(Received for publication, July 8, 1996, and in revised form, February 4, 1997)
From the In mammalian tissues two types of
cGMP-dependent protein kinase (cGK) have been identified.
In contrast to the dimeric cGK I, cGK II purified from pig intestine
was shown previously to behave as a monomer. However, recombinant rat
cGK II was found to have hydrodynamic parameters indicative of a
homodimer. Chemical cross-linking studies showed that pig cGK II in
intestinal membranes has a dimeric structure as well. However, after
purification, cGK II was found to be partly proteolyzed into C-terminal
monomeric fragments. Phosphorylation studies in rat intestinal brush
borders revealed that the potency of cGMP analogs to stimulate or
inhibit native cGK II in vitro (i.e.
8-(4-chlorophenylthio)-cGMP > cGMP > Cyclic GMP-dependent protein kinases
(cGKs)1 play an important role in the
signaling of various cGMP-linked hormones and neurotransmitters including nitric oxide (NO), natriuretic peptides, and guanylin (1, 2).
In mammalian tissues two types of cGK have been identified. Type I cGK,
consisting of In the intestine, cGMP is involved in the regulation of ion and water
transport. It inhibits the uptake of NaCl and stimulates the secretion
of Cl Sequence comparison and biochemical analysis revealed a large degree of
similarity in the structural organization of cGK I and II (2-5). Both
isotypes possess two cGMP binding domains on one polypeptide chain
which is covalently coupled to a catalytic domain. Their N terminus
contains an autoinhibitory region, one or more autophosphorylation
sites, and a leucine zipper motif, and both proteins are devoid of
hydrophobic trans-membrane domains. Despite these similarities, cGK II
was shown to differ from soluble dimeric type I cGK in that it behaved
as a membrane- and cytoskeleton-associated protein in intestinal brush
borders and as a monomer following its solubilization and purification
(3). However, recombinant mouse brain cGK II was reported to be soluble
and dimeric after expression in mammalian and insect cells (5, 16). In
contrast, recombinant rat intestinal cGK II was found to be tightly
bound to the membrane after expression in similar cell systems (4, 17,
18), but its oligomerization state was not established. Furthermore,
the Ka values for cGMP and cGMP analogs of
recombinant rat intestine cGK II purified from Sf9 cells (17) differed
considerably from those of recombinant mouse brain cGK II, which
contained an additional N-terminal histidine tag (16).
Since rat and mouse cGK II have very similar sequences, are
endogenously both membrane-bound in intestine, and are activable by
similar concentrations of cGMP after expression in COS-1 cells (5, 18),
the observed kinetic and structural differences among purified cGK II
preparations may reflect artifacts of the expression system and/or
purification rather than fundamental differences. To circumvent such
potential artifacts, we characterized the oligomerization state of cGK
II and its kinetic properties in an environment most relevant for its
physiological functioning, i.e. intestinal brush border
membranes. We also compared the stimulatory or inhibitory potency of
various cGMP analogs toward cGK II in vitro, with their
effects on Cl We report here that the endogenous intestinal membrane-bound type II
cGMP-dependent protein kinase exists as a dimer and
displays a distinct activation profile with respect to cGMP analogs,
which permits discrimination between cGK II and cGK I effects in
physiological processes.
Disuccinimidyl suberate was obtained from Pierce,
and 3-isobutyl-1-methylxanthine was from Sigma. Polyclonal cGK II or
cGK I antibodies, raised against recombinant cGK II or cGK I Confluent HEK-293 or NIH-3T3 cells, stably transfected
with cGK II (18), were washed twice with ice-cold phosphate-buffered saline, scraped with a rubber policeman in buffer A (150 mM
NaCl, 10 mM NaPO4, pH 7.4, 1 mM
EDTA, 100 µg/ml trypsin inhibitor, and 20 µg/ml leupeptin) and
processed directly or frozen in liquid N2 and stored at
Non-vesiculated brush border caps were prepared from jejunum and ileum
of male Wistar rats by vibration of everted intestine in hypotonic EDTA
buffer and low speed centrifugation, essentially as described
previously (20), and were finally resuspended to 0.4 mg of protein/ml
in buffer B.
Brush border membrane vesicles were isolated from everted pig intestine
by freeze-thawing, followed by Mg2+ precipitation according
to Ref. 21. cGK II was solubilized from HEK-cGK II membranes or pig
brush border membrane vesicles by addition of 0.5 M NaCl
and 1% Triton X-100 and subsequently purified by affinity
chromatography on 8-(2-aminoethyl)-amino-cAMP-Sepharose as described
(3).
Gel filtration was performed essentially as
described (22) using Superose 6 HR 10/30 or Superdex 200 HR 10/30
analytical gel filtration columns (Pharmacia Biotech Inc.) equilibrated
and subsequently eluted at 0.2 ml/min with 500 mM NaCl, 10 mM sodium phosphate buffer, pH 7.4, 10 mM EDTA,
and 0.1% Triton X-100. Fractions of 400 µl were collected and
analyzed by immunoblotting or protein kinase activity assays. For each
experiment the column was calibrated using standards.
Sucrose density centrifugation was performed with linear 7.5-25%
sucrose gradients in the same buffer used for gel filtration (22).
Apparent sedimentation coefficients
(sw, 20) were calculated by plotting
the distance from the top of the gradient against the position of the
following standards: catalase (11.3 S), bovine serum albumin (4.9 S),
and cytochrome c (1.9 S).
The molecular mass (m) of cGK I and cGK II was calculated
from: m = 6 Purified pig
cGK II was separated by SDS-PAGE and blotted to polyvinylidene
difluoride ProblottR membrane. Pieces of membrane
containing the 75- and 70-kDa forms of cGK II were cut out separately,
and both protein fragments were N-terminally sequenced by automatic
Edman degradation with a 473A protein sequencer (Applied
Biosystems).
For cross-linking, samples (2 mg of protein/ml) were incubated for 15 min at 0 °C in 10 mM phosphate buffer, pH 7.4, 100 mM NaCl, 10 mM EDTA with or without 0.6 mM disuccinimidyl suberate. The reaction was stopped by
addition of SDS-PAGE sample buffer and cGK II was analyzed by
immunoblotting.
Immunoblotting was
performed as described earlier (23). Immunoreactive proteins were
detected after incubation with cGK II or cGK I antibody (1:3000) by the
enhanced chemiluminescence method (Amersham Corp.) and quantitated by
densitometric scanning (Bio-Rad, model 620).
Protein kinase activity was determined by incubation of the samples
(4-10 µg of membrane protein in case of cGK II or 30 ng of purified
cGK I provided with 10 µg of bovine serum albumin) at 30 °C for
different times in 40 µl of 20 mM Tris/HCl, pH 7.4, 10 mM MgCl2, 5 mM A 1-cm
long segment of rat cecum was removed under light diethyl ether
anesthesia. The muscle layers were stripped off by blunt dissection,
and the mucosa was mounted in an Ussing chamber (0.3-cm2
area exposed) for measurements of short-circuit current
(Isc) across the tissue as described (24). Dose-response
curves were obtained by cumulative additions of agonists or
antagonists, and subsequent measurements of the plateau phase of the
ISC reached 10-20 min after each addition.
The values for EC50,
IC50, and apparent Ka (defined as the
concentration required for half-maximal activation) were determined
from dose-response curves fitted by the program Slidewrite. Km and Hill coefficients were calculated with the
program Enzfitter. Ki values were determined from
Dixon plots at half-maximal concentration of agonist (cGMP), assuming a
single site. The lipophilicities of cGMP analogs were determined by
gradient reversed phase chromatography essentially as described (25) and comparable to common log p values.
Immunoblots demonstrated that recombinant rat
cGK II solubilized from HEK 293 membranes was eluted from gel
filtration as a single peak with a Stokes radius of 6.6 which is
significantly larger than that of purified cGK I
Table I.
Hydrodynamic parameters of recombinant and endogenous cGK II
Departments of Biochemistry,
Laboratory of Clinical
Biochemistry,
-phenyl-1,N2-etheno-8-bromo-cGMP >
-phenyl-1,N2-etheno-cGMP and
Rp-8-(4-chlorophenylthio)-cGMPs > Rp-
-phenyl-1,N2-etheno-8-bromo-cGMPs,
respectively) correlated well with their potency to stimulate or
inhibit cGK II-mediated Cl
secretion across intestinal
epithelium but differed strikingly from their potency to affect cGK I
activity. These data show that the N terminus of cGK II is involved in
dimerization and that endogenous cGK II displays a distinct
activation/inhibition profile with respect to cGMP analogs, which
permits a pharmacological dissection between cGK II- and cGK I-mediated
physiological processes.
and
isoforms, is more generally expressed and
acts as a key regulator of cardiovascular homeostasis (1, 2). In
contrast, type II cGK was described originally as an intestine-specific
form (3). Molecular cloning demonstrated that cGK II is indeed a
distinct gene product expressed predominantly in epithelial cells of
the intestine (4), although its mRNA was also detected in kidney,
bone, and brain (4-7). Its widespread distribution in various areas of
the brain suggests an important role of cGK II in NO/cGMP signaling in
the central nervous system (6).
by activating the cystic fibrosis trans-membrane
conductance regulator, a Cl
channel that is mutated in CF
patients (8, 9). Guanylin, a small peptide derived from a larger
precursor protein released luminally by intestinal epithelial cells,
may function as the physiological activator of the cGMP-mediated
signaling pathway in intestine by activating guanylyl cyclase C (10,
11). Heat-stable enterotoxins secreted by various pathogenic strains of
Escherichia coli mimic the action of guanylin and elicit a
severe secretory diarrhea by hyperactivating guanylyl cyclase C (8, 9).
Electrophysiological and immunolocalization studies have provided
evidence for a key role for cGK II as a mediator of the cGMP-provoked
intestinal Cl
secretion (12-14). The critical role of
cGK II in STa/cGMP-induced diarrhea was recently confirmed by
pharmacological and gene disruption techniques (7, 15).
secretion in rat intestinal epithelium,
presumably a cGK II-mediated process (7, 9, 12-15).
Materials
expressed in E. coli, were prepared as described (12). The
cGK substrate peptide 2A3 (RRKVSKQE) and the Walsh inhibitor peptide
(protein kinase A inhibitor-(5-24)-amide) were synthesized by D. Palm
(University of Würzburg, Germany), and the cGMP analogs were
from Biolog (Bremen). cGK I (primarily the I
isoform; Ref. 13) was
purified from bovine lung as described (19).
80 °C. The cells were homogenized by brief sonication (three
bursts of 3 s, peak-to-peak amplitude 15-20 µm), and a crude
membrane fraction was prepared by centrifugation for 15 min at
20,000 × g. Membranes from HEK-cGK II cells were
resuspended in buffer A (2-3 mg of protein/ml), and membranes from
3T3-cGK II cells (which had lower basal protein kinase activity in cGMP activation assays) were resuspended to 1 mg of protein/ml in buffer B
(20 mM Tris/HCl, pH 7.4, 5 mM
-mercaptoethanol, 2 mM EDTA, 100 µg/ml trypsin
inhibitor, and 20 µg/ml leupeptin).
Nasw, 20/(1
)
where
= viscosity of medium, n = Avogadro's
number, a = Stokes radius,
sw, 20 = sedimentation coefficient,
= partial specific volume, and
= density of the medium. We assumed
for proteins to be 0.73 cm3/g (22).
-mercaptoethanol,
0.1 mM 3-isobutyl-1-methylxanthine, 25 mM
sodium
-glycerophosphate, 200 nM protein kinase A
inhibitor, 0.1 mg/ml cGK substrate peptide 2A3 (RRKVSKQE; Ref. 17), 1 µCi of [
-32P]ATP, and various concentrations of
nonradioactive Mg-ATP and cGMP or cGMP analogs as indicated. The
reaction was started by addition of 10-µl aliquots of the cGK II
preparations to 30 µl of prewarmed incubation buffer and quenched by
addition of 10 µl of 0.5 M EDTA. The samples were
subsequently centrifuged for 10 min at 20,000 × g. The
pellet fraction was resuspended in SDS-PAGE sample buffer for
determination of the autophosphorylation of cGK II (3), and 15 µl of
the supernatant was spotted in duplicate on sheets of P-81
chromatography paper (Whatman). After 4 washes with 1% phosphoric acid
the amount of label incorporated was quantitated with the Molecular
Imaging System GS-363 (Bio-Rad). The phosphotransferase activity of cGK
II as measured in intestinal brush border and 3T3 cell membranes and
the activity of cGK I were linear with the amount of enzyme up to the
concentration used in the kinetic study and with time up to 4 min in
the presence of relatively high ATP concentrations (
300
µM). However, in the presence of low ATP, the cGK II
kinase activity started to deviate from linearity after 2 min. This
nonlinearity was independent of the concentration of cGMP or cGMP
analogs and apparently resulted from ATP depletion caused by the action
of endogenous ATPases (data not shown). The Km of
cGK II for ATP was therefore determined from 2-min incubations, whereas
the kinetic parameters for the cGMP analogs were routinely determined
on the basis of 4-min incubations. The latter condition resulted in
similar kinetic values but higher signal to noise ratios in comparison
with 2-min incubations.
Oligomeric State of Recombinant Rat Intestine and Endogenous Pig
Intestine cGK II
(5.1) eluted under
the same conditions (Fig. 1A and Table
I). Velocity sedimentation analysis likewise revealed
single peaks for cGK II and I
with sedimentation coefficients of 6.8 and 7.2 S, respectively (Table I). From these hydrodynamic parameters
the molecular mass of cGK II was calculated as 190 kDa, more than twice
the monomeric mass of 87 kDa (4), indicating that cGK II most likely
exists as a homodimer in solution. cGK II immunoprecipitation or
purification by cAMP-Sepharose chromatography did not reveal any
associated proteins that could falsify the mass determination (data not
shown, Ref. 18).
Fig. 1.
Gel filtration profiles of crude recombinant
and purified endogenous cGK II. A, solubilized membrane
proteins from HEK 293 cells stably expressing rat intestine cGK II were
mixed with purified bovine cGK I and applied to a Superdex 200 HR gel filtration column. After elution, the presence of cGK was monitored in
each fraction by immunoblotting using specific antibodies against cGK
II and cGK I. B, purified pig intestine cGK II was subjected to gel filtration, and the cGK II content of the fractions eluted was
determined by immunoblotting using a specific antibody against cGK II.
On the left the molecular masses of the immunoreactive forms
of cGK II are indicated in kDa. Fraction numbers are indicated below the panels. BC denotes the original
preparation before it was injected on the column. The elution positions
of standards are indicated for ferritin (Stokes radius,
a = 6.1 nm), catalase (a = 5.2 nm), and
bovine serum albumin (BSA) (a = 3.5 nm).
[View Larger Version of this Image (50K GIF file)]
form) was added to the recombinant cGK II preparations as
an internal standard (Stokes radius = 5.0-5.3 nm, apparent
s20,w = 6.9-7.8 S and Mr = 150,000-178,000; Refs. 3, 13, and 27). Data represent means ± S.E. of three experiments, ND, not determined.
The Stokes radius was determined by gel filtration; the sedimentation
coefficient by sucrose density centrifugation, and the molecular weight
was calculated from these parameters as described under "Experimental
Procedures." Recombinant cGK II was analyzed either directly after
solubilization of membranes from HEK-293 cells stably transfected with
rat cGK II (Recomb. cGK II) or after subsequent purification (recomb.
cGK II pure). cGK II purified from pig intestine was found to consist
of a 86-kDa full-length form and two fragments of approximately 75 and
70 kDa (see Fig. 1). Purified bovine lung cGK I (consisting mainly of
the type 1
form) was added to the recombinant cGK II preparations as
an internal standard (Stokes radius = 5.0-5.3 nm, apparent
s20,w = 6.9-7.8 S and Mr = 150,000-178,000; Refs. 3, 13, and 27). Data represent means ± S.E. of three experiments, ND, not determined.
Sample
Stokes radius
Apparent
s20,w
Mr × 10
3 calculated
nm
S
Recomb. cGK II (rat)
6.6
± 0.2
6.8 ± 0.3
190
Recomb. cGK II
pure
6.4 ± 0.2
7.0 ± 0.2
190
cGK II pure (pig
intestine)
86-kDa form
6.3
± 0.2
ND
75-70-kDa forms
3.5
± 0.1
ND
cGK I pure (bovine
lung)
5.1 ± 0.2
7.2 ± 0.4
157
In previous studies cGK II purified from pig intestine was found to behave as a monomer in gel filtration and sucrose density sedimentation analyses, although kinase activity, not immunoblotting as above, was used for cGK II detection (3). We therefore compared the hydrodynamic parameters of recombinant rat cGK II with those of native pig cGK II after purification. As shown in Table I, purification had no effect on the dimeric state of recombinant rat cGK II. However, purified pig cGK II was recovered in two major peaks after gel filtration as detected by immunoblotting (see Fig. 1B). In the first peak, full-length cGK II (86 kDa) eluted at a position indicative of a dimer. Subsequently, 75- and 70-kDa cGK II fragments eluted at a position corresponding to a Stokes radius of 3.5 nm, similar to that found for monomeric cGK II previously (3). In accordance with the previous study, most of the kinase activity (>90%) was recovered in the second peak (data not shown). The proteins in the second peak were identified by protein sequencing as C-terminal, proteolytic fragments of cGK II. The N-terminal sequences VPLDV and PPEF obtained from the 75- and the 70-kDa form, respectively, matched those of a 75-kDa cGK II fragment beginning at Val101 and of a 70-kDa fragment starting at Pro139 (4).
To exclude the possibility that cGK II dimerization could be an
artifact of the solubilization process, we compared the oligomeric structure of cGK II in membranes with that of solubilized cGK II by
chemical cross-linking. As shown in Fig. 2, addition of the cross-linker disuccinimidyl suberate to pig brush border membrane vesicles or to membranes of HEK cells stably transfected with recombinant cGK II, either before or after solubilization, resulted in
the appearance of a cGK II complex at the position of a dimer (170 kDa)
in SDS-PAGE in all cases, indicating that cGK II is also dimeric in
membranes. The additional cross-linked bands observed at 80- and
210-kDa positions may represent intrachain or multiple interchain
cross-linked forms of cGK II.
Characterization of cGK II Activity in Rat Intestinal Brush Border Membranes
Non-vesiculated brush border caps freshly isolated from
rat intestine were found to offer a suitable model for measuring cGK II
activity in its native membrane environment. They are enriched in cGK
II (0.5-1 µg of kinase/mg of protein; Refs. 3, 21), contain no
detectable levels of cGK I (data not shown, cf. Refs. 3, 9,
and 12), and are fully accessible to ATP and exogenous peptide
substrates (21). Furthermore, cGMP could stimulate the phosphorylation
of exogenous substrate by brush border preparations to a similar extent
(7-fold) as reported for purified recombinant cGK II (Fig.
3; Refs. 16 and 17). The apparent Km of membrane-bound cGK II for the substrate peptide 2A3 was found to be
similar to that of cGK I (0.10 and 0.12 mM, respectively; data not shown). However, endogenous cGK II at saturating cGMP (10 µM) has a relatively high Km for ATP
(0.40 mM) in comparison to cGK I (0.066 mM)
(see Fig. 4A). Moreover, ATP was observed to
decrease the sensitivity of cGK II for cGMP. As shown in Fig.
4B, the apparent Ka for cGMP shifted more
than 10-fold, from 50 nM at 10 µM ATP to 560 nM at 1 mM ATP. A similar shift in the apparent
Ka for cGMP by ATP in the range of 10-100
µM was noticed in case of cGK I. In the following
experiments we routinely determined cGK kinase activity in the presence
of 300 µM ATP, since this condition ensured a relatively
high signal-to-noise ratio and was comparable with that used in similar
studies characterizing purified cGK II (16, 17). At 300 µM ATP using 2A3 as substrate, the apparent
Ka for cGMP of endogenous cGK II was found to be
almost identical to that of purified bovine lung cGK I (0.36 and 0.38 µM, respectively; Fig. 4B and Table
II). Using the same conditions, a similar apparent
Ka for cGMP (0.31 ± 0.03 µM;
n = 3) was found for recombinant membrane-bound cGK II
expressed in NIH 3T3 cells (not shown), indicating that the sensitivity of recombinant rat cGK II for cGMP is similar to that of endogenous rat
cGK II and is not critically dependent on its native environment. However, solubilization of recombinant cGK II with Triton X-100 and 0.5 M NaCl caused a small increase (1.6 ± 0.2-fold;
n = 3) in its apparent Ka for cGMP
(not shown). We determined the Ka for cGMP in
solubilized 3T3-cGK II membranes, because cGK II in solubilized rat
intestinal brush borders was found to be partially converted into
75-70-kDa proteolytic fragments during protein kinase assays (data not
shown), similar to pig cGK II during purification (Fig.
1B).
|
To further characterize native cGK II, we determined the apparent Ka for various cGMP analogs. The membrane-permeant analog 8-pCPT-cGMP appeared 4-5 times more potent than cGMP in activating cGK II, whereas PET-cGMP and 8-Br-PET-cGMP, in comparison with cGMP, were relatively poor agonists, showing 12- and 5-fold higher apparent Ka values, respectively (Fig. 3 and Table II). Both cGMP and 8-pCPT-cGMP, but not the two PET-cGMP derivatives, showed cooperative kinetics, as apparent from their Hill coefficient of 1.5 ± 0.2 (n = 3 for both; not shown).
In addition to stimulating substrate phosphorylation, cGMP was shown previously to increase the autophosphorylation of cGK II in brush border membranes (3). To establish whether both processes have similar kinetic characteristics, we compared their ATP and cGMP analog sensitivity. Surprisingly, under the same assay conditions the Km for ATP of cGK II autophosphorylation was 50-fold lower (5-10 µM) than for 2A3 phosphorylation (data not shown). We therefore determined apparent Ka values for cGMP and cGMP analogs for cGK II autophosphorylation at low concentrations of ATP (5 µM). As shown in Table II, there is a close correlation between the relative potency of the cGMP analogs to stimulate auto- and substrate phosphorylation, although 2A3 phosphorylation required 5-10-fold higher concentrations of cGMP or 8-pCPT-cGMP than self-phosphorylation (in the presence of 2A3) when determined at the same (low) concentration of ATP (data not shown; cf. Table II and Fig. 4B).
Another possible difference between autophosphorylation and substrate
phosphorylation by cGK II may be their response to
Rp-diastereoisomers of phosphorothioate-modified
cGMP analogs. These compounds were previously shown to stimulate
autophosphorylation of cGK II in rat intestinal brush border membranes
(9) but to inhibit substrate phosphorylation by purified recombinant
cGK I and II (16, 17). To exclude the possibility that the different
responses to Rp-cGMPS analogs were caused by
differences in the origin/environment of cGK II (purified recombinant
versus membrane-bound endogenous), we tested these analogs
on cGK II activity in brush border membranes. As shown in Fig.
5 and Table II, both
Rp-8-pCPT-cGMPS and
Rp-8-Br-PET-cGMPS inhibited 2A3 phosphorylation
by endogenous membrane-bound cGK II in a competitive fashion. At a
half-maximal concentration of cGMP, the Ki values
for Rp-8-pCPT-cGMPS and
Rp-8-Br-PET-cGMPS determined from Dixon plots
were 0.15 and 0.45 µM, respectively (Fig. 5).
To allow a direct comparison between the kinetic properties of cGK II
and cGK I, we determined the analog specificity of purified bovine lung
cGK I under the same conditions as used for cGK II. As shown in Fig.
6, pCPT-cGMP was a more potent stimulator of cGK II
compared with cGK I, whereas the 8-Br-PET and PET derivatives of cGMP
were relatively selective activators of cGK I. Similarly, Rp-Br-PET-cGMPS was found to preferentially
inhibit cGK I, whereas Rp-pCPT-cGMPS displayed
an opposite selectivity. Furthermore, the
Rp-cGMPS analogs failed to completely inhibit
the cGMP-stimulated activity of cGK I (maximal inhibition 70%, data
not shown) due to agonistic effects of these partial antagonists
occurring at higher concentrations. A similar agonistic effect of
Rp-pCPT-cGMPS concentrations in the millimolar
range has been reported previously for both cGK I and II (17).
Effects of cGMP Analogs on Cl
To evaluate whether the analog specificity of cGK II
determined in vitro has any relevance for its activation of
physiological substrates in intact epithelial cells, we examined the
effects of membrane-permeant cGMP analogs on intestinal
Cl secretion as monitored by short-circuit current
(ISC) measurements in rat cecum in Ussing chambers. In this
intestinal segment cGK II was found to be colocalized with the cystic
fibrosis trans-membrane conductance regulator Cl
channel
in the apical membrane of the crypt cells, in line with the relatively
large increase in Cl
secretion provoked by the
cGK-specific agonist 8-Br-cGMP in this tissue (12). As expected from
the relatively low apparent Ka of cGK II for
8-pCPT-cGMP, micromolar concentrations of this analog could stimulate
ISC (Fig. 7 and Table II), whereas
8-Br-PET-cGMP and PET-cGMP, relatively poor agonists of cGK II in
vitro, showed EC50 values in the submillimolar range
(Table II). The difference between the cGMP analogs cannot be caused by
a difference in membrane permeability since 8-Br-PET-cGMP and PET-cGMP
have a higher or equal lipophilicity, respectively, compared with
8-pCPT-cGMP (Table II). Furthermore, if the different lipophilicities
of the Rp cGMPS analogs were taken into account,
their inhibitory effects on 8-pCPT-cGMP-stimulated ISC (see
Fig. 7) correlated well with their IC50 values found in
in vitro cGK II protein kinase assays (Table II), further
corroborating the notion that the analog specificity of cGK II in
intact cells and in isolated membranes is very similar.
Type II and type I cGK isoenzymes are likely to play distinct physiological roles, as judged from their differences in tissue expression, subcellular localization, and functional properties in vitro (2-5, 12-18). These different characteristics could in principle be exploited to assess the involvement of a specific cGK isoform in a cGMP-regulated cellular process. However detailed analyses of the molecular properties of cGK II have so far been primarily confined to recombinant and/or purified enzyme preparations, and the results may not be representative for cGK II in its native membrane environment, e.g. due to possible differences in posttranslational modifications, loss of modulatory components, proteolysis or protein oxidation. Indeed pronounced differences in the membrane localization, oligomerization status, and kinetic properties of various preparations of cGK II have been reported (3-5, 16-18). These apparent discrepancies urged us to re-evaluate some of the structural and functional properties of cGK II in its native environment, the apical membrane of intestinal epithelial cells.
Pig cGK II was characterized as a dimer in brush border membranes, as
was soluble mouse cGK II expressed in Sf9 cells (16). The monomeric
behavior of purified pig cGK II described previously (3) appeared to
result from the presence of catalytically active C-terminal 75- and
70-kDa proteolytic fragments in this preparation, most likely generated
during purification by a specific endogenous protease characterized
previously in intestinal brush borders (26). The observation that these
proteolytic forms are monomeric indicates that dimerization of cGK II
is critically dependent on a structural domain located within the first
100 amino acids. This is consistent with a dimerization role for the
leucine zipper motif recognized in this region of cGK II (4). The
leucine-isoleucine repeat is conserved in all cGK sequences examined
(4) and has been shown to dimerize an N-terminal fragment of cGK I
in vitro (27). Both the detergent-solubilized recombinant
and endogenous cGK II observed in the present study were found to have
significantly larger Stokes radii and the recombinant cGK II a smaller
sedimentation coefficient as compared with either cGK I (Table I; 3, 16, 28) or the soluble recombinant mouse cGK II described by Gamm
et al. (16). This suggests that the molecular shape of
solubilized particulate cGK II is more elongated than that of cGK I and
very different from the almost spherical form of soluble mouse cGK II.
However, the primary amino acid sequences of rat intestinal and mouse
brain cGK II (4, 5, correction in Ref. 16) are nearly identical,
suggesting that the higher degree structures of these proteins should
be quite similar. Perhaps their shapes may be influenced by different
N-terminal modifications, since rat intestinal cGK II appears to be
myristoylated (18), whereas recombinant mouse brain cGK II was modified
with an N-terminal histidine tag (16). Certainly these differences may
be one explanation for the membrane localization of endogenous pig and
rat intestinal cGK II (3, 12), as well as rat intestinal cGK II
expressed in HEK 293, Sf9, and COS cells (4, 17, 18), whereas mouse brain cGK II expressed in COS and Sf9 cells remained soluble (5, 16).
Another discrepancy between the various preparations of cGK II is their sensitivity to cGMP and cGMP analogs. Part of these differences might be explained by variations in assay conditions, in particular the ATP concentration, which could shift the apparent Ka of cGK II for cGMP by as much as 10-fold (Fig. 4). A similar modulatory effect of ATP was observed for cGK I (Fig. 4B), in line with previous reports describing downward shifts in binding affinity of cGK I and cAMP-dependent protein kinase type I for cGMP and cAMP, at concentrations of ATP around the Km (29-31). This suggests a similar mode of interaction between the ATP-bound catalytic domain and the regulatory domain of all three cyclic nucleotide-dependent protein kinases, resulting in an apparent competition between ATP and cyclic nucleotides (31). However, the effect of ATP on cGMP affinity in kinase activity measurements was particularly prominent in case of cGK II which displayed an unusually high Km for ATP (400 µM) in comparison to cAMP-dependent protein kinase (5 µM; Ref. 31) and cGK I (66 µM, this study; 20-50 µM; Refs. 29 and 30).
After correction for the different ATP concentrations employed in each assay (using the data plotted in Fig. 4B), the apparent Ka for cGMP of endogenous rat cGK II was found to be approximately 3-fold higher than that of recombinant rat cGK II purified from Sf9 cells (17), however, 3-fold lower than that of either recombinant rat cGK II solubilized from COS-1 cells (18) or histidine-tagged recombinant mouse cGK II purified from Sf9 cells (16), and similar to the apparent Ka for recombinant mouse cGK II expressed in COS cells (5). The lack of consistent or major differences between native cGK II and the various preparations of cGK II obtained after solubilization or purification, suggests that the characteristics of cGK II are not dramatically influenced by the membrane environment. Indeed when tested directly, solubilization of rat cGK II expressed in 3T3 fibroblasts resulted only in a slight (1.6-fold) increase in its apparent Ka for cGMP.
A reasonable but not perfect correspondence was also found between the effects of cGMP analogs on endogenous cGK II and the purified recombinant enzymes described previously. For instance PET-cGMP had a low affinity for both endogenous rat cGK II (4.2 µM; Table II) and for recombinant mouse cGK II purified from Sf9 cells (4.7 µM; Ref. 16). In contrast, PET-cGMP had the same high affinity (0.06 µM) as had cGMP (0.04 µM) for the recombinant rat enzyme purified from Sf9 cells (17). However, the latter preparation, similar to endogenous rat cGK II, was stimulated by relatively low levels of 8-pCPT-cGMP (3.5 and 80 nM, respectively; Ref. 17). Furthermore, the endogenous rat enzyme, similar to recombinant rat and mouse cGK II purified from Sf9 cells, was inhibitable by RP analogs of cGMPS.
The inhibition of exogenous substrate phosphorylation by
Rp analogs clearly differs from their
stimulatory effects on cGK II autophosphorylation in brush border
membranes (9). The apparent Ka for cGMP was also
different for substrate- and autophosphorylation, the latter being
5-10-fold lower. Since the apparent Ka of cGMP for
autophosphorylation was close to the Kd of cGMP
determined previously for the high affinity site of cGK II (3), it is
tempting to suggest that autophosphorylation is stimulated by occupancy
of only one of the cyclic nucleotide binding sites available
(i.e. the "high affinity" or "slow" site), whereas substrate phosphorylation apparently requires both sites for optimal activity. Whether autophosphorylation has any effect on the kinetic properties of cGK II as described for cGK I and cGK I
(32, 33)
remains to be investigated. However, a substantial influence of
autophosphorylation on the kinetic parameters determined in this study
is unlikely considering the poor stoichiometry of cGK II
autophosphorylation within the relatively short incubation periods used
(3) and the observed linearity of the kinase activity with time,
i.e. the absence of a lag time (see "Experimental
Procedures").
The correlation observed here between the effects of cGMP analogs on
substrate phosphorylation and on intestinal Cl secretion
indicates that the characteristics of cGK II determined in
vitro are relevant for its functioning under physiological conditions. This implies that the cGMP analog specificity of cGK II,
which is clearly distinct from that of cGK I (according to this and
previous studies; see refs. 14, 16, 17, 34, and 35), can fortuitously
be used to discriminate between the functional effects of each isotype
both in in vitro assays and in intact cells. A potency order
8-pCPT-cGMP
8-Br-PET-cGMP > PET-cGMP would be diagnostic
for cGK II, whereas the order 8-Br-PET-cGMP
PET-cGMP > 8-pCPT-cGMP would indicate a role for cGK I (
or
).
Finally our observation that Rp-cGMPS analogs,
with some degree of specificity, can inhibit cGK II-controlled
functions in intact tissues, as well as cGK I action described earlier
(36, 37), indicates that these compounds are more generally applicable as tools to demonstrate the involvement of cGK I or II in specific physiological processes. For example, in a separate study of the mechanism of action of heat-stable enterotoxin (STa) in rat intestinal epithelium, Rp-8-pCPT-cGMPS appeared also able
to inhibit the STa-induced Cl secretion (15). These
results, together with recent cGK II knockout mouse data (7),
strengthen the concept that in native intestine, cGK II plays an
important role in the salt and water secretion initiated by activation
of guanylyl cyclase C by STa, or guanylin, the putative endogenous
ligand of GC-C (9-11).