Download Structure–function relationships in calpains1

Survey
yes no Was this document useful for you?
   Thank you for your participation!

* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project

Document related concepts

Enzyme kinetics wikipedia , lookup

Inositol-trisphosphate 3-kinase wikipedia , lookup

Transcript
Biochem. J. (2012) 447, 335–351 (Printed in Great Britain)
335
doi:10.1042/BJ20120921
REVIEW ARTICLE
Structure–function relationships in calpains1
Robert L. CAMPBELL and Peter L. DAVIES2
Department of Biomedical and Molecular Sciences, Queen’s University, Kingston, ON, Canada, K7L 3N6
Calpains are a family of complex multi-domain intracellular
enzymes that share a calcium-dependent cysteine protease core.
These are not degradative enzymes, but instead carry out
limited cleavage of target proteins in response to calcium
signalling. Selective cutting of cytoskeletal proteins to facilitate
cell migration is one such function. The two most abundant
and extensively studied members of this family in mammals,
calpains 1 and 2, are heterodimers of an isoform-specific 80 kDa
large subunit and a common 28 kDa small subunit. Structures of
calpain-2, both Ca2 + -free and bound to calpastatin in the activated
Ca2 + -bound state, have provided a wealth of information about
the enzyme’s structure–function relationships and activation. The
main association between the subunits is the pairing of their
C-terminal penta-EF-hand domains through extensive intimate
hydrophobic contacts. A lesser contact is made between the
N-terminal anchor helix of the large subunit and the penta-EFhand domain of the small subunit. Up to ten Ca2 + ions are co-
operatively bound during activation. The anchor helix is released
and individual domains change their positions relative to each
other to properly align the active site. Because calpains 1 and 2
require ∼30 and ∼350 μM Ca2 + ions for half-maximal activation
respectively, it has long been argued that autoproteolysis, subunit
dissociation, post-translational modifications or auxiliary proteins
are needed to activate the enzymes in the cell, where Ca2 + levels
are in the nanomolar range. In the absence of robust support for
these mechanisms, it is possible that under normal conditions
calpains are transiently activated by high Ca2 + concentrations in
the microenvironment of a Ca2 + influx, and then return to an
inactive state ready for reactivation.
INTRODUCTION
Origin of the name ‘calpain’
The present review will focus on insights about calpains and
their functioning that have largely been derived from recombinant
protein approaches, calpain crystal structures and homology
models based on these structures. There are excellent reviews
of calpain referenced in the present review. In particular, we
recommend reading the comprehensive text by Goll et al. [1]
to acquire the necessary background to the calpain field.
The first use of ‘calpain’ (referring to a calcium-dependent papainlike enzyme) was by Murachi et al. [7]. Around this time it became
apparent that mammalian calpain could be resolved into two main
isoforms (subsequently called μ- and m-) that differed in their
calcium requirement [8]. The designations μ- and m- are short
forms of micromolar and millimolar Ca2 + -requiring respectively.
These two isoforms were shown to have distinct (although similar)
large (80 kDa) subunits, but a common small (28 kDa) subunit [9].
The early cDNA sequencing of calpains revealed the fusion of a
papain-like cysteine protease with a calcium-binding protein and
was a major step forward in appreciating the organization and
complexity of these large proteases [10].
BRIEF HISTORY OF CALPAINS
First discovery
The first reports on calpain came from two different groups
in the 1960s who noted the presence of a calcium-activated
proteolytic activity in soluble extracts from rat brain [2] and
skeletal muscle [3]. In the 1970s, the enzyme was purified to
homogeneity from skeletal muscle [4,5] and was called CANP
(calcium-activated neutral protease). There ‘neutral’ referred to
its optimal pH for activity, which is compatible with a cytoplasmic
location. However, even at this early stage in calpain research,
concern was expressed that “since this enzyme requires an
extremely high and unphysiological concentration of Ca ions its
physiological role remains unclear” [6]. This thread of concern
that the [Ca2 + ] needed for half-maximal activation of calpain is
orders of magnitude higher than the resting [Ca2 + ] inside the cell
has been woven through the literature for three decades without
conclusive resolution.
Key words: calcium-binding, domain structure, enzyme
activation, inhibitor complex, proteinase, proteolysis, X-ray
crystallography.
Genome sequencing reveals extent of the calpain family
Tissue-specific calpains, for example the muscle-specific isoform
[11], were subsequently discovered. But it was not until the
human genome sequencing effort approached fruition that the full
extent of the mammalian calpain family was revealed [12,13]. A
total of 15 human calpain genes have been listed, CAPN1–15, one
of which (CAPN4) codes for the small subunit (Table 1). However,
the criterion for designation as a calpain is generally agreed to be
possession of a cysteine protease core sequence [14], even if some
of these (such as the CAPN6 gene product) are not enzymatically
functional because they are missing one or more of the catalytic
triad residues. By this measure, the small subunit is not a calpain
Abbreviations used: C2L, C2-like; FRET, fluorescence resonance energy transfer; GR domain, glycine-rich domain; PC domain, protease core domain;
LGMD2A, limb girdle muscular dystrophy type 2A; NS, N-terminal sequence; IS, insertion sequence; PEF, penta-EF-hand; RMSD, root mean square
deviation.
1
This review is dedicated to the memory of Darrel Goll and Koichi Suzuki, two luminaries of the calpain field who lit the way for others to follow.
2
To whom correspondence should be addressed (email [email protected]).
c The Authors Journal compilation c 2012 Biochemical Society
336
Table 1
R.L. Campbell and P.L. Davies
Calpain nomenclature [120]
T-S, tissue-specific; U, ubiquitous; X, no; Y, yes.
Calpain
Gene products
Aliases
PEF
Tissue distribution pattern
Expression (highest)
Calpain-1
Calpain-2
Calpain-3
Calpain-5
Calpain-6
Calpain-7
Calpain-8
Calpain-9
Calpain-10
Calpain-11
Calpain-12
Calpain-13
Calpain-14
Calpain-15
Calpain small subunit 1
CAPN1 and CAPNS1
CAPN2 and CAPNS1
CAPN3 dimer?
CAPN5
CAPN6
CAPN7
CAPN8
CAPN9 plus CAPNS1
CAPN10
CAPN11 plus ?
CAPN12 plus ?
CAPN13 plus ?
CAPN14 plus ?
CAPN15
CAPNS1
μ-Calpain, calpain I, μCANP
m-Calpain, calpain II, mCANP
p94, nCL-1 (Lp82, Lp85, Rt88)
nCL-3, htra-3, CAPN5
CAPN6
CAPN7, palBH
CAPN8, nCL-2
CAPN9, nCL-4
CAPN10
CAPN11
CAPN12
CAPN13
CAPN14
CAPN15, SolH
CAPN4, CPNS1*
Y
Y
Y
X
X
X
Y
Y
X
Y
Y
Y
Y
X
Y
U
U
T-S
U
T-S
U
T-S
T-S
U
T-S
T-S
U
U
U
U
Placenta, oesophagus, trachea
Kidney, lung, trachea
Skeletal muscle (lens, retina)
Brain, kidney, liver
Placenta
and therefore there are 14 calpain genes in humans with complete
protease cores. An additional gene, CAPN16 for calpain-16, has
been designated, but its predicted product only includes the first
half of the protease core [15].
NOMENCLATURE
Isoform nomenclature
The two abundant calpain isoforms found in nearly all tissues and
cell types of mammals have for a long time been referred to as μand m-calpain to reflect the difference in the [Ca2 + ] required for
half-maximal activation [16]. Unfortunately, this designation is
somewhat misleading, because the difference in activating [Ca2 + ]
is only one order of magnitude, being 3–50 μM for μ-calpain
and 400–800 μM for m-calpain isolated from bovine skeletal
muscle [1]. The [Ca2 + ] required for half-maximal activation of
recombinant rat m-calpain with an N-terminally truncated small
subunit, referred to extensively in the present review, was 350 μM
[17]. Complications could arise in identifying orthologues on
the sole basis of their Ca2 + requirement, and a more rational
sequence-based nomenclature has been needed for some time,
now that all calpain homologues have been identified in the
human genome. Following extensive online discussion by the
calpain community, we suggest in the present review that μand m-calpain be referred to as calpain-1 and -2 respectively.
Calpain-1 is a heterodimer of the CAPN1 gene product, CAPN1,
and the CAPNS1 (formerly CAPN4) gene product, CAPNS1.
Thus calpain-1 can be fully described as CAPN1 + CAPNS1
(Table 1) or CAPN1/S1 for short. Calpain-2 is a heterodimer of the
CAPN2 gene product, CAPN2, and CAPNS1, again CAPN2/S1
for short. Calpain-3 is the isoform found predominantly in skeletal
muscle [11]. It seems to exist as a homodimer of CAPN3
(CAPN3/3) [18]. It is widely known as p94, and splice
variants of this isoform exist, such as p82 in lens [19]. This
dual nomenclature, where p94 and p82 refer to molecular
mass×10 − 3 , might still be useful in combination with calpain3 for distinguishing these variants. Another example of a
calpain oligomer is G-calpain, which is associated with the
gastrointestinal tract [20]. This enzyme appears to be a dimer
of CAPN8 and CAPN9 and can be referred to as calpain-8/9 [21].
However, it is not presently clear whether other proteins, such as
the small subunit, are present in this complex.
c The Authors Journal compilation c 2012 Biochemical Society
Stomach, digestive tract
Digestive tract, heart
Heart
Testis
Hair follicle
Lung, testis
Brain
Heart, pancreas, kidney
There is a second small subunit homologue, S2, in the human
genome [22]. Although S2 could potentially form heterodimers
with PEF (penta-EF-hand)-containing CAPN gene products, these
variants have not been identified in vivo. A list of human calpain
isoforms and their tissue specificity is given in Table 1.
Interpretation of calpain domains on the basis of structure
The first crystal structures of calcium-free m-calpain from rat [23]
and human [24] (Figure 1) gave definition to the domains and
their boundaries (Figure 2), but has resulted in two contradictory
schemes for domain nomenclature. On the basis of sequence
comparisons, Domain I was originally conceived of as a region of
unknown function N-terminal to the papain-like region (Domain
II). It was presumed to include a putative propeptide region that
might need to be cleaved for enzyme activation. On the basis of the
crystal structures, the papain-like region or protease core is more
extensive than originally thought and was subsequently shown to
have elements that bind Ca2 + ions and are not present in papain
[25]. Indeed, the only portion of Domain I that is not structurally
part of the protease core is the 19-residue α-helical N-terminal
anchor [23], which is shown in Figure 1 making a contact with
the small subunit. To avoid renumbering downstream domains,
Hosfield et al. [23] called the two halves of the protease core
Domains I and II (Figure 2). In another scheme [24], the two
halves of the core are known as domains IIa and IIb.
A more detailed comparison of the structures of the protease
core of calpain with papain shows that there is only partial
structural overlap between them. The protease core of calpain-2
is approximately 140 residues larger than papain, and they each
contain structural elements not present in the other. Although the
two structures do contain regions of nearly identical topology,
they are not perfectly superimposable. The structural similarity
that is observed between these and other members of the clan CA
[26] cysteine protease family results in a substrate-binding site in
calpain that shares features, such as the hydrophobic S2 subsite,
with the smaller cysteine proteases. This active site similarity
makes it more difficult to design calpain-specific inhibitors.
There was unanimity about the proximal domains, with Domain
III referring to a β-sandwich structure that at the level of tertiary
structure resembles the C2L (C2-like) domains found in a number
of enzymes, such as protein kinase C and a phospholipase that
transiently bind to membranes [27,28]. The fact that most C2
Structure–function relationships in calpains
Figure 1
337
Crystal structure of human Ca2 + -free calpain-2
Domain structure of human calpain-2 (PDB code 1KFU) produced in baculovirus-infected insect cells with the GR domain present [24]. The N-terminus of the large (80 kDa) subunits begins with
the anchor helix (red) and leads into the protease core domains PC1 (orange) and PC2 (yellow), followed by the C2L domain (green) and then into the C-terminal PEF domain (light blue). This
PEF domain is paired with the C-terminal PEF domain of the small (28 kDa) subunit (dark blue). The N-terminal GR domain is unstructured and electron density is lost 11 residues in from the
C-terminal end of this domain. This structure is shown in preference to recombinant rat calpain-2 produced in E. coli (PDB code 1DF0) [23] which lacks some detail in flexible loop regions of
the PC2 and C2L domains that can be seen in the human enzyme.
domains contain Ca2 + and phospholipid-binding sites led to the
suggestion that the C2L domains of calpain might also bind these
ligands. The C2L domain in calpain-2 contains two clusters of
acidic residues: Glu392 , Glu393 , Glu394 , Asp395 , Glu396 , Glu481 and
Asp508 ; and Asp351 , Asp398 , Glu397 , Asp399 , Glu402 and Glu504 , which
supported the idea of a Ca2 + -binding site through co-ordination
and charge neutralization. However, no Ca2 + was observed to
be bound in this region in the crystal structures of Ca2 + -bound
calpain when in complex with calpastatin [29,30]. The only other
C2L calpain domain for which there is a crystal structure is that
of human calpain-7 (PDB code 2QFE), which does not have the
same cluster of acidic residues. Nor does this domain appear to
bind Ca2 + . Although the C2L domain does contain a similarlooking β-sandwich tertiary structure to the known C2 domains,
the topology of the C2L domain in calpains is completely different
from that of the previously described C2 domains (Figure 3). The
antiparallel β-sheet containing the N-terminal end of the C2L
domain of calpain is made up of β-strands 1, 8, 3 and 6. The
equivalent sheet from the C2 domain has strands 5, 2, 1 and 8 side
by side. The other sheets have strands 2, 7, 4 and 5, as opposed
to strands 4, 3, 6 and 7. Thus, given the absence of sequence
homology to C2 domains and the differences in topology, there is
no a priori reason why the C2L domain of calpains should bind
Ca2 + and phospholipids.
The C-terminal domains of the large and small subunits of
calpain (Domains IV and VI respectively) are PEF domains
[31]. These α-helical folds of approximately 170 amino acids
contain four EF hands that are able to bind Ca2 + and one EFhand that does not bind Ca2 + , but instead is involved in the
dimerization contacts between the two subunits. The two PEF
domains have a very similar sequence and structure. They show
approximately 45 % sequence identity and an RMSD (root mean
square deviation) of approximately 1.5 Å (1 Å = 0.1 nm) for Cα
atoms. By comparison, the superimposition of the Ca2 + -bound
rat small subunit PEF domain on that of the Ca2 + -free human
structure also gives an RMSD of 1.5 Å.
Domain V is the N-terminal region of the small subunit that
is structurally undefined but is notably rich in glycine. In the
human small subunit, the 95 amino acid residues that comprise this
domain contain 40 glycine and five proline residues, including two
contiguous stretches of 11 and 20 glycines. For the nine known
mammalian sequences, this region shows pairwise sequence
identities ranging from 93 % to 100 % and all contain similar
contiguous stretches of glycine residues.
c The Authors Journal compilation c 2012 Biochemical Society
338
Figure 2
R.L. Campbell and P.L. Davies
Rationalization of calpain domain boundaries and nomenclature
(A) The numerical domain nomenclature (I–IV) is displayed on a representation of the first complete sequence of a calpain large subunit, which is the chicken orthologue of human CAPN11 [10].
Domain colours are the same as those illustrated in Figure 1. (B) A redrawing of the domain boundaries necessitated by the crystal structure of rat calpain-2 [23]. Domain I was clearly not a structural
entity in its own right. The first 19 residues form a helix that anchors the N-terminal end of the large subunit to the PEF domain of the small subunit, while the C-terminal remainder packs against the
protease core. To avoid renumbering the downstream domains that continued into the small subunit, the two sub-domains of the protease core were designated D-I and D-II. These two sub-domains
do behave functionally and structurally as separate entities, as for example when one rotates around the other during activation (see Figure 4). The extended linker region, residues 514–531 between
D–III and D–IV, is represented as a thinner tube in grey. Following the structure determination of human calpain-2 [24], Strobl et al. [24] renamed the two protease core domains IIa and IIb, but
without including residues 16–93 in with IIa (C). They retained domain I for the anchor helix. To avoid having two conflicting numerical schemes for the domains, a new system has been developed
that identifies each domain by an acronym of its structural description [15,32]. Thus the protease core domains become PC1 and PC2; the C2-like domain is C2L; and the penta-EF-hand domains
of the large and small subunits are PEF(L) and PEF(S) respectively. The glycine-rich (GR) domain of the small subunit was previously Domain V. The domain boundaries are indicated by residue
numbers flanking each domain. The domain colour scheme (red to blue) borrows from the structural biology practice of representing a polypeptide chain in the colours of the rainbow to trace the
relationship of any one section to either terminus. However, it is the reverse of the PyMOL tradition of having blue at the N-terminus so that this colour coding can still be applied to calpains without
conflicting with domain colours.
Improved domain nomenclature
Following online discussion by the calpain community, a proposal
has emerged to give the calpain domains descriptive names
abbreviated as acronyms [15,32]. This renaming has the advantage
of eliminating two competing misaligned numerical systems,
while also providing a more informative clue to the identity of
the domain than a number. It will also allow for the subsequent
recognition and designation of domains N-terminal to the protease
core i.e. that precede Domain I, such as those in calpains 7 and 15
[33]. Thus the protease core domains are PC1 and PC2 rather than
I and II or IIa and IIb (Figure 2). Domain III becomes C2L for C2like domain, and the penta-EF-hand domains IV and VI become
PEF domains of the large (L) and small (S) subunits, respectively.
Finally, Domain V is referred to as the GR (glycine-rich) domain.
Is there such a thing as a typical calpain?
Another terminology that should be reviewed is the description
of some calpains as being ‘conventional’ or ‘typical’ (compared
with ‘atypical’), or ‘classical’ (compared with ‘non-classical’).
The perspective for ‘conventional’, ‘typical’ or ‘classical’ is
similar to the PEF-containing calpains 1 and 2, which were the
first calpain types to be characterized. The presence of a PEF
domain is a significant structural feature in that this domain
is typically involved in dimerization, either heterodimerization,
usually with the small subunit, CAPNS1, or homodimerization
[34] as in the case of calpain-3 [35,36]. Although the PEF
c The Authors Journal compilation c 2012 Biochemical Society
containing calpains make up eight of the 14 human calpains, they
are largely confined to animals (metazoans) [15]. The ‘atypical’
non-PEF-containing calpains are in fact far more widespread and
numerous (i.e. typical), occurring in animals, plants, fungi and
even some bacteria [37]. Under these circumstances, the term
‘classical’ is preferable to ‘conventional’ or ‘typical’, meaning
simply similarity to the first calpains that were characterized.
ROLES OF CALPAINS 1 AND 2
Physiological roles
The abundant heterodimeric calpains 1 and 2 are complex,
intracellular proteases that require high levels of Ca2 + for activity.
This presumably provides a safeguard against the overactivation
of these potent proteases within the cell. Indeed, the Ca2 +
concentrations for half-maximal activity of these two enzymes
are several orders of magnitude higher than the multi-nanomolar
resting [Ca2 + ] in the cell. Calpains 1 and 2 function in calcium
signalling by making specific limited cuts in proteins to effect
a change in the function of their targets rather than serving
as degradative enzymes. The target proteins and processes in
which calpains are implicated are too extensive to list in the
present review, but have been covered by Goll et al. [1]. Perhaps
the best documented of their many roles is in remodelling
of the cytoskeleton for processes such as cell movement [38,39].
Even with this role there is uncertainty about the division of
labour between the two main isoforms. For cells to move, the
Structure–function relationships in calpains
339
mass calpain inhibitors can bring about some amelioration of the
damage and has prompted the search and development of more
specific and efficacious calpain inhibitors [49–51].
ACTIVATION BY CALCIUM
The active site is misaligned in the absence of Ca2 +
The Ca2 + -free structures of rat and human calpain-2 [23,24]
revealed that the catalytic triad residues (Cys115 , His272 and Asn296 ;
calpain-1 numbering) in the protease core were not in a position to
produce an active enzyme. Unlike the structures of other cysteine
proteases, the cysteine and histidine side chains were much
too far apart (>8.5 Å) for the histidine to deprotonate the
cysteine and activate it for nucleophilic attack on a substrate.
The functional distance should be ∼3.5 Å.
The calpain core is activated by Ca2 + , even in the absence of other
domains
Figure 3 Domain topology diagrams for the synaptotagmin C2 domain [121]
(A) and the calpain C2L domain (B)
Strands of the β-sandwich are numbered and coloured in order from blue (N-terminus) through
cyan, green, yellow, orange to red (C-terminus). The cylinder between β-strands 1 and 2 in
(B) represents a short segment of α-helix.
trailing edge must be detached from the substrate. Here there is
a role for calpain(s) cutting the focal adhesion complexes for an
easier ‘get away’ [40,41]. At the leading edge of the cells, where
membrane ruffles, pseudopods and other protrusions occur, there
must be extensive reorganization of the cytoskeleton to achieve
movement, and again the major calpains are involved [42].
Calpains in disease
There is a case to be made that calpain-1 and -2’s high calcium
requirement for activity ensures that activation is temporally and
spatially limited, occurring only near the epicentre of a Ca2 +
influx, which is then quickly dissipated by diffusion and
Ca2 + removal by various pumps [43]. However, there are
situations in which Ca2 + levels in the cell rise out of control
and cause prolonged widespread overactivation of calpain with
off-target proteolysis leading to necrotic cell death [44]. This
contributes, for example, to reperfusion injury when a blockage
in the circulation is opened after heart attack, stroke or blunt
trauma [45]. When this previously anoxic tissue is re-supplied
with blood it undergoes a number of stresses that include the
loss of calcium homoeostasis. There are many other situations, as
in neurodegenerative diseases, muscular dystrophies, retinopathy
and cataract, where calcium homoeostasis can be compromised,
causing calpains to contribute to cell and tissue injury [46–48].
In various animal models, the administration of low-molecular-
The original concept of calpain activation by Ca2 + that flowed
from the protein sequence imagined the C-terminal PEF domains
as calmodulin-like regulators of the enzyme’s activity that would,
on binding Ca2 + , induce a conformational change in the dimer,
resulting in a propeptide cleavage event to autoproteolytically
activate the enzyme. The real situation is far removed from this
scenario. Unlike calmodulin, the PEF domains do not undergo
major conformational changes upon binding Ca2 + [29,30],
and calpains 1 and 2 do not have cleavable pro sequences.
Unexpectedly, the protease core has its own two calcium-binding
sites that act co-operatively to convert the core into a functioning
cysteine protease [25].
The existence of these Ca2 + -binding sites was suggested when
partial proteolysis of active-site-inactivated (C105S) calpain-2 in
the presence of divalent metal ions revealed that the protease
core was selectively stabilized by Ca2 + , even though this made
the whole enzyme far more susceptible to digestion by exogenous
proteases. No such effect was seen with Mg2 + [52]. Subsequently,
it was reported that the protease core of human calpain-2 had very
weak Ca2 + -dependent proteolytic activity [53]. The recombinant
calpain-1 core can be stably produced in Escherichia coli and
has 5–10 % of the activity of the whole enzyme [25]. Its crystal
structure shows two Ca2 + ions binding to non-EF-hand sites in
the core, one in each of the domains PC1 and PC2. Each Ca2 + binding site is made up of two flexible loops that supply backbone
carbonyl and side-chain carboxy groups for co-ordinating the
Ca2 + ions. The two Ca2 + -binding sites act co-operatively to cause
a large conformational change in the core that reorientates the two
domains. Reorientation enables the core to form a functional
active site cleft by a 25◦ rigid-body rotation of the two protease
core domains relative to each other [54]. As shown in a stereo
diagram, this movement brings the catalytic triad into register
for peptide bond cleavage (Figure 4). The key movement is
the repositioning of Cys115 (calpain-1 numbering) close enough
to His272 for deprotonation of the former to occur. At the
same time, a key tryptophan residue (Trp298 ) moves from an
exposed position in the cleft to tuck into a hydrophobic patch
formed by the rearrangement of the calcium-binding loops.
The shape of the cleft between the core domains is radically
different before and after activation (Figure 5). This process of
activation, which can be conveniently monitored by intrinsic
tryptophan fluorescence or hydrolysis of calpain substrates,
displays obvious co-operativity [25]. Despite the absence of
the other four domains, two of which bind eight Ca2 + ions,
the observed [Ca2 + ] for half-maximal activation of the core
c The Authors Journal compilation c 2012 Biochemical Society
340
Figure 4
R.L. Campbell and P.L. Davies
Movement of the protease core domains during activation
(A) Human calpain-2 protease core structure in the apo form (PDB code 1KFU) [24] shown as a stereo diagram in ribbon format. The catalytic triad residues and Trp288 are displayed to show their
side chains in green. (B) Stereo diagram of the rat calpain-2 protease core structure in the Ca2 + -bound form (PDB code 1KXR) [25] aligned to have its PC1 domain in the same orientation as in (A).
Calcium ions are shown as purple spheres. Note the movement of the PC2 domain on binding Ca2 + and the closing of the distance between the catalytic cysteine and histidine residues.
is similar to that for the whole enzyme. Indeed, the calpain-1
core becomes a functional protease (mini-calpain) with very
similar properties to the whole enzyme, including substrate
specificity, inhibitor sensitivity and calcium requirements, but
with a greatly reduced turnover number.
The role of other calcium-binding sites
What then are the roles of the eight Ca2 + ions that bind to the
two PEF domains? Rather than being the primary regulators of
calpain activity as originally thought [55], we suggest that the
PEF domains provide an additional level of safeguard against
casual activation of calpain. The Ca2 + -activated calpain structure
[29,30] shows that the protease core is stabilized by contacts with
the other domains, most notably the C2L and small subunit PEF
domains. The lack of support by these domains in the protease
core most probably explains why the enzymatic activity of the
core is only 5–10 % of that of the whole enzyme. In the protease
core, the PC1 and PC2 domains can rotate about a pivot at residue
Gly209 , where both the phi and psi angles change as a result of the
movement. Owing to this flexibility and the absence of supporting
domains, the Ca2 + -activated form is likely to be in equilibrium
with the apo form.
If the core is supported in its active conformation by the other
domains, it can be argued that these domains constrain the core
when Ca2 + is absent and prevent spurious activation. Indeed, it
is unlikely that the two core domains could rotate into the active
conformation without being released from contact with the other
domains. One constraint that is freed is the anchor helix, which
bridges the N-terminus of the large subunit to the small subunit.
It has been suggested that the co-ordination of Ca2 + by EF-hand
2 in the small subunit immediately opposite the anchor helix
causes a charge repulsion of the basic residues in the helix [24,56].
Consistent with this idea, the anchor helix is both displaced and
unstructured in the Ca2 + -bound calpastatin-inhibited calpain-2
c The Authors Journal compilation c 2012 Biochemical Society
structure [29,30]. This allows the core to nestle down on to, and
make even more intimate contacts with, the PEF domain of the
small subunit (Figure 6).
On the other side of the core there is an extensive series of
electrostatic contacts between domains PC2 and C2L and an
extended linker region that connects C2L to PEF (L). It had
been suggested that subtle conformational changes derived from
Ca2 + binding to the PEF domains might be transduced through
the linker to the domain PC2/C2L contacts [23]. Although there
is some experimental evidence for this mechanism from sitedirected mutagenesis [23,57], the full-length structures of Ca2 + bound calpastatin-inhibited calpain-2 lend little support to the
idea [29,30]. Instead, Ca2 + binding within PEF(L) may induce
changes in the C2L/PEF domain interface itself. For example,
binding of Ca2 + to the third EF-hand causes Glu626 to break an
electrostatic interaction with Lys629 . The latter is then free to form
a hydrogen bond with the backbone carbonyl of Ile446 of C2L.
The shift in the relative positions of C2L and PEF(L) upon Ca2 +
binding results in an increase in the contact area between the two
domains of approximately 25 %. This shift also causes the linker
region (residues 511–526) to become more flexible, such that
some residues were no longer visible in the electron density map.
In other words, there is no sign of tension in the linker, consistent
with a pull being exerted from the PEF domains.
How do in vivo calcium signals activate calpain?
In vitro studies clearly show that calpain-1 and -2 can be
fully activated without the involvement of extraneous factors
or post-translational modification, provided that sub-millimolar
to millimolar Ca2 + levels are present. However, it is not clear
whether (and how) these concentrations are met in vivo. Several
suggestions have been put forward for mechanisms that might
lower the Ca2 + requirement for calpain activation within the
cell. These include: the involvement of activator proteins [58,59],
Structure–function relationships in calpains
341
AUTOPROTEOLYSIS
Autoproteolysis of calpains 1 and 2 is an intermolecular reaction
Being a multi-domain heterodimer with exposed linkers and loops
(Figure 1), calpain is a prime target for digestion by proteases
[52,71], including calpain itself [72,73]. However, unlike calpain3, where there is clear evidence for at least one intramolecular cut
site, it is obvious from the crystal structures of apo- and Ca2 + bound calpain-2 that none of the large subunit sequences cut by
calpain-2 are within range of its own active site cleft [23,24,29,30].
The structural similarities between calpains 1 and 2 are so great
that this will probably be also true for calpain-1. Therefore,
although there is evidence that N-terminal autoproteolysis of
calpain-1 and -2 does lower the calcium requirement for enzyme
activation [16,66–69], this anchor peptide cleavage must be an
intermolecular event because the cleavage point is far removed
from the active site cleft. Thus at least one calpain would have to
be activated before it could activate others.
Autoproteolysis of calpains 1 and 2 is largely an artefact of
purification
Figure 5
Formation of the calpain-2 catalytic cleft
The catalytic cleft between the PC1 (orange) and PC2 (yellow) domains of human calpain-2
is shown (A) in the Ca2 + -free form [24], and that of rat calpain-2 in the Ca2 + -bound form is
shown in (B). In both Figures, a section of a calpastatin inhibitory domain (grey) is displayed
in the cleft for reference [29]. A section of the C2L domain (green) can be seen in the bottom
left-hand corner.
the effectiveness of which have not yet been substantiated by
other laboratories [60]; small subunit dissociation [61–64]; and
covalent modification of calpain, including phosphorylation [65]
and autoproteolysis [16,66–69]. In the latter case, there is at least
a structural explanation for how this might work. One of the ‘hotspots’ for autoproteolysis is the anchor helix [70]. If the anchor
helix were selectively cleaved off, then this modification might
lower the energy barrier to activation by removing a key restraint
on the movement of the core domains PC1 and PC2 (Figure 7).
The same effect could be achieved by small subunit dissociation.
However, given the many arguments in the next section against
autoproteolysis and small subunit dissociation as physiological
activation mechanisms, more consideration should be given to
the idea that calpain is simply activated in the immediate vicinity
of the epicentre of Ca2 + influx. This would both localize and limit
the amount of calpain that is activated in the cell. Rapid diffusion
of the localized high [Ca2 + ] down a huge concentration gradient
would return calpain to an inactive state and limit the time over
which calpain can be active once the signal ceases.
Other arguments can be made against autoproteolysis being
a physiological activation mechanism. In the inactive state of
calpain, the anchor peptide cleavage sites would be protected from
hydrolysis by being part of an α-helix. It is only after activation
and release from its contact with the small subunit PEF domain
that the anchor helix becomes disordered and hence susceptible
to proteolysis [29,30]. Although the anchor helix is cleaved early
on during autoproteolysis, so too are many sites within the C2L
domain that result in inactivation or greatly reduced activity of
the enzyme, making this a risky method for calpain activation
(Figure 7) [70]. Indeed, it could be argued that autoproteolysis
of calpain is an artificial situation that is promoted in vitro when
calpains are concentrated through purification. In vivo, the enzyme
would be surrounded by other proteins, including its substrates,
and would not naturally be concentrated to the same extent in the
cell.
If the anchor peptide is not cut during activation, then the
calpain molecules would be able to return to the ground state
for additional cycles of activation and inactivation in response
to calcium signals, rather than be consumed by autoproteolysis.
Indeed, the idea of activation by anchor peptide removal seems
to be a hangover from the time that the N-terminal region was
thought to be a propeptide blocking the active site cleft as in some
of the cathepsins, which the X-ray crystal structures have shown
is clearly not the case [23,24].
SMALL SUBUNIT DISSOCIATION
Does the small subunit dissociate under physiological conditions?
Another incompletely substantiated claim that is deeply
entrenched in the calpain literature is that the small ‘regulatory’
subunit dissociates from the large ‘catalytic’ subunit during
activation or even as a calpain activation mechanism [61–
63,74,75], although this finding is disputed by others [76,77].
In theory, subunit dissociation could be as equally effective as
autoproteolysis in releasing the constraints on realignment of the
protease core imposed by the anchor peptide. Some researchers
have even reported that N-terminal autolysis of calpain promotes
subunit dissociation [56,78]. However, the advantage of a subunit
dissociation activation mechanism without autoproteolysis would
be its potential reversibility.
c The Authors Journal compilation c 2012 Biochemical Society
342
Figure 6
R.L. Campbell and P.L. Davies
Calcium-dependent activation of calpain-2 produces a more condensed enzyme with new contacts between the small subunit and the protease core
In the space-filling view of the calpain-2 crystal structure in the apo-form (A), the anchor helix (red) makes contacts between the small subunit PEF (dark blue) and PC1 of the protease core (orange).
The C2L domain (green) is clearly visible between these two domains. After activation, the anchor helix has been displaced and is disordered. PEF(S) has made new contacts with PC1 of the protease
core that help stabilize the active conformation, and PC2 has moved up and over towards PC1. These new associations obscure that portion of C2L seen previously, although a new region projects
to the right of PEF(S). Overall, the activated enzyme is more condensed, consistent with its lower apparent molecular mass on size-exclusion chromatography.
One reason why this ‘subunit dissociation mechanism’ is hard
to accept from the perspective of structural biology is that
the interface between the PEF domains of the small and large
subunits is extensive and hydrophobic (Figure 8) [23,24]. It is
difficult to imagine this interaction being broken and reformed
without the involvement of a chaperone or a denaturing condition.
Crystal structures of the small subunit homodimer in the presence
[79,80] and absence [79] of Ca2 + only show small differences in
conformation or area of contact between the apo and Ca2 + -bound
forms. This is also true for the calpain-2 heterodimer when it
shifts from the calcium-free to the calcium-bound states [29,30].
Therefore, it seems unlikely that Ca2 + binding could perturb this
PEF domain interaction enough to disrupt the pairing up of their
fifth EF-hands. Furthermore, in the structure of the calpastatinbound Ca2 + -activated calpain-2, the protease core is nestled down
on to, and makes new specific contacts with, the small subunit.
This Ca2 + -dependent compression of the enzyme is seen even in
the absence of calpastatin, where the presence of both subunits
in a 1:1 stoichiometry has been documented by SDS/PAGE
([29]; R. A. Hanna and P. L. Davies, unpublished work). Thus
the heterodimer structure is stable in Ca2 + in the absence of
calpastatin and its dimeric state is not a consequence of tethering
by the extended calpastatin inhibitor contacting both subunits.
Some of the in vitro studies that show the stability of calpain2 have been done with two important structural modifications.
One is the absence of the N-terminal GR domain from the
small subunit. This domain has no assigned function as of yet
and, therefore, a role in activation cannot be ruled out. It was
deliberately omitted from these constructs because it is unstable
and rapidly proteolysed during production of the recombinant
enzyme in E. coli. Where efforts have been made to include
this domain in human calpain 2 expressed in insect cells, the
domain appears to be disordered in the crystal structure [24]. The
c The Authors Journal compilation c 2012 Biochemical Society
other modification has been the knockout of enzymatic activity
by converting the catalytic cysteine residue to serine or alanine
[81,82]. This ensures that there is no autoproteolysis. When
the native enzyme undergoes autoproteolysis or proteolysis by
exogenous proteases, the C2L domain is rapidly digested [52].
This releases the PEF domains of the two subunits as a stable
heterodimer that does not dissociate or undergo rearrangement
into homodimers [70]. This product is difficult to distinguish from
the small subunit homodimer and might have been mistaken for
it in some of the reports of subunit dissociation.
Having found fault with autoproteolysis and subunit
dissociation as activation mechanisms, and having doubted
protein activators and post-translational modifications because
these regulatory mechanisms have not been reproduced by
independent laboratories, there is an obligation to explain how
calpain can be activated in the cell when resting calcium levels
are three orders of magnitude lower than those need to achieve
half-maximal activation. Quite simply, as argued previously
[70,83,84], the extremely localized calcium concentration near
the inflow from channels and pumps should be high enough
to transiently activate calpain where needed – for example at
a focal adhesion complex. By the time this calcium influx has
diffused and registered cell-wide as a small increase in calcium
concentrations, calpain will already have returned to its inactive
state. We maintain that the physiological activation of calpain
occurs in an extremely localized fashion, such that at any one
time only a tiny fraction of calpain is activated in extremely
localized areas for very brief moments. Indeed, the argument has
been made that the requirement for high calcium levels acts as a
biological safety device that helps prevent calpain overactivation
under physiological conditions [85].
In support of this direct activation mechanism, we argue that
calpain is probably not activated long enough during physiological
Structure–function relationships in calpains
343
are high and monovalent ion concentrations are low [64]. Calpain
large-subunit aggregation was reported as a consequence of autoproteolysis and subunit dissociation [63]. However, aggregation
also occurs when calpain autoproteolysis is blocked by converting
the active-site cysteine residue to a serine [64]. In forming the
calpain–calpastatin complex for crystallography it was important
to mix calpastatin with calpain before the addition of calcium,
otherwise the enzyme precipitated [29]. It should be stressed that
the inhibitor has no affinity for calpain until the protease is activated by Ca2 + . Once calpastatin has bound to the Ca2 + -activated
calpain, the enzyme seems to be protected from aggregation and
precipitation, which we attribute to the coverage of hydrophobic
patches on calpain by the inhibitor. Aggregation of calpain in
the presence of Ca2 + poses a problem for crystallography of the
whole enzyme. Although physiological NaCl concentrations can
largely prevent aggregation, this additive places one limit on the
conditions under which the enzyme can be crystallized.
USEFULNESS OF ‘MINI-CALPAINS’
Inhibitor-bound calpain protease core structures
Figure 7
Hypotheses for the activation of calpains 1 and 2
(A) In this scenario, calpain becomes activated by autolytic cleavage of the anchor helix
(red rectangle). Loss of the helix (dotted red line) releases constraints on the protease core,
which then allow PC1 and PC2 to reorientate to form the active site (indicated by the black
‘V’) on binding Ca2 + . Domain colours are described in the legends to Figures 1 and 2. In
this Figure, cleavage of the GR domain (dotted red line) of the small subunit is collateral
damage and not obviously part of the activation process. For a movie of scenario A, please see
http://www.BiochemJ.org/bj/447/bj4470335add.htm. (B) In this scenario, release of constraints
on the calpain protease core is due to dissociation of the small subunit from the large subunit,
with the small subunit forming a homodimer. Dissociation may or may not be accompanied
by autoproteolysis of the anchor helix and GR domain. (C) Fully reversible Ca2 + -dependent
activation of calpain where the binding of Ca2 + to PEF(S) releases the anchor helix intact. Once
the calcium ions dissociate and diffuse away, the apo structure reforms with the anchor helix
constraining the core in its inactive form. (D) Consequences of prolonged activation in vivo or
calpain purification in vitro . Initial autoproteolysis produces contemporaneous cuts in the anchor
helix and C2L domain [70]. Extensive autoproteolysis leads to almost complete attenuation of
calpain activity, extensive digestion of the C2L domain (see Figure 8), but preservation of both
the protease core and a heterodimer of the PEF(L) and (S) domains. For an animated sequence
of this Figure, please see http://www.BiochemJ.org/bj/447/0335/bj4470335padd.htm.
calcium signalling to need inhibition by calpastatin. How else can
one explain the striking observation that a calpastatin-knockout
mouse is largely symptomless [86]? Calpastatin might only be
needed as a stop gap to prevent prolonged calpain activity under
conditions of cell stress [87,88]. This silencing of calpain appears
to break down in pathological conditions such as ischaemic injury,
because calpastatin is eventually proteolysed by its enzyme target
[1] or during apoptosis by caspases [89]. In the absence of robust
evidence for specific calpain activation mechanisms, it seems
reasonable to propose the localized activation of calpain at the
epicentres of Ca2 + (in)fluxes.
Aggregation of Ca2 + -activated calpain
Another complication of working with purified calpain is its tendency to aggregate, particularly when divalent ion concentrations
Problems with autoproteolysis and aggregation have complicated
work on whole calpains 1 and 2 and have retarded attempts to
obtain crystal structures of the Ca2 + -activated enzymes in the
absence of calpastatin. The discovery that domains PC1 and
PC2 were resistant to proteolysis in the presence of Ca2 + but
not in its absence [52] led to the expression and analysis of
the calpain protease core as a functional ‘mini-calpain’ [25]. In
particular, the protease core of calpain-1 has proved to be a useful
reagent, because it mimics the function of the whole enzyme,
but is not susceptible to autoproteolysis. This is illustrated when
comparing the time course of hydrolysis of a hexapeptide FRET
(fluorescence resonance energy transfer) substrate PLFAER by
the calpain-1 core compared with the whole enzyme (Figure 9).
The two enzymes have similar initial reaction rates. However, the
whole enzyme rapidly approaches a plateau of activity, whereas
the core maintains its enzyme activity for longer. As a result of
this stability, it has been possible to co-crystallize the rat [90–93]
and human [94] calpain-1 protease core with a series of covalent
cysteine protease inhibitors of varying degrees of reversibility and
calpain specificity (Table 2). These have included compounds
with epoxide [90,92], α-ketoamide [91,93] and aldehyde [90]
warheads. The same strategy was used to examine leupeptin in
the active site of human calpain-9 [95].
The established inhibitors and their derivatives typically occupy
the unprimed side of the active site cleft in the S1, S2 and S3
subsites, where it is difficult to distinguish calpains from other
cysteine proteinases. Inhibitors that could extend into the S4
subsite area would be in a region of calpain where contact with
the C2L domain could prove to be discriminatory, since this
domain is not present in other cysteine proteases such as the
cathepsins. An indication of this is given in the structure of Senju
Pharmaceutical’s inhibitor SNJ-1945 bound to the calpain-1 core
(Figure 10) [91]. Figure 10 shows that the long ether chain in
the P3 position is equally distributed between two sites in the
co-crystal structure, one of which would slightly clash if the C2L
domain were present. The nine structures of the protease core of
rat calpain-1 bound to a variety of inhibitors display a great degree
of similarity. Excluding the flexible gating loops (residues 64–84
and 253–262) [25], the RMSDs between all pairs of structures
vary between 0.09 Å and 0.5 Å for Cα atoms. After aligning
the structures, the RMSDs of the gating loops themselves range
from 0.21 Å for the most similar structures to 3.6 Å for the most
different.
c The Authors Journal compilation c 2012 Biochemical Society
344
Figure 8
R.L. Campbell and P.L. Davies
Contact surfaces between the calpain-2 subunits in the presence and absence of Ca2 +
Contact surfaces between the large (upper) and small (lower) subunits are colour-coded and compared in the absence (left) and presence (right) of Ca2 + . Atoms that make van der Waals contacts
with a domain in the other subunit are coloured. Those in the large subunit are coloured (using the colour scheme of previous Figures) by the domain they belong to and those in the small subunit
by the colour of the large subunit domain they touch. Purple spheres are Ca2 + ions. The small subunits (lower panels) have been rotated through 180◦ relative to the large subunits (upper panels).
Table 2
List of calpain structures with inhibitors bound at the active site
PDB code and reference
Calpain
Inhibitor abbreviation Inhibitor full name
1TL9 [90]
1TLO [90]
2G8E [91]
2G8J [91]
Rat calpain-1 protease core
Rat calpain-1 protease core
Rat calpain-1 protease core
Rat calpain-1 protease core
Leupeptin
E64
SNJ-1715
SNJ-1945
2NQG [92]
2NQI [92]
Rat calpain-1 protease core
Rat calpain-1 protease core
WR18(S,S)
WR13(R,R)
2R9C [93]
Rat calpain-1 protease core
ZLAK-3001
2R9F [93]
Rat calpain-1 protease core
ZLAK-3002
1ZCM [94]
2P0R
3BOW [29]
3DF0 [30]
Human calpain-1 protease core
Human calpain-9 protease core
Rat calpain-2 with calpastatin domain 4
Rat calpain-2 with calpastatin domain 1
ZLLYCH2 F
Leupeptin
CAST4
CAST1
c The Authors Journal compilation c 2012 Biochemical Society
N -Acetyl-Leu-Leu-Arg aldehyde
N -[N -[1-Hydroxycarboxyethyl-carbonyl]leucylamino-butyl]-guanidine
(2S )-4-Methyl-2-(3-phenylthioureido)-N -((3S )-tetrahydro-2-hydroxy-3-furanyl)pentanamide
((1S )-1-((((1S )-1-Benzyl-3-(cyclopropylamino)-2,3-dioxopropyl)amino)carbonyl)-3methylbutyl)carbamic acid 5-methoxy-3-oxapentyl ester
5-Azanylidyne-N -[(2S )-4-ethoxy-2-hydroxy-4-oxobutanoyl]-l-norvalyl-l-arginyl-l-tryptophanamide
N ∼2∼-[(2S )-2-{[(2R )-4-Ethoxy-2-hydroxy-4-oxobutanoyl]amino}pent-4-enoyl]-l-arginyl-ltryptophanamide
Benzyl (S )-1-((2S ,3S )-1-(3-(6-amino-9h-purin-9-yl)
propylamino)-2-hydroxy-1-oxopentan-3-ylamino)-4-methyl-1-oxopentan-2-ylcarbamate
Benzyl [(1S )-1-{[(1S ,2S )-1-Ethyl-2-hydroxy-3-{[3-(4-methylpiperazin-1-yl)propyl]amino}-3oxopropyl]carbamoyl}-3-methylbutyl]carbamate
N -[(benzyloxy)carbonyl]leucyl-N ∼1∼-[3-fluoro-1-(4-hydroxybenzyl)-2-oxoxpropyl]leucinamide
N -Acetyl-Leu-Leu-Arg aldehyde
Calpastatin domain 4
Calpastatin domain 1
Structure–function relationships in calpains
Figure 9
Effect of autoproteolysis on calpain reaction rate
Line A, rate of hydrolysis of 10 μM FRET substrate (EDANS)-EPLFAERK-(DABCYL) [98] by
10 nM calpain-1 (Calbiochem). Line B, rate of hydrolysis of the same FRET substrate (10 μM)
by 330 nM recombinant calpain-1 protease core [25]. The assays were performed at 22 ◦ C in
4 mM CaCl2 , 10 mM Hepes, pH 7.4, and 10 mM DTT (dithiothreitol).
The primed side of the active site cleft is more different
in calpain than its clan members, although this side is not
traditionally targeted by inhibitors. Several observations suggest
that there is potential for the development of primed-side calpain
inhibitors. When inhibitors have extended into this side of the cleft
they seem to have easily pushed aside the gating loops that initially
appeared to occlude this part of the substrate-binding site [93].
In the example of the inhibitor SNJ-1715 co-crystallized with the
calpain-1 core, a molecule of 2-(N-morpholino)ethanesulfonic
acid buffer was found in the primed side of the cleft [91]. The
series of primed side-extended α-ketoamide inhibitors developed
by the Powers laboratory included one with an adenyl group that
made a stacking interaction with the side chain of Trp298 [93].
Although this tryptophan residue is conserved in other cysteine
proteases, the regions around the residue are quite distinct to
Figure 10
345
calpain. The adenyl group could be a useful base for extending
functional groups outwards to contact these calpain-specific sites.
This arrangement with compounds occupying both sides of the
cleft presaged the fit of the inhibitory B-subdomain of calpastatin,
where peptide sequences on either side of a central loop make
specific intimate (but relatively independent) contacts with the
primed and unprimed sides of the cleft [29,30].
One of the features of most of the calpain inhibitors described
above that may be responsible for their poor specificity is
their inherent flexibility. An approach that has been used to
try to circumvent this problem is one in which the inhibitors
contain a macrocyclic moiety [96]. The goals of introducing the
macrocycle were to force the inhibitors into a predominantly βstrand conformation, and to rigidify the inhibitor, thus minimizing
the entropic penalty of binding. A side effect of this strategy
may be that a more rigid inhibitor will have a reduced ability
to accommodate different active site structures, thus improving
specificity. Various macrocyclic calpain inhibitors synthesized by
Abell et al. [96] were tested for their relative inhibition of ovine
calpains 1 and 2. The most selective inhibitor showed a 7-fold
lower IC50 value for calpain-2 than for calpain-1.
Use of the calpain core in screening for optimal substrate
sequences
Another important use of the calpain core has been to probe
the sequence preference of the enzyme. As calpain cleaves a
great variety of exposed sequences from many different proteins
with little obvious pattern, other than a strong preference for
leucine residues or other branched aliphatic amino acids in the
P2 position, it has been assumed that the enzyme has a minimal
cleavage sequence specificity. However, it is possible that these
sequences are cut at very different rates. When the calpain-1 core
was used to lightly digest a degenerate dodecapeptide library
[97], an analysis of the early cut sites revealed a distinct sequence
preference on the primed sides of the scissile bond [98]. When
The co-crystal structure of the protease core of rat calpain-2 with the inhibitor SNJ-1945 covalently bound in the active site cleft
The enzyme is shown in surface representation with C atoms in white, S atoms in yellow, charged O atoms in dark red, uncharged O atoms in light red, charged N atoms in dark blue, and uncharged
N atoms in light blue. The SNJ-1945 inhibitor (((1S )-1-((((1S )-1-benzyl-3-(cyclopropylamino)-2,3-dioxopropyl)amino)carbonyl)-3-methylbutyl)carbamic acid 5-methoxy-3-oxapentyl ester) bound
to Cys115 is shown in stick representation with C atoms in green. Its cyclopropyl group (CP) projects into the primed site of the cleft and forces the gating loops (G) into the open configuration
[91]. Hydrogen bond connections made by the inhibitor in the cleft are shown as blue dashed lines. The diethylene glycol group of SNJ-1945 in the P3 position is shown occupying two possible
conformations (i and ii), one of which (i) would clash with the C2L domain in whole calpain.
c The Authors Journal compilation c 2012 Biochemical Society
346
R.L. Campbell and P.L. Davies
this preference was built into the second stage degenerate library,
the scissile bond was readily cleaved with a characteristic
preference for branched aliphatic amino acids in the P2 position.
The use of the protease core by Cuerrier et al. [98] avoided any
contamination by internal autoproteolysis fragments that might
have confused the analysis. The effectiveness of the cleavage
sequence was shown by its rapid hydrolysis when presented as a
FRET substrate. This PLFAER consensus sequence was cleaved
much more rapidly than a consensus sequence based on a literature
survey of known calpain cleavage sequences and more rapidly
again than that of the calpain-specific cleavage site in spectrin.
The non-core domains have ancillary roles in calpain activity
The calcium-bound structure of the free calpain-1 core [25] is
virtually the same as that of the homologous calpain-2 core
within the calpastatin-bound Ca2 + -activated whole enzyme [29].
Indeed, the molecular replacement solution of the latter
successfully began with the calpain-1 protease core as a search
model. Since the core is functional in the absence of the
other calpain domains and is activated by the same calcium
concentrations as the whole enzyme, there is reason to question
the function of these other domains. It is likely that the C2L
domain and the PEF domains of the large and small subunit are
present to provide additional levels of regulation and control over
a potent intracellular protease. A particularly graphic example of
this is the inhibition of activated calpain where the endogenous
unstructured inhibitor, calpastatin, interacts with all five domains
of the enzyme. For this, domains C2L and PEF(L) and (S)
play various roles. They increase the specificity and strength of
calpastatin binding, and help direct the inhibitory portion of the
molecule into the active site cleft (see below).
Rather than pushing the calpain core towards an active form
when the calcium signal is registered, the adjacent domains seem
to be involved in restraining movement of the core domains
when they are in the inactive, apo, form. This restraint takes the
form of a circular arrangement of the five domains in which
the small subunit PEF domain makes contact with both ends
of the large subunit, through its N-terminal anchor helix and
through the pairing of the C-terminal fifth EF-hands. As
mentioned earlier, there is a logical mechanism by which the
anchor helix can be released during calcium signalling. As
the calcium ions bind to the small subunit, the Ca2 + that occupies
EF-hand 2 is directly opposite the anchor peptide binding site and
might help its release by charge repulsion of basic amino acids
in the helix [56]. In support of this model, the elution position of
the small subunit PEF dimer from an ion-exchange column shifts
enormously in a comparison with the Ca2 + -bound and Ca2 + -free
forms (Figure 11). The fully Ca2 + -loaded PEF(S) dimer fails to
bind to QAE-Sepharose at pH 7.6, but after being stripped of
Ca2 + by the chelating agent EDTA, the homodimer binds tightly
to the resin and is eluted by ∼0.2 M NaCl. The two PEF domains
can each bind four Ca2 + ions. The first few Ca2 + ions bind with
similar affinity to the Ca2 + pair binding to the protease core, as
judged by the smooth sigmoidal relationship between% activity
and calcium ion concentration [17]. Thus the whole activation
process seems to be a concerted, co-operative process involving
ten Ca2 + ions rather than just the two needed to activate the core.
Why the calpain-2 core is less active
When the protease core of calpain-2 was produced and tested,
it had very little proteolytic activity [99]. This core’s crystal
structure showed that a key α-helix, which supports the core
c The Authors Journal compilation c 2012 Biochemical Society
Figure 11 The influence of Ca2 + on the elution of PEF(S) dimer from an
anion exchanger
In the presence of millimolar Ca2 + at pH 7.6, a PEF(S) dimer fails to bind to a Mono-Q anion
exchange column (line A). After stripping the eight Ca2 + ions from the dimer using EDTA, the
apo form of the protein binds tightly to the column and is eluted (line B) by a salt gradient
(dotted line) at a NaCl concentration of 0.2 M.
domains in calpain-1, had denatured and caused distortion to the
PC1 domain and the active site cleft. Trp106 , which is a key residue
in maintaining the hydrophobic core of PC1, was displaced into
the active site cleft, thereby partially obstructing substrate access
to the catalytic cysteine residue. The basis for the structural
collapse is the presence of a glycine residue in the α-helix where
an alanine residue resides in calpain-1. The negative effects of
this helix breaker were demonstrated by mutating the glycine
residue to alanine in the calpain-2 core. The G203A mutant core
gained activity, although only approximately 10 % of that of the
calpain-1 core. Sequence comparisons of calpain isoforms show
that some have a glycine residue in this position and others have
alanine. It is not clear what the functional significance of the
sequence difference is, but it is likely that the activity of
the glycine-containing isoforms would be silenced by
autoproteolysis once the core was separated from the C2L domain,
where most of the autoproteolytic sites occur.
The protease cores of other calpains
Given the size and diversity of the calpain family, remarkably few
whole enzyme structures have been solved. This is despite the
human calpains having been a target of the Structural Genomics
Consortium (http://www.thesgc.org). The problem seems to lie
in the production of recombinant protein. Rat calpain-2 has been
produced in E. coli [82,100]; human calpain-2 was initially made
in insect cells by using a baculovirus vector [101], but has since
been produced in E. coli [102]. However, the GR domain has
either been deliberately deleted or remains unseen in the crystal
structure [24]. Although rat calpain-1 is poorly produced in
E. coli, a chimaera of rat calpains 1 and 2 was made in sufficient
quantities to solve its structure [103]. The strategy used to achieve
a useful yield of chimaeric enzyme was to replace both ends of
the calpain-1 large subunit that interacts with the common small
Structure–function relationships in calpains
subunit with the equivalent regions from calpain-2. In this way,
85 % of the calpain-1 large subunit structure was solved, albeit
with some uncertainty about whether or not formation of the
chimaera had caused some distortions.
Since the protease core and PEF domains of the major calpains
seem to be relatively stable proteins that are not prone to
proteolysis, the problem with producing the whole enzyme may
lie with the presence of the C2L domain, which is very susceptible
to proteolysis. Therefore one route to some structural information
about other calpain isoforms has been to produce individual
domains or domain combinations, particularly the protease core.
The protease core of calpain-3 was prepared in E. coli and was
stable when the active site cysteine residue was converted into a
serine [104]. In addition to the protease cores of rat calpains 1
and 2, structures of protease cores have been solved for: human
calpain-1 and -9 in the presence of Ca2 + [95]; the Ca2 + -bound
G203A mutant of human calpain-1 with the inhibitor ZLLYCH2F
bound to the active site cysteine residue [94]; and for Ca2 + -bound
human calpains 8 and 9 with leupeptin bound to the active site
cysteine residue [95]. The individual domain structures of all of
these protease cores are very similar to each other, with pairwise
RMSD values for C-α atoms ranging from 0.25 to 0.7 Å (omitting
the flexible gating loops). Similar to rat calpain-2, the protease
cores of human calpains 1 and 8 contain a glycine residue at
the position corresponding to Gly203 of rat calpain-2 and also
show a disturbance of the helix in that area. Alignment of the
complete protease core structures shows that they are all very
similar, with the exception of the structure of the human calpain-9
protease core. If the latter structure is omitted, the pairwise RMSD
values range from 0.38 to 0.85 Å, again without the flexible gating
loops being included. Curiously, although the conformation of
the leupeptin-bound protease core of calpain-9 is very similar
to the protease core of Ca2 + -activated calpain-1 with or without
leupeptin bound (RMSD for C-α atoms of approximately 0.5 Å),
the structure of the Ca2 + -bound protease core of calpain-9 in the
absence of leupeptin shows that the PC2 domain is rotated 48◦
away from the active conformation and in the opposite direction
from that demonstrated by the calcium-free structures (Figure 4).
In other words, the active site in this structure is even farther
from the active conformation than is the calcium-free calpain-2
structure. These observations reinforce the idea that the protease
core can be quite flexible in the absence of the other domains of
the full-length structure. This may help explain the relatively low
activity demonstrated by the protease cores – if only a subset of
the total population in solution is in the active conformation at any
instant. Also, it fits with the observation that the crystallization of
the protease core is assisted by the presence of inhibitors in the
active site cleft that serve to make contacts between the PC1 and
PC2 domains.
NO SOONER ACTIVATED THAN INHIBITED BY CALPASTATIN?
Domain structure of calpastatin
Calpastatin, the endogenous inhibitor of calpain, is the product
of a single gene with no known homologues [105]. A
number of isoforms are produced by alternative splicing and
transcription initiation. The archetypical calpastatin is a protein of
approximately 70 kDa. It is composed of an N-terminal L subunit
of unknown function followed by four independent inhibitory
domains, each of which contains three subdomains known as
A, B and C. The intact protein is capable of simultaneously
binding to and inhibiting four calpain molecules [106]. Calpastatin
has been shown to be intrinsically unstructured, although NMR
studies show that the A and C subdomains exhibit a propensity for
347
being α-helical [107] and the central region of the B subdomain
displays evidence for the formation of a β-turn [107,108]. The
B subdomain is the only one that exhibits any inhibitory activity.
A 27-residue fragment spanning this region, known as the B-27
peptide, has been shown to specifically inhibit calpain [109,110].
A structure of a 19-residue fragment of the C subdomain (peptide
C) in complex with a small subunit PEF homodimer showed that
the amphipathic α-helix of peptide C binds in a hydrophobic
groove on PEF(S) that becomes more open when the domain
binds Ca2 + [111]. By homology, the A subdomain was predicted
to bind to a similar groove on PEF(L). These structures facilitated
a model made by these authors in which the ends of calpastatin
are tethered by binding to the PEF domains in such a way that the
inhibitory B peptide could interact with the protease core.
Mechanism of action
The crystal structures of the first and fourth domains of calpastatin
in complex with Ca2 + -bound calpain-2 show exactly how
calpastatin is able to inhibit calpain without itself being cut
[29,30]. The A and C subdomains, as predicted and demonstrated
respectively, form α-helical regions that interact with the PEF
domains of the large and small subunits respectively. The extended
B subdomain binds across the surface of the C2L domain and
passes through the active site between the PC1 and PC2 domains.
The B subdomain sequence forms tight interactions with the
enzyme surface on either side of the catalytically active cysteine
residue. That is, the inhibitory peptide region lies within primed
and unprimed sides of the cleft, but in the vicinity of that cysteine
residue, the backbone of calpastatin loops out away from the
sulfhydryl group (Figure 12B). The loop contains a series of βturns corresponding to those identified by NMR of the isolated
inhibitor in solution [107,108]. It is these turns beginning with
one at the conserved glycine residue in the equivalent of the P1
position that enable the inhibitor to bend out of the cleft away
from the cysteine residue.
OTHER CALPAIN FAMILY MEMBERS
Calpain-3
This calpain [11] has been intensively studied, partly because of its
abundance in skeletal muscle and because inactivating mutations
in its gene (CAPN3) cause the genetic disease LGMD2A (limb
girdle muscular dystrophy type 2A) [112].
Three extra sequences that are not in CAPN2 increase the
CAPN3 molecular mass from 80 kDa to 94 kDa (hence the nickname p94). NS is a long N-terminal sequence that takes the
place of the N-terminal anchor helix in CAPN2. IS1 (insertion
sequence 1) is a 48-residue segment within the PC2 domain. IS2
lies in the region connecting the C2L and PEF(L) domains. It
appears to bind calpain-3 to titin of the sarcomere [113]. Despite
calpain-3’s abundance in muscle, it has not been possible to
produce the native enzyme without autoproteolysis [114]. The
same was true for attempts to produce recombinant enzyme
in COS cells [18]. Kinbara et al. [18] reported that the active
form of the enzyme was initially cut into two large fragments of
60 kDa and 58 kDa, with further processing to a 55 kDa fragment.
The insertion sequences IS1 and IS2 were identified as sites of
autoproteolysis. Consistent with this, a tissue-specific isoform
of calpain-3 (Lp82) that lacks these insertion sequences has
been characterized from rat lens; and this form of the enzyme
is relatively resistant to autoproteolysis [19]. Studies performed
with the protease core of calpain-3 suggest that IS1 is a propeptide
and that the first proteolytic cut made by this calpain is an
c The Authors Journal compilation c 2012 Biochemical Society
348
R.L. Campbell and P.L. Davies
Table 3
Conservation of calcium-binding residues in human calpains
The conservation of Ca2 + -binding residues in human calpains (compared with rat calpain-2)
was evaluated on the basis of a sequence alignment. Residues that contribute side-chain oxygen
atom(s) are indicated with an asterisk, others contribute only a main-chain oxygen atom. Italic
residues are conserved, bold residues contribute one side-chain oxygen atom (asparagine or
glutamine), underlined residues are not conserved. Note: since the two aspartate residues (2nd
and 4th positions) in the PC2 Ca2 + -binding site only contribute one oxygen atom each, if a
sequence had a glutamate, asparagine or glutamine instead of an aspartate, that residue was
still considered conserved.
PC1 domain
Calpain
Rat calpain-2
Human calpain-1
Human calpain-2
Human calpain-3
Human calpain-5
Human calpain-6
Human calpain-7
Human calpain-8
Human calpain-9
Human calpain-10
Human calpain-11
Human calpain-12
Human calpain-13
Human calpain-14
Human calpain-15
Figure 12 Binding and inhibition of calpain-2 by an inhibitory domain of
calpastatin
(A) The structure of Ca2 + -activated rat calpain-2 [29] is shown in surface representation with
CAST4 represented in ribbon format (purple). Ca2 + atoms are shown as mauve spheres. (B)
Close-up view of the active site cleft region bounded by the dotted lines in (A) showing the
calpastatin B peptide looping out around the active site cysteine. N and O atoms of calpastatin
are shown in blue and red respectively. Key residues from the core and inhibitor are indicated
in stick representation. Note the stacking between Pro620 (calpastatin) and Trp288 (calpain).
intramolecular autolytic cleavage near the N-terminal end of IS1
[104,115]. Once this cut is made, the calpain-3 core can hydrolyse
exogenous substrates and it also makes a distal cut towards the
C-terminal end of IS1. The residues that constitute the two nonEF-hand Ca2 + -binding sites of the core are conserved in calpain-3
(Table 3) [25], and the rate of autolytic cleavage is accelerated by
increasing the Ca2 + concentration. However, it has been reported
that Na + can also activate the enzyme [116].
There is evidence that calpain-3 forms a homodimer during
purification [18]. Given the natural tendency for PEF domains
to dimerize [31], we anticipated that these domains might be the
site of dimerization. Expression of the calpain-3 PEF domain
in E. coli produced a very stable homodimer and a model was
developed with the two active sites located at opposite ends of
the molecule [36]. Once the structure of calpain-2 was solved,
it was possible to homology model most of calpain-3 (all except
NS, IS1 and IS2, for which there are still no homologues). One
reason for doing the modelling was to see if any of the more
obscure LGMD2A point mutations could be explained in terms of
the three-dimensional structure [117]. After the calcium-activated
structure became available [29,30], this modelling exercise was
repeated [118]. Some of the LGMD2A mutations that could not
be rationalized in the context of the apo structure made sense
c The Authors Journal compilation c 2012 Biochemical Society
I
V
I
I
V
V
I
V
L
P
M
C
L
F
F
G
G
G
G
G
N
T
G
G
D
G
D
D
K
T
PC2 domain
∗
∗
∗
∗
D
D
D
D
D
Q
S
D
D
Q
D
D
D
D
D
E
E
E
E
E
E
E
E
E
V
E
E
E
L
Q
E
E
E
E
E
E
R
E
E
C
E
E
E
E
S
D
D
D
D
D
E
E
D
D
E
D
D
D
D
D
∗
Q
M
H
V
V
M
T
V
A
L
T
K
R
D
G
D
D
D
D
D
D
N
D
D
E
D
D
D
D
E
E
E
E
E
E
E
I
E
E
E
E
E
E
E
V
when the apo- and Ca2 + -bound structures were compared. The
two modelling studies showed that many of the LGMD2A point
mutations are to residues that provide contacts between domains
and are likely to be important for the folding of calpain-3 and its
transitioning between the active and inactive forms of the enzyme.
One of the best examples of this type became apparent when the
structure of the Ca2 + -bound calpain-1 core structure was solved
[25]. The human enzyme is inactivated by an arginine to glycine
mutation. In the Ca2 + -bound structure of the rat calpain-1 core,
Arg104 makes a double salt bridge to Glu333 , which helps to stabilize
the activated conformation.
Is calcium signalling involved in the activation of most calpains?
The discovery of the two calcium-binding sites in the calpain
protease cores of calpains 1 and 2 is also significant in the sense
that they provide an activation mechanism for the majority of
calpains that lack PEF domains. Most of the calcium co-ordinating
residues in PC1 and PC2 are conserved throughout the human
calpains (Table 3). In some cases, an aspartate or glutamate residue
is replaced with a side chain that can offer one co-ordination
position (asparagine or glutamine). The presence of calciumbinding sites in the protease core domains is particularly important
in linking their activation to calcium signalling, because the only
other domain common to most calpains is the C2L domain.
Although it was long speculated that this domain might bind
Ca2 + [119], there is no evidence for this in Ca2 + -activated CAST
(calpastatin inhibitory domain)-bound rat calpain-2 [29,30]. Also,
when the crystal structure of the C2L domain of human calpain-7
(PDB code 2QFE) was solved in the presence of Ca2 + , there were
no calcium ions bound to the protein.
FUTURE STUDIES
Many more structures to be done
Calpain structures and those of their constituent domains have
been extremely useful in elucidating the enzyme’s activation by
Structure–function relationships in calpains
calcium and inhibition by calpastatin. Structures of the calpain
core bound to low molecular mass inhibitors and of whole
calpain bound to calpastatin are paving the way for the
structure-guided design of more potent inhibitors with increased
specificity for calpains over other cysteine proteases. However,
few new structures have emerged in recent years despite a
considerable effort expended on human calpains by the Structural
Genomics Consortium in Toronto. One of the difficulties lies in
producing recombinant calpains. Until recently, the only isoform
which has been reliably produced in high yield from bacteria is
rat calpain-2 [82]. Even then it was not possible to make it with an
intact GR domain. The human orthologue differs by only 7 % and
yet until recently it has been difficult to produce in E. coli [102].
Unlike its calpain-2 paralogue, rat calpain-1 was produced in E.
coli with very poor yield, although as stated earlier the amount
produced could be increased by making a chimaera of the two
enzymes where the N- and C-terminal regions of CAPN1 were
replaced by CAPN2 [103].
The absence of the GR domain from the present structures
and most in vitro studies leaves considerable uncertainty about
its role. Because of this omission, one cannot be completely
confident about any aspect of calpain activation. Interestingly, the
GR domain’s presumed location in the structure of calpastatinbound active calpain would place it in position to make additional
contacts with that portion of calpastatin lying between the B and
C sub-domains that is presently not visible in the electron density
because of polypeptide flexibility (Figure 12)
CONCLUSION
Calpains are complex, multi-domain, calcium-dependent cysteine
proteases. They are intracellular enzymes responsible for limited
cleavage of target proteins during calcium signalling, and as such
their activity is tightly regulated. The recently published structures
of calcium-activated calpain-2 bound to different repeats of its
natural inhibitor, calpastatin, has been an important milestone
in the quest for knowledge about calpains. Together with the
earlier solved apo-structure of this calpain isoform, we now have
an appreciation for how Ca2 + switches the protease from
the inactive to the active form, and how the latter is then
inhibited by calpastatin. Along the way, the ability to produce
and crystallize an active calpain protease core free from the
complications of autoproteolysis has paved the way for structurebased design of calpain-specific low molecular mass inhibitors.
These compounds are needed to better understand the cellular
functions of calpain and as leads in drug development to combat
the over-activation of calpain seen in many diseases.
ACKNOWLEDGEMENTS
We thank past and present members of the Davies lab for their contributions to the
structural biology of the calpain field. Kristin Low, Jordan Chou and Olivia Macleod
supplied previously unpublished work and/or assisted with the production of Figures.
FUNDING
We thank the Canadian Institutes for Health Research for financial support of our research.
P.L.D. holds a Canada Research Chair in Protein Engineering.
REFERENCES
1 Goll, D. E., Thompson, V. F., Li, H., Wei, W. and Cong, J. (2003) The calpain system.
Physiol. Rev. 83, 731–801
2 Guroff, G. (1964) A neutral, calcium-activated proteinase from the soluble fraction of rat
brain. J. Biol. Chem. 239, 149–155
349
3 Huston, R. B. and Krebs, E. G. (1968) Activation of skeletal muscle phosphorylase
kinase by Ca2 + . II. Identification of the kinase activating factor as a proteolytic enzyme.
Biochemistry 7, 2116–2122
4 Dayton, W. R., Goll, D. E., Zeece, M. G., Robson, R. M. and Reville, W. J. (1976) A
Ca2 + -activated protease possibly involved in myofibrillar protein turnover. Purification
from porcine muscle. Biochemistry 15, 2150–2158
5 Dayton, W. R., Reville, W. J., Goll, D. E. and Stromer, M. H. (1976) A Ca2 + -activated
protease possibly involved in myofibrillar protein turnover. Partial characterization of the
purified enzyme. Biochemistry 15, 2159–2167
6 Ishiura, S., Sugita, H., Nonaka, I. and Imahori, K. (1980) Calcium-activated neutral
protease. Its localization in the myofibril, especially at the Z-band. J. Biochem. 87,
343–346
7 Murachi, T., Tanaka, K., Hatanaka, M. and Murakami, T. (1980) Intracellular
Ca2 + -dependent protease (calpain) and its high-molecular-weight endogenous
inhibitor (calpastatin). Adv. Enzyme Regul. 19, 407–424
8 Mellgren, S. I. (1980) [Alzheimer-type dementia. Possible relation to hypofunction of the
cholinergic central nervous system]. Tidsskr. Nor. Laegeforen. 100, 1355–1356
9 Wheelock, M. J. (1982) Evidence for two structurally different forms of skeletal muscle
Ca2 + -activated protease. J. Biol. Chem. 257, 12471–12474
10 Ohno, S., Emori, Y., Imajoh, S., Kawasaki, H., Kisaragi, M. and Suzuki, K. (1984)
Evolutionary origin of a calcium-dependent protease by fusion of genes for a thiol
protease and a calcium-binding protein? Nature 312, 566–570
11 Sorimachi, H., Imajoh-Ohmi, S., Emori, Y., Kawasaki, H., Ohno, S., Minami, Y. and
Suzuki, K. (1989) Molecular cloning of a novel mammalian calcium-dependent protease
distinct from both m- and mu-types. Specific expression of the mRNA in skeletal
muscle. J. Biol. Chem. 264, 20106–20111
12 Dear, N., Matena, K., Vingron, M. and Boehm, T. (1997) A new subfamily of vertebrate
calpains lacking a calmodulin-like domain: implications for calpain regulation and
evolution. Genomics 45, 175–184
13 Dear, T. N. and Boehm, T. (2001) Identification and characterization of two novel calpain
large subunit genes. Gene 274, 245–252
14 Croall, D. E. and Ersfeld, K. (2007) The calpains: modular designs and functional
diversity. Genome Biol. 8, 218
15 Sorimachi, H., Hata, S. and Ono, Y. (2011) Impact of genetic insights into calpain
biology. J. Biochem. 150, 23–37
16 Cong, J., Goll, D. E., Peterson, A. M. and Kapprell, H. P. (1989) The role of autolysis in
activity of the Ca2 + -dependent proteinases (mu-calpain and m-calpain). J. Biol. Chem.
264, 10096–10103
17 Graham-Siegenthaler, K., Gauthier, S., Davies, P. L. and Elce, J. S. (1994) Active
recombinant rat calpain II. Bacterially produced large and small subunits associate both
in vivo and in vitro . J. Biol. Chem. 269, 30457–30460
18 Kinbara, K., Ishiura, S., Tomioka, S., Sorimachi, H., Jeong, S. Y., Amano, S., Kawasaki,
H., Kolmerer, B., Kimura, S., Labeit, S. and Suzuki, K. (1998) Purification of native p94,
a muscle-specific calpain, and characterization of its autolysis. Biochem J. 335,
589–596
19 Ma, H., Fukiage, C., Azuma, M. and Shearer, T. R. (1998) Cloning and expression of
mRNA for calpain Lp82 from rat lens: splice variant of p94. Invest. Ophthalmol. Vis. Sci.
39, 454–461
20 Hata, S., Doi, N., Kitamura, F. and Sorimachi, H. (2007) Stomach-specific calpain,
nCL-2/calpain 8, is active without calpain regulatory subunit and oligomerizes through
C2-like domains. J. Biol. Chem. 282, 27847–27856
21 Hata, S., Abe, M., Suzuki, H., Kitamura, F., Toyama-Sorimachi, N., Abe, K., Sakimura, K.
and Sorimachi, H. (2010) Calpain 8/nCL-2 and calpain 9/nCL-4 constitute an active
protease complex, G-calpain, involved in gastric mucosal defense. PLoS Genet. 6,
e1001040
22 Schad, E., Farkas, A., Jekely, G., Tompa, P. and Friedrich, P. (2002) A novel human small
subunit of calpains. Biochem. J. 362, 383–388
23 Hosfield, C. M., Elce, J. S., Davies, P. L. and Jia, Z. (1999) Crystal structure of calpain
reveals the structural basis for Ca2 + -dependent protease activity and a novel mode of
enzyme activation. EMBO J. 18, 6880–6889
24 Strobl, S., Fernandez-Catalan, C., Braun, M., Huber, R., Masumoto, H., Nakagawa, K.,
Irie, A., Sorimachi, H., Bourenkow, G., Bartunik, H. et al. (2000) The crystal structure of
calcium-free human m-calpain suggests an electrostatic switch mechanism for
activation by calcium. Proc. Natl. Acad. Sci. U.S.A. 97, 588–592
25 Moldoveanu, T., Hosfield, C. M., Lim, D., Elce, J. S., Jia, Z. and Davies, P. L. (2002) A
Ca2 + switch aligns the active site of calpain. Cell 108, 649–660
26 Rawlings, N. D., Barrett, A. J. and Bateman, A. (2012) MEROPS: the database of
proteolytic enzymes, their substrates and inhibitors. Nucleic Acids Res. 40, D343–D350
27 Corbalan-Garcia, S. and Gomez-Fernandez, J. C. (2010) The C2 domains of classical
and novel PKCs as versatile decoders of membrane signals. Biofactors 36, 1–7
28 Rizo, J. and Sudhof, T. C. (1998) C2-domains, structure and function of a universal
Ca2 + -binding domain. J. Biol. Chem. 273, 15879–15882
c The Authors Journal compilation c 2012 Biochemical Society
350
R.L. Campbell and P.L. Davies
29 Hanna, R. A., Campbell, R. L. and Davies, P. L. (2008) Calcium-bound structure of
calpain and its mechanism of inhibition by calpastatin. Nature 456, 409–412
30 Moldoveanu, T., Gehring, K. and Green, D. R. (2008) Concerted multi-pronged attack by
calpastatin to occlude the catalytic cleft of heterodimeric calpains. Nature 456, 404–408
31 Kretsinger, R. H. (1997) EF-hands embrace. Nat. Struct. Biol. 4, 514–516
32 Ono, Y. and Sorimachi, H. (2012) Calpains: an elaborate proteolytic system. Biochim.
Biophys. Acta 1824, 224–236
33 Sorimachi, H. and Suzuki, K. (2001) The structure of calpain. J. Biochem. 129, 653–664
34 Ravulapalli, R., Campbell, R. L., Gauthier, S. Y., Dhe-Paganon, S. and Davies, P. L.
(2009) Distinguishing between calpain heterodimerization and homodimerization.
FEBS J. 276, 973–982
35 Kinbara, K., Sorimachi, H., Ishiura, S. and Suzuki, K. (1998) Skeletal muscle-specific
calpain, p49: structure and physiological function. Biochem. Pharmacol. 56, 415–420
36 Ravulapalli, R., Diaz, B. G., Campbell, R. L. and Davies, P. L. (2005) Homodimerization
of calpain 3 penta-EF-hand domain. Biochem. J. 388, 585–591
37 Sorimachi, H., Hata, S. and Ono, Y. (2011) Calpain chronicle – an enzyme family under
multidisciplinary characterization. Proc. Jpn. Acad. Ser. B 87, 287–327
38 Franco, S. J. and Huttenlocher, A. (2005) Regulating cell migration: calpains make the
cut. J. Cell Sci. 118, 3829–3838
39 Lebart, M. C. and Benyamin, Y. (2006) Calpain involvement in the remodeling of
cytoskeletal anchorage complexes. FEBS J. 273, 3415–3426
40 Chan, K. T., Bennin, D. A. and Huttenlocher, A. (2010) Regulation of adhesion dynamics
by calpain-mediated proteolysis of focal adhesion kinase (FAK). J. Biol. Chem. 285,
11418–11426
41 Wells, A., Huttenlocher, A. and Lauffenburger, D. A. (2005) Calpain proteases in cell
adhesion and motility. Int. Rev. Cytol. 245, 1–16
42 Cortesio, C. L., Perrin, B. J., Bennin, D. A. and Huttenlocher, A. (2010) Actin-binding
protein-1 interacts with WASp-interacting protein to regulate growth factor-induced
dorsal ruffle formation. Mol. Biol. Cell 21, 186–197
43 Brini, M. and Carafoli, E. (2009) Calcium pumps in health and disease. Physiol. Rev. 89,
1341–1378
44 McCall, K. (2010) Genetic control of necrosis - another type of programmed cell death.
Curr. Opin. Cell Biol. 22, 882–888
45 Inserte, J., Barrabes, J. A., Hernando, V. and Garcia-Dorado, D. (2009) Orphan targets
for reperfusion injury. Cardiovasc. Res. 83, 169–178
46 Bertipaglia, I. and Carafoli, E. (2007) Calpains and human disease. Subcell. Biochem.
45, 29–53
47 Biswas, S., Harris, F., Dennison, S., Singh, J. P. and Phoenix, D. (2005) Calpains:
enzymes of vision? Med. Sci. Monit. 11, RA301–RA310
48 Paquet-Durand, F., Johnson, L. and Ekstrom, P. (2007) Calpain activity in retinal
degeneration. J. Neurosci. Res. 85, 693–702
49 Pietsch, M., Chua, K. C. and Abell, A. D. (2010) Calpains: attractive targets for the
development of synthetic inhibitors. Curr. Top. Med. Chem. 10, 270–293
50 Carragher, N. O. (2006) Calpain inhibition: a therapeutic strategy targeting multiple
disease states. Curr. Pharm. Des. 12, 615–638
51 Donkor, I. O. (2011) Calpain inhibitors: a survey of compounds reported in the patent
and scientific literature. Expert Opin. Ther. Pat. 21, 601–636
52 Moldoveanu, T., Hosfield, C. M., Jia, Z., Elce, J. S. and Davies, P. L. (2001)
Ca2 + -induced structural changes in rat m-calpain revealed by partial proteolysis.
Biochim. Biophys. Acta 1545, 245–254
53 Hata, S., Sorimachi, H., Nakagawa, K., Maeda, T., Abe, K. and Suzuki, K. (2001) Domain
II of m-calpain is a Ca2 + -dependent cysteine protease. FEBS Lett. 501, 111–114
54 Moldoveanu, T., Jia, Z. and Davies, P. L. (2004) Calpain activation by cooperative Ca2 +
binding at two non-EF-hand sites. J. Biol. Chem. 279, 6106–6114
55 Suzuki, K., Emori, Y., Ohno, S., Imahori, S., Kawasaki, H. and Miyake, S. (1986)
Structure and function of the small (30K) subunit of calcium-activated neutral protease
(CANP). Biomed. Biochim. Acta 45, 1487–1491
56 Nakagawa, K., Masumoto, H., Sorimachi, H. and Suzuki, K. (2001) Dissociation of
m-calpain subunits occurs after autolysis of the N-terminus of the catalytic subunit, and
is not required for activation. J. Biochem. 130, 605–611
57 Alexa, A., Bozoky, Z., Farkas, A., Tompa, P. and Friedrich, P. (2004) Contribution of
distinct structural elements to activation of calpain by Ca2 + ions. J. Biol. Chem. 279,
20118–20126
58 Melloni, E., Averna, M., Salamino, F., Sparatore, B., Minafra, R. and Pontremoli, S.
(2000) Acyl-CoA-binding protein is a potent m-calpain activator. J. Biol. Chem. 275,
82–86
59 Melloni, E., Michetti, M., Salamino, F. and Pontremoli, S. (1998) Molecular and
functional properties of a calpain activator protein specific for mu-isoforms. J. Biol.
Chem. 273, 12827–12831
60 Farkas, A., Nardai, G., Csermely, P., Tompa, P. and Friedrich, P. (2004) DUK114, the
Drosophila orthologue of bovine brain calpain activator protein, is a molecular
chaperone. Biochem. J. 383, 165–170
c The Authors Journal compilation c 2012 Biochemical Society
61 Yoshizawa, T., Sorimachi, H., Tomioka, S., Ishiura, S. and Suzuki, K. (1995) Calpain
dissociates into subunits in the presence of calcium ions. Biochem. Biophys. Res.
Commun. 208, 376–383
62 Anagli, J., Vilei, E. M., Molinari, M., Calderara, S. and Carafoli, E. (1996) Purification of
active calpain by affinity chromatography on an immobilized peptide inhibitor. Eur. J.
Biochem. 241, 948–954
63 Li, H., Thompson, V. F. and Goll, D. E. (2004) Effects of autolysis on properties of muand m-calpain. Biochim. Biophys. Acta 1691, 91–103
64 Pal, G. P., Elce, J. S. and Jia, Z. (2001) Dissociation and aggregation of calpain in the
presence of calcium. J. Biol. Chem. 276, 47233–47238
65 Glading, A., Bodnar, R. J., Reynolds, I. J., Shiraha, H., Satish, L., Potter, D. A., Blair,
H. C. and Wells, A. (2004) Epidermal growth factor activates m-calpain (calpain II), at
least in part, by extracellular signal-regulated kinase-mediated phosphorylation. Mol.
Cell. Biol. 24, 2499–2512
66 Baki, A., Tompa, P., Alexa, A., Molnar, O. and Friedrich, P. (1996) Autolysis parallels
activation of mu-calpain. Biochem. J. 318, 897–901
67 Brown, N. and Crawford, C. (1993) Structural modifications associated with the change
in Ca2 + sensitivity on activation of m-calpain. FEBS Lett. 322, 65–68
68 Elce, J. S., Hegadorn, C. and Arthur, J. S. (1997) Autolysis, Ca2 + requirement, and
heterodimer stability in m-calpain. J. Biol. Chem. 272, 11268–11275
69 Suzuki, K., Sorimachi, H., Yoshizawa, T., Kinbara, K. and Ishiura, S. (1995) Calpain:
novel family members, activation, and physiologic function. Biol. Chem. Hoppe Seyler
376, 523–529
70 Chou, J. S., Impens, F., Gevaert, K. and Davies, P. L. (2011) m-Calpain activation in vitro
does not require autolysis or subunit dissociation. Biochim. Biophys. Acta 1814,
864–872
71 Thompson, V. F., Lawson, K. R., Barlow, J. and Goll, D. E. (2003) Digestion of mu- and
m-calpain by trypsin and chymotrypsin. Biochim. Biophys. Acta 1648, 140–153
72 Crawford, C., Willis, A. C. and Gagnon, J. (1987) The effects of autolysis on the
structure of chicken calpain II. Biochem. J. 248, 579–588
73 DeMartino, G. N., Huff, C. A. and Croall, D. E. (1986) Autoproteolysis of the small
subunit of calcium-dependent protease II activates and regulates protease activity.
J. Biol. Chem. 261, 12047–12052
74 Yoshizawa, T., Sorimachi, H., Tomioka, S., Ishiura, S. and Suzuki, K. (1995) A catalytic
subunit of calpain possesses full proteolytic activity. FEBS Lett. 358, 101–103
75 Suzuki, K. and Sorimachi, H. (1998) A novel aspect of calpain activation. FEBS Lett.
433, 1–4
76 Zhang, W. and Mellgren, R. L. (1996) Calpain subunits remain associated during
catalysis. Biochem. Biophys. Res. Commun. 227, 891–896
77 Dutt, P., Arthur, J. S., Croall, D. E. and Elce, J. S. (1998) m-Calpain subunits remain
associated in the presence of calcium. FEBS Lett. 436, 367–371
78 Kitagaki, H., Tomioka, S., Yoshizawa, T., Sorimachi, H., Saido, T. C., Ishiura, S. and
Suzuki, K. (2000) Autolysis of calpain large subunit inducing irreversible
dissociation of stoichiometric heterodimer of calpain. Biosci. Biotechnol. Biochem. 64,
689–695
79 Blanchard, H., Grochulski, P., Li, Y., Arthur, J. S., Davies, P. L., Elce, J. S. and Cygler, M.
(1997) Structure of a calpain Ca2 + -binding domain reveals a novel EF-hand and
Ca2 + -induced conformational changes. Nat. Struct. Biol. 4, 532–538
80 Lin, G. D., Chattopadhyay, D., Maki, M., Wang, K. K., Carson, M., Jin, L., Yuen, P. W.,
Takano, E., Hatanaka, M., DeLucas, L. J. and Narayana, S. V. (1997) Crystal structure of
calcium bound domain VI of calpain at 1.9 Å resolution and its role in enzyme assembly,
regulation, and inhibitor binding. Nat. Struct. Biol. 4, 539–547
81 Arthur, J. S., Gauthier, S. and Elce, J. S. (1995) Active site residues in m-calpain:
identification by site-directed mutagenesis. FEBS Lett. 368, 397–400
82 Elce, J. S., Hegadorn, C., Gauthier, S., Vince, J. W. and Davies, P. L. (1995) Recombinant
calpain II: improved expression systems and production of a C105A active-site mutant
for crystallography. Protein Eng. 8, 843–848
83 Davies, P. L., Campbell, R. L. and Moldoveanu, T. (2004) Calpain. In The Handbook of
Metalloproteins (Messerschmidt, A., ed.), pp. 489–500, John Wiley & Sons, Chichester
84 Mellgren, R. L. and Huang, X. (2007) Fetuin A stabilizes m-calpain and facilitates
plasma membrane repair. J. Biol. Chem. 282, 35868–35877
85 Friedrich, P. (2004) The intriguing Ca2 + requirement of calpain activation. Biochem.
Biophys. Res. Commun. 323, 1131–1133
86 Takano, J., Tomioka, M., Tsubuki, S., Higuchi, M., Iwata, N., Itohara, S., Maki, M. and
Saido, T. C. (2005) Calpain mediates excitotoxic DNA fragmentation via mitochondrial
pathways in adult brains: evidence from calpastatin mutant mice. J. Biol. Chem. 280,
16175–16184
87 Higuchi, M., Tomioka, M., Takano, J., Shirotani, K., Iwata, N., Masumoto, H., Maki, M.,
Itohara, S. and Saido, T. C. (2005) Distinct mechanistic roles of calpain and caspase
activation in neurodegeneration as revealed in mice overexpressing their specific
inhibitors. J. Biol. Chem. 280, 15229–15237
Structure–function relationships in calpains
88 Rao, M. V., Mohan, P. S., Peterhoff, C. M., Yang, D. S., Schmidt, S. D., Stavrides, P. H.,
Campbell, J., Chen, Y., Jiang, Y., Paskevich, P. A. et al. (2008) Marked calpastatin
(CAST) depletion in Alzheimer’s disease accelerates cytoskeleton disruption and
neurodegeneration: neuroprotection by CAST overexpression. J. Neurosci. 28,
12241–12254
89 Wang, K. K., Posmantur, R., Nadimpalli, R., Nath, R., Mohan, P., Nixon, R. A., Talanian,
R. V., Keegan, M., Herzog, L. and Allen, H. (1998) Caspase-mediated fragmentation of
calpain inhibitor protein calpastatin during apoptosis. Arch. Biochem. Biophys. 356,
187–196
90 Moldoveanu, T., Campbell, R. L., Cuerrier, D. and Davies, P. L. (2004) Crystal structures
of calpain-E64 and -leupeptin inhibitor complexes reveal mobile loops gating the active
site. J. Mol. Biol. 343, 1313–1326
91 Cuerrier, D., Moldoveanu, T., Inoue, J., Davies, P. L. and Campbell, R. L. (2006) Calpain
inhibition by α-ketoamide and cyclic hemiacetal inhibitors revealed by X-ray
crystallography. Biochemistry 45, 7446–7452
92 Cuerrier, D., Moldoveanu, T., Campbell, R. L., Kelly, J., Yoruk, B., Verhelst, S. H.,
Greenbaum, D., Bogyo, M. and Davies, P. L. (2007) Development of calpain-specific
inactivators by screening of positional scanning epoxide libraries. J. Biol. Chem. 282,
9600–9611
93 Qian, J., Cuerrier, D., Davies, P. L., Li, Z., Powers, J. C. and Campbell, R. L. (2008)
Cocrystal structures of primed side-extending α-ketoamide inhibitors reveal novel
calpain-inhibitor aromatic interactions. J. Med. Chem. 51, 5264–5270
94 Li, Q., Hanzlik, R. P., Weaver, R. F. and Schonbrunn, E. (2006) Molecular mode of action
of a covalently inhibiting peptidomimetic on the human calpain protease core.
Biochemistry 45, 701–708
95 Davis, T. L., Walker, J. R., Finerty, Jr, P. J., Mackenzie, F., Newman, E. M. and
Dhe-Paganon, S. (2007) The crystal structures of human calpains 1 and 9 imply diverse
mechanisms of action and auto-inhibition. J. Mol. Biol. 366, 216–229
96 Abell, A. D., Jones, M. A., Coxon, J. M., Morton, J. D., Aitken, S. G., McNabb, S. B.,
Lee, H. Y., Mehrtens, J. M., Alexander, N. A., Stuart, B. G. et al. (2009) Molecular
modeling, synthesis, and biological evaluation of macrocyclic calpain inhibitors. Angew.
Chem., Int. Ed. Engl. 48, 1455–1458
97 Turk, B. E., Huang, L. L., Piro, E. T. and Cantley, L. C. (2001) Determination of protease
cleavage site motifs using mixture-based oriented peptide libraries. Nat. Biotechnol. 19,
661–667
98 Cuerrier, D., Moldoveanu, T. and Davies, P. L. (2005) Determination of peptide substrate
specificity for mu-calpain by a peptide library-based approach: the importance of primed
side interactions. J. Biol. Chem. 280, 40632–40641
99 Moldoveanu, T., Hosfield, C. M., Lim, D., Jia, Z. and Davies, P. L. (2003) Calpain
silencing by a reversible intrinsic mechanism. Nat. Struct. Biol. 10, 371–378
100 Elce, J. S. (2000) Expression of m-calpain in Escherichia coli . Methods Mol. Biol. 144,
47–54
101 Masumoto, H., Yoshizawa, T., Sorimachi, H., Nishino, T., Ishiura, S. and Suzuki, K.
(1998) Overexpression, purification, and characterization of human m-calpain and its
active site mutant, m-C105S-calpain, using a baculovirus expression system.
J. Biochem. 124, 957–961
102 Hata, S., Ueno, M., Kitamura, F. and Sorimachi, H. (2012) Efficient expression and
purification of recombinant human m-calpain using an Escherichia coli expression
system at low temperature. J. Biochem. 151, 417–422
103 Pal, G. P., De Veyra, T., Elce, J. S. and Jia, Z. (2003) Crystal structure of a micro-like
calpain reveals a partially activated conformation with low Ca2 + requirement. Structure
11, 1521–1526
104 Diaz, B. G., Moldoveanu, T., Kuiper, M. J., Campbell, R. L. and Davies, P. L. (2004)
Insertion sequence 1 of muscle-specific calpain, p94, acts as an internal propeptide.
J. Biol. Chem. 279, 27656–27666
351
105 Wendt, A., Thompson, V. F. and Goll, D. E. (2004) Interaction of calpastatin with calpain:
a review. Biol. Chem. 385, 465–472
106 Hanna, R. A., Garcia-Diaz, B. E. and Davies, P. L. (2007) Calpastatin simultaneously
binds four calpains with different kinetic constants. FEBS Lett. 581, 2894–2898
107 Kiss, R., Kovacs, D., Tompa, P. and Perczel, A. (2008) Local structural preferences of
calpastatin, the intrinsically unstructured protein inhibitor of calpain. Biochemistry 47,
6936–6945
108 Ishima, R., Tamura, A., Akasaka, K., Hamaguchi, K., Makino, K., Murachi, T., Hatanaka,
M. and Maki, M. (1991) Structure of the active 27-residue fragment of human
calpastatin. FEBS Lett. 294, 64–66
109 Betts, R. and Anagli, J. (2004) The β- and γ -CH2 of B27-WT’s Leu11 and Ile18 side
chains play a direct role in calpain inhibition. Biochemistry 43, 2596–2604
110 Betts, R., Weinsheimer, S., Blouse, G. E. and Anagli, J. (2003) Structural determinants of
the calpain inhibitory activity of calpastatin peptide B27-WT. J. Biol. Chem. 278,
7800–7809
111 Todd, B., Moore, D., Deivanayagam, C. C., Lin, G. D., Chattopadhyay, D., Maki, M.,
Wang, K. K. and Narayana, S. V. (2003) A structural model for the inhibition of
by calpastatin: crystal structures of the native domain VI of calpain and its
complexes with calpastatin peptide and a small molecule inhibitor. J. Mol. Biol. 328,
131–146
112 Richard, I., Broux, O., Allamand, V., Fougerousse, F., Chiannilkulchai, N., Bourg, N.,
Brenguier, L., Devaud, C., Pasturaud, P., Roudaut, C. et al. (1995) Mutations in the
proteolytic enzyme calpain 3 cause limb-girdle muscular dystrophy type 2A. Cell 81,
27–40
113 Hayashi, C., Ono, Y., Doi, N., Kitamura, F., Tagami, M., Mineki, R., Arai, T., Taguchi, H.,
Yanagida, M., Hirner, S. et al. (2008) Multiple molecular interactions implicate the
connectin/titin N2A region as a modulating scaffold for p94/calpain 3 activity in skeletal
muscle. J. Biol. Chem. 283, 14801–14814
114 Sorimachi, H., Toyama-Sorimachi, N., Saido, T. C., Kawasaki, H., Sugita, H., Miyasaka,
M., Arahata, K., Ishiura, S. and Suzuki, K. (1993) Muscle-specific calpain, p94, is
degraded by autolysis immediately after translation, resulting in disappearance from
muscle. J. Biol. Chem. 268, 10593–10605
115 Rey, M. A. and Davies, P. L. (2002) The protease core of the muscle-specific
calpain, p94, undergoes Ca2 + -dependent intramolecular autolysis. FEBS Lett. 532,
401–406
116 Ono, Y., Ojima, K., Torii, F., Takaya, E., Doi, N., Nakagawa, K., Hata, S., Abe, K. and
Sorimachi, H. (2010) Skeletal muscle-specific calpain is an intracellular
Na + -dependent protease. J. Biol. Chem. 285, 22986–22998
117 Jia, Z., Petrounevitch, V., Wong, A., Moldoveanu, T., Davies, P. L., Elce, J. S. and
Beckmann, J. S. (2001) Mutations in calpain 3 associated with limb girdle muscular
dystrophy: analysis by molecular modeling and by mutation in m-calpain. Biophys. J.
80, 2590–2596
118 Garnham, C. P., Hanna, R. A., Chou, J. S., Low, K. E., Gourlay, K., Campbell, R. L.,
Beckmann, J. S. and Davies, P. L. (2009) Limb-girdle muscular dystrophy type 2A can
result from accelerated autoproteolytic inactivation of calpain 3. Biochemistry 48,
3457–3467
119 Tompa, P., Emori, Y., Sorimachi, H., Suzuki, K. and Friedrich, P. (2001) Domain III of
calpain is a Ca2 + -regulated phospholipid-binding domain. Biochem. Biophys. Res.
Commun. 280, 1333–1339
120 Ravulapalli, R. (2009) Regulatory domains of the human calpain family.
Ph.D. Thesis, Department of Biochemistry, Queen’s University, Kingston,
Canada
121 Fuson, K. L., Montes, M., Robert, J. J. and Sutton, R. B. (2007) Structure of human
synaptotagmin 1 C2AB in the absence of Ca2 + reveals a novel domain association.
Biochemistry 46, 13041–13048
Received 6 June 2012/11 July 2012; accepted 23 July 2012
Published on the Internet 5 October 2012, doi:10.1042/BJ20120921
c The Authors Journal compilation c 2012 Biochemical Society