Ataman Shea 07

Transcription

Ataman Shea 07

Structure
Article
The NMDA Receptor NR1 C1 Region Bound
to Calmodulin: Structural Insights into Functional
Differences between Homologous Domains
Zeynep Akyol Ataman,1 Lokesh Gakhar,1 Brenda R. Sorensen,1 Johannes W. Hell,2 and Madeline A. Shea1,*
1Department
of Biochemistry
of Pharmacology
Roy J. and Lucille A. Carver College of Medicine, University of Iowa, Iowa City, IA 52242-1109, USA
*Correspondence: [email protected]
DOI 10.1016/j.str.2007.10.012
2Department
SUMMARY
Calmodulin (CaM) regulates tetrameric Nmethyl-D-aspartate receptors (NMDARs) by
binding tightly to the C0 and C1 regions of its
NR1 subunit. A crystal structure (2HQW; 1.96 Å)
of calcium-saturated CaM bound to NR1C1
(peptide spanning 875–898) showed that NR1
S890, whose phosphorylation regulates membrane localization, was solvent protected,
whereas the endoplasmic reticulum retention
motif was solvent exposed. NR1 F880 filled the
CaM C-domain pocket, whereas T886 was closest to the N-domain pocket. This 1-7 pattern
was most similar to that in the CaM-MARCKS
complex. Comparison of CaM-ligand wraparound conformations identified a core tetrad
of CaM C-domain residues (FLMMC) that contacted all ligands consistently. An identical
tetrad of N-domain residues (FLMMN) made variable sets of contacts with ligands. This CaMNR1C1 structure provides a foundation for
designing mutants to test the role of CaM in
NR1 trafficking as well as insights into how the
homologous CaM domains have different roles
in molecular recognition.
INTRODUCTION
The ionotropic N-methyl-D-aspartate receptor (NMDAR)
is a major source of calcium flux into neurons in the brain
and has a critical role in learning, memory, neural development, and synaptic plasticity (Mori and Mishina, 1995).
Mammalian NMDARs have two families of subunits designated NMDAR1 (NR1) and NMDAR2 (NR2). Understanding the roles of NR1 and NR2 subunits in brain function
is complicated by variable developmental and spatial
expression of their mRNA, and by the presence of eight
variants of NR1 arising from N-terminal sequence variations and alternative splicing of four exons encoding the
C0 (membrane-proximal), C1, C2, and C20 regions (Figure 1A). The C-terminal tail of NR1 binds several proteins,
including calmodulin (CaM). CaM has two highly homologous domains (N and C); each domain has two EF-hand
sites that bind calcium cooperatively. CaM regulates
a wide array of target proteins, including kinases, phosphatases, and ion channels (cf. Bhattacharya et al.,
2004; Chin and Means, 2000; Saimi and Kung, 2002; Vetter and Leclerc, 2003).
Upon calcium influx, CaM induces inactivation of
NMDAR (reducing its open rate and mean open time) by
binding to NR1 C0 and C1 (Ehlers et al., 1996b). Inactivation releases NMDAR from the neuronal cytoskeleton by
disrupting interactions between NR1 and a-actinin2
(Krupp et al., 1999; Zhang et al., 1998). CaM binding to
NR1 C1 significantly enhances NMDAR inactivation (Ehlers et al., 1996b), which absolutely requires C0 (Zhang
et al., 1998). Dissociation constants for (Ca2+)4-CaM
binding to peptides representing the CaM-binding domains (CaMBDs) of NR1 showed that its affinity for C1
(NR1C1p; aa 875–898) was 20-fold more favorable than
its affinity for C0 (NR1C0p; aa 838–863) (Ehlers et al.,
1996b). The KD values (4 and 87 nM, respectively) suggest
that (Ca2+)4-CaM binds both C0 and C1 intracellularly,
where [CaM]free is 50–75 nM (Wu and Bers, 2007). C1
has an endoplasmic reticulum (ER) retention motif (R893–
R895; Figure 1A) and a Protein Kinase C (PKC) phosphorylation site (S896) that are needed for proper maturation
and ER release (Scott et al., 2001; Standley et al., 2000).
NR1 S890 phosphorylation by PKC disrupts surfaceassociated clusters of NR1, causing an even distribution
throughout fibroblasts (Tingley et al., 1997), and affects
receptor potentiation (Zheng et al., 1999). The position of
CaM relative to these regulatory motifs in NR1 is not
known.
To explore the role of CaM in NR1 trafficking, it is not
possible to make a viable CaM knockout organism because CaM is essential and has multiple targets. Our strategy was to determine the binding interface to provide
a platform for targeted mutagenesis of NR1. Here, we report a crystal structure of (Ca2+)4-CaM bound to NR1C1p
(1.96 Å resolution) in which the CaM domains wrapped
around helical NR1C1p (Figures 2A and 2B). The ER retention motif (R893–R895) and S896 were solvent exposed,
whereas S890 was buried in the N domain of CaM.
NR1C1p was predicted to be a 1-12 CaMBD (Yap et al.,
2000), but it was found to be a 1-7 CaMBD. NR1 F880
Structure 15, 1603–1617, December 2007 ª2007 Elsevier Ltd All rights reserved 1603
Structure
Calmodulin Bound to NMDAR NR1 C1 Region
Figure 2. Crystal Structure of the CaM-NR1C1p Complex
(A) NR1C1p sequence and structure superimposed on its electron
density map contoured at 1.0s.
(B and C) Alternate views of CaM-NR1C1p (2HQW) showing the CaM
N-domain backbone (blue), the C domain (red), Ca2+ ions and binding
sites (yellow), and NR1C1p (gray). The figure was made with MacPymol.
(D and E) Alignment of 17 canonical CaM-target complexes by their Ca
atoms of the (D) N-domain (residues 5–72; 68 atoms) and (E) C-domain
(residues 84–146; 63 atoms) FLMM residues as described in Experimental Procedures.
Figure 1. CaM Binding to NMDAR NR1
(A) Schematic diagram of NR1 indicating relative positions of intracellular regions C0, C1, and C2/C20 , and sequences of C0 and C1. The
CaMBD sequence of C0 (residues 838–865) shows the single tryptophan (presumed anchor) residue boxed. The sequence of C1 (residues
875–898) is shown with the ER retention signal (RRR) underlined, the
PKC sites boxed and shaded, and residues F880 and T886 boxed.
(B) Binding of CaM to NR1C1p monitored by fluorescence anisotropy
of Fl-NR1C1p (intrinsic value of 0.04) to a final concentration of 51.5 mM
apo CaM (open; KD = 158 mM) or 0.76 mM (Ca2+)4-CaM (filled; KD =
1.99 nM). The asterisk indicates that the anisotropy of Fl-NR1C1p
titrated with apo CaM was normalized to the value (0.13) observed
after saturation with calcium.
(C) Simulation of apo CaM (dashed black) and (Ca2+)4-CaM (solid
black) binding to NR1C1p with equilibrium constants from (B). For
comparison, binding of CaM to NR1C0p (gray) simulated with a KD
of 87 nM (Ehlers et al., 1996b) for (Ca2+)4-CaM (solid) and a KD of
2.25 mM (Akyol et al., 2004) for apo CaM (dashed).
was anchored in the C domain of CaM, whereas T886
(rather than F891) contacted the highest number of residues in the CaM N domain. The same 1-7 motif and nearly
identical primary contact residues were identified for CaM
bound to a MARCKS (myristoylated, alanine-rich, PKC
substrate) peptide (1IWQ). In that case, CaM binding interrupts attachment of the actin cytoskeleton to the plasma
membrane (Aderem, 1992). Parallels between proteins
regulating MARCKS attachment to the cytoskeleton and
proteins binding C1 suggest a similar mechanism of regulation of C1 in the formation of NR1-rich clusters (Ehlers
et al., 1996b; Tingley et al., 1997).
The 1-7 motif found here is unusual among CaM-target
interfaces in 17 complexes in which both the N and C domains of CaM contacts the target. The only other known
case is CaM bound to MARCKS. To determine whether
the nature as well as the spacing of CaM residues critical
to molecular recognition were different among these
structures, we analyzed the CaM-target contacts in all
17 complexes and found that 4 residues in the C domain
(F92, L105, M124, and M144: FLMMC) were used consistently by CaM to contact targets. Although a structurally
equivalent tetrad (F19, L32, M51, and M71: FLMMN) was
observed in the CaM N domain, these CaM residues
were not used identically by all targets.
RESULTS
Binding of CaM to NR1C1p and NR1C0p
Titrations of NR1C1p with (Ca2+)4-CaM and apo (Ca2+-depleted) CaM (Figure 1B) yielded a KD of 2.0 ± 0.1 nM for
1604 Structure 15, 1603–1617, December 2007 ª2007 Elsevier Ltd All rights reserved
Structure
(Ca2+)4-CaM, which agreed well with the value of 4 nM
determined previously (Ehlers et al., 1996b). A KD of 158 ±
3 mM (KA of 6.33E3 M1) was resolved for apo CaM binding to NR1C1p. Compared to NR1 C0, (Ca2+)4-CaM binds
C0 with a KD of 87 nM (Ehlers et al., 1996b) and apo CaM
binds C0 with a KD of 2.25 mM (Akyol et al., 2004). Comparison of simulated equilibrium titrations of CaM binding to
NR1C1p and NR1C0p (Figure 1C) showed that the midpoints for apo and (Ca2+)4-CaM binding to NR1Cp differed
by four orders of magnitude.
Structure of the (Ca2+)4-CaM-NR1C1p Complex
The crystal structure of (Ca2+)4-CaM bound to NR1C1p
was determined to 1.96 Å resolution (Figures 2A–2C). It
adopted the canonical CaM-target conformation in which
both the N and C domains of CaM contacted peptide to
form a compact, ellipsoidal complex. CaM residues 1, 2,
75–80, and 148 were disordered and were not included
in the model. Figure 2A shows the electron density map
for NR1C1p (residues S897 and K898 were disordered);
the peptide was built into the final model manually. Refinement statistics are given in Table 1.
This structure was compared to 16 other similar and
nonredundant (Ca2+)4-CaM-target structures (listed in Experimental Procedures). To evaluate the overall structural
variability in these 17 compact CaM-target complexes, we
compared each one to an average structure as described
in Experimental Procedures. Among 17 structures, the average rmsd of the CaM N domain was 0.75 Å, whereas that
of the C domain was 0.59 Å (Figures 2D and 2E). The complexes that showed the highest deviation from the average
backbone conformation in the C domain were those determined by NMR (CaM with CNG channel [1SY9], CaMKK
[1CKK], and skMLCK [2BBM]); skMLCK also had the highest N-domain deviation. Two structures of the drug TFP
bound to CaM (1A29 for the N domain; 1LIN for the C
domain) had the smallest rmsd values.
Accessibility of NR1C1p Motifs
Processing and localization of C1-containing NR1 subunits is regulated by an ER retention motif (R893–R895)
and by phosphorylation of S896. These residues had
a high fractional solvent-accessible surface area (SASA):
R893, 89.4%; R894, 71.6%; R895, 92.4%; S896, 87.1%;
average SASA, 85% (Figure 3A). In contrast, S890 (implicated in subunit clustering and receptor potentiation by
PKC) was protected by the N domain of CaM, having
only 41% SASA. Two CaM N-domain residues (M36,
M51) were within 4.5 Å of NR1 S890 as determined by using Contacts of Structural Units (CSU) (Sobolev et al.,
1999). The hydroxyl of S890 was 3.22 Å from the sulfur
of M36 and 4.01 Å from that of M51, suggesting that these
residues interact in the complex (Figure 3B). No contacts
were observed between NR1 S890 and any residues in
the CaM C domain.
Interface Contacts
To explore the CaM-NR1C1p interface, CSU was used to
determine CaM residues within 4.5 Å of NR1C1p residues
Table 1. Crystallographic Data Collection and
Refinement Statistics
(Ca2+)4-CaM-NRC1p Crystal
Data Collection Statistics
Space group
P3221
Cell dimensions
a = 40.361 Å; b = 40.361 Å;
c = 175.765 Å; a = 90 ;
b = 90 ; g = 120
Resolution (Å)
19.691.90 (2.001.90)
Rsym or Rmerge
0.034 (0.255)
I/sI
17.36 (3.92)
Completeness (%)
98.4 (92.6)
Redundancy
3.01 (2.56)
Refinement Statistics
Resolution (Å)
8.571.90
Number of reflections
13,379
Rwork/Rfree (%)
20.6/24.8
B factor for protein (Å)
33.8
B factor for ligand
43.5
B factor for ions
31.4
B factor for water
41.8
Rmsd bond lengths (Å)
0.019
Rmsd bond angles ( )
1.585
Number of protein atoms
1,089
Number of ligand atoms
178
Number of ions
4
Number of water atoms
66
Ramachandran plot (% residues)
Most favored
94.2
Additionally allowed
5.8
Disallowed
0.0
Values in parentheses refer to the highest-resolution shell.
(Figure 4). Among the ordered side chains, this analysis
identified 36 residues in the N domain (residues 3–74)
and 34 in the C domain (residues 81–147) that met this criterion. As shown in Figures 4A and 4B, contacts with the
CaM N domain were well distributed across the length of
NR1C1p: 17 with the N-terminal half (residues 875–885;
gray) and 19 with the C-terminal half (residues 886–896;
black). In contrast, contacts with the CaM C domain
were skewed: 27 with the N-terminal half of NR1C1p
and only 7 with the C-terminal half (Figures 4B and 4C).
While CSU analysis showed that most NR1C1p residues contacted a single CaM domain, side chains of
NR1C1p K875, K876, T879, and L887 contacted 2 or
more residues in each domain. The 7 contacts of K875
are shown in purple and are underlined in Figure 5A.
NR1C1p F880 contacted the highest number of residues
Structure
main (i.e., FLMMN and FLMMC) appeared to adopt nearly
identical spatial conformations (Figure 5C). The electron
density overlap of side chains of FLMMC residues and
NR1C1p F880 is shown in Figure 5D. The perpendicular
orientation of NR1C1p F880 relative to CaM F92 allows
for a favorable p-p interaction between the two aromatic
rings (Singh and Thornton, 1992).
Figure 3. Solvent Accessibility of S890
(A) Surface (CaM) and stick (NR1C1p) diagram of 2HQW colored as
a gradient from blue (buried) to red (exposed) according to % SASA
values: S890, 41%; R893, 89%; R894, 72%; R895, 92%; and S896,
87%.
(B) Ball-and-stick diagram of S890 and CaM N-domain residues (M36,
M51, and Q41).
The figure was made with MacPymol.
(7) within a single domain of CaM (Figure 4B). F880 was
within 4.5 Å of F92, I100, L105, M124, A128, F141, and
M144 (red letters in Figure 5A). All have hydrophobic
side chains located in the C domain of CaM. The peptide
residue making the second highest number of contacts
within a single domain of CaM was T886. Its partners in
CaM were F19, L32, M36, M51, and M72, all hydrophobic
side chains in the N domain (blue letters in Figure 5A). In
the common parlance of CaM-peptide interactions, F880
and T886 qualify as peptide anchors in the hydrophobic
clefts of the C and N domains, respectively; however,
the anchors are usually both hydrophobic. Alignment of
the CaM sequence by its calcium-binding sites (Figure 5A)
illustrated that a tetrad of the CaM residues contacting
F880 and T886 residues were a set of identical side chains
(FLMM) in corresponding positions: F19/F92, L32/L105,
M51/M124, and M71/M144 (boxed, Figure 5A).
The domains of CaM in complex with NR1C1p were
aligned by minimizing the distance between the Ca atoms
of the FLMM tetrad residues in each domain; their backbone structures were closely aligned (Figure 5B). The
rmsd for the Ca atoms of the FLMM tetrad in each domain
was 0.208 Å, and this value was 0.573 Å for a comparison
of the whole domain. Thus, the FLMM residues in each do-
Identifying CaM Residues Commonly Used
for Target Interactions
There is only one other structure of CaM bound to a 1-7
CaMBD motif, but there are numerous compact, ellipsoidal CaM-target structures. To explore how the CaMNR1C1p interface related to those complexes, we used
CSU to conduct a statistical analysis of CaM residues
contacting targets in 16 other compact (Ca2+)4-CaM-target structures (12 CaM-peptide, 4 CaM-drug complexes;
listed in Experimental Procedures). In the set of 17 structures analyzed, 3 CaM residues (F92, M124, and M144)
contacted every target; in all but one structure, L105
also contacted the target (Figure 6A). Thus, FLMMC consistently serves as the vertices of the C-domain hydrophobic pocket in these structures. An overlay of domains
aligned according to the Ca atoms of the FLMMC tetrad
shows that the positions of these FLMM residues in all
structures is in Figure 6B.
A corresponding analysis of the N domain showed that
although residues in the FLMMN tetrad were contacted in
at least 12 of the analyzed structures, these residues were
not the 4 residues contacted most frequently (Figure 6C).
Instead, E11 was the only residue found to be within 4.5 Å
of the target peptide or drug in all 17 of the structures
that were examined. However, 2 of those 17 structures
(1CTR.pdb and 1A29.pdb) were CaM-drug complexes in
which the pocket of the N domain was vacant. Thus,
E11 interacted with target molecules bound exclusively
in the hydrophobic cleft of the C domain. The second
most commonly used N-domain residue, A15, also contacts the target associated with the C domain of CaM. In
the 15 structures that had the hydrophobic cleft of the N
domain occupied, F19 contacted the target in all of
them, as did E14. However, the frequency of use of other
FLMMN residues (L32, M51, and M71) was lower and was
dispersed among other hydrophobic N-domain residues
(L18, M72, M36, F68, and L39) that contacted the target
in as many or more structures. An overlay of 17 N domains
aligned according to Ca atoms of the FLMMN tetrad
(Figure 6D) shows that the pocket formed is very similar
in all structures.
A comparison of side chain orientations of each FLMM
residue in these 17 structures is shown in Figure 6E. Each
FLMM residue was aligned with the corresponding residue in 2HQW; rmsds ranged from 0.2 to >1.6 Å. All
FLMM residues, except F92, had side chain orientations
that deviated by <1.0 Å in most structures; the smallest
deviations were observed for M residues. For residues
F19, L32, and F92, deviations ranged from 1.2 to 1.4 Å.
In these, F19 and F92 were rotated by 90 relative to
the orientation observed in 2HQW (Figures 6F and 6H),
Structure
Figure 4. Distribution of CaM N- and C-Domain Contacts in the CaM-NR1C1p Complex
(A) N-domain residues % 4.5 Å of NR1C1p shown as sticks; 17 contacts were made with NR1 residues 875–885 (gray), and 19 contacts were made
with residues 885–896 (black).
(B) Sequence map of CaM residues % 4.5 Å of NR1C1p. Residues in NR1C1p that make the highest number of contacts exclusively with the C domain
(F880) and the N domain (T886) are boxed; the ER retention signal is underlined.
(C) C-domain residues % 4.5 Å of NR1C1p shown as sticks; 27 contacts were made with residues 875–885, and 7 contacts were made with residues
885–896. Ca2+ ions and binding sites (yellow in [A] and [C]) are designated I, II, III, and IV.
Structure
Figure 5. Comparison of FLMM Tetrads
in N and C Domains of CaM
(A) Sequence alignment of the CaM N domain
(1–75) and C domain (76–148). Blue boxes
highlight F19, L32, M51, and M71; red boxes
indicate F92, L105, M124, and M144. Yellow
boxes indicate calcium-binding sites. Residues contacting K875 are purple and underlined.
(B) The CaM N domain (blue; residues 8–73)
and C domain (red; residues 81–146) aligned
according to Ca atoms (green) of their FLMM
tetrad residues. Ca2+ ions and binding sites
are yellow.
(C) Comparison of FLMM residue side chains
(sticks) after alignment of Ca atoms (green
spheres).
(D) Electron density of FLMMC and F880 shown
at a contour level of 1.0s.
and L varied most at the Cg and Cd atoms (see Figure 6G;
Table S1, see the Supplemental Data available with this
article online).
Identifying Target Residues that Contact
the FLMM Tetrads of CaM
An analysis of the chemical characteristics of the target
residues that contact FLMMN and FLMMC revealed that
these tetrads were not used identically by the targets.
Figure 7A shows sequences of the peptide in 13 compact
CaM-target complexes. These were aligned according to
the peptide residue (red box) that contacted the highest
number of FLMMC residues. In 11 of 13 complexes, this
residue also made the highest number of contacts with
all C-domain residues of CaM (Figure 8, red bars). In two
cases, there were 2 residues (Y1627 and F1628 of the
Cav1.2 channel in 2BE6, and W3620 and L3623 of
hRyR1 in 2BCX) that each contacted 3 FLMMC residues;
F1628 in 2BE6 and W3620 in 2BCX had the highest number of contacts with all C-domain residues of CaM. In 12 of
these 13 structures, the residue that contacted the highest
number of FLMMC residues had a large aromatic moiety
(7 F, 5 W, 1 Y) and two (CaMKII [1CDM] and hRyR1
[2BCX]) had a leucine residue in the cavity defined by
the FLMMC tetrad. A structural alignment of these complexes according to the Ca atoms of the FLMMC (Figure 7B) revealed that the orientations of the residue contacting the majority of these FLMMC residues in all 13
structures were well conserved.
The residue in each peptide that contacted the highest
number of the FLMMN residues is boxed in blue in Figures
7A and 8. In the case of CaM bound to CaMKIIa (1CDM),
smMLCK (2BBM), and hRyR1 (2BCX), more than one residue contacted an equal number of FLMMN residues. In
each structure, the residue that contacted the majority
of FLMMN residues was also the one that had the highest
number of contacts with all N-domain residues of CaM,
with the exception of the CaM-CaMKI structure (1MXE),
in which R317 contacted one more residue than M316.
However, unlike contacts in the C domain, there were
other residues in the target that had the same number of
contacts as these. For example, in the CaM-CaMKIIa
structure (1CDM), R297, G301, L304, T305, and A309 all
contacted the same number of N-domain residues (4) in
CaM; however, of these, only T305 and A309 contacted
the majority of the FLMMN residues.
A large variation was observed in the size, chemical
characteristics, and spacing of the residues that
Structure
Figure 6. Statistical Analysis of CaMTarget Interfaces
(A–D) Histograms showing residues in the (A) C
domain (residues 84–146) or (C) N domain (residues 5–72) of CaM % 4.5 Å from a bound peptide or drug in more than 11 of 17 compact
CaM-target structures. Bars for residues in
the FLMM tetrads are black; others are gray.
Only residues defined in all 17 structures
were analyzed. Alignment of 13 CaM-peptide
complexes by the Ca atoms of their (B) FLMMC
(red) or (D) FLMMN (blue) residues; their side
chains are shown as sticks. The domain surface of 2HQW is colored pink (C domain) or
light blue (N domain).
(E) A histogram of rmsds for residue side
chains in FLMMN and FLMMC in 16 CaM-target
structures compared to the corresponding residue in 2HQW (see Table S3).
(F–H) Bins represent increments of 0.2 Å. Comparison of side chain orientations of representative FLMM residues with high rmsds from
2HQW (black): (F) F19 in 2BCX (blue), 1SY9 (orange), and 1MXE (green); (G) L32 in 2BCX
(blue), 2BE6 (red), 1CKK (orange), and 1CDL
(green); and (H) F92 in 1NIW (orange) and
1MXE (green).
contacted the majority of the FLMMN residues relative to
the primary anchor residue at the reference position of
‘‘1’’ (Yap et al., 2000). In 11 of the sequences, the residue
contacting FLMMN was a hydrophobic amino acid, but its
position varied from 10, 11, 14, 16, or 17. The structures of
CaM-NR1C1p (2HQW), CaM-MARCKS (1IWQ), and CaMCaMKIIa (1CDM) were unusual in that the residue contacting the majority of the FLMMN residues was polar (Ser or
Thr) and at position 7. A structural alignment of these complexes according to the Ca atoms of the FLMMN
(Figure 7C) revealed a much larger variation in the position
of the target residue in the FLMMN cavity than was observed for the FLMMC cavity. In some structures (i.e.,
complexes with peptides from MARCKS, CaMKIIa, NR1
C1, eNOS, and Myosin VI), the FLMMN cavity was empty
or only partially occupied, as illustrated in Figure 7D for
the CaM-NR1C1p structure. The C-domain primary contact residue of the peptide in this structure (F880) was observed to fill the FLMMC cavity of CaM. However, T886,
which contacts the majority (3) of the FLMMN residues
and makes the highest number of contacts with the N domain of CaM, appears to contact only the rim of the cavity
defined by the FLMMN tetrad (Figure 7D).
To explore the general availability of the FLMM tetrads
to binding of a hydrophobic moiety that is not restricted
by the orientation and chemical linkage of residues in a
target peptide, this analysis of the FLMM tetrads was
focused to include those in four CaM-drug compact
complexes. Three of these have TFP (Trifluoperazine;
10-[3-(4-methyl-piperazin-1-yl)-propyl]-2-trifluoromethyl-10phenothiazine) bound in 3 CaM:drug ratios (1:1, 1:2, 1:4),
and the fourth has DPD (N-[3,3,-diphenylpropyl]-N0 -[1-R(3,4-BIS-butoxylphenyl)-ethyl]-propylenediamine) bound
in a 1:2 ratio of CaM:drug (1QIV.pdb). In the structures
with a 1:1 ratio of CaM:TFP (1CTR.pdb) and with a 1:2 ratio
(1A29.pdb), only the FLMMC tetrad was occupied, and in
the structure with a 1:4 ratio of CaM:TFP (1LIN.pdb)
both tetrads were occupied by TFP. A structural alignment
of the two domains of 1LIN.pdb according to the positions
of the Ca atoms of the FLMM tetrads illustrated that TFP
Structure
Figure 7. Distribution and Orientation of Target Residues Contacting CaM in Compact CaM-Peptide Complexes
(A) Sequences of 13 peptides aligned by the residue (red box) that contacted the majority of FLMMC residues; the peptide residue that contacted the
majority of FLMMN residues is boxed in blue. The numbers above the sequences denote the spacing between these 2 residues. Ten peptides bind to
CaM in an antiparallel orientation; sequences are shown by using the standard convention (the N-terminal residue is leftmost). Three peptides noted
by an asterisk (*) and listed last bind to CaM in a parallel orientation; their sequences are shown in reverse.
(B and C) Alignment of 13 CaM-peptide complexes by the Ca atoms of their FLMM residues. These residues in 2HQW are shown as red spheres; the
side chains of the primary contact residue of the target are shown as black sticks. The domain surface of 2HQW is colored pink (C domain) or light blue
(N domain).
(D) FLMM pocket occupancy in 2HQW. FLMMC (red) and FLMMN (blue) residues of CaM, as well as F880 (black), T886 (black), and F891 (gray) of
NR1C1p in 2HQW are shown as sticks.
(E) Drug occupancy of the CaM domains. Alignment of the N domain (blue, residues 8–73) and C domain (red, residues 81–146) of CaM in a 1:2
DPD:CaM (1QIV.pdb) by the Ca atoms of their FLMM residues. DPD bound to the CaM N domain is black, and DPD bound to the C domain is green;
the transparency of domains was 0.5.
was capable of binding both FLMM cavities in the same
orientation, and that the two domains of CaM bound to
TFP have similar structures (rmsd of 0.492 Å). This similarity between domains was also observed in the CaM-DPD
structure, in which both FLMM tetrads were occupied with
the same moiety of a DPD molecule in the same orientation (Figure 7E). The N and the C domains of CaM in this
structure align nearly as well as the domains of CaM
when bound to NR1C1p (Figure 5B) (rmsd of 0.652 Å
in the DPD complex versus 0.573 Å in the NR1C1p
complex).
DISCUSSION
The C1 region of the NR1 subunit of the NMDA receptor
has been shown to regulate receptor trafficking and
decrease PKC-induced receptor potentiation. The structure of (Ca2+)4-CaM bound to NR1C1p (2HQW) presented
Structure
Figure 8. Interface Analysis of 13 CaM-Peptide Complexes
Residues in the N domain (gray) and C domain (black) of CaM within 4.5 Å of a peptide residue determined with CSU. Red indicates the peptide residue contacting the highest number of C-domain residues; blue indicates the peptide residue contacting the highest number of N-domain residues.
here demonstrates that C1 is a 1-7 motif, and it indicates
which residues contribute to the interface between
CaM and NR1C1. It provides a foundation for the interpre-
tation of NR1C1 mutants that would disrupt association
with CaM and serve to test the role of CaM in NR1
trafficking.
Structure
Physiological Significance of the
CaM-NR1C1p Complex
NR1 subunits are expressed in an 10-fold excess over
NR2 subunits in the cell; however, only 40%–50% of these
NR1 subunits reach the cell surface in cultured hippocampal neurons (Okabe et al., 1999). There are eight splice
variants of NR1, and the major variant in the brain contains
the C0-C1-C2 regions (Mori and Mishina, 1995); however,
this variant has the lowest fraction of cell-surface expression (Okabe et al., 1999). ER retention of the NR1 subunit
is mediated by a sequence of three contiguous R residues
on the C1 region and is controlled by the phosphorylation
of S896 after the ER retention signal (Scott et al., 2001).
The structure of (Ca2+)4-CaM bound to NR1C1p showed
that the ER retention signal and S896 were not occluded
by CaM. While this structure alone cannot rule out the possibility of a direct role of CaM in ER retention, it is more
likely that CaM might serve as an indirect modulator by
interacting with a kinase or other protein that has a role
in ER retention.
Studies of NR1 splice variants containing the C1 region
showed that this region was necessary and sufficient for
the formation of discrete subcellular receptor clusters
that are associated with the plasma membrane when expressed in fibroblast cells (Ehlers et al., 1995). Phosphorylation of S890 within the C1 region by PKC disrupts the
receptor-enriched clusters, resulting in an even distribution of the NR1 subunit (Tingley et al., 1997). This residue
had only 41% solvent accessibility when in complex with
(Ca2+)4-CaM, and its hydroxyl group was within 4 Å from
the sulfur groups of M36 and M51 of CaM. We have shown
that (Ca2+)4-CaM binds NR1C1p with a high affinity (KD =
2 nM). Ehlers and coworkers (Ehlers et al., 1996b) reported
comparable affinities for NR1C1p and a larger fusion protein of NR1C1p (150 kDa; KD = 4 nM). Together, these
data suggest that CaM can protect this region from other
proteins in the presence of calcium.
Common Features of Hydrophobic Cavities
in CaM Domains
Early studies of CaM-peptide interactions demonstrated
the importance of hydrophobic residues found on both
the target and CaM, particularly the role of tryptophan in
a peptide and the methionine ‘‘puddles’’ of the CaM hydrophobic clefts (O’Neil and DeGrado, 1990). Subsequent
structural comparisons have concluded that large hydrophobic residues of the target bind the hydrophobic
pockets of CaM, and that orientation of binding is determined by the electrostatic characteristics of the domains
of CaM (cf. Bhattacharya et al., 2004; Hoeflich and Ikura,
2002; Ishida and Vogel, 2006; Vetter and Leclerc, 2003).
The roles of individual Met residues on CaM in enzyme activation have been investigated via mutagenesis (Chin
et al., 1997). Dynamics simulations further support the thesis that a set of conserved methionines in each domain of
CaM are flexible and allow for accommodation of variable
peptide residues (Fiorin et al., 2006). A structural overlay of
the domains of CaM in seven wrap-around CaM-target
structures available in the PDB (Ishida and Vogel, 2006)
highlighted a group of 7 hydrophobic residues (4 M, 2 F,
1 L) in each domain of CaM that surround the hydrophobic
pockets in these structures. On the basis of a comprehensive statistical analysis of the contact distances in 17
wrap-around CaM-target structures, we have identified
the FLMMC tetrad as prongs that hold a hydrophobic residue in all canonical CaM-target complexes studied here,
whereas the corresponding FLMMN tetrad is not as well
contacted.
Conservation of FLMM Tetrads in CaM Sequences
Given the prevalence of the FLMM tetrads in the targetbinding pockets of CaM, it was expected that these residues would be conserved in the sequence of CaM across
species. Comparison of 102 CaM sequences (Figure 9) revealed that 4 of the 8 FLMM residues (F19, L32, F92, and
M124) were completely conserved, and 2 (M51, L105)
were 99.98% conserved: position 51 was M in all but
two sequences (Calm_Yeast and Calm_KLULA have
L), and position 105 was L in all but two sequences
(Calm2_Pethy has V and Calm_Mousc has W; see Table
S2 for complete analysis and accession numbers). In the
102 CaM sequences analyzed, this was the only occurrence of W at any position. It is possible that a sequencing
ambiguity could account for the two substitutions
observed at this position. Codons for W (UGG) and V
(GUG) differ by a single base from the sequence for L
(UUG), raising the possibility that L105 is also completely
conserved.
The hydrophobicity of the remaining 2 FLMM tetrad residues (M71, M144) corresponding to the fourth position of
the tetrad in both domains of CaM (Figure 5A) was highly
conserved among the 102 sequences compared; however, these 2 residues showed higher sequence variation
than the other 6 FLMM residues. Position 71 was M in
62 sequences and L in the other 40. Position 144 was M
in 67 sequences, V in 23, L in 10, and I in 2. The smaller
size of the variant side chains may allow the pocket to
accommodate larger anchor residues.
Conservation of FLMM Tetrads
A search of the SWISS-PROT knowledgebase and the
Protein Data Bank (PDB) for all proteins with identical
spacing of the primary sequence for FLMM (e.g., F(x12)-L-(x18)-M-(x19)-M) with PROSITE (Sigrist et al.,
2002) on the ExPASy Proteomics Server (http://ca.
expasy.org) (Gasteiger et al., 2003) identified 361 nonredundant sequences in these databases (i.e., for sequences found both in the PDB and the SWISS-PROT
knowledgebase, the PDB sequence was omitted; Table
S3). Of these, 132 were sequences of CaM with 2 listings
per accession number, 1 for each domain. There were six
structures available in the PDB of non-CaM sequences
that contained an identical FLMM primary sequence:
CaM-like protein 3 (1GGZ-A), the C domain of CaM-like
protein 5 (2B1U-A), the N domain of centrin (caltractin;
2AMI-A), cytochrome P450 (2IJ5-A), aspartyl-rTNA synthase (1C0A-A), and the M chain of the photosynthetic reaction center (1AIJ-M). Alignment of Ca atoms of the
Structure
Figure 9. Comparison of CaM Sequences
from 102 Eukaryotes
(A) The number of species different from human CaM versus residue position; residues
1–75 (blue), 76–79 (gray), 80–148 (red).
(B) FLMM tetrad residues compared by BLAST
(Altschul et al., 1997) on ExPASy server (Gasteiger et al., 2003) (http://ca.expasy.org) searching the SWISS-PROT knowledgebase (see
Table S1 for accession numbers, molecular
identifications, and sequence substitutions
for each species). Conservation of residues of
FLMMN and FLMMC is shown in black bars;
in gray are substitutions for M51 (L), M71 (L),
L105 (V, W), and M144 (V, I, L).
FLMM residues in these structures with Ca atoms of the
FLMMC residues in 2HQW indicated that only the two
CaM-like proteins and centrin form an FLMM pocket similar to that observed in CaM (see Figure S1). The rmsds for
the remaining three structures ranged from 9.5 to 13.0 Å,
compared to 0.36–2.0 Å observed for the two CaM-like
proteins and centrin. Notably, other EF-hand proteins that
bind calcium, but cannot substitute for CaM in cellular
function (e.g., troponin C, S100B, calbindin, and parvalbumin), lack the FLMM tetrad. An alignment of these proteins
with rat CaM (3CLN) by using ClustalW (Chenna et al.,
2003) indicated that Troponin C (4TNC) has I51 and
L124; S100B (1PSB) has Y19, V51, and F71; calbindin
has Y19, L51, and L71; and parvalbumin (1B8R) has
C19, V32, A51, and F71.
Common Features of Residues that Occupy
the Hydrophobic Pockets of CaM
The diversity among sequences that represent the
CaMBDs of target proteins makes it difficult to classify
them and predict precisely how they will interact with do-
mains of CaM. Common classification schemes rely on
identifying hydrophobic side chain residues presumed to
anchor the peptide to the N and C domains of CaM and
matching the number of residues between the presumed
anchors to a known CaMBD. However, the prediction
that NR1C1p was a 1-12 motif (Yap et al., 2000), with
F880 as the bulky hydrophobic residue at position 1 and
F891 at position 12, did not correspond to the observed
structure of CaM bound to NR1C1p. Although F880 was
found in the C-domain hydrophobic pocket, F891 was not
in or near the corresponding pocket of the N domain. F891
made contacts with only 4 CaM residues (Q41, E84, E87,
A88), 3 of which were in the C domain rather than the N domain, and only one of which was aliphatic. This emphasizes the difficulty in predicting which residues in a CaMBD
will make close contacts with CaM and occupy the pocket
of each domain. On the basis of this structure, we propose
that mutation of NR1 F880 to any residue other than Trp
would decrease the affinity of NR1 for CaM. Mutation of
NR1 K875 (which also contacts 7 CaM residues) is anticipated to affect binding orientation and/or affinity.
Structure
Variable Use of FLMMN and FLMMC by Targets
An alignment of 17 CaM-target complexes according to
the Ca atoms of FLMMN and FLMMC illustrated that the
backbone structures of the N and C domains of CaM
when bound to these targets differed by less than 1.0 Å
(see Figures 2D and 2E). Our analysis showed that despite
the high degree of homology between the structures and
sequences of the N and C domains of CaM, they are not
used in the same way by the 13 canonical CaM-peptide
structures studied here. The cavity formed by the FLMMC
residues was consistently occupied by a large aromatic
moiety, whereas that formed by the FLMMN residues
was occupied by a variety of smaller, hydrophobic or polar
residues. However, this does not seem to be due to
a structural restraint of the N domain in binding a large
aromatic residue because structural alignments of the N
domain and C domain of CaM that have the same
drug molecules in their hydrophobic pockets show that
the drugs can bind both domains in a similar manner (Figure 7E). These findings suggest that a reason for the different usages of the FLMMN and the FLMMC pockets could
be because of the tertiary constraints that the peptides
have, such as the distances between the two potential
anchors or their orientation on the helix (i.e., NR1 F880
and F891 both face toward the C domain).
Summary
We present a crystal structure of (Ca2+)4-CaM in complex
with a peptide corresponding to the NR1 C1 subunit of the
NMDA receptor. The complex displays the canonical
wrap-around CaM-target-binding conformation and illustrates the position of CaM with respect to ER retention
and phosphorylation sites. This study also sheds light on
the structural basis for functional differences between
the highly homologous domains of CaM. The C domain
of CaM provides the major anchoring site for NR1C1p,
as was observed in 16 other compact structures of
(Ca2+)4-CaM bound to a peptide, protein fragment, or
drug. A detailed comparison of 17 CaM-target structures
demonstrated that a cavity defined by a tetrad of residues (FLMM) is spatially conserved in both domains,
but used consistently only by the C domain of CaM.
The energetic consequences for these observed domain-specific differences in CaM-target interactions
have yet to be fully understood. Despite the increasing
number of high-resolution structures of CaM bound to
peptides or drugs, it is not yet possible to reliably predict
the primary CaM-peptide contacts or the precise orientation of the two CaM domains connected by a flexible
linker region.
While results obtained from a crystal structure do not
provide direct evidence for the role of CaM association
with C1 in NMDAR function in the cell, it provides a foundation for mutagenic exploration to test the hypothesis that
CaM contributes to protein-protein interactions that regulate the cellular distribution of NR1 in vivo. A comparison
to similar systems in which the role of CaM is known can
provide meaningful insights. Of the 13 CaM-peptide complexes studied, the CaM-MARCKS complex (1IWQ) had
a CaM-target-binding interface most similar to that in the
CaM-NR1C1p complex. Both targets have a Phe residue
that occupies the FLMMC pocket, and equally distant
from this residue is either a Thr (NR1C1p) or a Ser
(MARCKS) that contacts the majority of FLMMN residues.
MARCKS is a membrane-associated protein that functions in the attachment of the cytoskeleton to the plasma
membrane (Aderem, 1992). Parallels have been drawn between the regulation of MARCKS and that of NMDARs
(Ehlers et al., 1996b). The effector domain of MARCKS
binds actin and CaM, and this binding is mutually exclusive, and is affected by PKC phosphorylation. Association
with CaM inhibits this phosphorylation. PKC phosphorylation of the C1 domain of NR1 prevents membrane-associated NR1 clustering, perhaps by disrupting an interaction
with the cytoskeleton. In addition, CaM association with
NR1 C1 decreases PKC phosphorylation (Ehlers et al.,
1996a). Although studies have yet to demonstrate a direct
association of NR1 C1 with actin, high receptor activity requires binding of a-actinin, an actin-binding protein essential for crosslinking actin filaments, to the C-terminal
portion of the neighboring C0 region of NR1 (residues
850–862, just 13 residues from the N terminus of C1)
(Leonard et al., 2002). Therefore, PKC phosphorylation
and CaM association with the NR1 C1 may regulate
the formation of NR1-enriched clusters in the plasma
membrane.
The comparison of 17 canonical CaM-target complexes
presented here enriches our understanding of domainspecific differences in CaM-peptide recognition. Despite
extraordinary conservation of backbone geometry, the
distribution of contacting residues varies enormously
between the N and C domains.
EXPERIMENTAL PROCEDURES
Materials
NR1C1p (KKKATFRAITSTLASSFKRRRSSK) and fluorescein-labeled
NR1C1p (Fl-NR1C1p) were purchased from Keck Biotechnology Resource Center (New Haven, CT). Recombinant mammalian CaM was
prepared as described previously (Pedigo and Shea, 1995; Sorensen
and Shea, 1998). CaM was depleted of Ca2+ by successive dialyses
at 4 C against 50 mM HEPES, 100 mM KCl, 1 mM MgCl2, and 0.5
mM NTA with 5 mM EGTA and then with 0.05 mM EGTA.
Optimization of Crystallization Conditions
Crystals of the CaM-NR1C1p complex grew at 4 C with a reservoir
solution of 7% PEG6000, 5 mM CaCl2, 100 mM Na-acetate (pH 4.0)
by using the hanging-drop vapor-diffusion technique. Two microliters
of a solution containing 0.9 mM NR1C1p and 0.7 mM (Ca2+)4-CaM
was mixed with an equal volume of the reservoir solution.
Crystals were cryoprotected in mineral oil and flash cooled in liquid
nitrogen.
Diffraction Data Collection and Processing
Crystals diffracted to 1.96 Å resolution. Data were collected at the Advanced Photon Source (APS; Argonne National Laboratory) at beamline 17-ID, Industrial Macromolecular Crystallography Association Collaborative Access Team (IMCA-CAT). Data were processed by using
d*TREK (Pflugrath, 1999) from Rigaku/MSC and REFMAC5 (Murshudov et al., 1997); the starting model was 1IWQ (Yamauchi et al.,
2003). NR1C1p was modeled with O (Jones et al., 1991). Water was
Structure
added by using Arp/Warp (part of CCP4 suite [CCP4, 1994]) and was
verified in O. TLS refinement used groups defined as N domain (chain
A, residues 3–74 and Ca2+ I and II), C domain (chain A, residues 81–146
and Ca2+ III and IV), and peptide (chain B, residues 875–896). Electron
densities for side chains of residues 893–896 of NR1C1p were not well
defined; rotamers that minimized electrostatic clashes for 893–896
were selected for the final model, which was verified with PROCHECK
and accepted for deposit by the PDB. Table 1 summarizes crystallographic statistics.
Fl-NR1C1p in the absence of calcium or 3.33–6.22 nM Fl-NR1C1p in
the presence of 10 mM CaCl2 in 50 mM HEPES, 100 mM KCl, 50 mM
EGTA, 5 mM NTA, 1 mM MgCl2 (pH 7.4) at 22 C were titrated with
CaM stocks ranging from 1.33 to 732 mM. Each titration was replicated
at least three times.
The Fl-NR1C1p-CaM complex had a 1:1 stoichiometry. To estimate
the affinity of CaM for Fl-NR1C1p, data were fit to a one-site binding
isotherm in Equation 1,
Y1 =
Solvent-Accessible Surface Area Calculations
A 1.4 Å probe in WHAT IF (Vriend, 1990) and the VACACC command
were used to calculate % solvent-accessible surface area (SASA)
for each residue relative to its accessibility in an unfolded
peptide (see Write-up and ACCESS links at http://swift.cmbi.kun.nl/
whatif/). The %SASA reported was an average value for the side chain
atoms. The model was colored by using PyMOL script color_b.py
(Queens University, http://adelie.biochem.queensu.ca/rlc/work/
pymol/).
Structure Alignment and Analysis
MacPymol 0.99rc6 (DeLano Scientific, CA) was used to align Ca atoms
of the FLMM residues of domains in a single structure (e.g., 2HQW
[NR1C1p; Figures 5B–5D], 1LIN [TFP, 1:4], and 1QIV [DPD; Figure 7E]) or between multiple structures (see Figures 6 and 7) in which
the four Ca atoms of the FLMM within another structure were aligned
with the corresponding atoms of 2HQW. Rmsd values for domain
alignments within a single structure were calculated for residues 8–
73 (66 atoms) and 81–146 (66 atoms) by using MacPymol. These
were chosen to minimize contributions from highly flexible regions
(e.g., termini and linker) of the complexes.
For CaM-peptide complexes compared to 2HQW, the peptide, resolution, and PDB ID were CaM-dependent protein kinase II (CaMK II,
2 Å; 1CDM), smooth muscle myosin light-chain kinase (smMLCK,
2.2 Å; 1CDL), nematode CaM-dependent protein kinase kinase
(CaMKK, 1.8 Å; 1IQ5), myristoylated alanine-rich C-kinase substrate
(MARCKS, 2 Å; 1IWQ), endothelial nitric oxide synthase (NOS, 2.05 Å;
1NIW), myosin VI motor protein (Myosin VI, 2.4 Å; 2BKH), skeletal muscle myosin light-chain kinase (MLCK, NMR; 2BBM), olfactory CNG
channel (Cng-2, NMR; 1SY9), CaM-dependent protein kinase I
(CaMKI, 1.7 Å; 1MXE), CaM-dependent kinase kinase (CaMKK,
NMR; 1CKK), voltage-dependent L-type calcium channel a-1C subunit
(Ca1.2, 2 Å; 2BE6) and the ryanodine receptor Type 1 (RYR1, 2.0 Å;
2BCX). Four other structures were CaM-drug complexes: CaM:TFP
1:1, 2.45 Å (1CTR); CaM:TFP 1:2, 2.74 Å (1A29); CaM:TFP 1:4, 2 Å
(1LIN); and CaM:DPD 1:2, 2.64 Å (1QIV). Distances between residues
in CaM and a target were calculated by using CSU (Sobolev et al.,
1999). Protons were added to crystal structures by using WHAT IF
(Vriend, 1990) to optimize the hydrogen-bonding network. A separation % 4.5 Å qualified as a contact.
Calculation of a Reference Average Structure
and Root-Mean-Square Deviations
Ca atoms for N-domain residues 5–72 (68 aa) or C-domain residues
81–145 (62 aa) were superimposed with corresponding atoms of
2HQW by using MacPymol. Resulting Ca positions were averaged
with AVEPDB for OS X (Kleywegt et al., 2001) available from http://
xray.bmc.uu.se/usf. The rmsd of the N and C domains of each structure from the average structure was normalized for the number of residues (62 for N, 68 for C). The average rmsd value of all 17 structures
was the average of the individual rmsd values.
CaM Titration of NR1C1p
The fluorescence anisotropy of Fl-NR1C1p (Figure 1B) was monitored
by using a Fluorolog 3 (Jobin Yvon, Horiba) spectrofluorimeter with
a lex of 496 nm, a lem of 520 nm, and bandpasses of 2 nm excitation
and 10 nm emission (Akyol et al., 2004). Averages of three readings
with a 1 s integration time were recorded. Samples of 25 nM
Ka ½CaMfree ;
1 + Ka ½CaMfree (1)
where KA represents the association constant, and [CaMfree] is unbound CaM, calculated from the independent variable (total concentration of CaM, [CaMTotal]) (Figure 1B) according to the quadratic equation in Equation 2,
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
b ± b2 4Ka ð½CaMTotal Þ
½CaMfree =
;
(2)
2Ka
where b = (1 + KA[F1 NR1C1pTotal] KA[CaMTotal]). Under apo
conditions, the affinity for Fl-NR1C1p was weak and [CaMfree] z
[CaMTotal]. However, calcium increased the affinity such that [FlNR1C1pTotal] z KD (1/KA). Under those conditions, [CaMfree] was limiting and estimated iteratively in the nonlinear least-squares analysis as
the difference between [CaMTotal] and [CaMbound] ([Fl NR1C1pTotal] Y1 ). A binding constant estimated this way is highly correlated with the
numerical value of [Fl NR1C1pTotal]; therefore, limiting values of KD
were reported. Experimental variations in endpoints of individual titrations were accounted for by Equation 3,
Signal = fðXÞ = Y½Xlow + Y1 Span;
(3)
where Y1 is the average fractional saturation of Fl-NR1C1p (Equation
2), YX[low] is the anisotropy of Fl-NR1C1p alone, and Span (Yhigh Ylow) accounts for the magnitude and direction of signal change
upon titration. The upper and lower endpoints of titrations in saturating
calcium were well defined; but, in the absence of calcium, the upper
endpoint was not well defined at the final [CaMTotal] tested (52 mM).
In the nonlinear least-squares analysis, Yhigh was fixed to be equivalent
to the anisotropy after addition of calcium to the apo CaM:Fl-NR1C1p
sample.
Supplemental Data
Supplemental Data include a figure illustrating the conservation of the
FLMM tetrads in six non-CaM proteins in the SWISS-PROT knowledgebase and PDB (Figure S1) and three tables showing the sequence
homology of 101 CaM species (Table S1), the occurrences of the
FLMM tetrad in identified protein sequences (Table S2), and rmsd
values for variations in the side chain orientations of individual FLMM
residues relative to those in 2HQW (Table S3) and are available at
http://www.structure.org/cgi/content/full/15/12/1603/DC1/.
ACKNOWLEDGMENTS
This work was supported by AHA 0410078Z (Z.A.A.), National Institutes of Health (NIH) R01 GM57001 (M.A.S.), and NIH R01
NS046450 (J.W.H.). We acknowledge use of the University of Iowa
Protein Crystallography Facility; S. Ramaswamy for resources and
training; and D. Ferraro and E. Brown for advice and data collection
at the Advanced Photon Source (APS). Use of the IMCA-CAT beamline
17-ID at the APS was supported by the Industrial Macromolecular
Crystallography Association through a contract with the Center for
Advanced Radiation Sources at the University of Chicago. Use of the
APS was supported by the U.S. Department of Energy, W-31-109Eng-38.
Structure
Received: November 14, 2006
Revised: October 1, 2007
Accepted: October 5, 2007
Published: December 11, 2007
REFERENCES
Aderem, A. (1992). The MARCKS brothers: a family of protein kinase C
substrates. Cell 71, 713–716.
Akyol, Z., Bartos, J.A., Merrill, M.A., Faga, L.A., Jaren, O.R., Shea,
M.A., and Hell, J.W. (2004). Apo-calmodulin binds with its COOH-terminal domain to the N-methyl-D-aspartate receptor NR1 C0 region. J.
Biol. Chem. 279, 2166–2175.
Altschul, S.F., Madden, T.L., Schaffer, A.A., Zhang, J., Zhang, Z.,
Miller, W., and Lipman, D.J. (1997). Gapped BLAST and PSI-BLAST:
a new generation of protein database search programs. Nucleic Acids
Res. 25, 3389–3402.
Bhattacharya, S., Bunick, C.G., and Chazin, W.J. (2004). Target selectivity in EF-hand calcium binding proteins. Biochim. Biophys. Acta
1742, 69–79.
CCP4 (Collaborate Computational Project, Number 4) (1994). The
CCP4 suite: programs for protein crystallography. Acta Crystallogr.
D Biol. Crystallogr. 50, 760–763.
NR1 subunit modulate Ca2+-dependent inactivation of NMDA receptors. J. Neurosci. 19, 1165–1178.
Leonard, A.S., Bayer, K.U., Merrill, M.A., Lim, I.A., Shea, M.A., Shulman, H., and Hell, J.W. (2002). Regulation of calcium/calmodulin-dependent protein kinase II docking to N-methyl-D-aspartate receptors
by calcium/calmodulin and a-actinin. J. Biol. Chem. 277, 48441–
48448.
Mori, H., and Mishina, M. (1995). Structure and function of the NMDA
receptor channel. Neuropharmacology 34, 1219–1237.
Murshudov, G.N., Vagin, A.A., and Dodson, E.J. (1997). Refinement of
macromolecular structures by the maximum-likelihood method. Acta
Crystallogr. D Biol. Crystallogr. 53, 240–255.
O’Neil, K.T., and DeGrado, W.F. (1990). How calmodulin binds its targets: sequence independent recognition of amphiphilic a-helices.
Trends Biochem. Sci. 15, 59–64.
Okabe, S., Miwa, A., and Okado, H. (1999). Alternative splicing of the
C-terminal domain regulates cell surface expression of the NMDA receptor NR1 subunit. J. Neurosci. 19, 7781–7792.
Pedigo, S., and Shea, M.A. (1995). Quantitative endoproteinase GluC
footprinting of cooperative Ca2+ binding to calmodulin: proteolytic susceptibility of E31 and E87 indicates interdomain interactions. Biochemistry 34, 1179–1196.
Chenna, R., Sugawara, H., Koike, T., Lopez, R., Gibson, T.J., Higgins, D.G., and Thompson, J.D. (2003). Multiple sequence alignment
with the Clustal series of programs. Nucleic Acids Res. 31, 3497–
3500.
Pflugrath, J.W. (1999). The finer things in X-ray diffraction data collection. Acta Crystallogr. D Biol. Crystallogr. 55, 1718–1725.
Chin, D., and Means, A.R. (2000). Calmodulin: a prototypical calcium
sensor. Trends Cell Biol. 10, 322–328.
Scott, D.B., Blanpied, T.A., Swanson, G.T., Zhang, C., and Ehlers,
M.D. (2001). An NMDA receptor ER retention signal regulated by phosphorylation and alternative splicing. J. Neurosci. 21, 3063–3072.
Chin, D., Winkler, K.E., and Means, A.R. (1997). Characterization of
substrate phosphorylation and use of calmodulin mutants to address
implications from the enzyme crystal structure of calmodulin-dependent protein kinase I. J. Biol. Chem. 272, 31235–31240.
Ehlers, M.D., Tingley, W.G., and Huganir, R.L. (1995). Regulated subcellular distribution of the NR1 subunit of the NMDA receptor. Science
269, 1734–1737.
Ehlers, M.D., Mammen, A.L., Lau, L.F., and Huganir, R.L. (1996a).
Synaptic targeting of glutamate receptors. Curr. Opin. Cell Biol. 8,
484–489.
Ehlers, M.D., Zhang, S., Bernhardt, J.P., and Huganir, R.L. (1996b). Inactivation of NMDA receptors by direct interaction of calmodulin with
the NR1 subunit. Cell 84, 745–755.
Fiorin, G., Pastore, A., Carloni, P., and Parrinello, M. (2006). Using
metadynamics to understand the mechanism of calmodulin/target
recognition at atomic detail. Biophys. J. 91, 2768–2777.
Gasteiger, E., Gattiker, A., Hoogland, C., Ivanyi, I., Appel, R.D., and
Bairoch, A. (2003). ExPASy: the proteomics server for in-depth protein
knowledge and analysis. Nucleic Acids Res. 31, 3784–3788.
Hoeflich, K.P., and Ikura, M. (2002). Calmodulin in action: diversity in
target recognition and activation mechanisms. Cell 108, 739–742.
Ishida, H., and Vogel, H.J. (2006). Protein-peptide interaction studies
demonstrate the versatility of calmodulin target protein binding. Protein Pept. Lett. 13, 455–465.
Jones, T.A., Zou, J.Y., Cowan, S.W., and Kjeldgaard, M. (1991). Improved methods for building protein models in electron density maps
and the location of errors in these model. Acta Cryatallogr. A 47,
110–119.
Kleywegt, G.J., Zou, J.Y., Kjeldgaard, M., and Jones, T.A. (2001).
Around O. In International Tables for Crystallography, Volume F.
Crystallography of Biological Macromolecules, M.G. Rossmann, and
E. Arnold, eds. (Dordrecht, The Netherlands: Kluwer Academic Publishers), pp. 353–367.
Krupp, J.J., Vissel, B., Thomas, C.G., Heinemann, S.F., and Westbrook, G.L. (1999). Interactions of calmodulin and a-actinin with the
Saimi, Y., and Kung, C. (2002). Calmodulin as an ion-channel subunit.
Annu. Rev. Physiol. 64, 289–311.
Sigrist, C.J., Cerutti, L., Hulo, N., Gattiker, A., Falquet, L., Pagni, M.,
Bairoch, A., and Bucher, P. (2002). PROSITE: a documented database
using patterns and profiles as motif descriptors. Brief. Bioinform. 3,
265–274.
Singh, J., and Thornton, J.M. (1992). Atlas of Protein Side-Chain Interactions (Oxford: IRL Press).
Sobolev, V., Sorokine, A., Prilusky, J., Abola, E.E., and Edelman, M.
(1999). Automated analysis of interatomic contacts in proteins. Bioinformatics 15, 327–332.
Sorensen, B.R., and Shea, M.A. (1998). Interactions between domains
of apo calmodulin alter calcium binding and stability. Biochemistry 37,
4244–4253.
Standley, S., Roche, K., McCallum, J., Sans, N., and Wenthold, R.J.
(2000). PDZ domain suppression of an ER retention signal in NMDA
receptor NR1 splice variants. Neuron 28, 887–898.
Tingley, W.G., Ehlers, M.D., Kameyama, K., Doherty, C., Ptak, J.B.,
Riley, C.T., and Huganir, R.L. (1997). Characterization of protein kinase
A and protein kinase C phosphorylation of the N-methyl-D-aspartate
receptor NR1 subunit using phosphorylation site-specific antibodies.
J. Biol. Chem. 272, 5157–5166.
Vetter, S.W., and Leclerc, E. (2003). Novel aspects of calmodulin target
recognition and activation. Eur. J. Biochem. 270, 404–414.
Vriend, G. (1990). WHAT IF: a molecular modeling and drug design program. J. Mol. Graph. 8, 52–56.
Wu, X., and Bers, D.M. (2007). Free and bound intracellular calmodulin
measurements in cardiac myocytes. Cell Calcium 41, 353–364.
Yamauchi, E., Nakatsu, T., Matsubara, M., Kato, H., and Taniguchi, H.
(2003). Crystal structure of a MARCKS peptide containing the calmodulin-binding domain complex with Ca2+-calmodulin. Nat. Struct. Biol.
10, 226–231.
Yap, K.L., Kim, J., Truong, K., Sherman, M., Yuan, T., and Ikura,
M. (2000). Calmodulin target database. J. Struct. Funct. Genomics 1,
8–14.
Structure
Zhang, S., Ehlers, M.D., Bernhardt, J.P., Su, C.T., and Huganir, R.L.
(1998). Calmodulin mediates calcium-dependent inactivation of
N-methyl-D-aspartate receptors. Neuron 21, 443–453.
Zheng, X., Zhang, L., Wang, A.P., Bennett, M.V.L., and Zukin, R.S.
(1999). Protein kinase C potentiation of N-methyl-D-aspartate receptor
activity is not mediated by phosphorylation of N-methyl-D-aspartate
receptor subunits. Proc. Natl. Acad. Sci. USA 96, 15262–15267.
Accession Numbers
Atomic coordinates were deposited in the Protein Data Bank under accession code 2HQW.

Ataman Shea 07

Transcription

Similar documents

Share the Thrill

`Matt thought he knew` fake profile activity

Can I use the document camera as a webcam for Skype (and other

Dynamic Range

online catalog

ExSum Willow Pass bay Point.xlsx

inapasl n ialya hy

engine - Custom Chrome Europe

Mikuni Cruise Control

cam connection - unioncrisis.org