(Received for publication, April 17, 1997)
From the Department of Pharmacology and Neuroscience Program, University of Colorado Health Sciences Center, Denver, Colorado 80262
Ca2+-sensitive adenylyl cyclases are key integrators of Ca2+ and cAMP signaling. To selectively probe dynamic changes in [Ca2+]i at the plasma membrane where adenylyl cyclases reside, a full-length, Ca2+-inhibitable type VI adenylyl cyclase/aequorin chimera has been constructed by a two-stage polymerase chain reaction method. The expressed adenylyl cyclase/aequorin chimera was appropriately localized to the plasma membrane, as judged by biochemical fractionation and functional analysis. The chimera retained full adenylyl cyclase activity and sensitivity to inhibition by physiological [Ca2+]i elevation. The aequorin portion of the chimeric construct was also capable of measuring changes in [Ca2+] both in vitro and in vivo. When the plasma membrane-tagged aequorin and cytosolic aequorin were compared in their measurement of [Ca2+]i, they showed contrasting sensitivities depending on whether the [Ca2+]i originated from internal stores or capacitative entry. This is the first full-length enzyme-aequorin chimera that retains the full biological properties of both aequorin and a Ca2+-sensitive adenylyl cyclase. This novel chimeric Ca2+ sensor provides the unique ability to directly report the dynamics of [Ca2+]i that regulates this Ca2+-sensitive enzyme under a variety of physiological conditions. Since this chimera is localized to the plasma membrane, it can also be used to assess local changes in [Ca2+]i at the plasma membrane as distinct from global changes in [Ca2+]i within the cytosol.
The jellyfish protein aequorin with its co-factor, coelenterazine, has been used for over 3 decades to measure cytosolic calcium ([Ca2+]i)1 (1). Although microinjection of the purified protein in early studies made its use technically challenging, the dynamic range of aequorin has always recommended it for studies of physiological transitions in [Ca2+]i (1). The cloning of one of the members of this family of proteins made it possible to transiently transfect cells with cDNAs encoding aequorin and thereby circumvent the need for microinjection (2, 3). Recently, aequorin has been creatively applied to the measurement of Ca2+ in discrete cellular subdomains, such as the near mitochondrial membrane (4-6) and the endoplasmic reticulum (7, 8), by creating chimeras of targeting sequences of intracellular organelle marker proteins with aequorin. This offers a clear advantage of selective measurement of [Ca2+]i in various subcellular domains over the current alternative of fluorescent dyes that are distributed uniformly in the cytosol.
Ca2+-sensitive adenylyl cyclases (even when transfected heterologously) are directly regulated by changes in [Ca2+]i (9, 10). The regulation by Ca2+ is due predominantly to capacitative Ca2+ entry; Ca2+ released from internal stores or nonspecific entry of Ca2+ via ionophore is unable to regulate Ca2+-sensitive adenylyl cyclases (11, 12). However, somewhat paradoxical is the fact that measurements of global cytosolic [Ca2+]i, using the fluorescent Ca2+-indicator fura-2, reveal that Ca2+ released from internal stores or nonspecific entry of Ca2+ via ionophore reports far higher levels of [Ca2+]i than are achieved by capacitative Ca2+ entry (11, 12). Such findings have led us to propose that Ca2+-sensitive adenylyl cyclases may be in close proximity to capacitative channels and that the actual [Ca2+]i that the enzyme encounters may be significantly different from cytosolic [Ca2+] (12). Therefore, in the present study, we have constructed a full-length chimera of aequorin and the Ca2+-inhibitable type VI adenylyl cyclase (ACVI) with the purpose of measuring [Ca2+]i, under a variety of physiological conditions, in the microdomain in which Ca2+-sensitive adenylyl cyclases reside.
Thapsigargin was from LC Services Corp. (Woburn,
MA). Forskolin and RO 20-1724 were from Calbiochem.
[2-3H]Adenine, [3H]cAMP, and
[-32P]ATP were obtained from Amersham Corp.
Coelenterazine was from Molecular Probes, Inc. (Eugene, OR). Other
reagents were from Sigma.
The mammalian expression vector pcDNA3
(Invitrogen, San Diego, CA) was used to express mouse ACVI, HA1-tagged
aequorin (cytAEQ), and the ACVI- and HA1-tagged aequorin (ACVI/AEQ)
chimera. The construction of ACVI, which contains the coding sequence
of mouse type VI adenylyl cyclase in pBluescript II SK() (Stratagene, La Jolla, CA), was described previously (13). This cDNA was excised
with KpnI and BamHI and ligated into pcDNA3.
The cytAEQ cDNA in pBS(+), kindly provided by Dr. R. Rizzuto
(University of Padova, Padova, Italy) (4), was released by
SalI and EcoRI digestion and ligated into
pBluescript II SK(
). This cDNA was excised with XbaI
(a flanking vector site) and introduced into pcDNA3. The
cytomegalovirus promoter of pcDNA3 directs expression of the
protein.
ACVI/AEQ
chimeric cDNA was engineered by two-stage PCR, combining two
separate PCR products with overlapping sequence into one longer product
(Ref. 14; see Fig. 1). The sequences of the oligonucleotide primers
were as follows: primer 1, ctcgctcgcccaTGACCAGGTCGGCCGCTCACACAT, 5
sense to ACVI including an AspI restriction site located at nucleotide 3235 (uppercase letters) and the sequence of the pDirect vector (CLONTECH Laboratories, Palo Alto, CA)
(lowercase letters); primer 2, ataatcaggaacatcataACTGCTGGGGCCCCCGTT, 3
antisense to ACVI, including the C terminus (uppercase letters),
overlapped with cytAEQ (lowercase letters); primer 3, AACGGGGGCCCCAGCAGTtatgatgttcctgattat, 5
sense to cytAEQ, including the
N terminus (lowercase letters), overlapped with ACVI (uppercase
letters), complementary to primer 2; primer 4, ctggttcggcccaTTAGGGGACAGCTCCACCGTAGA, 3
antisense to cytAEQ (uppercase
letters) and the pDirect sequence (lowercase letters). The first stage
PCR generated two fragments separately. One was the 3
segment of ACVI
cDNA encoding amino acids 1045-1165, including the AspI
restriction site, and also cDNA encoding the first 6 amino acids of
cytAEQ, obtained by using primers 1 and 2 and ACVI cDNA as the
amplification target. The other was the cytAEQ cDNA encoding the
whole photoprotein and cDNA encoding the last 6 amino acids of
ACVI, obtained by using primers 3 and 4, and cytAEQ cDNA as the
amplification target. In the second stage PCR, after gel purification,
the two overlapping fragments were mixed in equimolar amounts along
with primers 1 and 4. The PCR construct was subcloned into pDirect
(CLONTECH), and the sequence was confirmed by DNA
sequencing, using the dideoxynucleotide chain termination method with
modified T7 DNA polymerase (Sequenase, U.S. Biochemical Corp.) (15).
The PCR construct was then excised by double digestion with
AspI and XbaI and ligated with the 5
part of
ACVI cDNA in the pBluescript II SK(
) vector, cut with AspI and XbaI. Finally, the ACVI/AEQ chimeric
cDNA was excised from pBluescript II SK(
) with EcoRI
and XbaI and introduced into pcDNA3.
Cell Culture and Transfection of HEK 293 Cells
HEK 293 cells maintained as described previously (12) were transiently transfected at about 50% confluency, using the calcium phosphate method (16) with 26 µg of plasmid DNA for each transfection. Seventeen hours after transfection, the cells were harvested and replated either (i) onto 24-well culture plates and incubated for 2 days before cAMP measurements were made, or (ii) onto 100-mm culture plates for 48 h for aequorin measurements.
Cell FractionationHEK 293 cells, transfected with vector
alone, ACVI, or ACVI/AEQ plasmids, were lysed using a modification of a
glycerol-stabilized lysis method (17, 18). The lysate from six
75-cm2 flasks was adjusted to 3% sucrose, layered on top
of linear sucrose gradients (5-45%), and centrifuged at 27,000 rpm (1 h, 4 °C, SW28 rotor; Beckman Instruments, Palo Alto, CA). The
gradients were fractionated and analyzed for adenylyl cyclase activity
(19, 20), marker enzymes (cytochrome c oxidase,
mitochondria; 5-nucleotidase, plasma membrane; lactate dehydrogenase,
cytosol; Refs. 21 and 22), protein concentration, and sucrose.
Adenylyl
cyclase activity was measured in aliquots of the fractionated gradients
as described previously (19, 20). The incubation mixture included 100 mM Tris, pH 7.6, [-32P]ATP (1 × 106 cpm), 5 mM phosphocreatine, 80 units/ml
phosphocreatine kinase, 0.1 mM ATP, 2.0 mM
MnCl2, 0.1 mM 5
-cyclic-AMP, 1.0 mM
3-isobutyl-1-methylxanthine, 4 µM forskolin.
The Ca2+ sensitivity of transfected adenylyl cyclases was determined as described previously (19, 20). Purified membranes from transfected HEK 293 cells or crude cerebral cortical membranes (4-10 µg of protein/assay) were incubated (30 °C, 30 min, 100 µl) in a mixture similar to that described above but also containing a range of (60 µM) EGTA-buffered free [Ca2+] at pH 7.4. Adenylyl cyclase activity in HEK 293 cells was measured in the presence of 10 µM prostaglandin E1 and 20 µM forskolin, while cortical AC activity was assayed with only 1 µM calmodulin. Data (mean ± S.E. of triplicate determinations) are expressed in pmol/min and are corrected by the concentration of protein.
cAMP accumulation in intact cells was measured as described previously (11, 12) with some modifications. The ATP pool of HEK 293 cells on 24-well plates was labeled in minimal essential medium (60 min, 37 °C) with [2-3H]adenine (1.5 µCi/well). After washing, the cells were incubated with a nominally Ca2+-free Krebs buffer (12). The use of Ca2+-free Krebs buffer in experiments denotes the addition of 0.1 mM EGTA to the nominally Ca2+-free Krebs buffer. All experiments were carried out at 37 °C in the presence of the phosphodiesterase inhibitors 3-isobutyl-1-methylxanthine (500 µM) and Ro 20-1724 (100 µM), which were preincubated with the cells for 10 min prior to a 1-min assay. Assays were terminated, and samples were processed as described previously (23).
Aequorin MeasurementsAequorin measurements were carried
out 48 h after transfection. The in vitro
[Ca2+] calibration curves were determined by exposing
cell lysates of HEK 293 cells transfected with either the cytAEQ or
ACVI/AEQ plasmids to solutions of EGTA-buffered [Ca2+] as
follows. The cells were resuspended in a lysis buffer containing 100 mM HEPES, 0.2 mM EGTA, 0.2% bovine serum
albumin, 0.3 mM phenylmethylsulfonyl fluoride, protease
inhibitor mixture as described previously (24), pH 7.0, and lysed
through a 25-gauge syringe. Cell membranes were separated from cell
lysate by centrifugation (20 min, 4 °C, 12,000 rpm, Sorvall SS34).
The supernatant (cell lysate) was utilized for in vitro
calibration of cytAEQ, and the cell pellet (cell membranes) was
utilized for in vitro calibration of ACVI/AEQ. Apoaequorin
was reconstituted with 5 µM coelenterazine (Molecular Probes) for 2-2.5 h at 4 °C in the presence of 1%
-mercaptoethanol. Cell lysate or cell membranes were diluted into an
intracellular-like buffer containing the following final
concentrations: 130 mM KCl, 10 mM NaCl, 1 mM MgSO4, 0.5 mM
K2HPO4, 1 mM ATP, 0.2% bovine
serum albumin, 100 mM HEPES, pH 7.0, at 22 °C,
supplemented with 0.1 mM EGTA and different amounts of
total CaCl2, as in the cyclase assays, above.
Ca2+ buffers of various concentrations were injected into
the chamber, and the light emission was recorded. For in
vivo measurements, transfected cells were washed and loaded with 5 µM coelenterazine (Molecular Probes) in minimal essential
medium supplemented with 1% fetal calf serum and 0.5 mM
EGTA for 2 h at 37 °C in 5% CO2 in the dark. Cells
were resuspended in nominally Ca2+-free Krebs buffer at pH
7.4 supplemented with 75 µM Ca2+, and 0.2 mM EGTA was added immediately prior to measurements. Aequorin luminescence was measured in a Berthold Lumat LB 9501 luminometer (EG & G Berthold, Bad Wildbad, Germany) and collected in
ascii format on an 80486 microprocessor. The
Lmax was found by integrating a continuous
recording of aequorin-mediated light emission from transfected cells in
the presence of 0.2% Triton X-100 and 10 mM
Ca2+ until the light emission returned to basal levels, and
subsequent injections of Triton X-100 and Ca2+ were unable
to produce further increases in light emission. This back-calculation
strategy also corrected for any variability in the degree of expression
of the cytAEQ or ACVI/AEQ proteins (6). Luminescence values determined
in vivo were transformed into [Ca2+]i
values using an algorithm based on the reordering of an equation (25)
described by Allen et al. (26). The light output from
transfected cells loaded in the absence of coelenterazine or from
mock-transfected cells loaded with coelenterazine was not significantly
higher than background.
The design chosen for the adenylyl cyclase/aequorin chimera was to
place aequorin, linked at the N terminus with the HA1-epitope, downstream of the murine ACVI. It seemed that the adenylyl cyclase might tolerate the inclusion of a large addition at the C-terminal, since the Drosophila ACI homolog carries a substantial
C-terminal addition with no deleterious consequence for its activity
(27). In addition, substitution at the N terminus of aequorin does not affect its activity, whereas substitution at the C terminus eliminates Ca2+ binding (28). The strategy used to construct the
full-length ACVI-aequorin chimera was first to adopt a two-stage PCR
protocol to ligate the 3 356 base pairs of the ACVI (13) upstream of, and in frame with, HA1-aequorin (25). This initial construct could then
be readily ligated to the major 5
component of the ACVI cDNA.
After confirming the desired sequence of the cloned PCR fusion product,
the partial cyclase/aequorin chimera was religated to the remainder of
the ACVI cDNA (Fig. 1), and the full-length ACVI/AEQ
cDNA was expressed in HEK 293 cells.
Since the chimera included the full coding sequence of an adenylyl
cyclase, it was possible to determine whether a functional adenylyl
cyclase activity was expressed and to evaluate the subcellular localization of any such activity. Thus, vector alone (control), ACVI,
and ACVI/AEQ were transfected into HEK 293 cells. Cells were lysed and
fractionated on a continuous sucrose gradient (Fig. 2),
and adenylyl cyclase and marker enzyme activities were assessed. Both
wild type- and ACVI/AEQ-transfected cells showed exactly the same
sedimentation profile as control-transfected cells, although the
adenylyl cyclase-transfected cells had 3-fold higher activities than control-transfected cells (Fig. 2). Adenylyl cyclase activity sedimented with 5-nucleotidase activity (an enzyme known to reside at
the plasma membrane) and at a lighter density than the mitochondrial fraction (indicated by cytochrome c oxidase activity), which
supports its appropriate plasma membrane localization.
The foregoing biochemical fractionation data indicate appropriate
subcellular localization of the adenylyl cyclase chimera by gross
criteria and apparently unchanged overall
activity.2 The next series of experiments
asked whether the full functional properties associated with ACVI were
retained in the chimera. ACVI is an adenylyl cyclase that is inhibited
by [Ca2+] in the submicromolar range in vitro
and inhibited strictly by capacitative Ca2+ entry in
vivo (11-13, 24). The fractions corresponding to the peak
adenylyl cyclase activity from sucrose gradients (viz.
32-44% sucrose; see Fig. 2) were pooled and sedimented, and their
response to a range of EGTA-buffered Ca2+ concentrations
was evaluated. When the adenylyl cyclase activity in plasma membranes
from either the ACVI- or the ACVI/AEQ-transfected HEK 293 cells are
compared, it is clear that a significant (approximately 3-fold)
increment in activity is obtained over that of cells transfected with
vector alone (Fig. 3A). Strikingly, the
chimeric adenylyl cyclase shows exactly the same concentration
dependence for inhibition by Ca2+ as the unmodified ACVI.
Maximal inhibition of approximately 40% is observed at 1 µM Ca2+. As an internal control for the free
[Ca2+] established in this assay, cerebral cortical
membrane adenylyl cyclase activity was assayed and yielded a
stimulatory response over the same range of free [Ca2+],
in keeping with the predominance of Ca2+-stimulated
isoforms of adenylyl cyclase in this tissue (9, 10).
The critical biological hallmark of Ca2+-sensitive adenylyl cyclases, whether they are stimulated or inhibited by [Ca2+]i, is a strict dependence on capacitative Ca2+ entry for their regulation in vivo, rather than release from calcium stores or nonspecific cytosolic elevation of [Ca2+]i (9-12). Therefore, the most discriminating test of the appropriate cellular localization of the chimeric construct was whether it retained sensitivity to capacitative Ca2+ entry. Thus, the sensitivity to capacitative Ca2+ entry of the chimeric construct and the native ACVI was compared (Fig. 3B). HEK 293 cells transfected with control vector, ACVI, or ACVI/AEQ were treated with the sarco-/endoplasmic reticulum calcium ATPase inhibitor, thapsigargin, in the absence of extracellular Ca2+ to deplete inositol 1,4,5-trisphosphate-sensitive internal stores and to activate capacitative Ca2+ entry mechanisms. Such a regimen has been shown previously to generate robust capacitative Ca2+ entry, which in turn regulates both endogenous and transfected Ca2+-sensitive adenylyl cyclases (11, 12). When a range of extracellular [Ca2+] were introduced to the bathing medium to initiate capacitative Ca2+ entry and the cAMP response measured, it was clear that both ACVI- and ACVI/AEQ-transfected cells showed the same (approximately 40%) maximal inhibition (Fig. 3B). In the presence of 4 mM Ca2+, the activity of ACVI and ACVI/AEQ was reduced to 61 ± 7.9% and 58 ± 4%, respectively (Fig. 3B). These values were not statistically different from each other but were quite clearly statistically different from controls, p < 0.005. These data then unequivocally establish the cellular disposition of the chimeric adenylyl cyclase and predict its probable ability to serve as a sensor of Ca2+ at the plasma membrane.
The final series of experiments evaluated the ability of the adenylyl
cyclase/aequorin chimera to measure [Ca2+], both in
vitro and in vivo compared with cytAEQ. An in
vitro calibration curve of cytAEQ and ACVI/AEQ luminescence was
fitted to EGTA-determined free [Ca2+] (Fig.
4). After a crude separation of cell lysate from the
cell membrane fraction for use in the in vitro calibration
assay, 89% of the cytAEQ luminescence activity was in the cell lysate,
while 89% of the ACVI/AEQ luminescence activity was in the cell
membrane fraction. The results shown in Fig. 4 demonstrate that the
aequorin portion of the ACVI/AEQ chimera is active and can measure a
range of [Ca2+]free in vitro. The
relationship between [Ca2+] and
log(L/Lmax) for ACVI/AEQ was not
significantly different from that of cytAEQ. Together, these data
indicate that fusing the type VI adenylyl cyclase to the N terminus of
aequorin does not significantly alter its Ca2+-sensing
activity. Nonlinear curve fitting of the untransformed data yielded
macroscopic Kd values of 1.71 × 105 M and 1.78 × 10
5
M and Hill coefficients of 2.38 and 2.45 for cytAEQ and
ACVI/AEQ, respectively.
To investigate whether the plasma membrane-targeted ACVI/AEQ could
report changes in [Ca2+]i in vivo, the
luminescence of ACVI/AEQ and cytAEQ were evaluated in transfected HEK
293 cells under conditions of agonist-induced Ca2+-release
from intracellular stores or capacitative Ca2+ entry (Fig.
5). The addition of the muscarinic cholinergic agonist, carbachol, to a suspension of HEK 293 cells transfected with either the
cytAEQ or the ACVI/AEQ in the absence of extracellular Ca2+
produced a rapid rise (<3 s to peak) in aequorin luminescence, which
was converted to [Ca2+]i (Fig. 5A).
Luminescence did not increase in response to carbachol in cells where
the intracellular Ca2+ stores had been depleted by a
pretreatment with 200 nM thapsigargin, which demonstrated
that the aequorin-expressing cells were actually measuring increases in
[Ca2+]i due to Ca2+ release (data not
shown; these conditions mimic earlier treatments of these cells using
fura-2 to measure [Ca2+]i; Ref. 12). Cells
transfected with the cytosolically distributed cytAEQ reported a
significantly larger peak in [Ca2+]i upon
stimulation by 200 µM carbachol than cells expressing the
plasma membrane-bound ACVI/AEQ: 2105 ± 364 nM
versus 1160 ± 136 nM (n = 3), respectively (Fig. 5A). The likely explanation for this
difference is the disparate location of the two Ca2+
sensors within the cell, since there was no difference in their ability
to measure Ca2+ in vitro (Fig.
4).3 The plasma membrane-bound
Ca2+-sensor (ACVI/AEQ) detects Ca2+, which is
released from intracellular stores at a level that is considerably
lower than the levels detected by the uniformly cytosolically
distributed Ca2+-sensor (cytAEQ), presumably due to the
reuptake into intracellular Ca2+ stores and the activation
of Ca2+ extrusion mechanisms. If this explanation is
correct, it would be predicted that the plasma membrane-bound ACVI/AEQ
would detect higher [Ca2+]i during
Ca2+ entry than the cytosolic aequorin. To evaluate this
prediction, capacitative Ca2+ entry was activated in HEK
293 cells transfected with either the ACVI/AEQ or the cytAEQ by adding
4 mM Ca2+ to cells that had been pretreated
with 100 nM thapsigargin for 10 min in nominally
Ca2+-free buffer (Fig. 5B). In support of the
prediction, cells expressing the plasma membrane-bound ACVI/AEQ
reported a rapid rise in [Ca2+]i, which was
consistently higher than was detected by the cytosolically distributed
cytAEQ: 1225 ± 117 nM versus 660 ± 61 nM (n = 5), respectively.
By a wide variety of criteria, the presently described adenylyl cyclase/aequorin chimera is fully functional and retains the properties of both parent molecules. The adenylyl cyclase/aequorin is localized in the same cellular regions as native ACVI; the activities of the two forms of the enzyme and their sensitivity to Ca2+, both in vitro and in vivo, are identical. Aequorin, whether free in the cytosol or tagged onto the C terminus of ACVI, has unaltered affinities and properties for responding to Ca2+ and can measure [Ca2+]i. In the intact cell, the ACVI/AEQ reports lower concentrations of released Ca2+ reaching the plasma membrane than does the cytosolically distributed cytAEQ. During capacitative Ca2+ entry, however, the ACVI/AEQ reports a much higher [Ca2+]i at the plasma membrane than is seen by the cytosolically distributed cytAEQ. This finding is quite compatible with the reuptake, buffering, and extrusion mechanisms that are expected to reduce the amount of released Ca2+ reaching microdomains within the plasma membrane and to reduce the amount of Ca2+ reaching the cytosol during Ca2+ influx. These data provide an explanation for the fact that capacitative influx preferentially regulates Ca2+-sensitive adenylyl cyclases rather than Ca2+ released from internal stores, although global increases in [Ca2+]i due to released Ca2+ are much greater (11, 12).
A wide variety of precise information can be anticipated from the application of this construct to a range of physiological situations. For instance, the dynamics of [Ca2+]i in the immediate environment of a Ca2+-sensitive adenylyl cyclase and how this relates to the regulation of the enzyme can be determined. This is impossible to do with a cytosolically distributed Ca2+ indicator. Along with the limitation of not being able to resolve Ca2+ dynamics within a microdomain, cytosolically distributed fluorescent EGTA derivatives such as fura-2 can reach intracellular concentrations of 20 µM and greater. At these concentrations, fluorescent Ca2+-indicators such as fura-2 can actually alter the true kinetics of the response by acting as a Ca2+ buffer (25). However, the concentration of a recombinantly expressed protein would be considerably lower than 1 µM and would not be expected to contribute to the intracellular Ca2+ buffering capacity. In addition, the cyclase chimera can serve as a novel plasma membrane-embedded [Ca2+]i sensor, which, given the extreme dependence of Ca2+-sensitive cyclases on capacitative entry rather than release, could allow the monitoring of a very selective subpool of [Ca2+]i.
We thank Dr. R. Rizzuto for the HA1-aequorin cDNA, Drs. J. W. Karpen and A. Sorkin for useful comments on the manuscript, and Dr. P. H. Cobbold for early suggestions. We also thank Donna Detmar-Hanna and Dr. John G. Gerber for technical assistance and use of the luminometer and Dr. Nicole Mons for unpublished immunohistochemical data.