Báo cáo khoa học: Irregular dimerization of guanylate cyclase-activating protein 1 mutants causes loss of target activation ppt

Irregular dimerization of guanylate cyclase-activating protein 1
mutants causes loss of target activation
Ji-Young Hwang
1,
*, Ramona Schlesinger
2
and Karl-Wilhelm Koch
1
1
Institut fu
¨
r Biologische Informationsverarbeitung 1 and
2
2, Forschungszentrum Ju
¨
lich, Germany
Guanylate cyclase-activating proteins (GCAPs) are neur-
onal calcium sensors that activate membrane bound
guanylate cyclases (EC 4.6.1.2.) of vertebrate photoreceptor
cells when cytoplasmic C a
2+
decreases during illumination.
GCAPs contain f our EF-hand Ca
2+
-binding motifs, but the
first EF-hand is nonfunctional. It was concluded that for
GCAP-2, the loss of Ca
2+
-binding ability of EF-hand 1
resulted in a region that is crucial for targeting guanylate
cyclase [Ermilov, A.N., Olshevskaya, E.V. & Dizhoor, A.M.

(2001) J. Biol. Chem. 276, 48143–48148]. In this study we
tested the consequences of mutations in EF-hand 1 of
GCAP-1 w ith respect to Ca
2+
binding, C a
2+
-induced con-
formational c hanges and target activation. When the non-
functional first EF-hand in GCAP-1 is replaced by a
functional EF-hand the chimeric mutant CaM–GCAP-1
bound four Ca
2+
and showed similar Ca
2+
-dependent
changes i n t ryptophan fluorescence as the wild -type. CaM–
GCAP-1 neither activated nor interacted with guanylate
cyclase. Size exclusion c hromatography revealed that the
mutant tended to form inactive dimers instead of active
monomers like the wild-type. Critical amino acids in EF-
hand 1 of GCAP-1 are cysteine at position 29 and proline at
position 30, as changing these to glycine was sufficient to
cause l oss of t arget activation w ithout a l oss of Ca
2+
-
induced conformational c hanges. The latter mutation also
promoted dimerization of the protein. Our results show that
EF-hand 1 in wild-type GCAP-1 is critical for providing the
correct conformation for target activation.
Keywords: GCAP; guanylate cyclase; neuronal Ca

2+
sensor;
phototransduction.
Light triggers the activation of a G-protein coupled cascade
in photoreceptor cells that ultimately leads to the hydrolysis
of 3¢,5¢-cyclic GMP (cGMP) [1,2]. Cyclic nucleotide-gated
(CNG) channels in the plasma membrane are kept open by
cGMP in the dark and close upon light-induced cGMP
hydrolysis. CNG channels are permeable for Ca
2+
and
provide the main entrance route for Ca
2+
in photoreceptor
cell outer segments [3]. Extrusion of Ca
2+
is catalysed by a
Na
+
/Ca
2+
,K
+
-exchanger. Thus the cytoplasmic [Ca
2+
]in
the dark is balanced by these t wo transport mechanisms to
around 700 n
M
. Closure of CNG channels prevents Ca

2+
from entering the outer segment; however, by the continous
operation of the e xchanger cytoplasmic [Ca
2+
]decreases
to < 100 n
M
[1–3]. Changes in c ytoplasmic [Ca
2+
]are
detected by Ca
2+
-binding proteins such as calmodulin,
recoverin and the guanylate cyclase-activating proteins
(GCAPs), which in turn regulate their targets in a Ca
2+
-
dependent manner [4–6]. GCAPs serve a key function in
rod and cone cells as they activate two forms of a
membrane-bound rod outer segment guanylate cyclase
(ROS-GC), ROS-GC1 and ROS-GC2 (a lso n amed ret-
GC1, retGC2 and GC-E, GC-F) at low cyto plasmic [Ca
2+
]
and thereby participate in one of several negative feedback
loops that restore the dark state of the cell and help to
adjust the cell’s light sensitivity [6,7].
GCAPs belong to a group of proteins named neuronal
Ca
2+

sensor (NCS) proteins which are mainly expressed in
the nervous system [8,9]. Properties of GCAPs have been
intensively i nvestigated in recent y ears and the main focus
wasonGCAP-1andGCAP-2[4–6].Commonfeaturesof
these proteins are four EF-hand Ca
2+
-binding sites, but
only EF-hands 2,3 and 4 are functional. GCAPs are
myristoylated at the N-terminus, but they do not undergo a
Ca
2+
–myristoyl switch as do other NCS proteins such as
recoverin and hippocalcin [10,11]. However, they change
their conformation in response to b inding of Ca
2+
.
Conformational t ransition in GCAPs has been investigated
by different methods s uch as site-directed mutagenesis, gel
shift assays, tryptophan fluorescence, CD spectroscopy,
limited prot eolysis and thiol reactivity o f cysteines [11–18].
Although GCAP-1 and GCAP-2 share some properties,
they also differ in several aspects. For example, GCAP-1
and GCAP-2 regulate guanylate cyclase with different Ca
2+
Correspondence to K W. Koch, Institut fu
¨
r Biologische Informa-
tionsverarbeitung 1, Forschungszentrum Ju
¨
lich, D-52425 Ju

¨
lich,
Germany. Fax: +49 2461 614216, Tel.: +49 2461 61 3255,
E-mail:
Abbreviations:cGMP,3¢,5¢-cyclic guanosine monophosphate; CNG
channel, cyclic nucleotide-gated channel; GCAP, guanylate cyclase-
activating protein; CaM–GCAP-1 and CPG–GCAP-1, mutants of
GCAP-1; DTT, dithiothreitol; dibromo-BAPTA, 1,2-bis [2-bis(o-
amino-5-bromophenoxy]ethane-N,N,N¢,N¢-tetraacetic acid; GC,
guanylate cyclase; NCS, neuronal Ca
2+
sensor proteins; ROS, rod
outer segments; ROS-GC, rod outer segment guanylate cyclase;
Rh, rhodopsin; SEC, size exclusion chromatography.
Enzymes: guanylate cyclases (EC 4.6.1.2.).
*Present address: Genetic s & Molecular Biology Branch Nationa l
Human Genome Research Institute (NIH), Bethesda,
MD 20892-4442, USA.
(Received 24 June 2004, revised 26 J uly 2004, a ccepted 3 August 2004)
Eur. J. Biochem. 271, 3785–3793 (2004) Ó FEBS 2004 doi:10.1111/j.1432-1033.2004.04320.x
sensitivities and the myristoyl group has a strong impact on
regulatory properties of GCAP-1, but almost no effect on
GCAP-2 [19]. Further, reversible dimerization as a function
of Ca
2+
was demonstrated for GCAP-2 and it was
proposed that dimer formation in the absence of Ca
2+
is
one mechanistic s tep in the activation of guanylate c yclase

[20,21]. In c ontrast, G CAP-1 c an form dimers irrespective of
the Ca
2+
concentration, but only the GCAP-1 monomer is
active, whereas the dimer is not [17].
Sites in G CAP-1 a nd GCAP-2 tha t int eract with ROS-
GC1 h ave been mapped by testing chimeric GCAP
mutants, single point mutations in GCAPs, and by the
use of peptide libraries in competition assays [13,21–25].
Thereby several distant regions were identified that parti-
cipate in the activation or inhibition of ROS-GC1. The
N-terminus seems to have a critical function in both GCAPs
and it was proposed that EF-hand 1 in GCAP-2 has lost the
ability to bind Ca
2+
in order to target to guanylate cyclase
[21]. A similar function was also discussed for EF-hand 1 in
GCAP-1, but was not tested experimentally. Furthermore,
an amino acid sequence comparison of the first (i.e.
nonfunctional) EF-hand in NCS proteins with the func-
tional EF-hand in the prototypical C a
2+
-binding protein
calmodulin unveils differences of hypothetical importance.
In the present report we investigated whether manipulations
in the r egion of EF- hand 1 o f GCAP-1 can change its
Ca
2+
-dependent conformational transitions and its regula-
tory features.

Experimental procedures
Construction of mutants CPG–GCAP-1 and CaM–GCAP-1
To create the mutants of CPG–GCAP-1 the amino acids
cysteine at position 29 and proline a t position 30 in EF-1 of
GCAP-1 were both substituted by glycine. In CaM–GCAP-
1theEF-1regionofGCAP-1wasreplacedwiththeEF-1
region of human calmodulin (data base accession number
J04046). Oligonucleotide-directed mutagenesis was per-
formed by PCR using the cDNA of GCAP-1 as a template
[26]. To generate mutant CPG–GCAP-1, the inner muta-
genesis primer (5¢-GGAACCCTCTGTC ATGAACTTC
TTGTACC-3¢) and the flanking primers (upstream primer
with an NdeI site, indicated by italics: 5 ¢-GCCATATGGG
TAACATTATGGACGGTAAGTCGA-3¢; downstream
primer introducing a BamHI site, indicated by ital-
ics: 5¢-CGGGATCCTAGCCGTCGGCCTCCGCGGC-3¢)
were used in a two-step PCR.
For creating mutant CaM–GCAP-1 two parallel PCR
were performed. One PCR contained fusion primer A
(5¢-GACAAGGATGGAGATGGCACTATCACCACCAA
GGAGTTCCGCCAGTTCTTCGGCC-3¢)andthesame
downstream primer as before whereas the other reaction
contained fusion primer B (5¢-CTTGGTGGTGA
TAGTGCCATCTCCATCCTTGTCGAACTTCTTGTAC
CACTGGTGC-3¢) a nd the upstream primer. In a s econd
step, the two PCR products were interlinked by PCR
by their overlapping r egions (bold s equence in p rimers
A and B). This procedure results in a hybrid gene w here
12 amino acids (residues 26–37) in GCAP were
replaced by 11 amino acids (residues 21–31) of human

calmodulin.
Preparation of ROS
Bovine ROS were prepared according to a standard
protocol using dim red light. This involves a sucrose
density centrifugation in the presence of moderate salt
concentration to minimize loss of cytoplasmic proteins
[19]. R OS were at all t imes stored and handled in the
dark. Rhodopsin (Rh) concentration was determined
spectrophotometrically at 498 nm using a molar extinc-
tion coefficient of 40 000
M
)1
Æcm
)1
. Preparation of
washed ROS membranes a nd guanylate cyclase activity
measurements were performed as described previously
[18,19,26].
Expression of GCAPs
GCAPs were e xpressed i n Escherichia coli as described
[18,19]. Plasmid pET-11a/GCAP was construced by sub-
cloning DNA fragments with the GCAP1 gene with NdeI/
BamHI into the vector pET-11a. Bacterial strains Epicurian
ColiÒ BL21-CodonPlus
TM
(DE3)-RIL (Stratagene) were
used for o verexpression of GCAPs. Cells were cultured in
dYT medium (16 g bacto-tryptone, 10 g bacto-yeast extract
and 5 g NaCl p er 1 L) at 37 °C. Expression of GCAPs w as
induced by 1 m

M
isopropyl thio-b-
D
-galactoside at 37 °C.
After 4 h cells were harvested by centrifugation for 20 min
at 10 000 g at 4 °C and then resuspend ed in 20 m
M
Tris
pH 7.4, 150 m
M
NaCl, 2 m
M
EGTA or 2 m
M
CaCl
2
,2m
M
dithiothreitol (DTT) and proteinase inhibitor cocktail m ix
(Boehringer Mannheim).
Purification of GCAP-1 wild-type and mutants
The overexpressed GCAPs were released from the BL21-
CodonPlus
TM
(DE3)-RIL cells by passing through a French
press (SLM Aminco; American Instrument Exchange, Inc.,
Haverhill, MA). Subsequent purification of GCAPs by size
exclusion and anion exchange chromatography using an
A
¨

KTA FPLC system (Pharmacia Biotech) was exactly as
described [11,22]. Purified G CAPs were dialys ed against
50 m
M
ammonium bicarbonate buffer. Aliquots of 1 mg
were lyophilized by a Speedvac concentrator and then
stored at )80 °C until further use. SDS/PAGE was
performed as described [11,22].
45
Ca
2+
-binding assay
Binding of Ca
2+
to GCAPs was performed as previously
described for binding of Ca
2+
to recoverin [ 27,28]. Briefly,
all buffers and p rotein solutions used in Ca
2+
-titration
experiments were p assed over a Chelex column (Bio-Rad)
to remove residual a mounts of Ca
2+
. Chelex resin was
prepared and e quilibrated according to the manufacturer’s
instructions. Samples (0 .3–0.5 m L) that contained 5 0 or
100 l
M
protein were dissolved in 20 m

M
HEPES pH 7.5,
100 m
M
NaCl and 1 m
M
DTT and were transferred to
Centricon 10 devices (Amicon). A solution of 10 lLof
0.2 m
M
45
CaCl
2
(0.2–0.3 lCi) solution was added and
centrifuged for 1 min at 3 0 °C (16 000 g) in a tabletop
centrifuge (Beckman model TJ-6). After centrifugation,
the r adioactivity in 10 lL of the filtrate (free Ca
2+
)and
an equal volume of protein sample (total Ca
2+
)were
3786 J Y. Hwang et al. (Eur. J. Biochem. 271) Ó FEBS 2004
determined by liquid scintillation counting. In the next
steps nonradioactive CaCl
2
was added and the above
centrifugation procedure was repeated. Protein-bound
Ca
2+

vs. free Ca
2+
was determined from the excess
Ca
2+
in the protein sample over that present in the
ultrafiltrate. The data were analysed as follows:
½Ca
2þ

free
¼ðR
f
=R
p
Þ½Ca
2þ

total
where R
f
is the radioactivity in the filtrate, R
p
is radioactivity
in protein sample, Ca
2+
total
and Ca
2+
free

are the total and
free Ca
2+
concentration, respectively.
Bound Ca
2+
thatwasretainedonwild-typeGCAP-1
after Chelex treatment, was determined as follows: GCAP-1
was denatured by heating a t 95 °C in 0.5% SDS for 5 min.
Next, the sample was added to Ca
2+
-free buffer (see above)
that contained 10 l
M
of 1,2-bis [2-bis(o-amino-5-bromo-
phenoxy]ethane-N,N,N¢,N¢-tetraacetic acid, (dibromo-
BAPTA) [29] and the absorption was m easured at 264 nm
against a suitable control. Free [Ca
2+
] was determined from
aCa
2+
/EGTA-calibration curve using dibromo-BAPTA as
indicator.
Size exclusion chromatography
Protein samples were chromatographed on a BioSep–S EC-
S2000 column (Phenomenex, Aschaffenburg, Germany) b y
injection of 2 0 lL of molecular mass standards o r G CAP
solutions. M olecular mass s tandards for gel filtration were
aldolase (Stokes’ radius: 48.1 A

˚
, molecular mass 158 kDa) ,
BSA (Stokes’ radius: 35.5 A
˚
, 66 kDa), ovalbumin (Stokes’
radius 30.5 A
˚
, 43 kDa), chymotrypsinogen (Stokes’ radius:
20.9 A
˚
, 25 kDa) and ribonuclease A (Stokes’ radius:
16.4 A
˚
, 13.7 k Da). Standards were dissolved at 1.2–
5mgÆmL
)1
in size exclusion chromatograpy (SEC) buffer
(20 m
M
Tris/HCl pH 7.5 , 50 m
M
KCl, 10 m
M
NaCl,
10 m
M
MgCl
2
,1m
M

DTT and either 0.3 m
M
CaCl
2
or
0.4 m
M
EGTA). Void volume ( V
o
)ofthecolumnwas
determined with dextran blue (2000 kDa). Elution volumes
(V
e
) of s tandards were d etermined by single r uns or in
different combinations. GCAPs were dissolved at a
concentration of 1.8–3.9 m gÆmL
)1
in SEC buffer contain-
ing either 0.3 m
M
CaCl
2
or 0.4 m
M
EGTA. S tokes’ radii of
GCAPs were obtained from a plot according to Laurent &
Killander [30], where (–logK
AV
)
1/2

was plotted vs. the
Stokes’ radius. K
AV
¼ V
e
) V
0
/V
t
) V
0
;whereV
t
is the
total column volume. Apparent molecular m asses w ere
obtained from a plot of logM
r
vs. K
AV
. For analysis we
assumed a similar compact structure for GCAP-1 as that
reported for GCAP-2 [31] and therefore we based the
determination of Stokes’ radii on a calibration curve
obtained with protein standards o f spherical shape.
Fluorescence spectroscopy
Fluorescence experiments were performed with a Shimadzu
RS-1501 fluorescence spectrometer. GCAPs were dissolved
in 50 m
M
HEPES/K OH pH 7.4, 100 m

M
NaCl and 1 m
M
DTT at a concentration of 2 l
M
. The excitation wavelength
was 280 nm. The trypto phan fluorescence emission spec-
trum was recorded between 290 and 450 nm. The free
[Ca
2+
] in the buffer was adjusted by a Ca
2+
/EGTA buffer
system as described a nd varied between 10
)3
and 10
)9
M
free Ca
2+
[19].
Limited proteolysis of GCAPs
GCAPs ( 0.2–1 mgÆmL
)1
) w ere incubated with trypsin at a
ratio of 300 : 1 in a total volume of 50 or 60 lL. CaCl
2
or
EGTA were added to yield a final concentration of 0.1 m
M

.
Incubation was performed at certain time intervals at 30 °C,
proteolysis was stopped by removing 10 lLfromthe
incubation mixture and adding 1 m
M
phenylmethylsulfonyl
fluoride a nd 1 m
M
benzamidine. Results were a nalysed by
SDS/PAGE and staining with Coomassie blue.
Results
Ca
2+
-binding to GCAP-1 mutants
The first EF-han d in all known NCS proteins is nonfunc-
tional. It is mainly distorted from an effective Ca
2+
-binding
loop by a Cys and Pro in the position between y and z of the
EF-hand loop region (Fig. 1 ). A NMR derived three-
dimensional structure of Ca
2+
-bound GCAP-2 demon-
strates that the bulky sulfhydryl group at the third position
in the loop of EF-hand 1 prevents entry of Ca
2+
[31]. In
addition the Pro at the fourth position in the loop leads to a
distortion from a favourable Ca
2+

-binding geometry. A
sequence alignment of the first EF-hand in GCAP-1,
GCAP-2 and calmodulin unveils the critical difference a t
these two particular positions, c almodulin has a Gly (G)
insteadofCysPro(CP).MutantsofGCAP-1were
constructed in which either the first EF-hand was r eplaced
by the first EF-hand of calmodulin (CaM–GCAP-1) or
the CP in position 29 a nd 30 was changed to a G (CPG–
GCAP-1). Mutants were heterologously expressed in E. coli
and purified by chromatographic procedures. It is a
reasonable assumption that the mutant CaM–GCAP-1
could bind four Ca
2+
, as it contains four intact EF-hands.
We tested this hypothesis by a direct
45
Ca
2+
-binding assay.
In fact, at saturating f ree [Ca
2+
] (> 180 l
M
), we measured
a stoichiometry of 4.0 ± 0.5 Ca
2+
bound per CaM–
GCAP-1. The mutant CPG–GCAP-1 showed a lower
stoichiometry of 3.3 ± 0.5 Ca
2+

bound. Surprisingly, we
Fig. 1. Sequence alignment of the loop region in the first (nonfunctional)
EF-hand of GCAP-1 and GCAP-2 in comparison to the first EF-hand i n
calmodulin. Data base accession num bers for sequences are; bovine
GCAP-1, P46046; bovine GCAP-2, U 32856; bovine calmodulin,
MCBO. The am ino acid sequence Trp21–Met26 of G CAP-1 that is
located N-terminally to the loop regionhasbeensuggestedtoforman
interaction domain [ 24] a nd is shown in comparison to th e corres-
ponding sequences of GC AP-2 and calmodulin. Positions of oxygen -
containing amino acid side chains that participate in complexing Ca
2+
in a canonical EF-hand are marked x, y, z, -y, -x and -z. The arrow
highlights the position CysPro in neuronal Ca
2+
sensor proteins, that
is not present in calmodulin.
Ó FEBS 2004 Irregular dimerization of GCAP-1 mutants (Eur. J. Biochem. 271) 3787
measured for wild-type GCAP-1 a pproximately two Ca
2+
(1.8 ± 0.3) at saturating free [Ca
2+
] and not as expected
three bound Ca
2+
. However, denaturing Chelex treated
wild-type GCAP-1 led to the additional release of
1.2 ± 0.75 Ca
2+
per GCAP-1 molecule indicating the
presence of a high affinity Ca

2+
-binding site.
CaM–GCAP-1 and CPG–GCAP-1 sense changes in [Ca
2+
]
Both mutants exhibited a decrease in electrophoretic
mobility when Ca
2+
in the sample buffer was complexed
by EGTA (Fig. 2). This electrophoretic mobility shift of
% 4.5 kDa was nearly identical to that observed with wild-
type GCAP-1. It indicated that the GCAP-1 mutants
underwent a similar Ca
2+
-induced conformational change
as wild-type GCAP-1.
This conclusion was further supported by tryptophan
fluorescence studies. wild-type GCAP-1 contains three
tryptophan residues that caused fluorescence emission
spectra as shown in Fig. 3 (upper part, left panel; see a lso
[16]). Varying the free [Ca
2+
]from10
)3
to 10
)9
M
caused an
increase in maximum fluorescence intensity. When the
fluorescence intensity is plotted as a function of [Ca

2+
]
the resulting curve saturated below 10
)7
M
Ca
2+
and the
change in intensity was half-maximal at 300 n
M
(Fig. 3 A,
right panel), which i s consistent with th e known activation
profile of guanylate cyclase by GCAP-1 in the submicro-
molar range. The CaM–GCAP-1 chimera showed a similar
Ca
2+
-dependent change in tryptophan fluorescence inten-
sity (Fig. 3B, left panel), which was half-maximal at 600 n
M
Ca
2+
(Fig. 3 B, right panel). The CPG mutant also showed
an increase in tryptophan fluorescence emission, although
the relative changes in emission intensity were not as high as
those observed with wild-type and CaM–GCAP-1 (Fig . 3 C,
left panel). However, the Ca
2+
titration curve revealed a
similar EC
50

of 350 n
M
for the Ca
2+
-induced change in
emission intensity (Fig. 3C, right panel). W e c oncluded
from the r esults obta ined from t he gel shift assay and the
tryptophan fluorescence spectroscopy study that the Ca
2+
-
sensing properties of the mutants do not differ significantly
from those of wild-type GCAP-1. Therefore, we conclude
that mutants have retained characteristic properties of wild-
type GCAP-1 with respect to Ca
2+
binding.
GCAP-1 mutants have lost activating properties
Both mutants were then tested in guanylate cyclase (GC)
activity assays using washed native ROS membranes that
lacked endogeneous GCAPs (Fig. 4) [11,18,19]. T he con-
centration of CaM–GCAP-1 and CPG–GCAP-1 was 2 l
M
in the p resence of 1 m
M
Ca
2+
(black bars) or 2 m
M
EGTA
(grey bars). A control incubation was performed with wild-

type GCAP-1 or buffer. Basal activity in t he presence of
1m
M
Ca
2+
was a round 2 nmol c GMPÆmin
)1
Æmg
)1
Rh
(Fig. 4 ) and was identical to the GC activity of washed
GCAP-depleted R OS membranes. Increase in GC activity
by wild-type GCAP-1 at low [Ca
2+
] was fivefold, but
neither CaM–GCAP-1 nor CPG–GCAP-1 were able to
stimulate GC activity above the basal level (Fig. 4 ).
The apparent failure of the mutants to activate GC could
either reflect no interaction with the target or could result
from interaction without transition to the activating state. A
simple way t o distinguish betwee n these possibilities is a
competition experiment: guanylate cyclase wass activated
by 2 l
M
wild-type GCAP-1 or wild-type GCAP-2 at low
free [Ca
2+
] and a fivefold excess (10 l
M
)ofCaM–GCAP-1

or CPG –GCAP-1 were added ( grey bars in Fig. 5). When
the concentration of wild-type GCAP-1 was decreased to
0.9 l
M
(at which activation is half-maximal) [11] we
observed no interference b y more than 11-fold excess of
CaM–GCAP-1 or CPG–GCAP-1 (data not shown). Thus,
high concentrations of the mutants did not interfere with
the ac tivation of ROS-GC1 by either GCAP-1 or GCAP-2
indicating that the mutants could not compete with wild-
type GCAPs for the same binding site. However, we cannot
entirely exclude that GC AP-1 mutants associate with a
different region in ROS-GC1 than wild-type GCAP-1,
because the cross-linking experiments gave no conclusive
results.
Mutations in EF-hand 1 of GCAP-1 causes dimerization
What causes the failure of GCAP-1 mutants to interact
properly with GC? An answer came from size exclusion
experiments. Active GCAP-1 exists as a monomer both in
the p resence o f C a
2+
and EGTA. This is different from
GCAP-2 which forms dimers when Ca
2+
is chelated by
EGTA and t hus needs the dimerization step in order to
switch from the inhibitor to activator state. We performed
size exclusion experiments to test wild-type and mutants of
GCAP-1 for irregular shape or dimerization. GCAP-1
eluted on a size exclusion column mainly as a protein with a

Fig. 2. Ca
2+
-dependent electrophoretic mobility shift of GCAPs. Wild-
type GCAP-1 (a and a¢), CPG–GCAP-1 (b and b¢)andCaM–GCAP-1
(c and c¢) were se parated by electrophoresis in the presence of CaCl
2
or EGTA in a 15% polyacrylamide gel. GCAPs exhibited an
electrophoretic mobility shift of % 4.5 kDa, when CaCl
2
was replaced
by EGTA. The low molecular mass (LMW) standard consisted o f
phosphorylase b (97.4 kDa), BSA (66 kDa), ovalbumin (45 kDa),
carbonic anhydrase (29 kDa), soybean trypsin inhibitor (20.1 kDa)
and a-lactalbumin (14.2 kDa). Th e gel was stained with Coomassie
blue.
3788 J Y. Hwang et al. (Eur. J. Biochem. 271) Ó FEBS 2004
Stokes’ radius of 26.5 A
˚
and apparent molecular mass of
36 kDa (Fig. 6A and B, l eft p anels; t hin tra ce). A small
fraction containing < 10% of the total GC AP-1 that was
applied to the column eluted with a Stokes’ radius of 35 A
˚
and apparent molecular mass of 62 kDa. This elution
pattern d id not change when Ca
2+
was replaced b y EGTA
in the column buffer (Fig. 6A and B, right panels; thin
trace). These values are consistent with the main fraction
being a GCAP-1 monomer and the smaller fraction being a

dimer. However, the hydrodynamic properties of the
mutants d iffered significantly f rom t hese wild-type proper-
ties. For instance, the chimeric protein CaM–GCAP-1
eluted mainly as a dimer with a Stokes’ radius of 32.2 A
˚
(Fig. 6 A, right panel; thick trace). In the presence of Ca
2+
only 20% of the protein eluted as a monomer (Stokes’
radius 25.3 A
˚
). When the run w as performed in the presence
of EGTA the main portion of CaM–GCAP-1 remained in
the dimeric form. A second peak appeared on the shoulder
of the dimer peak (Fig. 6B, left panel; thick trace) that was
situated between wild-type monomer and dimer. This peak
would correspond to a protein with a Stokes’ radius of 29 A
˚
and an apparent molecular mass of 45 k Da. It p robably
represented a monomer of CaM–GCAP-1 with a different
shape (see Discussion). CPG–GCAP-1 eluted almost
entirely as a dimer in the presence of Ca
2+
(Fig. 6 B, left
panel; thick trace, Stokes’ radius 33.5 A
˚
), a sma ll fraction
eluted in a peak at V
e
¼ 7 .2 mL r epre senting a h igher
oligomeric or aggregated state of CPG–GCAP-1. A com-

pletely different picture was obtained when CPG–GCAP-1
was chromatographed in t he presence of EGTA (Fig. 6B,
right panel; thick trace). The dimer peak became much
[Ca
2+
] (M)
10
-9
10
-8
10
-7
10
-6
10
-5
10
-4
10
-3
10
-2
Intensity
100
105
110
115
120
125
130

CPG-GCAP-1
300 350 400
0
50
100
150
Fluorescence (arbitrary units)
Wavelength (nm)
[Ca
2+
] (M)
10
-9
10
-8
10
-7
10
-6
10
-5
10
-4
10
-3
10
-2
Intensity
60
65

70
75
80
85
90
95
WT
300 350 400 450
0
50
100
150
Fluorescence (arbitrary units)
Wavelength (nm)
[Ca
2+
] (M)
10
-9
10
-8
10
-7
10
-6
10
-5
10
-4
10

-3
10
-2
Intensity
80
90
100
110
120
130
140
150
CaM-GCAP-1
300 350 400 450
0
50
100
150
Fluorescence (arbitrary units)
Wavelen
g
th (nm)
A
B
C
Fig. 3. Tryptophan fluorescence spectra of
GCAPs at different [Ca
2+
]. In each case 2 l
M

of wild-type or mutant G CAP-1 was ana lysed.
Emission spectra for each GCAP-1 form are
displayed on the left panels and the corres-
ponding maximum relative fluorescence
intensity values at different [Ca
2+
] are shown
on the right panels. (A) Wild-type GCAP-1.
(B) CaM–GCAP-1. (C) CPG–GCAP-1.
Fig. 4. Guanylate cyclase activity in the presence of GCAPs. Washed
ROS membranes were reconstituted w ith 2 l
M
of either wild-type
GCAP-1 (WT) , buffer (RM ), CaM–GCAP-1 o r CPG–GCAP-1.
Guanylate cyclase (GC) activity was measured in the presence of 1 m
M
CaCl
2
(black bars) or 2 m
M
EGTA (grey bars) and is expressed a s
nmol cGMPÆmin
)1
Æmg
)1
Rh.
Ó FEBS 2004 Irregular dimerization of GCAP-1 mutants (Eur. J. Biochem. 271) 3789
smaller a nd the m ajority of C PG–GCAP-1 e luted as a
protein of apparent molecular mass of 45 k Da and is similar
to the behaviour of the CaM–GCAP-1 mutant in the

presence of EGTA (Fig. 6A, right panel).
Limited proteolysis of GCAPs
The Ca
2+
-bound form of wild-type GCAP-1 can be
envisaged as a rather co mpact core s tructure with less
compact parts at the N- and C-termini. It is protected from
limited proteolytic degradation, whereas the Ca
2+
-free form
undergoes more rapid proteolysis by trypsin [14]. We
compared the protein/peptide pattern of wild-type GCAP-1
with that of CaM–GCAP-1 and CPG–GCAP-1 after
limited proteolysis (Fig. 7). As reported previously for
wild-type GCAP-1, the C a
2+
-free form was rapidly d egra-
ded (Fig. 7, upper part) and the Ca
2+
-bound form was
better protected: proteolysis yielded large fragments
between 14 and 20 kDa (b, c, d in th e upper part of
Fig. 7). The relative amounts of proteolytic fragments
varied, but the pattern was similar in different trials (data
not shown). The mutants CaM–GCAP-1 and CPG–
GCAP-1 exhibited stronger protection ) in particular when
the proteolysis was compared with that of t he wild-type i n
the absence of Ca
2+
(Fig. 7 , upper part in comparison to

middle and lower parts). Proteolysis of CaM–GCAP-1 in
the presence of Ca
2+
yielded six main fragments between 14
and 20 kDa (b)k, Fig. 7 ) that were a lso seen in the absence
of Ca
2+
. However, staining intensity of the large fragment b
was lower in the absence of Ca
2+
indicating that progessive
trypsin d igestion of the Ca
2+
-free f orm o f CaM–GCAP-1
had already occurred. A similar observation was made with
CPG–GCAP-1, fragments I–VIII (Fig. 7) were produced in
the presence and absence of Ca
2+
, but proteolysis pro-
ceeded slightly faster in the absence of Ca
2+
.
Discussion
Peptide competition and truncation studies have shown that
the N-terminal 20–25 amino acids of GCAP-1 are import-
ant f or regulation of ROS-GC1 [13,22,25] and a study with
recoverin/GCAP-1 chimera c oncluded that t he sequence
Trp21–Thr27 is required for activation of ROS-GC1 [24].
In addition, other regions in GCAP-1 contribute to the
activation or represent contact s urfaces [13,24]. However,

the role of the nonfunctional EF-hand 1 in GCAP-1
remains unclear, in particular whether it acts as an essential
target region as it does in GCAP-2 [21]. Our s tudy was
focused on the loop region of EF-hand 1 in GCAP-1, where
a CysPro sequence i s supposed to prevent binding of Ca
2+
.
Fig. 5. Competition experiments. G uanylate cyclase in washe d ROS
membranes was in cubated with 2 l
M
of either GCAP-1 or GCAP-2 at
2m
M
EGTA. The corresponding maximal activity is set as 100%
(black bars). Addition of 10 l
M
CaM–GCAP-1 or 10 l
M
CPG–
GCAP-1 did n ot ch ange significantly the maximal GC activity (grey
bars). Error bars are within the size of the columns.
time (min)
0246810121416
Ca
2+
monomer
(WT)
dimer
A
280

CaM-GCAP-1
time (min)
0246810121416
A
280
EGTA
CaM-GCAP-1
monomer
(WT)
dimer
time (min)
0246810121416
A
A
B
280
EGTA
CPG
monomer
(WT)
dimer
time (min)
0246810121416
Ca
2+
monomer
(WT)
dimer
A
280

CPG
Fig. 6. Size exclusion chromatography of
CaM–GCAP-1 and CPG–GCAP-1 on a Bio-
Sep–SEC-S2000 column in the presence of
Ca
2+
(left panels) or EGTA (right panels).
Chromatograms of w ild-type GCAP-1 (WT,
thin traces) are always included to allow a
direct comparison with the elution profiles of
the mutants. Chromatograms of GCAP-1
mutants are presented as thick lines.
Monomer and dimer forms are indicated.
3790 J Y. Hwang et al. (Eur. J. Biochem. 271) Ó FEBS 2004
The functional EF-hand 1 in calmodulin contains a Gly at
this position. A GCAP chimera in which the loo p region of
EF-hand 1 is replaced by the corresponding region of
calmodulin had lost completely any cyclase-activating
property. In fact, loss of activity was even observed i n
a point mutant where CysPro was substituted by Gly
(CP fi G; CPG). T hese results highlight the importance of
Cys29andPro30inGCAP-1fortheactivationprocessand
are consistent with previous findings t hat substitu tion of
Cys29 b y Ser, Gly or Asp produced inactive mutants [17].
However, Cys29 can be deleted or changed to Asn, Tyr and
Ala without significant loss of GCAP activity [17,18].
Interestingly, the mutant C29S was described as a potent
competitor of a constitutively active GCAP-1 mutant
showing that it binds to the target without activating it
[17]. In contrast, our CPG mutant did not bind to the

cyclase (Fig. 5), which indicates that Pro30 in EF-hand 1
might be necessary for the interaction with the target.
However, it is also possible that the short sequence CysPro
is important for p roviding the proper c onformation for
target activation.
Why were the mutants CaM–GCAP-1 and CPG–
GCAP-1 without activity? Reasonable explanations are
that they do not bind Ca
2+
or do not change conforma-
tion in response to C a
2+
. We can exclude these possibil-
ities, as the mutants exhibited similar Ca
2+
-induced
conformational changes as wild-type GCAP-1 a s probed
by gel shift and tryptophan fluorescence assays. Further-
more, stoichiometries of Ca
2+
bound per GCAP-1
mutants were consistent with four (CaM–GCAP-1) or
three (CPG–GCAP-1) functional EF-hands. However,
wild-type GCAP-1 bound only two Ca
2+
under the
conditions used and no t three as e xpected. Due t o its
activation profile and because of the intermediate affinity
for Ca
2+

of its EF-hand 3, wild-type GCAP-1 must have
at least one Ca
2+
-binding site of very high affinity [18]. We
think t hat t reating GCAP-1 solutions with Chelex did not
remove Ca
2+
from the high affinity site and therefore we
measured only b inding of Ca
2+
to the remaining two sites.
This assumption was verified by th e r elease of one
additional C a
2+
from denatur ed wild-type G CAP-1. In
the case of CaM–GCAP-1 we were able to observe the
binding of four Ca
2+
, because the introduction of a
complete functional EF-hand might have lowered the
affinity of the other EF-hand(s).
Size exclusion chromatography revealed a plausible
reason why these EF-hand 1 mutants had lost their
activating p roperties. Sokal et al. r eported r ecently that
GCAP-1 dimers are inactive [17]. Our results showed that
mutations in EF-hand 1, like those in CaM–GCAP-1 a nd
CPG–GCAP-1, promote the formation of inactive dimers.
Dimerization of GCAP-1 apparently prevents the access of
the target to the target binding site. Therefore, the target site
must not necessarily be lo cated within EF-hand 1. Alter-

natively, dimer formation could allosterically change the
contact surface of the interaction site. So far we cannot
distinguish between these two possibilities.
GCAP-2 undergo es a Ca
2+
-controlled re versible dimeri-
zation in order to activate g uanylate cyclase [20,21]. But
point mutations in EF-hand 1 of GCAP-2 led to inactive
mutants that were still able to dimerize in the absence of
Ca
2+
[21]. The difference to GCAP-1 in our study is
significant: in one case (GCAP-1) changes in EF-hand 1
induce i rregular dimerization of the protein, in the other
case (GCAP-2) the Ca
2+
-induced transition to the ÔactiveÕ
dimer still works, but the protein lacks activity.
Fig. 7. Limited p roteolysis of GCAPs by trypsin. GCAPs were incu-
batedwithtrypsininthepresenceofCa
2+
(Ca) or in its absence
(EGTA) and the fragmentation pattern was analysedbySDS/PAGE.
Time of incubation is indicated in minutes. Molecular mass standards
are shown on the left site of each gel (MW), numbers refer to kDa of
corresponding standards. Upper part: 5 lg of undigested wild-type
GCAP-1 (a) and 10 lg of tryptic fragments were loaded on the
gel. Main fragments after staining withCoomassieblueareindicated
as b–g. Middle part: CaM–GCAP-1, amount of protein as indicated
above. Undigested protein is labelled a and main fragments are

labelled b–k. Lower part: CPG–GCAP-1, am ount of protein a s
indicated above. Undigested protein is labelled I and main fragments
are labelled I I–VIII. Ca
2+
-dependent differences in electrophoretic
mobilities as shown in Fig. 2 are not visible, because the final EGTA
concentration of the sam ples was too l ow.
Ó FEBS 2004 Irregular dimerization of GCAP-1 mutants (Eur. J. Biochem. 271) 3791
A s pecial case was observed with t he CPG mutant in the
presence of the Ca
2+
chelator EGTA. A lthough dimeriza-
tion was observed, the main portion of the mutant eluted as
a monomer (Fig. 6B, right panel) but with a shifted
retention volume. This indicated that the overall shape of
the mutant deviates from the assumed c ompact shape o f
wild-type G CAP-1 and probably has a more elongated
form. If w e assume that the three-dimensional structure of
GCAP-1resemblesthatofGCAP-2, one can s peculate
about the structural impact o f the CPG replacement. The
CysPro region is located before a short b-strand that is fixed
by interaction with EF-hand 2. The entering helix of EF-
hand 1 has hydrophobic c ontact with the exiting helix of
EF-hand 2 [31]. Shortening of the distance in the loop
region by the mutation could trigger something like a lever
switch which would c ause a movement o f the N-terminus
away from the centre of the molecule leading to a more
elongated form.
A similar obse rvation was also made with CaM–GCAP-1
in the presence of EGTA, but here t he dimer is t he prominent

form of the mutant (Fig. 6A, right panel). Thus, the
mutation CPG changed mainly the hydrodynamic v olume
of the protein in addition to promoting the transition to the
inactive dimer form. Furthermore, the CPG exchange could
have lowered the Ca
2+
affinity of the high affinity EF-hand
in CPG–GCAP-1, which would explain why we observed
three bound Ca
2+
in CPG–GCAP-1.
Proteolysis studies showed that the mutants exhibited a
lower accessibility for tryptic dige stion. Of particular interest
was the differen t pattern we observed for the digestion of the
Ca
2+
-free GCAP-1 form, which is the cyclase-activating
conformation. Removing Ca
2+
triggered a conformational
change that opens the i nterior of the protein a nd allows
tryptic cleavage at several interior trypsin cleavage sites [14].
Fragments f and g from the wild-type GCAP-1 digest
(Fig. 7 ) probably c orrespond to fragments consisting of
residues 9–91 (% 10 kDa) and 121–172 (% 6kDa)thathad
been identified previously [14]. While these fragments
appeared immediately after start of the digestion of wild-
type GCAP-1 (2 min), digestion of the mutants produced
larger fragments at higher molecular mass (see also Results)
and to a lesser extent a few faint b ands of the appropriate

size of g and f (6 and 10 kDa, Fig. 7, CaM–GCAP-1, below
k and CPG–GCAP-1 one band at VIII). These critical
digestions occur at t he tryptic c leavage sites L ys91 and
Arg120 which are located at the N-terminal and C-terminal
flanks of EF-hand 3. Dissociation of Ca
2+
from EF-hand 3
triggers the conformational change in GCAP-1 leading to
an opening of the c ompact core structure [14,16,18]. Our
data show that these opening steps are hindered i n the
mutants CaM–GCAP-1 a nd CPG–GCAP-1. However, for
each mutant a different reason might account for this
behaviour. CaM–GCAP-1 forms a dimer in the a bsence of
Ca
2+
(Fig. 6 ), which probably protects the site around EF-
hand 3 f rom tryptic digestion, although CaM–GCAP-1
undergoes Ca
2+
-induced conformational changes (Figs 2
and 3). CPG–GCAP-1 also forms a dimer in the absence of
Ca
2+
, but a m ain fraction is p resent as a distorted monomer
(see above). Thus, the CPG point mutation in the
nonfunctional EF-hand 1 exerts an effect o n the opening
of the GCAP-1 structure around EF-hand 3 when Ca
2+
is
released from this site.

In summary, our results show that EF-hand 1 in wild-
type GCAP-1 is critical for forming an inactive dimer and
mutations in EF-hand 1 can hinder the opening of the core
structure around EF-hand 3 .
Our results highlight the differences between GCAP-1
and GCAP-2 [11,19]. Both proteins operate in the same type
of cell, but they differ significantly in their regulatory
functions.
Acknowledgements
We thank D. Ho
¨
ppner-Heitmann for excellent technical a ssistance,
Prof. Georg Bu
¨
ldt (IBI-2, F Z Ju
¨
lich) for continuous support a nd Dr
Oliver Weiergra
¨
ber (IBI-2) for helpful discussions. This work was also
supported by a grant from the De utsche Forschungsgemeinschaft to
K W.K. (Ko948/5-3) and a DAAD fellowship to J Y.H.
References
1. Pugh, E.N. Jr & Lamb, T.D. (2000) Phototransduction in verte-
brate rods and cones: Molecular mechanisms of amplification,
recovery and light adaptation. In Handbook of Biological Physics
(Stavenga, D.G., DeGrip, W.J. & Pugh, E.N. Jr, eds), pp. 183–
255. Elsevier Science BV, North-Holland.
2. Burns, M.E. & Baylor, D.A. (2001) Activation, deactivation, and
adaptation in vertebrate phot oreceptor c ells. Annu. Rev. Neurosci.

24, 779–805.
3. Kaupp, U .B. & Seifert, R. (2002) Cyclic nucleotide-gated ion
channels. Physiol. Rev. 82, 769–824.
4. Palczewski, K., Polans, A .S., Baehr, W. & Ames, J.B. (2000)
Ca
2+
-binding proteins in the retina: structure, function, and the
etiology of human visual diseases. Bioessays 22, 337–350.
5. Dizhoor, A.M. (2000) Regulation of cGMP synthesis in
photoreceptors: role in signal transduction and congenital diseases
of the retina. Cell. Signal. 12 , 711–719.
6. Koch, K W., D uda, T . & Sharma, R.K. (2002) Photoreceptor
specific guanylate cyclases in verteb rate p hototransductio n. Mol.
Cell. Biochem. 23 0, 97–106.
7. Pugh, E.N. Jr, Nikonov, S. & Lamb, T.D. (1999) Molecular
mechanisms of vertebrate photoreceptor light adaptation. Curr.
Opin. Neurobiol. 9, 410–418.
8. Braunewell, K H. & G undelfinger, E.D. ( 1999) Intracellular
neuronal calcium sensor p roteins: a family of EF-hand calcium-
binding proteins in search of a function. Cell Tissue Res. 295,
1–12.
9. Burgoyne, R.D. & Weiss, J.L. (2001) The neuronal calcium sensor
family of Ca
2+
-binding proteins. Biochem. J. 353, 1–12.
10. Olshevskaya, E.V., Hughes, R.E., Hurley, J.B. & D izhoor, A.M.
(1997) Calcium binding, but not a calcium-myristoyl switch,
controls the ability of guanylyl cyclase-activating protein GCAP-2
to regulate photorec eptor guan ylyl cyclase. J . B iol. C hem . 272,
14327–14333.

11. Hwang, J Y. & K och, K W. (2002) The myristoylation o f the
neuronal Ca
2+
-sensors guanylate cyclase-activating protein 1 and
2. Biochim. Biophys. Acta 1600, 111–117.
12. Dizhoor, A.M. & H urley, J.B. (1996) Inactivation of E F-han ds
makes GCAP-2 (p24) a c onstitutive activato r of p hotorecep tor
guanylyl cyclase by p reventing a Ca
2+
-induced Ôactivator-to-
inhibitorÕ transition. J. Biol. Chem. 271, 19346–19350.
13. Otto-Bruc, A., Buczylko, J., S urgucheva, I., Subbaraya, I., Rud-
nicka-Nawrot, M., Crabb, J.W., Are ndt, A., Hargrave, P.A.,
Baehr, W. & Palczewski, K . (1997) Functional reconstitution of
photoreceptor guanylate cyclase with native and mutant forms
of guanylate cyclase-activating protein 1. Biochemistry 36,
4295–4302.
3792 J Y. Hwang et al. (Eur. J. Biochem. 271) Ó FEBS 2004
14. Rudnicka-Nawrot, M., Surgucheva, I., Hulmes, J.D., Haeseleer,
F., Sokal, I., Crabb, J.W., Baehr, W. & Palczewski, K. (1998)
Changes i n biological activity and folding of g uanylate cyclase-
activating protein 1 as a function of calcium. Biochemistry 37,
248–257.
15. Hughes, R.E., Brzovic, P.S., Dizhoor, A.M., Klevit, R.E. &
Hurley, J.B. (1998) Ca
2+
-dependent conformational changes in
bovine GCAP-2. Protein Sci. 7, 2675–2680.
16. Sokal, I., Otto-Bruc, A.E., Surgucheva, I., Verlinde, C.L.M.J.,
Wang,C K.,Baehr,W.&Palczewski,K.(1999)Conformational

changes in guanylyl cyclase-activatin g protein 1 (GCAP1) and its
tryptophan mutants a s a function of calcium concentration.
J. Biol. Chem. 274, 19829–29837.
17. Sokal, I., Li, N., K lug, C.S., Filipek, S., Hubbell, W.L., Baehr, W.
& Palczewski, K. (2001) Calcium-sensitive regions of GCAP1 as
observed by chemical modifications, fluorescence, and EPR
spectroscopies. J. Biol. Chem. 276, 43361–43373.
18. Hwang, J Y., Schlesinger, R. & K och, K W. ( 2001) Calcium-
dependent c ysteine reac tivities in the neuronal calcium sensor
guanylate cyclase-activating protein 1. FEBS Lett. 508, 355–359.
19. Hwang, J Y., Lange , C., Helten, A., Ho
¨
ppner-Heitmann , D.,
Duda,T.,Sharma,R.K.&Koch,K W.(2003)Regulatorymodes
of rod outer seg ment membrane guanylate cyclase differ in catalytic
efficiency and Ca
2+
-sensitivity. Eur. J. Biochem. 270, 3814–3821.
20. Olshevskaya, E.V., Ermilov, A.N . & Dizhoor, A.M. (1999)
Dimerization of guanylyl cyclase-activating protein and a
mechanism of photorecep tor guanylyl cyclase activation. J. Biol.
Chem. 274, 25583–25587.
21. Ermilov, A.N., O lshevskaya, E.V. & Dizhoor, A.M. (2001)
Instead of b inding calcium, o ne of the E F-hand structures i n
guanylyl cyclase activatin g protein-2 is required for targeting
photoreceptor guanylyl cyclase. J. Biol. Chem. 276, 48143–48148.
22. Schrem, A., Lange, C., Beyermann, M. & K och, K W. (1999)
Identification of a domain in guanylyl cyclase-activating protein 1
that interacts with a complex of guanylyl cyclase and tubulin in
photoreceptors. J. Biol. Chem. 274, 6244–6249.

23. Olshevskaya, E.V., Boikov, S., Ermilov, A., Krylov, D., H urley,
J.B. & Dizhoor, A.M. (1999) Mapping functional domains of the
guanylate cyclase regulator protein, GCAP-2. J. Biol. Chem. 274,
10823–10832.
24. Krylov, D.M., Niemi, G.A., Dizhoor,A.M.&Hurley,J.B.(1999)
Mapping sites in guanylyl cyclase activating protein-1 reguired for
regulation of photoreceptor membrane guanylyl cyclases. J. Biol.
Chem. 274, 10833–10839.
25. Li, N., Sokal, I., Bronson, J.D., Palczewski,K.&Baehr,W.(2001)
Identification of functional reg ions of guanylate c yclase-activating
protein 1 (GCAP1) using GCAP1/GCIP chimeras. Biol. Chem.
382, 1179–1188.
26. Frins, S., Bo
¨
nigk, W., Mu
¨
ller, F., Kellner, R. & Koch, K W.
(1996) Functional characterization of a guanylyl cyclase-activating
protein from vertebrate rods. J. Biol. Chem. 271, 8022–8027.
27. Ames, J.B., Porumb, T., Tanaka, T., Ikura, M. & Stryer, L. (1995)
Amino-terminal myristoylation induces cooperative calcium
binding to recoverin. J. Biol. Chem. 270, 4526–4533.
28. Senin, I.I., Fischer, T., Komolov, K.E., Zinchenko, D .V., Phili-
pov, P.P. & Koc h, K W. (2002) Ca
2+
-myristoyl switch in the
neuronal calciu m sensor recoverin requ ires different functions o f
Ca
2+
-binding sites. J. Biol. Chem. 277, 50365–50372.

29. Tsien, R.Y. (1980) New calcium indicators and buffers with high
selectivity against magnesium and protons: design, synthesis, and
properties of prototype structures. Biochemistry 19, 2396–2404.
30. Winzor, D.J. (1969) Analytical gel filtration. In Physical Principles
and Techniques of Protein Chemistry, Part A (Leach, S.J., ed.),
pp. 451–495. Academic Press, New York.
31. Ames,J.B.,Dizhoor,A.M.,Ikura,M.,Palczewski,K.&Stryer,L.
(1999) Three-dimensional structure of guanylyl cyclase activating
protein-2, a calcium-sensitive modulator of photoreceptor gua-
nylyl cyclases. J. Biol. Chem. 274, 19329–19337.
Ó FEBS 2004 Irregular dimerization of GCAP-1 mutants (Eur. J. Biochem. 271) 3793