FEBS Letters 582 (2008) 2149–2154

Calcium-induced tripartite binding of intrinsically disorderedcalpastatin to its cognate enzyme, calpain

Robert Kissa, Zoltan Bozokyb, Denes Kovacsb, Gergely Ronab, Peter Friedrichb, Peter Dvortsakc,Rudinger Weisemannc, Peter Tompab, Andras Perczela,d,*

a Institute of Chemistry, Laboratory of Structural Chemistry and Biology, Eotvos Lorand University, P.O. Box 32, H-1538 Budapest, Hungaryb Institute of Enzymology, Biological Research Center, Hungarian Academy of Sciences, Budapest, Hungary

c Bruker BioSpin GmbH, Silberstreifen, 76287 Rheinstetten, Germanyd Protein Modeling Group MTA-ELTE, Institute of Chemistry, Eotvos Lorand University, P.O. Box 32, H-1538, Budapest, Hungary

Received 29 March 2008; revised 2 May 2008; accepted 6 May 2008

Available online 2 June 2008

Edited by Stuart Ferguson

The authors would like to dedicate this work to Prof. Medzihradszky Kalman on the occasion of his 80th birthday.

Abstract The activity of calpain is controlled by the free intra-cellular calcium level and by the protein�s intrinsically disorderedendogenous inhibitor, calpastatin, mediated by short conservedsegments: subdomains A–C. The exact binding mode of calpast-atin to the enzyme has until now been unclear. Our NMR data ofthe 141 amino acid long inhibitor, with and without calcium andcalpain, have revealed structural changes and a tripartite bindingmode, in which the disordered inhibitor wraps around, and con-tacts, the enzyme at three points, facilitated by flexible linkers.This unprecedented binding mode permits a unique combinationof specificity, speed and binding strength in regulation.

Structured summary:

MINT-6549073:

Calpain-2 catalytic subunit (uniprotkb:P04632), Calpain-2 cat-

alytic subunit (uniprotkb:P17655) and calpastatin (uni-

protkb:P20810) physically interact (MI:0218) by nuclear

magnetic resonance (MI:0077)

� 2008 Federation of European Biochemical Societies.Published by Elsevier B.V. All rights reserved.

Keywords: Intrinsically disordered protein; Supramolecularcomplex by NMR; Ligand binding specificity

1. Introduction

Calpain is an intracellular calcium-activated cysteine prote-

ase present in all eukaryotic cells [1]. It controls the activity

of over one hundred target proteins [2] by limited proteolysis

in the regulation of cell division, differentiation and cell motil-

ity. It occurs together with its endogenous protein inhibitor,

Abbreviations: IDP, intrinsically disordered protein; hCSd1, humancalpastatin domain 1; hCSd1 Æ Ca2+, human calpastatin domain 1 withequivalent CaCl2; complex, human calpastatin domain 1 and m-calpain complex; cCS, combined chemical shift; A, subdomain A; B,subdomain B; C, subdomain C

*Corresponding author. Address: Institute of Chemistry, Laboratoryof Structural Chemistry and Biology, Eotvos Lorand University, P.O.Box 32, H-1538 Budapest, Hungary. Fax: +36 1 3722 620.E-mail address: perczel@chem.elte.hu (A. Perczel).

URL: http://www.chem.elte.hu/departments/protnmr.

0014-5793/$34.00 � 2008 Federation of European Biochemical Societies. Pu

doi:10.1016/j.febslet.2008.05.032

calpastatin [3]. The loss of the control of different calpain

forms is implicated in a wide array of diseases ranging from

cataract through muscle dystrophy to diabetes [4]. Despite its

immense biomedical interest, the atomic details of the interac-

tion of calpastatin with calpain are not yet understood. This

paper aims to close this gap.

Calpastatin is an intrinsically disordered [5] protein, com-

posed of an N-terminal L-domain and four equivalent inhibi-

tory domains, each of about 140 amino acids in length [6].

Every inhibitory domain contains three short conserved seg-

ments, termed subdomains, which are primarily responsible

for calpastatin binding to calpain. Subdomain A and C (or

A and C) potentiate inhibition, as they tether the inhibitor to

the enzyme in a calcium-dependent manner, but do not inhibit

enzyme activity on their own [3,7]. The middle segment, B,

binds in the active-site region of the enzyme and carries inhib-

itory potential in itself. In its unbound form, calpastatin lacks

a well-defined structure and falls into the class of intrinsically

disordered proteins, IDPs [8,9].

IDPs play important roles in signal transduction and tran-

scription regulation [10]. Their functions stem from molecular

recognition, in which a part of the IDP becomes ordered in a

process of folding induced by its partner [11], but part of it re-

mains disordered [12]. Uncovering their bound states and the

process of recognition is key to understand IDP action at the

molecular level, as well as to extend the structure–function par-

adigm [13]. Most often, IDPs only bind their partner via a

short recognition element [14], and less frequently by virtue

of two binding elements separated by a disordered linker re-

gion [15]. Only in one case so far has the tripartite binding

of an IDP been described: protein phosphatase 1 (PP1) by

inhibitor-2 (I2) [9,16].

The uniqueness of the interaction of calpastatin with calpain

stems from biochemical data also suggesting its tripartite bind-

ing, with the involvement of the three subdomains and the

intervening disordered linker regions. Due to technical difficul-

ties, however, the structure of the calpain–calpastatin complex

has not yet been solved. The only part of the structure seen is

C, which adopts a helical structure when bound to the small

subunit of calpain [17]. The binding of other segments of cal-

pastatin can only be inferred from the homology of calpastatin

A–C, and the inhibitory action of B. Here we address the

structural state of calpastatin in the bound form by multi-

blished by Elsevier B.V. All rights reserved.

2150 R. Kiss et al. / FEBS Letters 582 (2008) 2149–2154

dimensional NMR. It is to be emphasized that NMR charac-

terization of the complexes of IDPs is limited by size, and thus

the longest IDP segment structurally characterized by NMR in

its partner-bound state is 80 amino acids (PDB_entry: 1kxr).

Thus, our attempt to delineate the structural state of the 141

amino acids hCSd1 bound to calpain (MW > 120 kDa) ven-

tures into uncharted territory in terms of size.

D1H and D13C secondary chemical shifts in the presence of

calcium show that the acidic segments on the C-terminal side

of A and C bind Ca2+. Previous models of calpain–calpastatin

binding have suggested that calcium binds calpain, and brings

about its activation by a conformational change [18], which re-

sults in the binding of calpastatin [3]. One study, however, has

suggested that calpastatin can also bind calcium in a way that

may structurally prime it for calpain binding [19]. This novel

finding suggests that calcium-binding of calpastatin may di-

rectly regulate calpastatin action. In addition, titration of la-

beled calpastatin with calpain has shown that interaction of

the two proteins is limited to three short regions, correspond-

ing to A–C. This is an example of a tripartite binding by an

IDP, which binds its partner via three short recognition ele-

ments separated by two linker regions that remain disordered.

We suggest that this mode of binding, which has so far been

observed in the case of the I2-PP1 complex only, may provide

a unique combination of specificity and binding strength.

2. Materials and methods

2.1. Preparation of calpastatin domain 115N- and 15N-, 13C-labeled human calpastatin domain 1 (hCSD1)

(corresponding to A137–K277 of hCSD1, SwissProt P20810) was pro-

cCS ¼ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiðdðHNHÞSam2 � dðHNHÞSam1Þ

2 þ 0:17ðdðNNHÞSam2 � dðNNHÞSam1Þ20:32ðdðCaÞSam2 � dðCaÞSam1Þ

2

q

duced by using the pEThCSD1 vector (courtesy of Professor Masato-shi Maki) and the Escherichia coli BL21(DE) strain. Transformed cellswere grown in a minimal medium (5 g glucose, 6 g Na2HPO4, 3 gKH2PO4, 1 g NH4Cl, 0.5 g NaCl, 0.12 g MgSO4 and 0.01 g CaCl2per liter of medium) at 37 �C to A600 = 0.4–0.5, induced with0.5 mM isopropyl-b-DD-thiogalactopyranoside and grown overnight at30 �C to allow protein expression. The 15N- and 15N–13C-labeled pro-tein was expressed similarly. The protein was initially purified as de-scribed previously [5], and then on a C-18 HPLC column: finalpurity >98%.

2.2. Preparation of calpainThe BL21(DE3) E. coli strain was transformed with expression vec-

tors coding for the inactive (C105S) m-calpain large and small subunitsprovided by Prof. JS. Elce. Cells were grown on NZYM medium (1%N-Z-Case Plus I (Sigma), 0.5% select yeast extract (Sigma), 0.5% NaCland 0.2% MgSO4) containing the appropriate antibiotics (carbenicillin50 lg ml�1, kanamycin 20 lg ml�1 at 37 �C, 250 rpm until A600 = 0.6–0.8). Inactive m-calpain was prepared from the bacteria soluble frac-tion. Expression was induced by 0.4 mM isopropyl-1-thio-b-DD-gala-lctopyranoside at 30 �C for 3 h. The culture was cooled and the cellswere collected by centrifugation at 2500 · g, 20 min, 4 �C and resus-pended in lysis buffer (50 mM TRIS Æ HCl pH 7.50; 300 mM NaCl;1 mM EDTA; 5 mM benzamidine; 1 mM phenylmethylsulphonyl-fluo-ride; 10 mM b-mercaptoethanol). The sample was sonicated for

6 · 15 s at 16 lM on ice and the lysate was centrifuged at100000 · g; 60 min; 4 �C. The supernatant (added 10 mM MgCl2 toblock EDTA) was applied to a 5 ml Ni-NTA resin (equilibrated with50 mM TRIS Æ HCl pH 7.50; 300 mM NaCl). The column was washedwith 15 ml buffer (50 mM TRIS Æ HCl pH 7.50; 300 mM NaCl; 15 mMimidazole; 5 mM benzamidine; 1 mM phenylmethylsulphonyl-fluoride;10 mM b-mercaptoethanol) and the protein was eluted with imidazole(250 mM) containing buffer. The purified protein was dialyzed into cal-pain buffer (50 mM TRIS pH 7.50; 150 mM NaCl; 1 mM EDTA;5 mM benzamidine; 1 mM phenylmethylsulphonyl-fluoride; 10 mMb-mercaptoethanol) at 4 �C. A total of 17 mg calpain was obtainedfrom 9 dm3 medium. The protein solution was concentrated on Ami-con Ultra-15 column (Millipore) and dialyzed into the NMR measurebuffer (9 mM PIPES pH 6.18; 15 mM NaCl; 10 lM CaCl2; 0.5 mMbenzamidine; 0.1 mM phenylmethylsulphonyl-fluoride).

2.3. NMR assignment and relaxation measurementsNMR samples contained �1 mM hCSD1 dissolved in:Sample 1 (hCSD1 Æ Ca2+): 90% H2O/10% D2O with equivalent CaCl2

(10 lM) and 15 mM NaCl.Sample 2 (hCSD1): in 90% H2O/10% D2O, with 9 mM PIPES buffer

without Ca2+ ion.All NMR measurements were performed at 300 K. Resonance

assignment of backbone carbon and nitrogen atoms was obtained from[1H, 15N]-HSQC, HNCA, HN(CO)CA, HNCACB, HN(CO)CACBand CC(CO)NH type experiments (a total of 2048 complex points inthe 1H- and 128 and 64 complex points in the 15N- and 13C-dimensionswere acquired). Typical spectral widths were 14 ppm in the direct,32 ppm in the 15N- and 75 ppm in the Ca-, Cb-, and 32 ppm in theCa dimension. Data were acquired for hCSD1 Sam1 and last step oftitration of Calpastatin with calpain, by using a spectral width of10 ppm · 23 ppm (2k · 256 complex points). For NOE enhancementpeak intensities were used.

To characterize the Ca2+ dependence of hCSD1, N-, NH- and Cachemical shifts derived from HSQC and HNCA spectra of Sam1 andSam2 were used, the combined chemical shifts (cCS) are calculatedas follows:

Calpain titration was performed by adding small amounts of calpainto hCSD1 in Sam2. Calpain stock contained 9 mM PIPES pH 6.18;15 mM NaCl; 10 lM CaCl2; 0.5 mM benzamidine; 0.1 mM phenylm-ethylsulphonyl-fluoride. In first three titration steps 30 ll aliquots ofcalpain stock were added to 330 ll of calpastatin solution and then60 ll–60 ll in the last two steps. Ligand–substrate ratio in each titra-tion steps were: 0%, �60%, �120%, �240%, �360% and 460%. Duringtitration steps with calpain HSQC spectra (Fig. S1) were collected ineach step and peak intensities were used to characterize binding to cal-pain. The correction for dilution was performed using the signal inten-sity of K141, residue surely unaffected by bounding. Data processed byNMRPipe [20], resonance assignments completed within Sparky 3.113[21].

3. Results

The high flexibility of the backbone leads to narrower lines,

enabling full resonance assignment of both hCSD1 and

hCSD1 Æ Ca2+ even at a sample concentration <0.5 mM. A to-

tal of 123 spin-systems were identified for the hCSD1 Æ Ca2+

and 121 for the Ca2+ free form of hCSD1. An unresolved ques-

tion of the calpain–calpastatin system is whether hCSD1 itself

Fig. 1. The effect of Ca2+ on hCSD1 as reported by combined {NNH–HNH–Ca} chemical shifts (see Section 2). The NNH, HNH and Ca data wererecorded for the Ca2+ bond (Sample1) and free (Sample2) states of hCSd1 by using specific triple resonance NMR experiments. Subdomains A and Chave a tendency to form helix.

Fig. 2. Relative signal intensity changes as recorded by 1H–15N HSQC spectra for the free hCSD1 and hCSD1 gradually titrated with unlabeledcalpain; hCSD1:calpain ratios are 1:0, 1:0.6, 1:1.2, 1:2.4, 1:3.6, 1:4.6. Note that even in the complex, except residues belonging to the core ofsubdomains A–C, backbone amide groups of hCSd1 are sufficiently separated from calpain to relax independent of the supramolecular complex,preserving their fragmental molecular motion. Interestingly, 1H–15N intensity of A20, D22 and T27 residues of subdomain A and D93, S101, F103 ofsubdomain C drops after the second titration step, indicating their tighter binding to calpain.

Table 1The original and the fine-tuned boundaries of subdomain A–C of hCSd1

Subdomain of hCSD1 A B C

Conserved regions A1–P11 S12–G30 P31–P49 M50–R70 E71–P89 G91–T104 T104–K141

Functional regions A1–K14 S15–G29 G30–K60 R61–L72 L73–P89 I90–T109 A110–K141

NMR signal intensity at 100%a 49.2 2.4 40.6 27.4 50.1 8.6 47.1

aReassessment was completed by using the present hCSd1:calpain complex.

R. Kiss et al. / FEBS Letters 582 (2008) 2149–2154 2151

Tab

le2

het

-NO

Evalu

eso

fse

lect

edre

sid

ues

of

15N

lab

eled

hC

Sd

1in

its

free

(a)

an

dca

lpa

inb

on

d(b

)fo

rmin

the

pre

sen

ceo

f1

0l

MC

a2+

AA

#V

2E

5S

12

D23

G30

E35

T40

E45

K68

K75

I79

G81

A84

D97

A110

S120

S128

V132

A136

NO

Ea

(Sa

mp

le1

)�

1.0

�0

.80

.7n

.a0

.10

.70

.90

.80

.80

.80

.80

.40

.2n

.a.

0.7

0.8

0.6

�0

.7�

0.8

NO

Eb

ou

nd

b�

1.5

�1

.0�

1.7

�1

.7�

1.3

�1

.1�

2.4

�2

.3�

2.1

�2

.9�

2.9

�1

.5�

2.9

�2

.5�

2.0

�2

.7�

1.8

�1

.0�

1.1

No

teth

at

ba

ckb

on

efl

exib

ilit

yd

ecre

ase

sin

the

bo

un

dfo

rm(p

osi

tiv

eN

OE

sb

etw

een

sub

do

ma

ins

Aa

nd

C)

wit

hre

spec

tto

the

free

form

of

hC

Sd

1.

2152 R. Kiss et al. / FEBS Letters 582 (2008) 2149–2154

binds Ca2+. By using CD-spectroscopy, it was shown for a

fragment of C that it binds Ca2+ [19]. The combined chemical

shift (cCS) of each amino acid, except Pro, reveals the residue-

specific calcium ion binding capacity of hCSd1 (Fig. 1). The

largest shift is detected for E33–E37 positioned right after the

C-terminal part of A, where the COO� groups can catch

Ca2+. The glutamic acid-rich region around E119 (C-end of

C) also present signs of Ca2+ binding. In both cases, cCS

changes are either due to the direct influence of Ca2+ or to

ion-induced structural shifts, i.e. local backbone rearrange-

ment(s), as seen for Y54, Y69 and K60, where direct binding

of Ca2+-ions is less probable.

Based on the estimated rotational diffusion correlation time

(sc) of a supramolecular complex of 123 kDa, such as hCSd1/

calpain, no detectable NMR signal is expected. The gradual

titration of the non-selectively 15N-labeled hCSd1 by calpain

has revealed the immediate H-NH signal intensity loss of se-

lected residues. As expected due to the tight binding of calpain

to calpastatin (Kd = 4.5 · 10�12), most residues of A and C

bind at high affinity to calpain, and thus lose fragmental back-

bone motion typical of IDPs (Fig. 2). However, other residues

located between the binding sites and preceding A and follow-

ing C preserve their characteristic IDP�s motion and still relax

slowly. Thus, they remain detectable by a conventional sensi-

tivity-enhanced HSQC (Figure S1) and no TROSY-type exper-

iment was required. During titration, no significant chemical

shift changes were observed, but signals of free hCSD1 are de-

tected only. The absolute signal intensities obtained (Figure

S2), after correction for dilution (see Section 2) (Fig. 2), show

that the boundaries of the binding sites overlap almost per-

fectly with values derived from primary sequence homology

[1] (Table 1). In the absence of Ca2+, no binding of calpastatin

to calpain was seen, i.e. fast exchange was observed in the

whole concentration range (data not shown), in accord with

previous observations (reviewed in Refs. [1,3]).

In general, the most affected signals in the presence of Ca2+

correspond to S15–G29 and I90–T109 residues, typical bound-

aries of A and C. The sCS analysis of hCSd1 in its free form

has revealed for these residues a limited tendency of forming

residual or nascent helices. Detailed analysis of the complex

shows that within these regions G16–L28 in A and G91–F103in

C bind the strongest (Figure S2). At both ends of A and C,

Pro-rich regions are detected (P and P–X–Y–P), most probably

for breaking both helices and for ‘‘elevating’’ the forthcoming

residues, preventing their aspecific binding to calpain. This

molecular strategy is rather efficient as a few residues away

from the tightly bound S15–G29 and I90–T109, segments relax

independent of the complex, presenting a flexible and frag-

mented backbone motion typical of IDPs. B interacts with cal-

pain, but not exactly in the expected M50–R70 region. The

present titration experiment shows that the interaction is

shifted toward the C-terminus of the molecule to the region

R61–L72, and it is less tight than previously thought. Further-

more, the two residues involved most in the binding of B, Y54

and Y69, bind less tightly than those of A and C (Figs. 2 and

S2).

The het-NOEs of calpain–bound calpastatin were measured

and compared with those of the free hCSD1. Values of selected

and functionally important amino acid (Table 2) show that the

C- and N-terminal residues of the bound calpastatin have al-

most the same flexibility as in the free form. (The NOEs of

the tightly bound residues can not be measured.) However,

R. Kiss et al. / FEBS Letters 582 (2008) 2149–2154 2153

in the bound form amino acids of the region between A and C

show a somewhat more ordered state (positive NOEs) than in

free hCSd1.

Fig. 3. Schematic representation of calpain: calpastatin complexdepicting the tripartiate binding mode.

4. Discussion

Details of the binding mechanism of the IDP calpastatin are

important for a full understanding of the regulation of cal-

cium-activated protease, and to extend our knowledge on the

structure–function relationship of IDPs. These proteins often

carry out their functions by molecular recognition [8,9,11], in

which one or two short regions bind to the partner and become

locally ordered, while flanking regions remain disordered [12].

Our studies throw new light on both calpastatin structure and

the mechanism of binding of calpastatin to calpain.

The first intriguing finding of our studies is that calcium has

a direct effect on calpastatin. It is known, and corroborated by

our observations, that in the absence of calcium calpastatin

does not bind calpain [3,22]. This lack of binding has been

interpreted as a result of the calcium-stimulated conforma-

tional change in calpain, resulting both in activation of the en-

zyme [3] and structural competence with respect to calpastatin

binding. The observation that calpastatin binding requires

lower calcium concentration than enzyme activity [23] has been

explained as suggesting that calpastatin binding shifts the con-

formation of the enzyme to the active form [7]. Our results here

suggest an alternative explanation, namely that A and C of

hCSd1 are responsible for the calcium-induced tethering of

the inhibitor to the enzyme by positioning B next to its active

site [24]. Both A and C are delimited by carboxy-terminal

acidic regions, which directly bind Ca2+ as shown by pro-

nounced changes in sCS. The large negative charge of these re-

gions (E32ETEEE37 and E116KEESTE122) might be neutralized

by Ca2+, which is favorable for the formation of a helix at the

subdomains due to stabilizing the electrostatic dipole that

develops upon helix formation. A similar effect is seen in the

case of helix formation promoted by phosphorylation [25].

As A and C form transient helices in isolation [19], and bind

calpain as amphipathic helices [17], their stabilization by

Ca2+ may be an important mechanistic step in the regulation

of calpastatin action.

The second important result is the binding mode of calpast-

atin to calpain. Prior experiments have suggested that calpast-

atin may only bind via the three subdomains, which can bind

calpain on their own [1,24], and mutations of which results in a

significant reduction of inhibitory potential. The exact mode of

binding could not be elucidated, because of the size of the com-

plex (122 kDa) and resistance of calpain and calpain–calpasta-

tin complex to attempts of crystallization in the presence of

Ca2+. The structure of full-length calpain could only be solved

in the absence of calcium, whereas the conformational change

brought about by calcium binding could be inferred from stud-

ies of a truncated version [18]. There have been attempts to

assemble the model of calpastatin binding from separate stud-

ies of its conserved regions, i.e. A and C [26] and B [27]. From

these, the picture has emerged that binding is limited to the

short segments, with the connecting linker regions, which

may even remain disordered in the bound state, of secondary

importance. Our results for the first time provide evidence

for this tripartite binding mode. Titration of labeled calpasta-

tin with calpain unequivocally shows that regions S15–G29and

I90–T109 (Table 1) bind calpain the strongest, and become fully

immobilized in the complex. The middle region R61–L72 (Table

1) corresponding to B, is also in direct contact with the en-

zyme, but its binding is weaker, transiently sampling unbound

states even in the complex. The linker regions connecting the

subdomains, i.e. G30–K60and L73–P89, remain in a state similar

to the unbound form of high mobility. The terminal regions,

A1–K14 and A110–K141, also remain disordered, not establish-

ing specific contacts with calpain. Such a binding mode is

rather uncommon for IDPs. Most often, they bind via a short

recognition segment with the flanking region remaining disor-

dered [14]. Sometimes IDPs employ two recognition elements

connected by a linker region, which remains disordered in

the bound state. Such bi-partite binding has been described

for Ste5, Oct1 [15], and bacterial cellulose [28]. Tripartite bind-

ing of an IDP has only been described for I2 binding to PP1

[16], and its general occurrence may be limited by the large

and unfavorable decrease in configurational entropy, to be

compensated by advantages of excess specificity.

The present results confirm the previous definition of subdo-

mains in calpastatin, but refine their exact locations (Table 1).

For the first time, subdomains were defined as three well-con-

served regions encoded by separate exons [26,29]. Other se-

quence alignment studies have suggested that A is located

between S12 and G30 and C is between G91 and T104 [3]. These

segments have been suggested as forming a-helices when

bound to calpain, and interference with their helix-forming po-

tential by point mutations does cause a drastic decrease in

binding strength [30]. Our results define the functional regions

at A as S15–G29, and at C as I90–T109 directly involved in bind-

ing. With respect to B, the borders of the corresponding exon

are D48–A74, and the corresponding peptide shows strong

inhibitory potency [29], though M50–R70 segment is established

as B. Several studies have addressed the core region of inhibi-

tory potential and found that the carboxy-terminal T64–R70

segment is the one primarily responsible for inhibition. In

our experiments the binding of B is not uniform. The edges

(S52–I55 and E62–L72) show stronger binding capacity than

the middle (E56–R61), which might help establish the correct

position of the inhibitory segment.

Our results suggest a tripartite binding mode, in which the

disordered inhibitor wraps around and contacts calpain at

three distinct points (Fig. 3). This binding makes it possible

to combine calcium-dependent recognition by ensuing inhibi-

2154 R. Kiss et al. / FEBS Letters 582 (2008) 2149–2154

tion at distinct sites, enabled by the flexibility of the linker re-

gions. Calpastatin is known to be the only inhibitor of calpain

that is highly specific to the enzyme. We suggest that the ob-

served unprecedented binding mode enables the unique combi-

nation of specificity, high speed and reversibility in binding

and calpain regulation. Due to the apparent functional bene-

fits, other IDPs may also turn out to utilize similar binding

strategies.

Acknowledgements: This work was supported by OTKA (HungarianScientific Research Fund) Grants NK72973, NK71582, K60694, NI-68466, and by Wellcome Trust ISRF067595.

Appendix A. Supplementary material

Supplementary data associated with this article can be

found, in the online version, at doi:10.1016/j.febslet.2008.

05.032.

References

[1] Goll, D.E., Thompson, V.F., Li, H.Q., Wei, W. and Cong, J.Y.(2003) The calpain system. Physiol. Rev. 83, 731–801.

[2] Tompa, P., Buzder-Lantos, P., Tantos, A., Farkas, A., Szilagyi,A., Banoczi, Z., Hudecz, F. and Friedrich, P. (2004) On thesequential determinants of calpain cleavage. J. Biol. Chem. 279,20775–20785.

[3] Wendt, A., Thompson, V.F. and Goll, D.E. (2004) Interaction ofcalpastatin with calpain: a review. Biol. Chem. 385, 465–472.

[4] Zatz, M. and Starling, A. (2005) Mechanisms of disease: calpainsand disease. New Engl. J. Med. 352, 2413–2423.

[5] Csizmok, V., Bokor, M., Banki, P., Klement, T., Medzihradszky,K.F., Friedrich, P., Tompa, K.A. and Tompa, P. (2005) Primarycontact sites in intrinsically unstructured proteins: the case ofcalpastatin and microtubule-associated protein 2. Biochemistry-USA 44, 3955–3964.

[6] Emori, Y., Kawasaki, H., Imajoh, S., Minami, Y. and Suzuki, K.(1988) All 4 repeating domains of the endogenous inhibitor forcalcium-dependent protease independently retain inhibitory activ-ity – expression of the CDNA fragments in Escherichia coli. J.Biol. Chem. 263, 2364–2370.

[7] Tompa, P., Mucsi, Z., Orosz, G. and Friedrich, P. (2002)Calpastatin subdomains A and C are activators of calpain. J.Biol. Chem. 277, 9022–9026.

[8] Tompa, P. (2002) Intrinsically unstructured proteins. TrendBiochem. Sci. 27, 527–533.

[9] Dunker, A.K., Brown, C.J., Lawson, J.D., Iakoucheva, L.M. andObradovic, Z. (2002) Intrinsic disorder and protein function.Biochemistry 41, 6573–6582.

[10] Iakoucheva, L.M., Brown, C.J., Lawson, J.D., Obradovic, Z. andDunker, A.K. (2002) Intrinsic disorder in cell-signaling andcancer-associated proteins. J. Mol. Biol. 323, 573–584.

[11] Dyson, H.J. and Wright, P.E. (2002) Coupling of folding andbinding for unstructured proteins. Curr. Opin. Struct. Biol. 12,54–60.

[12] Tompa, P. and Fuxreiter, M. (2008) Fuzzy complexes: polymor-phism and structural disorder in protein–protein interactions.Trend Biochem. Sci. 33, 2–8.

[13] Wright, P.E. and Dyson, H.J. (1999) Intrinsically unstructuredproteins: re-assessing the protein structure–function paradigm. J.Mol. Biol. 293, 321–331.

[14] Fuxreiter, M., Simon, I., Friedrich, P. and Tompa, P. (2004)Preformed structural elements feature in partner recognition byintrinsically unstructured proteins. J. Mol. Biol. 338, 1015–1026.

[15] vanLeeuwen, H.C., Strating, M.J., Rensen, M., deLaat, W. andvanderVliet, P.C. (1997) Linker length and composition influencethe flexibility of Oct-1 DNA binding. EMBO J. 16, 2043–2053.

[16] Hurley, T.D., Yang, J., Zhang, L.L., Goodwin, K.D., Zou, Q.,Cortese, M., Dunker, A.K. and DePaoli-Roach, A.A. (2007)Structural basis for regulation of protein phosphatase 1 byinhibitor-2. J. Biol. Chem. 282, 28874–28883.

[17] Todd, B., Moore, D., Deivanayagam, C.C.S., Lin, G.D., Chatto-padhyay, D., Maki, M., Wang, K.K.W. and Narayana, S.V.L.(2003) A structural model for the inhibition of calpain bycalpastatin: crystal structures of the native domain VI of calpainand its complexes with calpastatin peptide and a small moleculeinhibitor. J. Mol. Biol. 328, 131–146.

[18] Moldoveanu, T., Hosfield, C.M., Lim, D., Elce, J.S., Jia, Z.C. andDavies, P.L. (2002) A Ca2+ switch aligns the active site of calpain.Cell 108, 649–660.

[19] Mucsi, Z., Hudecz, F., Hollosi, M., Tompa, P. and Friedrich, P.(2003) Binding-induced folding transitions in calpastatin subdo-mains A and C. Protein Sci. 12, 2327–2336.

[20] Delaglio, F., Grzesiek, S., Vuister, G.W., Zhu, G., Pfeifer, J. andBax, A. (1995) NMRPIPE – a multidimensional spectralprocessing system based on Unix pipes. J. Biomol. NMR 6,277–293.

[21] Goddard, T.D. and Kneller D.G. SPARKY 3, University ofCalifornia, San Francisco.

[22] Crawford, C., Brown, N.R. and Willis, A.C. (1993) Studies of theactive-site of m-calpain and the interaction with calpastatin.Biochem. J. 296, 135–142.

[23] Barnoy, S., Zipser, Y., Glaser, T., Grimberg, Y. and Kosower,N.S. (1999) Association of calpain (Ca2+-dependent thiol prote-ase) with its endogenous inhibitor calpastatin in myoblasts. J. CellBiochem. 74, 522–531.

[24] Croall, D.E. and McGrody, K.S. (1994) Domain-structure ofcalpain – mapping the binding-site for calpastatin. Biochemistry-USA 33, 13223–13230.

[25] Waygood, E.B. (1998) The structure and function of HPr.Biochem. Cell Biol. 76, 359–367.

[26] Maki, M., Takano, E., Osawa, T., Ooi, T., Murachi, T. andHatanaka, M. (1988) Analysis of structure–function relationshipof pig calpastatin by expression of mutated CDNAS in Esche-richia coli. J. Biol. Chem. 263, 10254–10261.

[27] Maki, M., Bagci, H., Hamaguchi, K., Ueda, M., Murachi, T. andHatanaka, M. (1989) Inhibition of calpain by a syntheticoligopeptide corresponding to an exon of the calpastatin gene.J. Biol. Chem. 264, 18866–18869.

[28] von Ossowski, I. et al. (2005) Protein disorder: conformationaldistribution of the flexible linker in a chimeric double cellulase.Biophys. J. 88, 2823–2832.

[29] Ishima, R., Tamura, A., Akasaka, K., Hamaguchi, K., Makino,K., Murachi, T., Hatanaka, M. and Maki, M. (1991) Structure ofthe active 27-residue fragment of human calpastatin. FEBS Lett.294, 64–66.

[30] Suzuki, K., Imajoh, S., Emori, Y., Kawasaki, H., Minami, Y. andOhno, S. (1988) Regulation of activity of calcium activatedneutral protease. Adv. Enzym. Regul. 27, 153–169.