(Received for publication, November 11, 1996, and in revised form, February 11, 1997)
From the Institut für Pharmakologie und Toxikologie der Technische Universität München, Biedersteiner Straße 29, D-80802 München, Germany
The cGMP-dependent protein kinases
(cGK) I and I
have identical cGMP binding sites and catalytic
domains. However, differences in their first 100 amino acids result in
15-fold different activation constants for cGMP. We constructed
chimeras to identify those amino acid sequences that contribute to the
high affinity cGK I
and low affinity cGK I
phenotype. The cGK
I
/I
chimeras contained permutations of six amino-terminal regions
(S1-S6) including the leucine zipper (S2), the autoinhibitory domain
(S4), and the hinge domain (S5, S6). The exchange of S2 along with S4
switched the phenotype from cGK I
to cGK I
and vice
versa, suggesting that the domains with the highest homology
between the two isozymes determine their affinity for cGMP. The high
affinity cGK I
phenotype was also obtained by a specific
substitution within the hinge domain. Chimeras with the sequence of cGK
I
in S5 and cGK I
in S6 were activated at up to 6-fold lower cGMP
concentrations than cGK I
. Based on the activation constants of all
chimeras constructed, empirical weighting factors have been calculated that quantitatively describe the contribution of the individual amino-terminal domains S1-S6 to the high affinity cGK I
phenotype.
The amino terminus of the cGK1
regulates several important functions such as the dimerization of the
two cGK subunits, the inhibition of the catalytic center in the
non-active enzyme, the affinity of the binding sites for cGMP, and the
concentration of cGMP required for activation of the enzyme (Refs.
1-5; for reviews see Refs. 6-8). The amino acid sequences responsible for these multiple functions of the amino terminus have been partially identified. The leucine zipper, a sequence of heptad repeats of leucines and isoleucines forming an -helical structure at the first
part of the amino terminus (9, 10), has been identified as the
dimerization domain. Removal of the leucine zipper along with the
autoinhibitory domain (residues 1-78) of cGK I
results in a
permanently active and monomeric enzyme, which still binds cGMP to its
two binding sites but with lower affinity than the native enzyme (11).
The removal of the leucine zipper in cGK I
(residues 1-62) results
in a monomeric cGK that is still activated by cGMP (2), suggesting that
the autoinhibitory domain resides in a sequence that is
carboxyl-terminal to the leucine zipper. The autoinhibitory sequences
contain the main in vitro autophosphorylation sites of cGK
I
(12) and cGK I
(13).
To date it has remained unclear which amino acid sequences of the amino
terminus regulate the binding affinity for cGMP (1, 3). The two
amino-terminal splicing forms of the cGK type I, I and I
, which
have been identified by cDNA cloning (14, 15) and protein
purification (16, 17), differ only in their first 89 and 104 amino-terminal residues, respectively. Although these amino termini do
not contain the cGMP binding sites, the two isozymes differ 15-fold in
their affinity for cGMP. Cyclic GMP binds to a high and a low affinity
site with Kd values of 10 and 150 nM in
cGK I
(3, 18), but to two low affinity sites with Kd values of 150 nM in cGK I
(3).
These differences in cGMP binding of the two isozymes are also
reflected in the dissociation kinetics. Cyclic GMP dissociates with a
fast and a slow rate from cGK I
, but only with a fast rate from cGK
I
. The decreased binding affinity of the cGK I
results in a
15-fold increase in the cGMP concentration needed for half-maximal
activation when compared with cGK I
(3, 19). To identify those amino
acid sequences within the amino terminus that determine the affinity of
the isozymes for cGMP, chimeras of cGK I
and cGK I
were
constructed. The analysis of these chimeras shows that the leucine
zipper and the autoinhibitory domain mainly affect the phenotype.
[Ser21]protein kinase inhibitor-(14-22) substrate peptide and protein kinase inhibitor-(5-24) amide were synthesized as described (20). The expression vector pMT3 was obtained from R. Kaufmann, Genetics Institute, Cambridge, MA.
Cloning of a Full-length Clone of cGK IAn oligo(dT)-primed, size-fractionated cDNA library
was constructed in the pcDNAII vector. Colonies (3 × 105) were screened with a 980-base pair cDNA fragment
(nucleotides 1635-2616) from mouse cGK II (21). Hybridization was
performed under low stringency (50 °C). A clone with an insert of
2.5 kilobase pairs was identified as cGK I. This clone contained the
complete coding region of cGK I
and confirmed the sequence composed
of partial clones (14).
The cGK
I cDNA containing the consensus sequence for optimal expression
in Sf9 cells was ligated into pFastBac1 yielding the bacmid transfer
vector pFB1/cGKI
. Transformation of pFB1/cGKI
into Max Efficiency
DH10Bac cells, identification of recombinant clones, and isolation of
the recombinant bacmid DNA were performed according to the instructions
of the Bac-to-Bac baculovirus expression kit (Life Technologies, Inc.).
Recombinant bacmid DNA was transfected into Sf9 cells. Recombinant
viruses were identified by their ability to direct the expression of
cGK I
. Recombinant virus was amplified without further purification
and viral titer estimated by end point dilution.
Suspension cultures (1.3 × 1010 Sf9
cells/7.2 liters) were infected at a multiplicity of infection of 10. After 72 h, cells were harvested by centrifugation, washed twice
with serum-free TC-100 medium, resuspended in buffer A (20 mM Tris/HCl, pH 7.4, containing 100 mM NaCl,
2.5 mM DL-dithiothreithol, 2.5 mM benzamidine, 0.1 mM EDTA, 0.1 mM
phenylmethylsulfonyl fluoride, 1 µg/ml leupeptin), and stored at
80 °C. Cells were lysed by freeze/thawing, homogenized, and
centrifuged. The supernatant was loaded onto an cAMP affinity column
packed with cAMP-agarose (Sigma, catalog no. A7775) equilibrated with
buffer A. The column was washed with buffer A prior to elution of bound
cGK I
with buffer A supplemented with 100 µM cGMP. The eluate was dialyzed against 20 mM TES buffer (pH 7.1),
containing 2.5 mM DL-dithiothreithol, 0.1 mM sodium azide, and 1 µg/ml leupeptin. The purified
protein was concentrated to ~4 mg/ml using Macrosep-10 concentrators
(Pall Filtron), made up 50% (v/v) in glycerol and stored at
20 °C. The cGK I
was expressed (22) and purified (23) as
described.
The coding DNA sequence of cGK I (nucleotides
6 to
2120) and cGK I
(nucleotides
23 to 2165) were constructed into the eukaryotic expression vector pMT3 (24). For constructing chimeras, replacement sequences containing amino-terminal domains of cGK I
and
cGK I
with the boundaries as given in Fig. 2 were generated by
standard polymerase chain reaction overlap extension techniques. All
chimeras exhibited the Koczak consensus sequence ACC (nucleotides
3
to
1) prior to the ATG start codon. COS-7 cells were transfected, and
cytosolic proteins containing recombinant cGKs were extracted as
described previously (3).
Ka Determination
The activity of the expressed
I, I
, and chimeric cGK was measured as described (3).
Ka values were obtained by a nonlinear fit of the
data according to the Hill equation.
Equilibrium binding of [3H]cGMP to the expressed cGK proteins was carried out as described (3). Binding of [3H]cGMP to sites A and B was determined as described (1). Calculation of the Kd values and the maximal binding capacities was performed by the LIGAND program (25). The stoichiometry of cGMP binding was calculated by dividing the maximal binding capacity obtained from Scatchard analysis by the amount of cGK present in the cytosolic extract from COS cells. The concentration of cGK in COS cell extracts was determined by the phosphotransferase activity using the specific activity of 2.4 ng of pure cGK added to an identical amount of cytosolic extract from control COS cells. The 2.4 ng of cGK yielded a specific activity of 4 µmol of phosphate transferred/min/mg of pure cGK. Dissociation rate constants were measured as described (3).
Calculation of Weighting FactorsWeighting factors were obtained by an iterative procedure using chimeras 1-19 and optimized by one-dimensional interpolation.
StatisticsData are means ± S.E. of 5-17
(Ka) and 3-5 (k1(fast),
k
1(slow)) determinations from at least three different expressions. Statistical significance was determined by
Student's t test for paired and unpaired data. A
p value less than 0.05 was considered to be significant.
A full-length cDNA for the cGK I was cloned from
bovine testis that confirmed the previously published cDNA sequence
of bovine cGK I
deduced from partial clones (14). The cloning of a
full-length cDNA excluded the possibility that the cGK I
differed from the cGK I
isozyme in regions other than the
amino-terminal domain. In contrast, the previously published human cGK
I
sequence (15) differed from the bovine full-length clone at
Lys280 and Asn290, which were replaced in the
human cGK I
by Thr280 and Ser290,
respectively.
To demonstrate that the differences in cGMP binding and activation
between cGK I and cGK I
(3) were intrinsic properties of the
isozymes, cGK I
and cGK I
were expressed in Sf9 cells, purified
to homogeneity, and subsequently characterized. The phosphotransferase activity of cGK I
was stimulated 7-fold by cGMP with apparent Ka and Vmax values of 59 nM and 6.2 µmol/min × mg, respectively (Fig.
1). The phosphotransferase activity of cGK I
was
stimulated 35-fold with apparent Ka and
Vmax values of 1210 nM and 7.6 µmol/min × mg, respectively. The dissociation kinetics of cGMP
from the cGK I
clearly defined a fast dissociation rate from a low
affinity site and a slow dissociation rate from a high affinity site
(Fig. 1). In contrast, cGK I
exhibited only fast dissociation from
two low affinity sites. The corresponding k
1 values were 6.2 and 0.009 min
1 for cGK I
and 5.3 min
1 for cGK I
. Equilibrium binding experiments
indicated that cGK I
and cGK I
bound 1.8 and 1.6 mol of cGMP/mol
of subunit, respectively. The apparent Kd values
were 98 and 7 nM for cGK I
and 172 nM for
cGK I
. These results suggest that the binding and catalytic
constants of pure cGK I
and cGK I
expressed in Sf9 cells were
similar to those measured in the cytosolic extract of COS cells
(3).
Alignment and Subdivision of the cGK I
To identify the structural basis of their 15-20-fold
differences in the Ka values, the amino-terminal
domains of cGK I and cGK I
were aligned and subdivided into six
domains, S1-S6 (Fig. 2): S1, the extreme amino-terminal
sequence; S2, the leucine zipper containing 5 heptad repeats of
leucines; S3, the short linker connecting the leucine zipper and the
autoinhibitory domain; S4, the autoinhibitory domain containing the
substrate-like sequence PRT59TR with the
autophosphorylation site (Thr59) of cGK I
(12) and the
putative pseudosubstrate sequence KRQAISAE of cGK I
(26); and S5,
the first part, and S6, the second part of the hinge domain, which
links the amino terminus to the other domains of the cGK. The hinge
domain was subdivided since proteolysis at
Arg78/Gln79 led to a monomeric and active cGK
I
, which had different cGMP binding properties than the intact cGK
I
(11), suggesting that the sequences amino-terminal from
Gln79 considerably influence the binding properties of the
enzyme.
For the construction
of chimeras, the six domains with the boundaries defined above were
exchanged between cGK I and cGK I
. Expression vectors carrying
the coding sequence of cGK chimeras or that of cGK I
and cGK I
were transfected into COS cells. The chimeric and wild type cGK
proteins were identified in the soluble extracts of transfected cells
by an antibody directed against a carboxyl-terminal sequence of the
enzyme (20). The mutant and wild type cGK had apparent
Mr values of 75 kDa and were expressed at
concentrations between 0.5 and 4.5 µg/mg of soluble protein,
respectively, as assessed by their maximal cGMP binding capacities. All
of the chimeras expressed comigrated with the pure cGK I
and cGK
I
in 5-15% sucrose density gradient centrifugation with
s20w values between 7.2 and 7.8, suggesting that they were dimers. Subsequently, the chimeras were
characterized by the cGMP concentration needed for half-maximal
activation (Ka) and the dissociation rate constants
(k
1) of cGMP from the high and low affinity
site.
The Ka values for cGMP of the cGK
I and cGK I
expressed in COS cells were 77 and 1748 nM, respectively (Fig. 3). Almost identical
values were obtained with the pure isozymes (Fig. 1). Cyclic GMP
dissociated with a slow rate (k
1 = 0.008 min
1) and a fast rate (k
1 = 6.5 min
1) from the high and low affinity site of the cGK I
(Fig. 3). In contrast, only fast dissociation from the two low affinity sites was observed for cGK I
(k
1 = 5.0).
The two cGMP binding sites of cGK I
could not be distinguished
kinetically. Studying several chimeras with randomized permutations in
the six amino-terminal domains, we identified the sequences necessary to generate the respective phenotype. Chimeras containing the sequence
of cGK I
in S2 and S4 (chimeras 1-5) exhibited
Ka values between 580 and 1136 nM. These
constants were only 1.5-3-fold lower than the Ka
value of cGK I
. The cGK I
phenotype of chimeras 1 to 5 was
further confirmed by monophasic dissociation with rate constants
between 4.4 and 10 min
1. Substitutions of cGK I
sequence in S1, S3, S5, and S6, alone or in combination, did only
marginally affect the kinetic constants of chimeras 1-5. Conversely,
chimeras 6-9 containing the cGK I
sequence in S2 and S4 had
Ka values between 112 and 169 nM. These
values were only 1.4- and 2.2-fold above that of cGK I
(Fig. 3). In
agreement with the cGK I
-like activation constants, these chimeras
showed biphasic dissociation kinetics with a fast and a slow component
(Fig. 3). The presence of cGK I
substitutions in S1, S3, S5, and S6
did not substantially alter the high affinity cGK I
phenotype of
these chimeras. These results suggested that the leucine zipper along
with the autoinhibitory domain were the main structural elements which
determine the affinity of the cGK isozymes. Interestingly, these two
domains are very homologous between cGK I
and cGK I
with 28 identical, 4 conservatively substituted, and 22 different amino acids
(Fig. 2), suggesting that 29% and 25% of the total residues of the
amino terminus (cGK I
= 89 and cGK I
= 104 amino acids) were
sufficient to generate the high affinity cGK I
and the low affinity
cGK I
phenotype, respectively.
Equilibrium Binding of cGMP to cGK Chimeras
Chimera 9 and 5 were exemplarily selected to measure the equilibrium binding of cGMP,
because they fulfilled the minimal requirements for the high affinity
cGK I and low affinity cGK I
phenotype, respectively. Both
chimeras bound about 4 mol of cGMP/mol of holoenzyme (Fig.
4). As expected, chimera 9 bound cGMP to sites 1 and 2 with apparent Kd values of 4.3 nM and
439 nM and a stoichiometry of 2.4 and 1.4 mol of cGMP
bound/mol of holoenzyme, respectively. The dissociation induced by the
addition of a 1000-fold excess of unlabeled cGMP yielded two rates of
0.008 and 6.5 min
1. Chimera 5 bound 4 mol of cGMP/mol of
holoenzyme with an apparent Kd of 195 nM. A distinct high affinity site was not detected. These
results suggested that chimera 5 contained two low affinity binding
sites. This finding was confirmed by monophasic dissociation of cGMP
with a rate constant of 4.8 min
1. These values are not
different from the wild type enzymes and indicate, therefore, that S2
and S4 determine the high affinity cGK I
and low affinity cGK I
phenotype.
Chimeras of Intermediate Phenotype
The results above did not
definitely rule out that the cGK I or cGK I
phenotype would also
be obtained by the substitution of S2 or S4 alone. Chimeras 10-17 with
substitution in only one domain exhibited an intermediate phenotype
with Ka values of about 300 nM. These
values were roughly 5-fold higher and lower than the values for cGK
I
and cGK I
, respectively (Fig. 3). This shift in the
Ka values was also reflected in the accelerated
dissociation of cGMP from the high affinity site in most of these
chimeras. However, dissociation was never monophasic as observed for
chimeras exhibiting the cGK I
phenotype. In addition to the effect
on the high affinity site, chimeras 10-13 showed also a slight,
statistically not significant deceleration of dissociation from the low
affinity site. The rate constants of chimera 14 are disconcordant with
those of the other chimeras in the group. The Ka of
363 nM is probably caused by a decrease in the association
rate. These results obtained with different substitutions in S2 and S4
confirmed the observation that only the combined exchange of both
domains alter the phenotype from cGK I
to cGK I
and vice
versa.
Exceptions from the above rule were observed with
some chimeras. Chimeras 18 and 19 containing the sequences of cGK I
in S2, S5, and S6 also exhibited the low affinity phenotype with Ka values of 1 µM, although the
sequence of cGK I
was present in the autoinhibitory domain S4 (Fig.
3). The structural basis of this phenomenon was limited to the short
linker sequence between S2 and S4. This linker varied in length between
cGK I
and cGK I
(Fig. 2). Apparently, the cGK I
linker exerted
this paradoxical effect since it facilitated the expression of the low
affinity cGK I
phenotype (Fig. 3). In fact, the exchange of the
linker sequence from cGK I
to cGK I
decreased 3-fold the
Ka from 1 µM in chimeras 18 and 19 to
about 380 nM in chimeras 12 and 13. These chimeras
suggested that the expression of the respective phenotype was
strengthened by inserting the linker sequence of the other cGK
isozyme.
Similar as found for the low affinity cGK I
phenotype, chimeras were identified that express the high affinity cGK
I
phenotype without fulfilling the requirements in S2 and S4.
Chimera 20 contained the sequence of cGK I
in S1-S5 and that of cGK
I
in S6 but exhibited a 6-fold lower Ka value
than cGK I
itself (Fig. 3). Even further cGK I
substitutions
additional to S6 in S1, S2, and S3 (chimera 21) or in S4 (chimera 22)
yielded cGK enzymes, which were stimulated at significantly lower cGMP
concentrations (p < 0.03) than cGK I
. Obviously, a
specific constellation, i.e. the combination of cGK I
substitution in S5 and that of cGK I
in S6, resulted in a cGK I
phenotype with increased affinity for cGMP (Fig. 3). The increased
affinity of chimera 20 was apparently caused by slowing the
dissociation of cGMP from the high affinity site. In contrast,
dissociation of cGMP in chimera 21 was accelerated 5-fold instead of
being decelerated when compared with cGK I
. The high affinity of
this chimera was due to a faster association rate for cGMP. Whereas
association of cGMP included a fast (k+1
0.05 min
1·nM
1) and a slow
component (k+1 = 0.00042 min
1·nM
1) in the cGK I
,
only the fast component was observed with chimera 21 resulting in
equilibrium binding within 30 s after addition of cGMP. Taken
together, the combination of cGK I
substitution in S5 and that of
cGK I
in S6 enhanced the Ka decreasing effect of
cGK I
substitutions in S2 and S4. Even the effect of cGK I
sequence in one domain alone, either S2 or S4, exceeded that of cGK
I
sequence in both domains as indicated by the significantly lower
Ka value of chimera 21 and 22 when compared with the
Ka of cGK I
. These results suggest that the
flexibility and anchoring function of the hinge domain increased the
interaction of S2 and S4 with the cGMP binding domain under the
condition described above. This effect was possibly mediated by
altering the position of the the amino-terminal domain relative to the other domains in the cGK holoenzyme structure.
The contribution
of each of the 6 amino-terminal domains, S1-S6, to induce the high
affinity cGK I phenotype could be best assessed by assigning
arbitrary weighting factors to the individual domains of cGK I
(Fig.
5). The weighting factors were obtained by an iterative
procedure. The factors for the leucine zipper and the autoinhibitory
domain were found to be most influential with values of 0.2 and 0.3, respectively, thereby confirming the importance of these domains for
the high affinity cGK I
phenotype. On the other hand, the individual
effect of the extreme amino-terminal sequence (S1) and the second part
of the hinge domain (S6) were marginal with factors of 0.9 and 0.8, respectively. The linker sequence (S3) contributed with the factor 1.8 to the phenotype indicating that it acted in a reverse manner,
i.e. cGK I
substitution therein increased
Ka 1.8-fold instead of decreasing it. The weighting
factors allowed the estimation of the Ka values for
most of the chimeras constructed with considerable accuracy (Fig. 5).
To obtain the Ka value of any chimera, the
Ka value of cGK I
(= 1748 nM) was
multiplied with the individual factors for S1 to S6 of each cGK I
and cGK I
domain. With three exceptions, the ratio between the
Ka values predicted and the Ka
values actually measured varied between 0.70 and 1.28 with a mean value
of 1.03 ± 0.05 suggesting that the factors reliably described the
impact of the individual amino-terminal domains on the high affinity
cGK I
phenotype. This arbitrary equation, however, failed to
describe the Ka decreasing effect of the specific
constellation having cGK I
substitution in S5 and that of cGK I
in S6. The specific requirements in S5 and S6 suggested that this
effect depended rather on synergistic than on individual contribution of the two domains. This idea was supported by the finding that the
reverse substitution, i.e. substitution of cGK I
in S5
and that of cGK I
in S6 did not generate a "super" cGK I
enzyme (see chimeras 1, 6, 11, and 15). Thus, the dramatic
Ka decreasing effect observed with chimeras 20, 21, and 22 could not be covered by the empirical equation.
Chimeras Constructed to Test the Validity of the Empirical Equation
Of the remaining possibilities for permutations, we
selected chimeras 23, 24, and 25 to test the validity of the equation. Chimera 23 was constructed to show that also large portions of the cGK
I sequence (in this case almost 70%) were able to generate the cGK
I
phenotype when the Ka increasing effect of cGK
I
substitution in S2 was partially neutralized by the reverse effect
of cGK I
substitution in S3. The Ka of 166 nM was similar to that of cGK I
. In contrast, chimera 10 with cGK I
substitutions in S2 but not in S3 had an intermediate
Ka of 305 nM (Fig. 6).
Furthermore, chimera 23 demonstrated that the weighting factors
contributed independently to the Ka, even when
clustered. Chimera 24 was constructed to show that cGK I
substitution in S4 alone did not produce an intermediate phenotype when
all other domains contained cGK I
substitutions. This result was
exactly predicted by the equation. The replacement of the cGK I
substitutions in S3 and S1 in this chimera yielded chimera 25. The
exchange decreased the Ka as expected, although the
proportion of cGK I
sequence increased. These results demonstrate that the empirical equation can be used to describe the individual impacts of amino-terminal domains on the high affinity cGK I
phenotype.
The present study demonstrates that the leucine zipper and the
autoinhibitory domain determine the affinity of cGK I and cGK I
for cGMP. These results assign a new role for the two domains. Up to
now the leucine zipper and the autoinhibitory domain were thought to
mediate holoenzyme formation and to inhibit phosphotransferase activity
in the absence of cGMP, respectively. The regulatory effect of the
leucine zipper (S2) and the autoinhibitory domain (S4) is mediated
indirectly, since the primary structure of the binding domains are
identical in cGK I
and cGK I
. The indirect effect mainly concerns
the binding site which represents the high affinity site in cGK I
.
Recent evidence suggests that this is the amino-terminal site A (27).
The presence of cGK I
substitutions in S2 and S4 confers high
affinity for cGMP to site A, whereas the presence of cGK I
substitutions herein confers low affinity to this site. This
interpretation is supported by the dissociation rate constants which
indicated that the S2 and S4 substitution affected mainly the slow
dissociation rate.
High affinity binding to site A in cGK I depends on cooperative
interaction of site A and B (3, 4, 28, 29). The dissociation from site
A of cGK I
obtained by dilution of the cGK/cGMP complex is about
100-fold faster than that observed by addition of an excess of
unlabeled cGMP (4). For cGK I
, however, the rate constants are
similar, regardless whether dissociation has been induced by addition
of excess unlabeled cGMP or by dilution of the cGK/cGMP complex (3). It
has therefore been concluded that cooperativity is diminished or even
abolished in cGK I
. This study identifies S2 and S4 as the sequences
that allow cooperativity between the binding site A and B. S2 and S4
enable high cooperativity when derived from cGK I
and low
cooperativity when derived from cGK I
as demonstrated by the
dissociation kinetics. The involvement of the autoinhibitory domain in
regulating cooperativity was postulated previously. Autophosphorylation
of the amino terminus of the purified lung cGK (
90% cGK I
)
decreases the cooperativity between sites A and B by accelerating the
dissociation from site A (4, 28).
The ability of S2 and S4 to affect cooperativity between cGMP binding
site A and B implies that these amino-terminal domains have to be
included in any structural model of the cGK and its binding domains.
Only the conformation of the cGK induced by S2 and S4 results in the
appropriate affinity of the cGMP binding sites. The structure of the
RI subunit of cAMP kinase provides evidence that an interdomain
hydrophobic interaction region carries out the communication between
the cAMP binding domains A and B in the cAMP kinase (30). This
interdomain region allows that structural alterations in domain B, the
site that is initially occupied by cAMP during activation of the
kinase, are transmitted to site A. This interdomain region is therefore
assumed to play a central role in the cooperativity between the two
cAMP binding sites in cAMP kinase. Assuming similar tertiary structure
of the cyclic nucleotide binding domains in cGK and cAMP kinase, it is conceivable that S2 and S4 interfere with this interdomain region in
the cGK, thereby affecting the cooperativity between the binding sites
A and B.
While the leucine zipper and the autoinhibitory domain control
holoenzyme formation and inhibition of catalytic activity individually, synergistic interactions of both domains are required to develop their
full regulatory potential on cGMP binding. The involvement of the two
domains in regulating affinity for cGMP is intriguing, as both the S2
and S4 domain represent the amino-terminal sequences with highest
homology between cGK I and cGK I
. Thus, the interaction of the
amino terminus with the cGMP binding domains is apparently based on a
conserved structure.
Non-homologous sequences of the cGK I and cGK I
amino termini are
less important for the regulation of binding affinity. For example, the
linker sequence (S3) between the leucine zipper and the autoinhibitory
domain, which is variable in length between cGK I
and cGK I
, is
not crucial for producing the cGK I
or cGK I
phenotype. Moreover,
this linker even exerts a reverse effect as it significantly decreases
affinity when it derives from cGK I
and increases affinity when it
derives from cGK I
. Specific residues in the linker may contribute
to this effect. Other non-homologous amino-terminal domains, such as S1
and S6, only negligibly affect the affinity of binding site A. The
domain S5, however, which has been regarded as a part of the hinge
domain, still exerts a considerable, but less pronounced effect than S2 and S4. Its contribution to the phenotype suggest that it may be
attributed structurally and functionally to the preceding
autoinhibitory domain.
The molecular basis underlying the effect of the leucine zipper and the
autoinhibitory domain is unclear at the moment. The nature of their
interactions with the binding site may depend on the net charge of the
leucine zipper. Both the cGK I and cGK I
zipper are interspersed
with positive and negative amino acid residues. The negative charges
predominate in the cGK I
zipper and the positive charges in the cGK
I
zipper (Fig. 2). Thus, the electrostatic interactions resulting
from a positioning of the zipper near the cGMP binding domain would be
different in cGK I
and cGK I
. The leucine zipper is thought to
form an
-helix that allows the dimerization of two cGK subunits (2,
9-11). The absolute length of the leucine zipper and, thus, the length of the putative helix are not crucial for the interaction with the
binding site, since two additional helical turns provided by S1 of cGK
I
do only marginally affect the high or low affinity cGK phenotype.
The additional need for the autoinhibitory domain (S4) may also be
explained by structural interactions. Autoinhibition of cGK I
is
achieved by a substrate-like sequence including the autophosphorylation
site -Thr59- (12), and that of cGK I
is achieved by the
autoinhibitory pseudosubstrate sequence -75KRQAI- (31). It
is conceivable that this unique features of cGK I
and cGK I
have
influence on the interaction between the autoinhibitory and catalytic
domain. As both domains are located on the same polypeptide chain,
having the cGMP binding domain in between, different interactions
between autoinhibitory and catalytic domain also differently affect the
structure of the binding pockets. Consequently, the affinity for cGMP
may be increased and decreased through the autoinhibitory domains of
cGK I
and cGK I
, respectively.
The alteration of the primary structure in the various chimeras changed
the structural context of the autoinhibitory domain (S4). Nevertheless,
all of the chimeras exhibited less than 10% basal activity, regardless
whether they contained the S4 from cGK I or cGK I
. This suggests
that both S4 domains included the essential sequence necessary for
autoinhibition. Only a cGK I
deletion mutant, which lacked the
sequence Thr59-Thr60-Arg61 of S4
had a 75% basal activity,2 thereby
confirming that the in vitro autophosphorylated
Thr59 (12) is essential for autoinhibition of cGK I
. The
in vitro autophosphorylated Ser64 of cGK I
(13), however, is apparently not involved in autoinhibition. This site
resides in S3 of cGK I
and is not present in S3 of cGK I
. The
exchange of S3 from cGK I
by that from cGK I
does not yield a cGK
with increased basal activity.
Finally, the identification of specific amino acid sequences that
regulate the affinity for cGMP may help to find analogues sequences in
other cyclic nucleotide regulated proteins. A specific -helical
segment that regulates the affinity of the cGMP binding site has
already been characterized in cGMP-regulated ion channels (32). This
segment shows considerable homology to part of the leucine zipper of
the cGK.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBankTM/EMBL Data Bank with accession number(s) Y08961[GenBank].
We thank L. Koblitz and Stefan Beck for technical help.