THE JOURNAL OF BIOLCXXCAL CHEMISTRY Vol. 249, No. 6, Issue of March 25, pp. 1568-1677, 1974

Printed in U.S.A.

Spin-labeled Sulfonyl Fluorides as Active Site Probes of Protease Structure

I. CO&IPARISON 0% THE ACTIVE SITE ENVIRONMENTS IN CY-CHYMOTRYPSIN AND TRYPSIN*

(Received for publication, July 23, 1973)

LAWRENCE J. BERLINER AND SHAN S. WONG

From the Department of Chemistry, The Ohio State University, Columbus, Ohio 43.210

SUMMARY

a-Chymotrypsin and trypsin were spin-labeled with 14 ortho-, meta-, and para-substituted phenylsulfonyl fluorides. These analogues of tosylfluoride were substituted with either pyrrolidinyl, pyrrolinyl, or piperidinyl nitroxide moi- eties. A complete description of their synthesis and the enzyme inhibition studies is presented in the following paper (WONG, S. S., QUIGGLE, K., TRIPLETT, C., AND BERLINER, L. J. (1974) J. Biol. Chem. 249, 1678-1682).

The studies were designed to test for “dynamic confor- mational homologies” in these closely related proteins. However, the spectral results at pH 3.5 clearly showed that the active site conformations of trypsin and a-chymotrypsin were different. Model-building studies indicated which “tosyl” spin labels could bind isomorphously to tosyl-ar- chymotrypsin. Several of those labels which were shown to be restricted structurally from binding in the chymotrypsin aromatic pocket were found upon exposure to saturated indole to give electron spin resonance spectra closely resem- bling those for trypsin. Those inhibitors which could the- oretically bind in the pocket were affected least or not at all. With trypsin, on the other hand, 1 mM benzamidine produced no spectral shifts.

A suggested model for cY-chymotrypsin based on steric arguments provides for three general binding modes: (a) one isomorphous to tosylchymotrypsin, (b) one reflecting partial binding at the tosyl hole, and (c) one in a general region of the active site. Indole can displace the partially bound State b toward the general binding Mode c. This general Mode c is probably common to trypsin as well.

The use of slkyl and aromatic sulfonyl lluorides as specific active sit.e-directed irreversible inhibitors for serine proteases was demonstrated several years ago by Fahrney and Gold (1). Of these rcngents, p-toluene sulfonyl fluoride (tosyl fluoride)

* This research was sponsored in part through grants from the National Science Foundation ((;B 16437), Research Corporation, and the Public Health Service (GM 19363). A preliminary report of this work has been presented at the IVth International Bio- physics Congress, Moscow, August 7 to 14, 1972.

and its heavy atom analogue, p-iodobenzene sulfonyl fluoride (pipsyl fluoride), proved to be invaluable aids in the crystal- lographic structure determination of Lu-chymotrypsin (2) and later y-chymotrypsin (3), elastase (4), and subtilisin BPN’ (5). Furthermore, it was shown crystallographically for Lu-chymo- trypsin that the tosyl moiety was bound in the aromatic specific- ity binding pocket (“tosyl hole”) of the enzyme (6).

In an attempt both to probe and to compare the active site conformations of Lu-chymotrypsin and related serine proteases, we have synthesized and tested several spin label derivatives of benzene sulfonyl fluorides of the general structure

0,F

& R

where R = a nitroxide (spin label). While it was recognized that bulky substitution on the aromatic ring, particularly at the para position, might preclude the binding of this derivative in the chymotrypsin binding pocket, substitution at the met,a or ortho position might allow specific binding. This reasoning was based on crystallographic and kinetic studies with ol-chymo- trypsin which place a limit on both the size and shape of the “tosyl hole” (7). For example, m-formyl-p-iodophenylalanine and @-(p-iodophenyl)propionate did not bind in the pocket in cr-chymotrypsin crystals (8)) while oxygen-alkylated tyrosine and p-iodo-n-phenylalanine substrate derivatives were hydro- lyzed considerably more slowly in solut.ion than the unsubstituted compounds (9).

The spin label experiment provides extremely sensitive in- formation about the motional freedom of the nitroxide moiety (10). Consequently, the rotational mobility of the spin label, R, is dictated in large part by its immediate physica environment and the nature of the interactions between the nitroxide ring and the protein structure. Assuming that a variety of topo- graphically variable spin labels were incorporated isomorphously at a specific site in a protein molecule, the corresponding mo- bilities could be analyzed to yield a “dynamic conformational map” which contains information about both physical structure and forces (irneractions). We t.hen define a “dynamic con- formational homology” as a case where two proteins give identical results for the same series of spin labels. A schematic diagram of the approach used in these studies is shown in Fig. 1. The spin labels utilized are shown in Fig. 2. In those cases where

1668

by guest on March 1, 2020

http://ww

w.jbc.org/

Dow

nloaded from

1669

Ser

l

L 195

His 57 dCH2

I o=s=o -

-

,,a)_ Spin Label

\\ $$c

Q -Chymotrypsin

FIG. 1. A schematic model for the “specific” binding of a spin- labeled sulfonyl fluoride. Obviously a para-substituted spin label would be sterically prohibited from binding in the specificity pocket (“tosyl hole”) and therefore must bind in an alternate mode outside the pocket.

the aromatic portion of the spin label may bind in the tosyl hole, the immediat.e structural region outside this pocket may be probed by monitoring the mobilit’y of the attached nitroside spin label moieties. On the other hand, those derivatives which do not bind in the pocket but nevertheless (covalently) inhibit the enzyme, must adopt some alternat.ive binding mode, perhaps one common to many serine proteases. In addition, the narrow dimensions of the tosyl hole (10 to 12 A by 5.5 to 6.5 A by 3.5 to 4.0 A) dictate that the aromatic group would always be that part of the inhibitor which binds in the pocket since the spin label moiety is too thick by virtue of t,he dimethyl groups which flank the nitroxide linkage (8). Consequently, for this latter case of secondary or “nonspecific” binding, a sensitive comparison may be made between the active site conformations of cu-chymo- trypsin and the relat.ed protease, trypsin. Furthermore, it had been suggested and recently reported that the many homologies in primary structure for these two enzymes were reflected in significant homologies in their tertiary structures (11, 12). The spin label approach outlined above potentially offers an extremely sensitive dynamic test of these conformations1 homologies in the active site.

MATERIALS Ah-D METHODS

The 15 spin labels synthesized are shown in Fig. 2. Details of their synthesis and purification as well as the enzyme inhibi- tion studies are presented in the following paper (13). ESR measurements were taken at X band on a Varian E-4 spectrom- eter at 26 f 2”.

Spin Label Nomenclature--In order for the reader to envision most easily the structure of a particular spin label under discus- sion, we always refer to a given inhibitor with two abbreviated codes as well as occasionally citing a more specific aspect of it.s structure. Labels which are identical except for the position of the -SO$ moiety are designated with the same Roman numeral; e.g. see the first column in Fig. 2 with the o, p, and m derivatives o-1 (o-NH-5=CO), p-1 (p-NH-5=CO), and m-1 (m-NH- 5=CO). The abbreviation in parentheses completely describes the structure. As usual, o-, m-, or p- specifies the position of the SO&’ group. The next symbol designates the functional group from the phenyl group: -NH, amido; CO, acyl; -NCO, carbamyl; or SO*-, sulfonyl. The next numerical symbol refers to the

ortho

poro

met0

met0

0 0 SOzF H-N

k d

m-1 (m-NH-51601

m- “I m-w, m-VIII Im-NCO-60Hl ,,,,-NCO-6NN, I,“-NH-5C0,

0 S%F

e -0

4

m-,x ,l7+-00-5C”p”,

FIG. 2. Structures of spin-labeled sulfonyl fluorides. Com- pounds which are isomers of the same structure are designated by the same R,oman numeral (e.g. o-1, p-1, m-1).

type of nitroxide ring: 5, the five-membered (saturated) pyr- rolidinyl ring; 5=, the five-membered (unsaturated) pyrrolinyl ring; or 6, the six-membered piperidinyl ring. Last, the remaining symbol refers to the functional group derived from the nitroxide spin label moiety in the Iinkage: CO, acyl (from the carboxylic acid) ; OH, an ester (from the alcohol) ; NH, amido (from the amine); and CHZOH, ester (from the primary alcohol). Therefore, an example might be p-IV (p-CO-6NH) which is the para-fluorosulfonyl benzamide of the six-membered piperidinyl amine; or m-VIII (m-NH-5CO), the meta-fluoro- sulfonyl anilide of the five-membered (saturated) pyrrolidine carboxylic acid.

RESULTS

ESR Spectra-The ESR spectra for cu-chymotrypsin and trypsin inhibited with 14 of the 15 spin label inhibitors of Fig. 2 are shown in Figs. 3 and 4.’ It was necessary to measure these spectra at pH 3.5, where hydrolysis and desulfonylation problems

1 Although o-11 inhibited both enzymes, the rate of intra- molecular hydrolysis of the label itself was so rapid that a spin- labeled enzyme was virtually impossible to measure. See the following paper for a more detailed discussion of this problem.

by guest on March 1, 2020

http://ww

w.jbc.org/

Dow

nloaded from

1870

CHYMOTRYPSIN TRYPSIN

FIG. 3. ESR spectra of cx-chymotrypsin and trypsin, respectively, spin-labeled with the o- and p-substituted inhibitors shown in the center column. Conditions were 26 =k 2”, pH 3.5 (-0.006 M acetic acid) and 0.1 M NaCl. and frequently 1 mM benzamidine.

Trypsin samples also contained 0.02 M CaClz

were at an absolute minimum (13). This should have no bearing on the results of these comparative studies.

enzyme (12).)2 An initial examination of these comparative (There was in fact spectra (Figs. 3 and 4) suggests that there are few, if any, simi-

evidence that no detectable conformational changes occurred larit.ies between cu-chymotrypsin and kypsin for each label

between pH 3.5 and neutral in the active site regions of either e L. J. Berliner and S. S. Wong, unpublished observations.

by guest on March 1, 2020

http://ww

w.jbc.org/

Dow

nloaded from

1671

CHYMOTRYPSI N

FIG. 4. ESR spectra of Lu-chymotrypsin and trypsin, respectively, labeled with m-substituted inhibitors. with those in Fig. 3. Conditions were identical

studied. It was also apparent in almost every case, except perhaps for p-11 (p-Sot-60H), that the spin-labeled trypsin

necessarily arise only from a structurally constricting environ-

exhibited broader line shape spectra than did the corresponding ment but could also arise from specific noncovalent attractive interactions.

cY-chymotrypsin derivative. This signifies that in trypsin, the nitroxide moiety is in a state of more hindered motion. It is

Model Building-An atomic model of the active site of tosyl-cu-

important to point out that this rotational hindrance need not chymotrypsin was constructed according to the coordinates of Birktoft and Blow (14) with Kendrew skeletal models (5 cm =

by guest on March 1, 2020

http://ww

w.jbc.org/

Dow

nloaded from

1672

1 A) purchased from Cambridge Repetition Engineers, Cam- bridge, England. For each inhibitor in Fig. 2, the spin label moiety was attached bo the fixed (“bosyl”) phenyl group and then t.ested for contact,s of less than 3 A wit,h the enzyme struc- ture. The assumptions made as to the conformation of the bound inhibitor were (a) the aromat,ic group was isomorphous wit.h t,he tosyl group in tosyl-cu.chymotrypsin, (b) amide linkages were assumed to be planar, however, not necessarily coplanar to the aromatic ring. Several reported crystallographic struc- tures indicate that the amide plane was out of the plane of the phenyl ring; e.g. acetanilide, 38” (15) or benzamide, 26” (16). Alt.hough the isomorphous binding of the aromatic spin label in the “tosyl hole” has not been firmly established, the assump- tion seems reasonable in light of the evidence we have obtained.

To refine the interpretation, the observed spin label mobilities were also considered when predicting the rot,ational freedom of a label in the enzyme model. A label which was apparently highly restricted in the model, yet showed a highly mobile (weakly immobilized) ESR spectrum, must therefore bind in an alterna- tive orient,ation, or exist in an equilibrium between t,he rigidly bound and alternative orientation. While all of the para deriva- tives were excluded from binding at the “tosyl” hole, several meta-substituted derivatives were also sterically prohibited from binding in this pocket. A summary of these results is given in Table I.

Indole Eifects-Since it was anticipated that some meta and ortho derivatives would bind specifically at the “tosyl hole” and since the resultant ESR spectra (Figs. 3 and 4) for both enzymes reflected almost no homologies in active site conformations (as sensed by the nitroside moiety), we used a method which would effect a displacement of the spill-labeled “tosyl” moiety from the chymotrypsin aromatic binding pocket. This involved the use of indole, a strong competitive inhibitor for cu-chymotrypsin which has been shown to bind specifically at the tosyl hole by crystallography (8), and, above pH 7, to affect the optical rota- tion of the native enzyme (17) or effect spectral changes for spin label moieties bound at or near this pocket (18). However, indole was shown not to effect optical rotation changes at lower pH values (17) where the studies above were undertaken. Therefore, a simple steric argument was chosen for the effects discovered here. We reasoned then that for those spin labels which bind at the tosyl hole, but exist in some equilibrium with state (or states) away from this binding pocket, and where a preferential (binding) affinity exists for indole in t,he “t’osyl hole,” this spin label equilibrium would be shifted toward those binding modes oubside the pocket. Thus, the indole could, in effect, displace the spin-labeled “tosyl” group from the pocket.

All of the cr-chymotrypsin derivatives were examined at pH 3.5 in the presence of saturated indole. The results obtained were striking in many respects. The majority of those deriva- tives shown from model building not to bind in the “tosyl hole” exhibited dramatic spectral changes upon exposure to indole. The effects were shown to be reversible by dialyzing away the indole. Specifically, several of those derivatives effected by indole gave spin resonance spectra which were either identical or very similar to those for trypsin in aqueous solution at the same pH. Those derivatives displaying these spectra “equiva- lences” in indole were m-1 (m-NH-5=CO), m-III (m-CO-GOH), m-IV (m-CO-GNH), m-VII (m-NCO-GNH), and m-VIII (m-NH-5CO). Two examples are shown in Fig. 5 for m-1 (m-NH-5=CO) and m-111 (m-CO-GOH). In every case a broad line component appeared when the labeled chymotrypsin was exposed to indole (center spectra). The relative amount of

this component was similar to that found for trypsin (right specfra). Another group, o-1 (o-NH-5=CO), p-11 (p-Sot-60H), p-111 (p-CO-GOH), p-IV (p-CO-SNH), and m-VI (m-NCO- 6OH), gave “indole effects” which shifted t,he spectra toward the same degree of mobility as for trypsin; however, they were not superimposable wit,h the corresponding trypsin spectrum. Additionally, the two derivatives p-1 (;D-NH5=CO) and m-IX (m-CO-5CH20H) gave spectral shifts in indole which were opposite to the corresponding trypsin spectrum.

The ortho-substituted derivative, (o-1) (o-NH-5=CO), ap- peared to bind in the aromatic pocket in a fairly unrestricted orient,ation, yet in solution it evidently occupied a second alterna- tive binding mode. This was suggested by the fact that the spectrum of o-1 (o-NH-5=CO) in Fig. 3 contained a small broad line component. This broad component was clearly dist.inguished from the more mobile component (narrow line spectrum) by observing the ESR spectrum at high sucrose con- centrations where the macromolecular contribution to the ob- served tumbling rate was reduced significantly.3 That is, the ESR spect.rum represent.ed primarily t.he motion of t.he label with respect to the protein. The clearly distinguishable two- component spectrum in sucrose was also shifted on exposure to indole in a fashion identical with that of the labeled enzyme in no sucrose.

The sulfonylated ester derivative p-11 (p-S02-60H) was also affected dramatically by indole, but in a somewhat different manner. Examination of p-11 (p-SOY60H) labeled cu-chymo- trypsin and trypsin in Fig. 3, indicated that both had two compo- nent spectra, the cu-chymotrypsin derivative possessing a higher percentage of the more hindered broad line component (or conversely, a smaller amount of narrow line component) than did the trypsin derivative. In indole, the labeled chymotrypsin spectrum shifted significantly toward the (more mobile) narrow line component.

Most of those meta-substituted derivatives shown by model building to bind in the “tosyl” hole were least affected upon exposure to indole: m-V (m-CO-5NH), m-VI (m-NCO-GOH), m-VII (m-NCO-6NH), m-VIII (m-NH-5CO), and m-IX (m-CO-5CH20H). Presumably their ‘(effective equilibrium constants” favored the spin label binding in rather than away from the pocket. A complete summary of the cu-chymotrypsin results is given in Table II.

The spectral shifts upon indole exposure were manifested in several cases by the formation of a motional component spectrally ident,ical with t,hat for trypsin. While t.hese shifts were not always complete, t,his was underst,andable in view of the probable binding constant for indole. Fersht and Requena found an essentially pH-independent dissociation constant for indole of 0.8 mM at. pH 7.8 and this was approximately the same at pH 4.5 (20). We would assume that t.he inhibited enzyme binds indole less effectively and that the value 0.8 mM would be an absolute lower limit for the dissociation constant of spin-labeled indole complexes at pH 3.5. It is interesting to note that Kosman obtained dissociation constant,s for modified chymo- trypsin-indole complexes of 3 and 2 mM for a methionine 192- alkylated and a histidine 57-alkylated derivative, respectively (21). That is, it was unlikely that the enzyme was completely saturated with indole at the concentrations attainable in these

3 Extraneous structural perturbations due to the exposure of the enzyme to high sucrose concentrations seems unlikely from the recent results of Timasheff and co-workers who found no changes in both circular dichroism and activity measurements with cu-chymotrypsin (19).

by guest on March 1, 2020

http://ww

w.jbc.org/

Dow

nloaded from

1673

TABLE I

Results from model building for a-chymotrypsin

Excluded from binding isomorphously in “tosyl hole”

Spin label

All para derivatives

m-1 (m-NH&GO)

m-111 (m-CO-6OH)

m-IV (m-CO-6NH)

m-V (m-CO-5NH)

m-VI (m-NCO-6OH)

m-VII (m-NCO-6NH)

m-VII1 (m-NH-5CO)

m-IX (m-CO-5CHzOH)

o-1 (o-NH-5=CO)

-

Probable contacts

p-1 (p-NH-5=CO), p-11 (p-S02-GOH), p-111 (p-CO- 60H), p-IV (p-CO-GNH), p-V (p-CO-5NH).

Pyrrolinyl ring within 0.2 to 0.5 A of Ser 218-Thr 219 peptide bond.

13 A contacts with Ser 217, Ser 218, Thr 219. Thr 219 peptide carbonyl oxygen sticks up and closest to piperidinyl ring methylene or methyl carbon atoms.

As above.

Comments

Excluded by residues 217 to 220 lining the “bottom” of the pocket.

If there was an assumed free rotation of the ester linkage itself, a few possible restricted orientations are possible; however, such rest,ricted binding does not appear to be consistent with the “highly mo- bile” spectrum obtained for this derivative (see Fig. 4).

If the amide linkage is held coplanar with the aro- matic ring, a single restricted orientation is pos- sible which is even less consistent with the ob- served spectrum t,han above.

Highly restrictive yet possible binding and minimally restricted binding in “tosyl hole”

S enantiomer situates with the pyrrolidinyl ring plane over Ser 218-Thr 219-Cys 220 main chain; about 30-40” rotation is the maximum allowed.

R enantiomer situates perhaps slightly better; Ser 218 ?-OH must point down via a rotation about the Ser 218 CL-C, bond. All contact distances are right at 3 A.

Situated over the peptide chain Ser 218-Thr 219 with some rotational freedom. Allowing also free ro- tation about the carbonyl-oxygen bond of the car- bamate, the piperidinyl ring may swing into the open cleft region where substantial rotational free- dom is allowed.

As above. However, if substantial restrictions are placed on the planarity of the urea linkage, a more restricted rotational freedom is found, as noted above.

S enantiomer, the only position of any plausibility is with the amide linkage almost parallel to the benzene ring leaving the pyrrolidinyl ring right over the Ser 218-Thr 219 peptide bond. The methyl groups p to the nitroxide are just at 3 A contact distances from this portion of the protein.

R enantiomer, similar restrictions to above. The pyrrolidinyl ring situates right over the Ser 218 C,- carbonyl bond. The Thr 219 C=O group point- ing up presents the most contacts.

Possible close contact with Ser 218 C, when under- going rotational motion. Met 192 must swing up and out of the way.

CPK models of this label also indicate an inherent restricted rotation about the NH-pyrrolidinyl ring bond.

The only difference between this label and m-V (m-CO-5NH) its “reversed” derivative, is the direction of the amide linkage. However, in this derivative (m-VIII) the pyrrolidinyl ring is placed even closer into contacts with enzyme than above (m-V).

The pyrroline ring situates in the open cleft portion of the active site near but more than 3 A from His 57.

experiments (17). The indole concentration (solubility) at this

pH was measured spectrophotometrically as 2 X 10m2 M.’

Addit,ionally, the reversible inhibitors uL-tryptophan and hippuric acid were investigated with a few of these derivatives. Saturated oL-tryptophan (approximately 0.056 M) affected the spectrum of m-1 (m-NH-5=CO) only slightly, m-V (m-CO- 5NH) and p-V (p-CO-5NH) less so, and m-VI (m-NCO-6OH)

and m-VII (m-NCO-6NH) not at all.2 Saturated hippuric acid

(about 0.018 M) gave similar results. All of these inhibitors

have comparable dissociation constants near neutral pH: at

pH 7.9, 1 M NaCl; u-tryptophan, KI = 10 mM; L-tryptophan, KI = 20 mM (22) ; at pH 8.28 (0.2 M Tris-HCl), hippurate, KI =

4.8 mM (23). Evidenbly, either t,heir binding dissociation con- stants are much larger at pH 3.5, or their specific binding orienta- tion at the tosyl hole differs from that for indole.

Trypsin-Trypsin derivatives were examined initially in the presence of 1 mM benzamidine at pH 3.5 (0.006 M acetic acid,

0.02 M CaCLJ in order to reduce the slow, yet detectable nuto- digestion at this pH (13, 24). These derivatives were later re-examined with much difficulty in the absence of beneamidine,

by guest on March 1, 2020

http://ww

w.jbc.org/

Dow

nloaded from

1674

TRYPSIN CHYMOTRYPSIN

t INDOLE

FIQ. 5. ESR spectra of selected spin label derivatives. Left, spin-labeled cu-chymotrypsin, pH 3.5; center, spin-labeled ti-chymotryp- sin at pH 3.5 in the presence of saturated indole; right, spin-labeled trypsin, pH 3.5.

and were found to give spectra which, within the limits of the experiment, were qualitatively identical with the spectra obtained earlier. Except for the six-membered piperidine carbamate and urea derivatives, m-VI (m-NCO-6OH) and m-VII (m-NCO- 6NH), respectively, the other derivatives could not be com- pletely freed of autolyzed material (0.5 to 1 y0 of the total protein in the sample). Without benzamidine present, the rate of autolysis increases in concentrated solutions of trypsin even at low pH and it is suspected that serine-blocked (i.e. sulfonylated) derivatives are even more susceptible than the native enzyme (25). Experimentally, the spectra for trypsin in the absence of benzamidine always contained a small overlapping narrow line component (that of t,he autolyzed form) which frequently ob- scured accurate spectral comparisons and measurement of some line shape parameters. What small differences arose with and without benzamidine were impossible to distinguish from the contribution due to limited autolysis. The trypsin spect,ra in Figs. 3 and 4, therefore, are judged to be the same within the accuracy of our spectral analysis, whether benzamidine was present or not (most of t,he spectra presented in those figures were for trypsin in 1 mM benzamidine at pH 3.5). The diffi- culties encountered in obtaining the clean, “nonautolyzed” spectra of Figs. 3 and 4 were many; a complete discussion of these problems and their resolution is the subject of the following paper (13). An esample of the spectral problems caused by autolyzed spinlabeled enzyme is shown in Fig. 6. In addition, all of the trypsin derivatives were examined in the presence of saturated indole (with and without 1 InM benzamidine) and were found, within esperiment,al error, to exhibit insignificant,, if any, effects on the spectra.

It seems appropriate to consider why the trypsin spectra did

not change dramatically in the presence of benzamidine. In some respects bensamidine and trypsin are analogous to indole and cY-chymotrypsin. Both are potent reversible inhibitors which bind in the respective specificity pockets (8, 12), although indole has been shown in special cases with some nonspecific ester substrates to activate deacylation by increasing the rate constant, kB (18, 22). This does not occur with benzamidine, since crystallographic work indicates that the benzamidine- binding region in benzamidine-trypsin overlaps the DIP-binding region in DIP4-trypsin.5 Nevertheless, it has been shown both in solution (26) and in the crystal5 that DIP-trypsin can bind benzamidine, even though “binding” in the reverse order is impossible (as evidenced by inhibit.ion) . Several experiments suggested that the binding of benzamidine had little if any effect on the conformational environment of these spin labels. In the past work of Berliner and Wong (24) with the fluorophosphonate spin label (DIPSL),

6

the presence of 1 mM benzamidine at, pH 3.5 or 7.1 had no effect on the spectra. As mentioned earlier with the spin label-

4 The abbreviat,ion used is : DIP, diisopropylphosphoryl. 6 M. Krieger, L. M. Kay and R. Stroud, to be published in J.

Mol. Biol. (1974).

by guest on March 1, 2020

http://ww

w.jbc.org/

Dow

nloaded from

1675

TABLE II

Summary of spin-labeled o-chymotrypsin derivatives and their indole effects

0) sianifies that the binding in the model was quite restrictive. . , Y

Label

O-I P-I

p-11 p-111

p-IV

P-V

m-1 m-111 m-IV m-V m-VI

m-VII m-VIII m-IX

(Abbreviation name)

o-NH&==CO p-NHZi==CO

p-SO*-6OH p-CO-6OH

p-CO-6NH

p-CO-5NH

m-NH-5=CO m-CO-6OH m-CO-6NH m-CO-5NH m-NCO-60H

m-NCO-6NH m-NH-5C0 m-CO-5CH20H

Binds in tosyl hole (model building)

-

+ -

- -

-

-

- - -

-I-, (3 +, 0)

-t, (?I +, (,I

+

- -- Indole effect

+ +

+ + (weak)

sulfonylated trypsin derivatives, the effects of benzamidine were much harder to distinguish, if they occurred at all. Further- more, it was probably unlikely that the spectra in Figs. 3 and 4 for spin-labeled trypsin were formed significantly in a complex with benzamidine (1 mM) at pH 3.5. Bechet and D’Albis (27) reported inhibition constants for native trypsin by benzamidine at pH 3.5 and pH 3.1 of 0.56 and 1.6 mM, respectively. It is reasonable to assume that the serine-blocked enzyme would bind even less benzamidine at this concentration. (For comparison, at pH 5.5, benzamidine-trypsin, KI g 3 X 10S5 M (27) whereas DIP-trypsin binds about 10 times less effectively, KI = 4.5 x 10d4 M (26).) That is, at the pH of these experiments, the benzamidine probably was bound fractionally to the labeled enzyme, yet effectively inhibited the unlabeled (native) trypsin in the samples.

A limited number of investigations were made at higher pH and somewhat higher benzamidine concentrations for one of the less “labile” inhibitors, m-VII (m-NCO-6NH). Trypsin in- hibited with this spin label showed no detectable spectral changes when compared at pH 3.5 and 5.67 in benzamidine concentrations from 0 to 4 mM, in the presence or absence of CaCh and in t,he presence or absence of saturated indole. At pH 6.5 the spectrum appeared to shift to slightly higher mobility (approximately a 10% increase in rotational correlation time) ; however, the accuracy of these measurements was plagued with the virtual impossibility of completely removing the small amounts of free label due to desulfonylation (13). In any case, it appeared at least for this label m-VII (m-NCO-6NH) and from the past work with the DIP analogue (DIPSL) (24) that these experi- ments were not sensitive to any (benzamidine) ligand-induced structural changes or the acidic pK, connected with this binding (26, 27). Both East and Trowbridge (26) and Bechet and D’Albis (27) discovered a pK, of about 4.5 associated with benzamidine binding to trypsin as studied by spectral and other techniques.

+ (weak)

+ (weak)

+ + + - +

+ + +

-

Comments

Two-component spectrum. Two-component spectrum; temperature sensitive;

shifts toward higher mobility in indole. Two-component spectrum. Shifts toward a more immobilized state in indole;

the effect is more apparent in saturated sucrose solutions.

Rotational correlation time increases about 8 to 10% (more immobilized) in indole; the effect is more pronounced in sucrose.

Very slight effect in either aqueous or sucrose solu- tions.

See Fig. 5. See Fig. 5. Indole effect quite similar to that for m-111 above. See Table I. Very slight effect in aqueous solution; indole effect

more pronounced in sucrose; only a highly re- strict.ed binding orientation in the tosyl hole was found from model building (see Table I).

See above for m-VI. See Table I. Indole shifts the spectrum toward higher mobility.

DISCUSSION

The results presented here emphasize several points. First, a more comp1et.e comparison of the active site conformations of two enzymes is accomplished by utilizing several structurally different conformational probes (e.g. spin labels). The spin labels described in this report offer a wide sensitivity in the range of rotational correlation times (~10~~ to lo-lo s) displayed in the spect.ra of Figs. 3 and 4.

Of most importance is the interpretation of the results with respect to the original model proposed in the introduction. It is quite clear that the two active sites probed by the 14 spin- labeled inhibitors were not conformationally identical. In fact, there generally appeared to be a more restrictive interaction in trypsin than in a-chymotrypsin. The binding of the aromatic portion of the label was probably the dominant feature account- ing for the “specificity” of these inhibitors with Lu-chymot.rypsin in contrast to the less specific binding of aromatic moieties in trypsin6 Although it was shown from model-building experi- ments that several of these spin labels were sterically restricted from binding in the same orientation as the tosyl group in tosyl- chymotrypsin, the extreme sensitivity of these labels to t.he binding of indole at the aromatic specificity pocket clearly reflected some linkage to this pocket, either proximally (steri-

6 This, of course, assumes that the aromatic group does not bind in the trypsin specificity pocket. A few of these inhibitors were tested in a trypsin model built from DIP-trypsin coordinates. It was not entirely impossible for the aromatic group to partially “dip” into the (benzamidine) specificity pocket. However, our solution results above suggest the contrary. A further thermo- dynamic argument disfavors the burying of a charged group (Asp 189) at the bottom of the pocket. On the other hand, assuming that some relationship may be attached between rate constant data and structure, Bender et al. (28) found that several acyl- trypsins and acyl-a-chymotrypsins had identical deacylation rate constants, k~, for the same (in many cases aromatic) acyl sub- stituent.

by guest on March 1, 2020

http://ww

w.jbc.org/

Dow

nloaded from

FIG. 6. Top, ESR spectrum of m-111 (m-CO-6OH) after reac- tion with trypsin and subsequent dialysis overnight against pH 3.5 (0.05 M CaCl2) dilute acetic acid. Mole percentage of au- tolyzed spin-labeled enzyme approximately 2 to 5 mole %. The autolyzed enzyme component is easily distinguished by the high field narrow line peak. Center, above after purification by SP-50 Sephadex chromatography under the conditions described in the followine article (13). The nresence of a small fraction of au- tolyzed ipin-labeled trypsin is apparent in this spectrum (about 0.1 to 0.5 mole %) as caused from subsequent autodigestion by any remaining native (active) trypsin in this sample or incom- plete chromatographic separation. This is obviated by the pres- ence of the small, but observable narrow line peak at the same field position as above. Bollom, immediately after chromato- graphic purification exceeding approximately 99.8y0 removal of autolyzed materials from the pure (a, 0) trypsin. The narrow high field line has disappeared.

tally) or allosterically. The latter mechanism is more difficult to support; in fact, crystallographic evidence indicates that no conformational changes were observed upon indole binding, within the experimental sensitivity of the measurement (about 0.2 A) (S).’ Our arguments are based purely on steric factors, although the less simple alternative choice would fit the observa- tions as well. We suggest the following scheme relating to the originally proposed model.

There are three principal binding modes for these spin-labeled inhibitors in ol-chymotrypsin, (a) binding which mimics tosyl-a- chymotrypsin; (b) partial binding near the tosyl hole, this could

7 Furthermore, serine 195-blocked DIPSL chymotrypsin, showed no changes in its ESR spectrum upon exposure to indole at pH 3.5 (B. Landis and L. J. Berliner, unpublished).

be either a partial insertion of the aromatic group into the pocket or binding in a mode just outside, “covering” the pocket; (c) binding in a “general” region of the active site, perhaps common to all “homologous” serine proteases. Due principally to steric problems, many of those derivatives which bind in the a or b mode will exist in a small yet finite equilibrium with the other binding modes. Therefore, an equilibrium should exist which describes the general binding of any inhibitor:

with equilibrium const.ant,s K,b and Kbe. Furthermore, the binding of the inhibitor indole (I) to the tosyl pocket

can, in effect, displace the spin label from a or b t.o c if the indole- binding constant, K,, is sufficiently competitive wit’h the “effec- tive” binding constant Kab or Kbc of the sulfonyl inhibitor. It is obvious that a label which binds in a might be more difficult to displace (i.e. binds more strongly) than a label at b. In the studies presented here it was found that those labels which could not bind at a (the tosyl hole) displayed dramatic indole effects in most cases. This corresponds in our model to a shift from b to c. In several cases, the chymotrypsin spectra became qualitatively quite similar to those for trypsin upon indole exposure, suggesting that in trypsin the labels bind at a site identical with c. A possible orientation for c is suggested by the structure of tosylelastase where the tosyl group is pointing almost 180” from its tosyl-cr-chymotrypsin orientation, almost sandwiched to histidine 57 (29).

The validity of this model may be difficult to confirm by other methods. On the other hand, the experimental evidence is sufficiently strong to fit such a model as a reasonable possibility. The significance of the labels giving indole effects with “trypsin- like” spectra is more difficult to confirm, since not all of the spin labels behaved similarly. A similar comparison with the enzyme elastase may yield corroborating results. This enzyme resembles chymotrypsin in several respects with the prime exception of a blocked entrance to its specificity pocket (29).

Residues Involved in Structural Discrimination-Although there are over-all homologies in a-chymotrypsin and trypsin conformation, some small differences do arise in the vicinity of the active site. These are principally an addition or deletion of a peptide in a portion of a chain or slight differences in back- bone folding. For instance, when fitting several of the spin- labeled inhibitors in the chymot.rypsin active site model, it was found that most interactions with the nitroxide ring took place along the chain Ser 217.Ser 21%Thr 219.Cys 220. Assuming that the aromatic sulfonyl group would bind similarly in trypsin, we note that its corresponding sequence is Ser 217-deletion (218) ,-Gly 219.Cys 220 (12). In cY-chymotrypsin this chain is directed more toward the observer if viewed from the a*-b (x-y) face near the dyad axis of the molecule. Such a difference is probably discounted in an overview of the two enzyme strut- tures, yet may possibly be the significant point of discrimination in the study presented above. It is interesting to consider further the general observation that in both this study and the DIPSL spin label work of Morrisett and Broomfield (25), the spin label was consistently more immobilized in trypsin than in cY-chymotrypsin. Although these two enzymes possess many homologies in their three-dimensional structures (12), there are obviously subtle differences which may only distinguish them- selves in a sensitive spin label experiment. One such possibility

by guest on March 1, 2020

http://ww

w.jbc.org/

Dow

nloaded from

was forwarded above. However, another related aspect is the well documented existence of an auxiliary hydrophobic binding site in trypsin. For example, Sanborn and Hein (30) showed the existence of a neutral molecule binding site which was sepa- rate from the primary substrate binding site. There was also further evidence for an auxiliary hydrophobic site from the work of Heidberg et al. (31) and Hartmann and Holler (32) on alkyl amine inhibition of trypsin catalysis, the substrate activation studies of Trowbridge et al. (33) and product activation work of Howard and Mehl (34) to mention only a few. Since all of the spin label moieties were of a hydrophobic nature, it cannot be ruled out that much of the immobilization was manifested in selective binding to this hydrophobic binding region, an aspect which will deserve further study.

Finally, in closing, we may note that some other features of the general approach utilized here become apparent. First, al- though one studies the structural features of an enzyme in solu- tion, it is imperative that the details of the crystallographic structure are considered and examined carefully, if only as a guideline. Conversely, the crystallographic structure is useful only as a possible static structure which cannot always be used to predict dynamic structural states in solution. In particular we are referring to the predict,ion of a dynamic picture of the binding of a small inhibitor molecule at the active site. Second, it became apparent in these studies that in choosing an aromatic probe for which one of the enzymes was more specific, one could no longer assume that the probe (spin label) u-as inert or impartial to the structural environments. As noted above, at least two classes of binding were possible in cr-chymotrypsin where the spin-labeled aromatic sulfonates adopted orientations in and near the specificity pocket, respectively. In any case, if one knew essent.ially nothing about the chymotrypsin and trypsin structures, obviously the same conclusion, that their active site structures were indeed different, could have been drawn, albeit with less structural detail. Perhaps t.he definition of conforma- tion here should be modified to include potential noncovalent interactions in addition to the relative juxtaposition of residues.

Acknowledgments-We are especially grateful to Dr. Robert Stroud and Mr. Monty Krieger, California Institute of Tech- nology, both for kind hospitality while examining their DIP- trypsin model, for many helpful discussions, and for their trans- mittal of prepublication results. We also thank Doctors David Macmurchie and Richard Henderson for helpful comments.

REFERENCES

1. FAHRNEY, D. E., .IND GOLD, A. M. (1963) J. Amer. Chem. sot. 86, 997-1000

2. SIGLER, P. B., JEFFERY, B. A., M.ITTHEWS, B. W., AND BLOW, D. M. (1966) J. Mol. Biol. 16, 175-192

3. SIGLER, P. B., SKINNICR, H. C. W., COULTI~R, C. L., K~LLOS, J., BRAXTON, H., AND DAVIES, D. R. (1964) Proc. Nut. Acad. Sci. U. S. A. 61, 1146-1151

1677

4. WATSON, H. C., SHOTTON, D. M., Cox, J. M., AND MTJIRHEAD, H. (1970) Nature 226.806-811

5. WRIGHT, d. S., ALDEN, R. A., AND KRAUT, J. (1969) Nature 231, 235-242

6. SIGLER, P. B., BLOW, D. M., MATTHEWS, B. W., AND HENDER- SON, R. (1968) J. Mol. Biol. 36. 143-164

7. BLOW, D. M. (1971) in The Enzymes (BOYER, P., ed) Vol. III, Ch. 6, Academic Press. New York

8. STEITZ,’ T. A., HENDERSON, R., AND BLOW, D. M. (1969) J. Mol. Biol. 46, 337-348

9. PETERSON, R. L., HUBELE, K. W., AND NIEMANN, C. (1963) Biochemistry !4,942-946

10. HAMILTON, C. L., AND MCCONNELL, H. M. (1968) in Structural Chemistry and Molecular Biology (RICH, A., AND DAVIDSON, N., eds) pp. 115-149, W. H. Freeman & Company, San Fran- cisco

11. WALSH, K. A., AND NEURATH, H. (1964) Proc. Nat. Acad. Sci. U. S. A. 62,884-889

12. STROUD, R. M., KAY, L. M., AND DICKERSON, R. E. (1971) Cold Sprins Harbor Sumv. &ant. Biol. 36. 125-140

13. WONG, S: S., &UIGGLE, k,.TR&LETT, C., AND BERLINER, L. J. (1974) J. Biol. Chem. 249, 16781682

14. BIRKTOFT, J. J., AND BLOW, D. M. (1972) J. Mol. Biol. 68, 187-240

15. B&wN,C. J., END CORBRIDGE, D. E. C. (1954) Acta Crystal- lop. 7, 711-715

16. PENFOLD, B. R., AND WHITE, J. C. B. (1959) Acta Crystallogr. 12, 130-135

17. GAREL, J.-R., AND LABOUESSS, B. (1970) J. Mol. Biol. 47, 41-56 -_ __

18. KOSMAN, D. J., HSIA, J. C., AND PIBTTF,, L. H. (1969) Arch. Biochem. Biovhus. 133. 29-37

19. LEE, J. C., FRITOS, R. P., HIRSH, J., THOMAS, J., AND TIMA- SHEFF, S. N. (1973) Fed. Proc. 32, 496

20. FERSHT, A. R., AND REQUENA. Y. A. (1971) J. Amer. Chem. . , Sot. 93, 7079-7087

21. KOSMAN. D. J. (1972) J. Mol. Biol. 67. 247-264 22. FOSTER,‘R. J. (1961) 2. Biol. Chem. 286; 2461-2466 23. KEZDY, F. J., FF,DER, J., AND BENDER, M. L. (1967) J. Amer.

Chem. Sot. 89. 1009-1016 24. BERLINER, L.-J.‘, AND WONG, S. S. (1973) J. Biol. Chem. 248,

1118-1120 25. MORRISETT, J. D., AND BROOMFIELD, C. A. (1972) J. Biot.

Chem. 247, 722447231 26. EAST, E. J., -&ND TROWBRIDGE, C. G. (1968) Arch. Biochem.

Biophys. 126, 334-343 27. BECHET, J.-J., AND D’ALBIS, A. (1969) Biochim. Biophys.

Acta 178, 561-576 28. BENDER, M. L., KILLHEFFER, J. V., JR., BND KEZDY, F. J.

(1964) J. Amer. Chem. Sot. 86, 5330-5331 29. SHOTTON, D. M., AND WATSON, h. C. (1970) Nature 226, 811-

816 30. SANBORN, B. M., AND HF,IN, G. E. (1968) Biochemistry 7,

3616-3624 31. HEIDBERG, J., HOLLER, E., AND HARTMANN, H. (1967) Ber.

Bunsenges. Phys. Chem. 71, 19-29 32. HARTMANN, H., AND HOLLER, E. (1970) Eur. J. Biochem. 16,

80-91 33. TROWBRIDGE, C. G., KREHBIEHL, A., AND L~SKOWSKI, M.,

JR. (1963) Biochemistry 2, 843-850 34. HO~.IRD, S. M., AND MEHL, J. W. (1965) Biochim. Biophys.

Acta 106, 594-596

by guest on March 1, 2020

http://ww

w.jbc.org/

Dow

nloaded from

Lawrence J. Berliner and Shan S. Wong-CHYMOTRYPSIN AND TRYPSIN

αCOMPARISON OF THE ACTIVE SITE ENVIRONMENTS IN Spin-labeled Sulfonyl Fluorides as Active Site Probes of Protease Structure: I.

1974, 249:1668-1677.J. Biol. Chem. 

  http://www.jbc.org/content/249/6/1668Access the most updated version of this article at

 Alerts:

  When a correction for this article is posted• 

When this article is cited• 

to choose from all of JBC's e-mail alertsClick here

  http://www.jbc.org/content/249/6/1668.full.html#ref-list-1

This article cites 0 references, 0 of which can be accessed free at

by guest on March 1, 2020

http://ww

w.jbc.org/

Dow

nloaded from