3Fe-4S]cluster in the inactive protein structure. Because this peak is2.7 A from each of the Fe atoms of the [3Fe-4S] cluster, thenew cluster was modeled as [4Fe-4S] (Fig. lb); however, noligand was modeled at the fourth coordination site of thecluster. Refinement of the inactive protein structure coordi-nates against the 2.5-A activated data reduced the R factorfrom 0.29 to 0.23. Refinement of individual temperaturefactors further reduced R to 0.20.

Phases calculated from the partially refined activatedstructure were used to compute an electron density map withthe activated data. The Fourier coefficients, 21FOI - IFcI, wereweighted by the method of Sim (30). This map confirmed the[4Fe-4S] model for the activated cluster and also showed adistinct lobe of electron density on the new Fe site (Fe4) (Fig.lc). The position of this density is within 1 A of a watermolecule placed in the 2.1-A refinement of the inactivestructure. This water molecule (W806) was not moved from itsposition in the inactive structure; a bound SO2- ion presentin both structures and additional water molecules observed inthe activated structure electron density map were also addedto the model. The model was subjected to additional refine-ment against the 2.5-A activated data. The refined structureincluding 226 waters and the SO2- with isotropic temperaturefactors on all 6076 atoms has an R factor of 0.182 for all data.O.OcF in the resolution range 5.0-2.5 A (Table 3). The rmsdeviations from ideality of bonds and angles are 0.027 A and4.360, respectively; 184 water molecules have B < 40 A2 andthe remaining 42 have B 55 A2; all waters have at least onehydrogen bond to the protein and were refined with unitoccupancy. Sections of the final 2.5-A resolution 21FOl - IFJIelectron density map are shown in Fig. 2 a-f.

tBrookhaven Protein Data Bank (Brookhaven Natl. Lab., Upton,NY), File 2ACN.

The position of the new, fourth Fe site in the activatedstructure of aconitase was confirmed with a Bijvoet differ-ence Fourier map (31). The anomalous scattering data fromthe activated crystals to 2.5 A was combined with phasesfrom the inactive structure to give an independent image ofthe Fe structure (Fig. ld). Because there is no Fe at the Fe4site in the inactive structure, this map independently con-firms the presence of a [4Fe-4S] cluster in activated aconi-tase.The activated enzyme structure is remarkably isomorphous

to the inactive structure. The Ca structure of inactive aconi-tase and the [3Fe-4S] cluster is shown in Fig. 3a in a view intothe enzyme cleft. A description of the protein structure will bepresented elsewhere. The 755 common pairs of Ca atoms ininactive and activated aconitase differ on average by 0.19 A inthe two structures (maximum deviation, 0.47 A). Becausethese small differences are randomly distributed throughoutthe molecule, and because the unit cell parameters of the twocrystal forms are the same, it is likely that the observeddifferences in the structures represent no more than randomerrors in the coordinates at 2.5- to 2.1-A resolution.The isomorphism of activated and inactive aconitase also

pertains to the side chains when the structures are examinedin three dimensions. Fig. 3b shows the N, Ca, C, and sidechain atoms of 14 active site residues in the two models. Bothstructures also have in common a bound SO2- and a numberof ordered solvent molecules modeled as waters. Four watermolecules adjacent to the Fe-S cluster in the cleft in theactivated structure (W789, W806, W807, W963) are observed atcommon sites in the inactive structure (Fig. 3 c and d); theinactive structure contains a fifth water (W892) in the imme-diate vicinity of the active site. B factors for these watermolecules range from 10 to 25 A2 and 15 to 26 A2 in theinactive and activated models, respectively. The positions ofthe four common pairs of waters differ by 0.95 A, 0.06 A, 0.48A, and 0.55 A, respectively, at W806, W807, W789, and W963.The shift of W806 is toward the Fe4 site from its startingposition in the inactive structure and into the density asso-ciated with Fe4 (Fig. lc). W806 was not constrained to bebonded to Fe4 in the refinement; the resulting Fe-O distanceis 1.6 A. The 0.95-A shift of W806 is significant in that thelargest shift at any Ca position is 0.47 A.Table 4 summarizes the bond distances and angles for the

[3Fe-4S] and [4Fe-4S] clusters in the two structures. B factorsare 14-23 A2 for [3Fe-4S] and 13-18 A2 for [4Fe-4S]. The[3Fe-4S] moieties in the two structures appear to be indis-

Table 2. Activated aconitase dataNo. of independent

Resolution reflectionsrange, A Observed Possible % .20.0F Average J/lo(I)

oo-4.47 5739 5783 99.2 18.24.47-3.55 5557 5565 99.8 14.63.55-3.10 5437 5481 99.2 8.93.10-2.82 4926 5469 90.1 6.32.82-2.62 4076 5449 74.8 3.92.62-2.46 2534 5433 46.6 3.2

28,269 33,180 85.2 12.7

Proc. Natl. Acad. Sci. USA 86 (1989)

Dow

nloa

ded

by g

uest

on

June

2, 2

021

Proc. Natl. Acad. Sci. USA 86 (1989) 3641

Table 3. R factor for the activated aconitase structureNo. of

reflections*

Resolution range, A2.45-2.552.55-2.662.66-2.802.80-2.972.97-3.203.20-3.523.52-4.014.01-5.00

TotalF.bW-amplitude range

0.02-8.448.44-16.8616.86-25.2825.28-33.7033.70-42.1342.13-50.5550.55-58.9758.97-67.39

Total

6082,3282,6663,0893,4%3,6%3,7823,817

23,482

10,1639,0073,2368202063974

23,482

R factort

0.30370.28330.26590.21860.20140.17200.14730.14060.1820

0.34670.17260.11670.08680.06140.06990.08230.04110.1820

*Includes all observed data .O.OcF.tR = >IhlFol - IFCLI/ZIFoI, where IFOI and IFcl are observed andcalculated structure amplitudes, respectively.

three cysteine Sy ligand atoms differ by 0.25 A. By compar-ison, the [3Fe-4S] clusters of inactive aconitase (2.1-Arefinement) and 7Fe ferredoxin (1.9-A refinement; ref. 24)agree within 0.09 A; in this case the Sy atoms agree within0.25 A. The similarity of [3Fe-4S] and [4Fe-4S] core geometryas observed in aconitase is also observed in 7Fe ferredoxin;in this protein the seven common atoms of the two differentclusters agree within 0.08 A. Variations in average S-Fe-S vs.S-Fe-Sy bond angles (Table 4) are also similar to thoseobserved in other Fe-S clusters (32).

FIG. 1. Stereo figures of electron density maps of activated andinactive aconitase. (a) Superposition ofthe 2.5-A resolution differenceFourier map with coefficients (IF(activated)I - IF(inactive)I) showing thefourth Fe site (heavy lines) and the 2.1-A resolution (21FOI - IFJI)Fourier map of the inactive protein (thin lines). The model of the[3Fe-4S](Sy)3 cluster in the inactive protein is shown. Phases for bothmaps were calculated from the 2.1-A refinement of the inactivestructure; each map is contoured at 0.45 of the maximum density. (b)Difference Fourier map as in a superposed on the model for [4Fe-4S]cluster in the activated structure. Also shown are the three cysteineSy atoms attached to the cluster and a water molecule (W8%) refinedadjacent to the fourth Fe site (Fe4). In this view W806 is at the tip ofthe vector attached to Fe4-i.e., above the cluster and differenceelectron density peak. Additional crosses are atoms within a s.o-Asphere of the center of the cluster. (c) Model for the [4Fe-4S] clusterwith three cysteine Sy ligands and a water ligand (WN6, above clusterin this view as in b superposed on the 2.5-A (21FO1 - IFcI) electrondensity map of the activated protein. The map is contoured at 0.27and 0.45 the maximum density. The cysteine ligands are residues 359(left), 422 (right), and 425 (behind cluster in this view). (d) Bijvoetdifference Fourier map using activated aconitase data to 2.5-Aresolution with coefficients (IF+ - IF-I) and phases from therefined inactive protein structure. The fourth Fe site is at the top ofthe cluster in this view. Only the four Fe atoms are shown (averageseparation, 2.62 A). The map is contoured at 0.45 of the maximumdensity.

tinguishable, consistent with the overall isomorphism of theprotein structures. The seven common atoms of the Fe-Score structures differ on average by 0.11 A in position; the

Table 4. Iron-sulfur geometry[4Fe-4S] cluster [3Fe-4S] cluster*

Fe...Fe distances6 Number 3

2.51-2.69 A Range 2.64-2.73 A2.62 Average 2.69

Fe...S distances12 Number 9

2.22-2.37 A Range 2.25-2.35 A2.30 Average 2.30

Fe...Sy distances3 Number 3

2.33-2.38 A Range 2.28-2.34 A2.36 Average 2.32

S...Fe...S angles12 Number 9

103-1130 Range 101-11301070 Average 1070

Sy...Fe...S angles9 Number 9

98-1210 Range 94-12301120 Average 1120

Fe...S ...Fe angles12 Number 6

66-730 Range 70-720690 Average 710

Restraints used in refinement: Fe...Fe 2.75 A; Fe ...S, Sy 2.31 A;S...Fe...S, Sy 109.40; Fe...S...Fe, 750*Coordinates taken from inactive structure refinement.

a

b **

*

c

*

d

Biophysics: Robbins and Stout

Dow

nloa

ded

by g

uest

on

June

2, 2

021

3642 Biophysics: Robbins and Stout

a

eb

f

FIG. 2. 21FOl - IFTI electron density maps of activated aconitase at 2.5-A resolution, contoured at 0.25 of the maximum density for clarity.(a-d) Portions of the parallel a-sheets in each of the four domains. (a) Domain 1, centered on residue 70. (b) Domain 2, centered on residue238. (c) Domain 3, centered on residue 355. (d) Domain 4, centered on residue 634. (e) a-Helix in domain 1, residues 17-29. (f) a-Helix in domain3, residues 362-382.

DISCUSSIONComparison of the structures of inactive and activatedaconitase confirms the model that the enzyme contains[3Fe-4S] and [4Fe-4S] clusters in these two states (6, 9). Thestructures provide direct crystallographic evidence for a Fe-Scluster interconversion. Each cluster has three cysteineligands. The crystal structure shows that the fourth ligand ofthe Fe inserted into the [3Fe4S] cluster is water or hydroxyl.This result is not surprising in view ofthe known ability of thefourth site (Fea) to coordinate water or hydroxyl (13-16, 18),the presence of trapped water molecules in the active sitecleft, and the absence of any amino acid-in particular, freecysteine-in a favorable position to be a ligand (Fig. 3b). Theabsence ofa fourth cysteine ligand in the [4Fe-4S] form is alsoconsistent with cysteine titration and labeling experiments(33, 34). The ability of the fourth Fe to coordinate a wateroxygen may depend on water molecules being trapped withinthe enzyme active site. One of three histidines (residue 102,148, or 168) or aspartate residue 166 could perhaps serve toabstract a proton from water (Fig. 3d). The pH dependenceof kinetic parameters for yeast aconitase suggests that acarboxylate residue may be involved in the mechanism (35).

Both activated and inactive aconitase structures contain aSO- bound in the active site (Fig. 3 c and d). The inhibitorused in crystallization, tricarballylate, is weakly binding and isnot observed in either crystal structure. The presence of abound SOt- and crystal packing forces may constrain theprotein so that the observed structure is not necessarily in theconformation of active [4Fe-4S] enzyme in solution. Theisomorphism of the two structures does imply, however, that[3Fe-4S] to [4Fe-4S] cluster conversion is a facile process; inaconitase it can occur without rearrangement of any aminoacid residue. On the other hand, the aconitase active site isable to rearrange atpH > 9.5 to accommodate a linear [3Fe-4S]cluster (36). Conformational change as monitored by anincrease in tryptophan fluorescence is observed upon reduc-tion of inactive aconitase preceding activation (37). Therefore,though a [3Fe4S] to [4Fe-4S] conversion and the presence ofa non-cysteinyl ligand are consistent with data for aconitase insolution, the conformation observed in the activated crystalstructure may not be the same as during catalysis.

We thank D. Aul and W. E. Brown for providing protein samples,H. Zalkin, J. B. Howard, and W. E. Brown for making available

Proc. Natl. Acad. Sci. USA 86 (1989)

Dow

nloa

ded

by g

uest

on

June

2, 2

021

Proc. Natl. Acad. Sci. USA 86 (1989) 3643

a c[3Fe-4S] Aconitase

b d[4Fe-4S) Aconitase

[3Fe-4S] Aconitase

(4Fe-4S] Aconitase

FIG. 3. Active site residues and Fe-S cluster in aconitase. (a) Ca structure of inactive aconitase and the [3Fe-4S](Sy)3 cluster viewed towardthe active site cleft. The three N-terminal domains are arranged around the Fe-S cluster in a threefold manner; the C-terminal domain is abovethe cluster in this view. (b) Superposition of 14 residues of the activated and inactive protein structures showing N, Ca, C, and side chain atomsand the [3Fe-4S](Sy)3 cluster. These residues surround a cavity at the base of a cleft leading from the surface of the protein to the Fe-S cluster.(c) The side chains of eight active site residues in the immediate vicinity of the [3Fe-4S](Sy)3 cluster, the bound SOJ- ion, and five watermolecules from the 2.1-A resolution refinement. The unlabeled side chains are H14 and S167. The SOJ- is hydrogen bonded to Q73 and also R581,Sw, and R645 (not shown, see b). (d) The amino acid side chains as in c, the [4Fe-4S](Sy)3 cluster, and four water molecules in the active sitein common with the inactive structure (c), which refined in the activated structure. A fifth water, W892, refined to B > 55 A2 and is not included.One of the common waters, W806 (no label), refined to a position 1.6 A from Fe4 of the [4Fe-4S] cluster and is shown bonded to Fe4. The sidechains of H148 and S167 are unlabeled.

sequence data prior to publication, and M. H. Emptage, J. B.Howard, and D. Case for discussions. This work was supported byNational Institutes of Health Grant GM-36325.

1. Villafranca, J. J. & Mildvan, A. S. (1971) J. Biol. Chem. 246, 772-779.2. Rose, I. A. & O'Connell, E. L. (1967) J. Biol. Chem. 242, 1870-1879.3. Ruzicka, F. J. & Beinert, H. (1974) Biochem. Biophys. Res. Commun.

58, 556-563.4. Kennedy, C., Rauner, R. & Gawron, 0. (1972) Biochem. Biophys. Res.

Commun. 47, 740-745.5. Ruzicka, F. J. & Beinert, H. (1978) J. Biol. Chem. 253, 2514-2517.6. Beinert, H., Emptage, M. H., Dreyer, J.-L., Scott, R. A., Hahn, J. E.,

Hodgson, K. 0. & Thomson, A. J. (1983) Proc. Nat!. Acad. Sci. USA 80,393-3%.

7. Ryden, L., Ofverstedt, L.-G., Beinert, H., Emptage, M. H. & Kennedy,M. C. (1984) J. Biol. Chem. 259, 3141-3144.

8. Beinert, H. & Thomson, A. J. (1983) Arch. Biochem. Biophys. 222, 333-361.

9. Kent, T. A., Dreyer, J.-L., Kennedy, M. C., Huynh, B. H., Emptage,M. H., Beinert, H. & Munck, E. (1982) Proc. Nat!. Acad. Sci. USA 79,1096-1100.

10. Kennedy, M. C., Emptage, M. H., Dreyer, J.-L. & Beinert, H. (1983) J.Biol. Chem. 258, 11098-11105.

11. Emptage, M. H., Dreyer, J.-L., Kennedy, M. C. & Beinert, H. (1983) J.Biol. Chem. 258, 11106-11111.

12. Kennedy, M. C., Emptage, M. H. & Beinert, H. (1984) J. Biol. Chem.259, 3145-3151.

13. Emptage, M. H., Kent, T. A., Kennedy, M. C., Beinert, H. & Munck,E. (1983) Proc. Nat!. Acad. Sci. USA 80, 4674-4678.

14. Kent, T. A., Emptage, M. H., Merkle, H., Kennedy, M. C., Beinert, H.& Munck, E. (1985) J. Biol. Chem. 260, 6871-6881.

15. Telser, J., Emptage, M. H., Merkle, H., Kennedy, M. C., Beinert, H. &Hoffman, B. M. (1986) J. Biol. Chem. 261, 4840-4846.

16. Kennedy, M. C., Werst, M., Telser, J., Emptage, M. H., Beinert, H. &Hoffman, B. M. (1987) Proc. Nat!. Acad. Sci. USA 84, 8854-8858.

17. Kuo, D. J. & Rose, 1. A. (1987) Biochemistry 26, 7589-75%.18. Emptage, M. H. (1988) ACS Symp. Ser. 372, 343-371.19. Moura, J. J. G., Moura, I., Kent, T. A., Lipscomb, J. D., Huynh, B. H.,

LeGall, J., Xavier, A. V. & Munck, E. (1982) J. Biol. Chem. 257, 6259-6267.

20. Moura, I., Moura, J. J. G., Munck, E., Papaefthymiou, V. & LeGall, J.(1986) J. Am. Chem. Soc. 108, 349-351.

21. Surerus, K. K., Munck, E., Moura, I., Moura, J. J. G. & LeGall, J.(1987) J. Am. Chem. Soc. 109, 3805-3807.

22. Stout, G. H., Turley, S., Sieker, L. C. & Jensen, L. H. (1988) Proc.Nat!. Acad. Sci. USA 85, 1020-1022.

23. Stout, C. D. (1988) J. Biol. Chem. 263, 9256-9260.24. Stout, C. D. (1989) J. Mol. Biol. 205, 545-555.25. Antonio, M. R., Averill, B. A., Moura, I., Moura, J. J. G., Orme-

Johnson, W. H., Teo, B.-K. & Xavier, A. V. (1982) J. Biol. Chem. 257,6646-6649.

26. Robbins, A. H. & Stout, C. D. (1985) J. Biol Chem. 260, 2328-2333.27. Howard, A. J., Nielsen, C. & Xuong, N. H. (1985) Methods Enzymol.

114, 452-472.28. Brunger, A. T., Kuriyan, J. & Karplus, M. (1987) Science 235, 458-460.29. Jones, T. A. (1978) J. Appl. Crystallogr. 11, 268-272.30. Sim, G. A. (1960) Acta Crystallogr. 13, 511-512.31. Strahs, G. & Kraut, J. (1968) J. Mol. Biol. 35, 503-512.32. Carter, C. W. (1977) in Iron-Sulfur Proteins, ed. Lovenberg, W. (Aca-

demic, New York), Vol. 3, pp. 157-204.33. Plank, D. W. & Howard, J. B. (1988) J. Biol. Chem. 263, 8184-8189.34. Kennedy, M. C. & Beinert, H. (1988) J. Biol. Chem. 263, 8194-8198.35. Schloss, J. V., Emptage, M. H. & Cleland, W. W. (1984) Biochemistry

23, 4572-4580.36. Kennedy, M. C., Kent, T. A., Emptage, M., Merkle, H., Beinert, H. &

Munck, E. (1984) J. Biol. Chem. 259, 14463-14471.37. Ramsay, R. R. & Singer, T. P. (1984) Biochem. J. 221, 489- 497.

Biophysics: Robbins and Stout

Dow

nloa

ded

by g

uest

on

June

2, 2

021