Dynamic Imaging of Endoplasmic Reticulum Ca²⁺ Concentration in Insulin-Secreting MIN6 Cells Using Recombinant Targeted Cameleons: Roles of Sarco(endo)plasmic Reticulum Ca²⁺-ATPase (SERCA)-2 and Ryanodine Receptors

BODIPY-TR-X ryanodine, Oregon Green 488 BAPTA-1 dextran, and Alexa 568/Texas red anti-goat secondary antibody were obtained from Molecular Probes (Eugene, OR). Goat polyclonal anti-SERCA2 and anti-SERCA3 antibodies were from Autogen Bioclear U.K. (Wiltshire, U.K.). Mouse monoclonal anti-RyR antibody and rabbit polyclonal anticalreticulin antibody were obtained from Upstate Biotechnology (Lake Placid, NY) and Sigma (Poole, Dorset, U.K.), respectively. Oligonucleotides were purchased from Cruachem (Glasgow, U.K.). Optimem I serum-free medium and LipofectAMINE were from Gibco (Life Science Research, Paisley, U.K.). All other chemicals and tissue culture materials were from Sigma.

MIN6 and INS-1 pancreatic β-cells were cultured in Dulbecco’s modified Eagle’s medium (DMEM) and RPMI 1640 tissue culture medium supplemented with 15% (vol/vol) and 10% (vol/vol) fetal calf serum, respectively, 100 units/ml penicillin, and 0.1 mg/ml streptomycin at 37°C in an atmosphere of humidified air (95%) and CO₂ (5%) as described previously (43). MIN6 cells were used between passages 19 and 35.

MIN6 cells were seeded at a density of 0.4–0.6 × 10⁶/ml on 24-mm diameter poly-l-lysine-coated coverslips and cultured overnight. Cells were transfected with 1 μg plasmid per coverslip encoding the untargeted yellow cameleon (Ycam-2) or ER-targeted yellow cameleon (Ycam-4ER) and 10 μg/ml lipofectamine in Optimem I medium for 4 h. The cells were cultured for 2–4 days in high-glucose complete growth medium, which was then replaced with a 3 mmol/l glucose-containing DMEM medium 12 h before Ca²⁺ imaging.

Intact MIN6 cells were perifused at a constant rate of 1–2 ml/min with Krebs-Ringer HEPES bicarbonate (KRH) buffer comprising 140 mmol/l NaCl, 3.6 mmol/l KCl, 0.5 mmol/l NaH₂PO₄, 0.5 mmol/l MgSO₄, 2.0 mmol/l NaHCO₃, 3 mmol/l glucose, 10 mmol/l HEPES (pH 7.4), and 0.2 mmol/l EGTA (Ca²⁺-free KRH) or 1.0 mmol/l CaCl₂ equilibrated with O₂/CO₂ (95:5, vol/vol). The ER store was depleted of Ca²⁺ by washing the cells twice with Ca²⁺-free KRH buffer and incubating in Ca²⁺-free KRH buffer containing 10 μmol/l cyclopiazonic acid (CPA) and 2 mmol/l EGTA for 10 min at 4°C. Cells were then imaged at 37°C on an inverted optics Leica DM/IRBE epifluorescence microscope (Leica, Heidelberg, Germany) by using a ×63 PL Apo 1.4 oil-immersion objective. Cells were illuminated with a 100-W mercury arc lamp, and the excitatory light was attenuated with a neutral density filter plus a 440 ± 21 nm excitation filter. The microscope was fitted with a 455-nm long-pass dichroic mirror and two emission filters (480DF30 for ECFP and 535DF25 for EYFP, both from Omega Optical, Glen Spectra, Middlesex, U.K.), alternated with a mechanical filter wheel (Ludl Electronic Products, Hawthorn, NY). Images were acquired on a 12-bit cooled charge-coupled device camera (Hamamatsu C4742-95) every 10–20 s and controlled by Openlab software (Improvision, Coventry, U.K.) running on an Apple Macintosh G3 Powerbook. For in situ calibration of [Ca²⁺], the following equation was used: [Ca²⁺] = K′_d[(R − R_min)/R_max − R)]^l/n, where K′_d is the apparent dissociation constant corresponding to the Ca²⁺ concentration at which R is midway between R_max and R_min, and n is the Hill coefficient. R_max was obtained in the presence of 10 μmol/l of Ca²⁺ ionophore ionomycin and 20 mmol/l CaCl₂. R_min was established by clamping the external Ca²⁺ to zero with 20 mmol/l EGTA in the presence of 10 μmol/l ionomycin. Bleaching was corrected by fitting a simple exponential equation (Microsoft Excel) to the ratio values preceding the first addition of CaCl₂, thus generating the predicted bleaches in the absence of further additions (see below, Fig. 1A, dashed trace, running top-left, bottom-right). Control experiments (without addition of CaCl₂) showed a good approximation of the actual bleaching observed when CaCl₂ was not added (not shown). Corrected traces were then generated by division of the observed ratio values by those predicted by the fitted exponential (see below, Fig. 1B).

Antisense and sense oligonucleotide probes were designed based on the mouse SERCA2 sequence (EMBL accession number AJ 223584), antisense (AS2) 5′-CTGTTTGACACCAGGAGTCATG-3′ (1818–1839) and the corresponding sense control (S2) 5′-CATGACTCCTGGTGTCAAACAG-3′, and SERCA3 sequence (EMBL accession number U49394), antisense (AS3) 5′-GAACAGCGTGTGATACAAGC-3′ (1877–1896) and the corresponding sense control (S3) 5′-GCTTGTATTCACACGCTGTTC-3′. The oligonucleotides were phosphothioate-modified on both starting and terminating nucleotides. A total of 5 μmol/l oligonucleotide and 0.3 mg/ml plasmid DNA encoding Ycam-2 or Ycam-4ER in 0.2× Tris EDTA buffer (2 mmol/l Tris, 0.2 mmol/l EDTA, pH 8.0) were centrifuged at 14,000g for 30 min and then comicroinjected into the nuclei of individual MIN6 cells (36). At 12, 24, 48, or 72 h after microinjection, cells were fixed using 4% (wt/vol) paraformaldehyde in 50 mmol/l phosphate buffer (pH 7.2) for 20 min at room temperature and then washed three times with phosphate-buffered saline (PBS). Cells were incubated with blocking solution (5% [vol/vol] fetal bovine serum, 3% bovine serum albumin [wt/vol] in PBS) in the presence of 1% (vol/vol) Triton X-100 for 30 min and then with primary antibodies against SERCA2 or SERCA3 (1:100 dilution in blocking solution) for 1.5 h at room temperature. Cells were washed three times with blocking solution, incubated with a Texas red-conjugated anti-goat antibody (1:250 dilution in blocking solution) for 30 min at room temperature, and washed four times in PBS. Coverslips were mounted in Vectashield mounting medium (Vector Laboratories, Peterborough, U.K.).

Images were captured on a Leica TCS-NT confocal laser-scanning microscope attached to a DM IRBE epifluorescence microscope using a ×63 PL Apo 1.4 NA oil-immersion objective. The antisense-treated cells expressing cameleons were identified by exciting at 488 nm the EYFP moiety of the cameleon and using fluorescein isothiocyanate filters for fluorescence emission. The immunostaining of the same cells was visualized by exciting at 568 nm and using tetramethylrhodamine isothiocyanate filters for fluorescence emission. Under the imaging conditions used, there was no detectable bleed-through of fluorescence from one channel to the other when we studied single-labeled specimens. The fluorescence intensity of SERCA2 or SERCA3 immunostaining in antisense- or sense-treated cells was quantified with the Leica software and compared with that of the neighboring untreated cells.

Oregon Green 488 BAPTA-1 dextran (2.5 mg/ml) in Tris buffer (pH 8.0) was microinjected into MIN6 cells in the absence (control) or presence of 1 mmol/l caged cADPr, 250 μmol/l caged IP₃, or 2 mg/ml caged nitrophenyl (NP)-EGTA. Where indicated, the ER/Golgi store was depleted of Ca²⁺ by washing the cells twice with Ca²⁺-free KRH buffer and then incubating in Ca²⁺-free KRH buffer containing 1 μmol/l thapsigargin and 2 mmol/l EGTA for 10 min at 4°C. Calcium transients were recorded 2–4 h after microinjection at 37°C on a Leica TSC-SP2 confocal laser-scanning microscope attached to a Leica DM IRBE inverted epifluorescence microscope using a ×63 PL Apo 1.4 NA oil-immersion objective. Oregon Green BAPTA-1 dextran was excited at 488 nm (argon laser). Fluorescence was detected at wavelengths longer than 515 nm (using a long-pass cutoff filter at this wavelength). A second laser (Coherent), which provided light at wavelengths of 351 and 364 nm (via the objective lens), was used for photolysis of the caged compounds. This uncaging light was applied for 1 s. Images were acquired before and after the ultraviolet (UV) pulse at 2.5- to 5-s intervals.

MIN6, INS-1, and primary mouse islet β-cells were incubated with 10–100 nmol/l BODIPY TR-X ryanodine (absorbance/emission = 589 nm/616 nm) in KRH buffer for 10–15 min at 37°C. The binding specificity of the dye was tested by incubating the cells with 250 μmol/l nonfluorescent ryanodine for 15 min before adding BODIPY TR-X ryanodine. After loading, cells were washed five times to remove unbound ligand, and confocal imaging was then performed as outlined above. For quantification of BODIPY TR-X fluorescence in ryanodine-pretreated or untreated cells, the microscope settings/conditions were kept constant.

We expressed Ycam-2, a cytosolic protein, and Ycam-4ER, which is targeted to the ER, in MIN6 pancreatic β-cells. These cameleon variants differ in both their localization signals and Ca²⁺ affinity. Thus, the latter contains an NH₂-terminal sequence encoding the signal peptide from calreticulin to achieve cotranslational insertion into the ER plus a four-amino acid ER retention signal (KDEL) at its COOH-terminus (37). Moreover, Ycam-4ER has a lower affinity for Ca²⁺ than the untargeted Ycam-2 to permit measurement of the expected higher Ca²⁺ levels within the ER. Expressed by transient transfection of MIN6 cells with the corresponding plasmid, the distribution of Ycam-2 protein was first examined. Ycam-2 was uniformly distributed in the cytosolic compartment but was excluded from the nucleus (Fig. 2A). By contrast, Ycam-4ER, examined in the same way, was localized to a reticular structure, consistent with targeting to the ER lumen (Fig. 2B). The fluorescence intensity of cells expressing Ycam-4ER was considerably lower than that of cells expressing Ycam-2, although the reasons underlying this difference are unknown.

Plasmid DNA encoding the cameleons could also be microinjected directly into the nucleus of MIN6 cells to achieve higher levels of expression (36). For Ycam-2, an identical cytosolic distribution was observed after microinjection as with transient transfection. However, Ycam-4ER was mistargeted to the cytosol after cDNA microinjection. The mechanisms involved in this mistargeting are not clear. However, one possibility is that the relatively high concentration of Ycam-4ER generated after cDNA microinjection buffers [Ca²⁺]_ER, leading to protein misfolding. Alternatively, it is feasible that at high expression levels, the protein may have been cleaved between the CaM and M13, generating split cameleons (as observed with mitochondrially targeted cameleons; A.V., G.A.R., unpublished data). Because of this targeting problem with the Ycam-4ER construct, we only used Ycam-2 for those experiments that required microinjection (see below).

Continuous monitoring of the fluorescence emission ratio (535 nm/480 nm) permitted ER Ca²⁺ to be followed in real time (Fig. 1). Fluorescence bleaching, which was minimized by attenuating the excitation light, was corrected mathematically offline (Fig. 1A). Re-addition of 1.0 mmol/l CaCl₂ to Ca²⁺-depleted cells and Ca²⁺ influx through store-operated Ca²⁺ entry channels (44) led to the apparent refilling of the ER with Ca²⁺. Thus, after a rapid increase in the fluorescence intensity (535 nm/480 nm) ratio, a prolonged steady state of stable [Ca²⁺]_ER followed (Fig. 1B). Estimated using the calibration approach described in research design and methods and using an apparent K_d for Ca²⁺ of Ycam-4ER of 700 μmol/l (37), [Ca²⁺]_ER was 247 ± 28 μmol/l at this plateau (n = 15 cells from six independent experiments).

We next studied the effects of several agents on the filling and the release of Ca²⁺ from the ER on the assumption that changes in intralumenal free Ca²⁺ concentration would largely reflect the flux of these ions across the ER lumenal membrane. As expected, the SERCA inhibitor CPA rapidly reduced [Ca²⁺]_ER (Fig. 1). CPA caused a complete emptying of the store in some cells (Fig. 1C), although in others, [Ca²⁺]_ER remained elevated by 10–15% with respect to the basal level (Fig. 1B). The removal of CPA resulted in a complete refilling of the store, and the increase in [Ca²⁺]_ER was most apparent underneath the plasma membrane (Fig. 1D), as expected during activated Ca²⁺ influx. Thapsigargin also effectively depleted the ER (data not shown) and (in a Ca²⁺-free medium) led to a transient increase in [Ca²⁺]_cyt, most likely reflecting the release of Ca²⁺ from the ER (see below, Fig. 4).

To study the specific role of the SERCA isoforms coexpressed in β-cells, we selectively suppressed their expression by nuclear microinjection of targeted antisense cDNAs (45). Antisense oligonucleotides were designed based on the mouse SERCA2 and SERCA3 cDNA sequences. The antisense oligonucleotides corresponded to a unique mRNA region of SERCA2 (1818–1839 bp; 45 and 0% identity with the corresponding SERCA3 cDNA and amino acid sequences, respectively) or SERCA3 (1877–1896 bp; 45 and 12.5% identity with the corresponding SERCA2 cDNA and amino acid sequences, respectively).

Expression level of the SERCA proteins was measured 12, 14, 48, or 72 h after microinjection using immunocytochemistry with SERCA2 or SERCA3 isoform-specific antibodies. The measurements were done in the cytosolic region of the cells excluding the nucleus in 16 SERCA2 antisense, 19 SERCA2 sense, and 45 control untreated cells, and in 18 SERCA3 antisense, 14 SERCA3 sense, and 48 control untreated cells. SERCA2b expression was reduced 2 days after microinjection of the SERCA2-selective antisense oligonucleotide. On the third day after injection, expression of SERCA2b had recovered to levels seen in cells injected with a control (scrambled) oligonucleotide (not shown). SERCA3 expression was inhibited with a similar time course by the SERCA3-selective antisense oligonucleotide. Measured 48 h after injection, the SERCA2-selective antisense oligonucleotide reduced SERCA2 protein expression by 40 ± 5.3% (n = 6 cells from three independent experiments) but had no effect on the expression of SERCA3. Similarly, the SERCA3-selective antisense oligonucleotide reduced the expression of SERCA3 by 32 ± 4.8% (n = 6 cells from three independent experiments; Fig. 3) without affecting SERCA2 expression. Neither of the sense oligonucleotides altered SERCA2 or SERCA3 expression levels.

Because the antisense oligonucleotide-injected cells also expressed Ycam-2 from the comicroinjected plasmid, changes in [Ca²⁺]_cyt could readily be monitored in these cells and compared with changes observed in cells injected with control oligonucleotide. The impact of SERCA2b or SERCA3 inactivation on ER Ca²⁺ content was determined 48 h after microinjection by estimating the size of the thapsigargin-sensitive Ca²⁺ pool, monitored as the increase in [Ca²⁺]_cyt observed in the presence of thapsigargin. Addition of 100 nmol/l thapsigargin to control (untreated SERCA2 or SERCA3 sense-treated) cells in a Ca²⁺-free medium led to an increase in [Ca²⁺]_cyt, reflecting the release of Ca²⁺ from the ER (Fig. 4). In SERCA3 antisense-treated cells, the amplitude of the thapsigargin-induced Ca²⁺ release was the same as that in control cells (Fig. 4). However, in SERCA2 antisense-treated cells imaged 48 h after microinjection, the [Ca²⁺]_cyt elevation was markedly reduced compared with control cells. This result was correlated with the marked decrease in SERCA2 expression at this time point. Furthermore, no impairment of the [Ca²⁺]_cyt increase in response to thapsigargin was observed in the SERCA2 or SERCA3 antisense-treated cells monitored 24 or 72 h after microinjection, at which time points SERCA2 levels were not significantly altered. Thus, suppressed expression of SERCA2b induced a marked reduction in [Ca²⁺]_ER, whereas SERCA3 reduction was without effect in this cell type.

Carbachol, an acetylcholine analog whose effects in islets are largely mediated by the activation of m₃ muscarinic receptors (46), elicited a marked decrease in [Ca²⁺]_ER (Fig. 5A). The addition of extracellular ATP, which activates P_2U and P_2Y purinergic receptors (47), also clearly reduced [Ca²⁺]_ER (Fig. 5B). Whereas the SERCA inhibitors caused around 90% emptying of the ER, carbachol mobilized only ∼30% and ATP ∼10–15% of the ER Ca²⁺ pool.

We next investigated the effects on [Ca²⁺]_ER of nutrient and other secretagogues. Glucose (20 mmol/l), 20 mmol/l leucine, or 20 mmol/l KCl each induced a substantial apparent increase in [Ca²⁺]_ER (Fig. 6A–C), as reported by the increase in intramolecular FRET and consequent elevation of the 535 nm/480 nm emission ratio. Interestingly, the magnitude of the increase in [Ca²⁺]_ER immediately beneath the plasma membrane was greater than that apparent in parts of the ER located deeper within the cell (Fig. 6A–C), presumably reflecting the activation of Ca²⁺ influx and the establishment of a gradient of [Ca²⁺]_cyt beneath the plasma membrane (48).

Caffeine (10 mmol/l; Fig. 5C) or 4-chloro-3-ethylphenol (4-CEP) (500 μmol/l; Fig. 5D) caused an almost complete emptying of the ER Ca²⁺ store in intact MIN6 cells. Although the effect of caffeine was fully reversible, the ER Ca²⁺ was only partially restored after the treatment with 4-CEP. Furthermore, under similar conditions to those used for monitoring [Ca²⁺]_ER, either caffeine or 4-CEP induced a parallel increase in [Ca²⁺]_cyt (Fig. 5E and F). Although caffeine-induced increases in [Ca²⁺]_cyt were relatively small (Fig. 5E), a clear [Ca²⁺]_cyt increase was consistently observed on perifusion with 500 μmol/l 4-CEP. [Ca²⁺]_cyt began to rise immediately on arrival of the compound in the cell chamber and remained elevated during the continued presence of 4-CEP, but returned to pre-addition values on washout of the drug. This response of [Ca²⁺]_cyt to caffeine or 4-CEP, as monitored with the cytosolic cameleon, was essentially identical to that observed using Fura-2 (A.V., G.A.R., unpublished data) and is most easily explained by a flux of Ca²⁺ ions from the ER into the cytosol.

We next studied the effect of cADPr, the putative physiological agonist of type 2 RyR, on [Ca²⁺]_cyt in intact MIN6 cells. Cells were comicroinjected with caged cADPr and the Ca²⁺-sensitive fluorescence dye Oregon Green 488 BAPTA-1 dextran. In these experiments, the concentration of the caged cADPr was 1 mmol/l in the pipette and resulted in a final intracellular concentration of the caged precursor of ∼25–75 μmol/l, given an injection volume of 2.5–7.5% total cell volume (49). Cells were then subjected to a brief (1-s) UV flash, with excitatory light led to the cells via the objective lens in three to four discrete regions of the cell, together comprising <10% of the total cell surface. The efficiency of conversion of caged cADPr to the active molecule during this approach is not known but is likely to be of the order of 5–20% at the relatively low light intensities used here (50,51). As shown in Fig. 7, Oregon Green BAPTA-1 dextran (K_d for Ca²⁺ 170 nmol/l) is optimized for excitation by the 488-nm argon laser, such that the photolysis flashes at wavelengths of close to 360 nm caused no artifacts (beyond a small degree of photobleaching during the UV pulse) in fluorescence recordings (Fig. 7A–C, Cont). (By contrast, targeted cameleons rapidly bleached out after the UV excitation and could not be conveniently used in these experiments.) Photorelease of cADPr led to an increase in Oregon Green fluorescence, which then gradually fell back to preflash levels over the next 100–110 s (Fig. 7A). The mean increase in Oregon Green fluorescence caused by photorelease of cADPr was 12 ± 3% (n = 6 cells). Photorelease of caged IP₃ (6–20 μmol/l) or Ca²⁺ (NP-EGTA; 50–150 μg/ml) also increased Oregon Green fluorescence (Fig. 7B and C) by 15 ± 2% (n = 12 cells) and 26 ± 5% (n = 12 cells), respectively, compared with an increase in response to the muscarinic receptor agonist carbachol (100 μmol/l) of 22 ± 2.3% (n = 5 cells). The effect of either photoreleased cADPr or IP₃ on [Ca²⁺]_cyt was abolished when cells were preincubated with 1 μmol/l thapsigargin (a concentration that blocked Ca²⁺ release from the ER in response to carbachol; not shown), under which conditions Oregon Green fluorescence remained essentially the same as in control cells microinjected with Oregon Green dextran alone (Fig. 7A– C).

The spatial distribution of RyRs in living primary islet mouse β-cells or in MIN6 and INS-1 pancreatic β-cells was directly visualized by staining with cell-permeate BODIPY TR-X ryanodine. The specific binding of BODIPY TR-X ryanodine to β-cells was observed using low concentrations (10–100 nmol/l) of the probe (Fig. 8B, D, and F). Fluorescence was strictly confined to an extranuclear compartment (Fig. 8B, D, and F), consistent with the localization of RyRs to intracellular organelles throughout the cytoplasm (most likely the ER). Moreover, immunostaining of RyRs in β-cells with a mouse monoclonal anti-RyR antibody (clone XA7B6) showed high colocalization with the ER marker calreticulin (data not shown). Competition experiments to determine the binding specificity of BODIPY TR-X ryanodine revealed that the fluorescence signal was reduced by 80–85% by treating the cells with a combination of BODIPY-ryanodine plus excess unlabeled ryanodine (Fig. 8H).

Recombinant cameleons can be efficiently targeted to either the cytosol or the ER compartment of MIN6 β-cells (Fig. 2) through the use of the cell’s own protein trafficking/sorting machinery. Thus, the subcellular localization of cameleons in β-cells observed here agrees with those described previously for HeLa and HEK293 cells (37,52,53). However, in the present study, we observed that care must be taken in the use of these probes to overcome the significant bleaching of fluorescence that occurs, especially at higher excitation intensities; this bleaching preferentially affects EYFP, causing an apparent Ca²⁺-independent decrease in the ratio of the 535 nm/480 nm emission ratio intensities. This progressive decrease means that measurements are most efficiently performed using conventional epifluorescence microscopy (rather than a laser-based confocal system) and with subsequent mathematical correction for signal drift. Care must also be taken to overcome the relatively small changes in fluorescence ratio compared with those observed using nontargetable fluorescent probes of the fura-2 family (32) and to overcome mistargeting of the ER cameleons, which can occur under some circumstances (e.g., coinjection of antisense oligonucleotides). These caveats aside, the use of cameleons provides the considerable advantage over the alternative targetable probe, aequorin (22,26), in permitting compartmentalized Ca²⁺ changes to be imaged at the subcellular level and in real time rather than simply sampled in large cell populations.

Using the ER-targeted cameleon, we measured a resting value of [Ca²⁺]_ER (at 3 mmol/l glucose) of 247 ± 28 μmol/l in MIN6 β-cells, corresponding well to the values observed using the same method of detection in HeLa cells (60–400 μmol/l) (37). This value is also similar to resting [Ca²⁺]_ER (∼300 μmol/l) in HeLa cells (28) and INS-1 β-cell populations (26) measured using recombinant aequorin.

In the present studies, the MIN6 cell ER Ca²⁺ store was efficiently emptied (by ∼90%) with the SERCA inhibitors CPA and thapsigargin, whereas the IP₃-generating agents carbachol and ATP caused only ∼30 and 10% to 15% mobilization of the releasable Ca²⁺, respectively. Again, these data are in good agreement with the results obtained on intact INS-1 β-cell populations with aequorin (26) or single permeabilized mouse pancreatic β-cells, imaged with furaptra (54).

We found that three insulin secretagogues—glucose, leucine, and KCl—each efficiently stimulated Ca²⁺ uptake into the ER of MIN6 β-cells. Tengholm et al. (23) previously proposed that this increase is in part due to an increase in intracellular free ATP concentrations (55) caused by the elevated glucose concentration as well as by the increase in cytosolic free Ca²⁺ concentration subsequent to K_ATP channel closure. However, it might be mentioned that the relative contributions of a glucose-induced change in [Ca²⁺]_cyt and in cytosolic [ATP] in the regulation of SERCA pump activity are difficult to differentiate because an increase in [Ca²⁺]_cyt per se leads to the activation of mitochondrial ATP synthesis and thus cytosolic ATP concentrations (55,56).

In this study, the expression of cytosolic (untargeted) Ycam-2 was used as a convenient tool to investigate the impact of selective SERCA isoform inactivation because Ycam-2 cDNA could be comicroinjected into single cells along with SERCA-specific antisense oligonucleotides (Fig. 3A). Although the reduced expression of SERCA3 was without effect on the apparent size of the ER Ca²⁺ store, suppression of SERCA2b markedly lowered the amount of Ca²⁺ releasable from the ER. It has previously been shown that SERCA2b is ubiquitously expressed and probably plays a housekeeping function that is critical for the long-term viability of most mammalian cell types (57). In a recent study, the physiological function of SERCA2b was investigated by developing SERCA2 knockout mice, which demonstrated that the absence of the pump is incompatible with life because no homozygous mutants were observed (58). In cardiomyocytes from heterozygous animals, SERCA2b protein levels were reduced by ∼35%, which resulted in decreased sarcoplasmic reticulum Ca²⁺ stores (by 40–60%) and Ca²⁺ release (by 30–40%), in good agreement with our present data. Through the microinjection of targeted antisense constructs, we achieved ∼40% reduction in SERCA2b protein level in MIN6 β-cells and observed a marked reduction in the Ca²⁺ storage capacity of the ER. These data suggest that the loss of SERCA2b pump protein is not compensated by other calcium-handling mechanisms and that SERCA2b pump activity is critical for normal Ca²⁺ handling in β-cells. Whether glycemic control or insulin secretion is affected in animals heterozygous for mutations in the SERCA2b gene (58) is presently unknown.

By contrast, we observed that a similar reduction of SERCA3 protein level did not affect Ca²⁺ release from the ER in MIN6 β-cells. This result was surprising given that SERCA3 expression is lowered in several models of type 2 diabetes in which insulin secretion and β-cell Ca²⁺ handling are impaired (12,13,59–61). However, the function and regulation of SERCA3 is poorly understood, especially in light of the unusually low Ca²⁺ affinity of this Ca²⁺ ATPase. Based on our present data in MIN6 cells, which express similar total levels of SERCA2 and SERCA3 as primary β-cells (12), it seems likely that reduced SERCA3 expression is not sufficient in itself to affect β-cell Ca²⁺ homeostasis and thus insulin secretion. Nevertheless, confirmation of this point will require studies in primary β-cells, which were beyond the scope of the present work.

In most earlier studies, the effect of RyR activators was indirectly investigated by monitoring changes in [Ca²⁺]_cyt using loaded intracellular dyes of the fura-2 type. Here, we have studied the effects of RyR activators by monitoring Ca²⁺ specifically inside the ER in intact cells. An important potential advantage of this approach is that the release of a relatively small number of Ca²⁺ ions, and thus relatively minor changes in the cytosolic Ca²⁺ concentration, may be detected if these produce larger changes in [Ca²⁺]_ER. We show first that essentially all of the Ca²⁺ stored in the ER by SERCA pump action can be released by cell stimulation with caffeine or 4-CEP, a lipophilic analog of 4-chloro-m-chresol (27). Two IP₃-generating agonists, carbachol and ATP, induced a smaller Ca²⁺ release from the ER than caffeine. It has been shown that in permeabilized INS-1 cells, however, direct addition of IP₃ produced nearly complete mobilization of ER Ca²⁺ (20,26). Nilsson et al. (62) and Rutter et al. (18) previously reported that IP₃ mobilized only a minor fraction of the total thapsigargin-sensitive Ca²⁺ pool in primary rat β-cells and INS-1 cells, respectively.

Under similar conditions to those used for imaging [Ca²⁺]_ER, either caffeine or 4-CEP increased [Ca²⁺]_cyt, measured with the untargeted Ycam-2. This in good agreement with recent data, which showed that, in intact single β-cells from ob/ob mice (24), INS-1 cells (25), and HIT-T15 β-cells (63), caffeine released Ca²⁺ from the ER. In other reports, however, caffeine failed to lower ER Ca²⁺ levels in digitonin or α-toxin-permeabilized ob/ob mouse β-cells (20,23) or INS-1 cells (18). From the accumulating data, it seems likely that the lack of an effect of caffeine might be due to the loss of protein kinases/phosphatases, or other cofactors (such as a regulatory subunit of the RyR), during the permeabilization procedure.

Type 2 RyR mRNA is expressed in ob/ob mouse islets, βTC3 cells (24), and rat islets of Langerhans (64). Moreover, type 2 RyR protein has been detected in INS-1 cells but not in the glucose-insensitive RINm5F β-cells (25). RyR protein was also present in human islets, rat islets, and C57BL/6J mouse islet cells (63). In this study, we visualized the cellular distribution of RyR binding sites in live primary mouse β-cells, MIN6 cells, and INS-1 β-cells using BODIPY TR-X ryanodine. Strongly suggesting the presence of high-affinity RyR in each cell type, the fluorescence staining was highly specific and was blocked with unlabeled ryanodine.

Because type 2 RyRs are sensitive to the putative physiological agonist cADPr (17,65), we studied its effect on Ca²⁺ mobilization in intact β-cells. Photorelease of cADPr, which allows the effect of this molecule to be examined in live cells, caused an increase in [Ca²⁺]_cyt, which was blocked by pretreatment of cells with thapsigargin. These data strongly indicate that cADPr is able to mobilize Ca²⁺ from the ER/Golgi in well-differentiated MIN6 β-cells. The lack of effect of cADPr on Ca²⁺ mobilization reported in many previous studies (18,23,31) can probably be attributed, at least in part, to the use of 1) permeabilized cells (18,23,31) (see above), 2) β-cell lines (e.g., RINm5F) that show poor responses to caffeine (at low glucose concentrations) (18) and express low levels of type 2 RyRs (25), or 3) relatively insensitive measurements of cytosolic (as opposed to ER lumenal) Ca²⁺ concentration in live β-cells.

Overall, our new results support the idea that the bulk ER is capable of releasing Ca²⁺ either via IP₃ receptors or via RyR in the highly differentiated MIN6 cell line. Given the demonstrated presence of RyR binding sites in human islets (64) and now INS-1 and primary isolated mouse β-cells (Fig. 8), it also seems likely that these receptors are important in the regulation of [Ca²⁺]_cyt and insulin secretion in primary islets. The ability of thapsigargin to eliminate the response to uncaged cADPr of intracellular [Ca²⁺]_cyt suggests that, in β-cells, as in most other cell types, the bulk of the RyR-gated intracellular Ca²⁺ store is localized to organelles expressing SERCA pumps (i.e., the ER and Golgi). However, recent data (A.V., K. Mitchell, G.A.R., unpublished data) suggest that these receptors may also be present on insulin secretory vesicles and could play an important role in allowing the localized release of Ca²⁺ from individual vesicles to modulate the exocytosis of individual granules. Importantly, at the signal-to-noise levels achieved in the present uncaging experiments, we cannot exclude the possibility that a small amount of Ca²⁺ (<30% of the total releasable pool) may be released from secretory vesicles by cADPr in the presence of thapsigargin.

Whether an increase in cADPr concentration is important in prompting an increase in β-cell [Ca²⁺]_cyt in response to an acute challenge with glucose remains to be established because, to date, measurements of cADPr content in isolated islets have failed to reveal any rapid increase in response to glucose (66). Nevertheless, the presence of functional RyRs in β-cells does support the view that these channels may participate in Ca²⁺-induced Ca²⁺ release and thus the propagation of intracellular [Ca²⁺] increases in response to nutrients and other secretagogues, as recently proposed (67).

This work was supported by project grants from the Biotechnology and Biological Sciences Research Council, the Medical Research Council (U.K.), the Wellcome Trust, DiabetesUK, the European Union, and the U.K. Joint Infrastructure Fund (Office of Science and Technology & Wellcome Trust).

We thank the Medical Research Council for providing an Infrastructure Award and Joint Research Equipment Initiative Grant to establish the School of Medical Sciences Cell Imaging Facility and Dr. Elek Molnar, Dr. Mark Jepson, and Alan Leard for assistance.