A Reappraisal of the Effects of L-type Ca²⁺ Channel Blockers on Store-Operated Ca²⁺ Entry and Heart Failure

Abstract

Dihydropyridines such as amlodipine are widely used as antihypertensive agents, being prescribed to ∼70 million Americans and >0.4 billion adults worldwide. Dihydropyridines block voltage-gated Ca²⁺ channels in resistance vessels, leading to vasodilation and a reduction in blood pressure. Various meta-analyses show that dihydropyridines are relatively safe and effective in reducing hypertension. The use of dihydropyridines has recently been called into question as these drugs appear to activate store-operated Ca²⁺ entry in fura-2-loaded nonexcitable cells, trigger vascular remodeling, and increase heart failure, leading to the questioning of their clinical use. Given that hypertension is the dominant “silent killer” across the globe affecting ∼1.13 billion people, removal of Ca²⁺ channel blockers as antihypertensive agents has major health implications. Here, we show that amlodipine has marked intrinsic fluorescence, which further increases considerably inside cells over an identical excitation spectrum as fura-2, confounding the ability to measure cytosolic Ca²⁺. Using longer wavelength Ca²⁺ indicators, we find that concentrations of Ca²⁺ channel blockers that match therapeutic levels in serum of patients do not activate store-operated Ca²⁺ entry. Antihypertensive Ca²⁺ channel blockers at pharmacological concentrations either have no effect on store-operated channels, activate them indirectly through store depletion or inhibit the channels. Importantly, a meta-analysis of published clinical trials and a prospective real-world analysis of patients prescribed single antihypertensive agents for 6 mo and followed up 1 yr later both show that dihydropyridines are not associated with increased heart failure or other cardiovascular disorders. Removal of dihydropyridines for treatment of hypertension cannot therefore be recommended.

Graphical Abstract

Graphical Abstract.

Introduction

Hypertension was either the primary or contributing cause of mortality for just over half a million deaths in the United States in 2019.¹ A total of 116 million Americans are thought to have hypertension or are taking medication for its treatment² and, in 37 millions of this cohort, hypertension is uncontrolled.² Hypertension leads to greater risk of myocardial infarction (MI), heart failure (HF), stroke, retinopathy, and kidney disease.³ These figures are not unique to the United States. Around 1 in 5 adults globally is diagnosed with hypertension, with some ethnic groups particularly prone to the disease.⁴

Current therapies for combating hypertension include thiazide-like diuretics, angiotensin-converting enzyme (ACE) inhibitors, angiotensin II receptor blockers (ARBs), voltage-gated Ca²⁺ channel blockers (CCBs), and α1 or β1 adrenoceptor antagonists. CCBs target plasmalemmal Cav1.2 channels (L-type) in resistance vessels, the sino-atrial node, and cardiac myocytes. Inhibition of these channels reduces blood pressure by causing vasodilatation and decreasing heart rate and stroke volume. Widely prescribed CCBs are members of the dihydropyridine family and include amlodipine, felodipine, and nifedipine.

In 2004, 37.7 million Americans were prescribed amlodipine and this increased to 73.5 million in 2019.⁵ Amlodipine is now amongst the 6 most prescribed drugs in the United States,⁶ it is sixth in the United Kingdom⁷, and is the single most prescribed antihypertensive agent in India.⁸ The use of CCBs in treating patients with hypertension has recently been called into question.⁹ Johnson et al.⁹ reported that amlodipine and other dihydropyridines and nondihydropyridine CCBs increased cytosolic Ca²⁺ in fura-2-loaded HEK293 cells, independent of Cav1.2 channels, through opening of store-operated Ca²⁺ release-activated Ca²⁺ (CRAC) channels in the plasma membrane.⁹ CRAC channels are activated by agonists that deplete the endoplasmic reticulum (ER) of Ca²⁺¹⁰ and gating involves ER-resident Stromal Interaction Molecule (STIM) proteins and plasma membrane Orai proteins, pore-forming subunits of the CRAC channel.¹¹ The fall in store Ca²⁺ content is sensed by ER Ca²⁺ sensors STIM1 and STIM2, which are integral proteins in the ER membrane. This leads to a conformational switch in STIM, resulting in oligomerization and migration toward specialized ER-plasma membrane junctions, where the C-terminal CRAC activation domain or STIM-Orai1 activating region binds to and gates open the Orai homologs Orai1, Orai2, and Orai3 in the plasma membrane.¹² CCBs such as amlodipine appeared to activate Orai channels in a manner independent of store depletion, and the subsequent rise in cytosolic Ca²⁺ triggered vascular smooth muscle cell proliferation.⁹ An analysis of patients’ records from the local healthcare network led to the conclusion that CCBs were more associated with HF than any other antihypertensive drug in patients, leading these authors to challenge the clinical use of CCBs for treating hypertension.⁹

CRAC channels are robustly expressed in immune cells.¹³ If CCBs activated CRAC channels, then they might be expected to induce a hyper-active immune system, resulting for example in autoimmunity, especially as patients often take dihydropyridines for years. However, there are no reports of autoimmunity or other consequences of a chronically stimulated immune system in patients on CCBs. We therefore hypothesized that other factors in immune cells might render their CRAC channels less susceptible to activation by CCBs. In this study, we have investigated the effects of CCBs on CRAC channels in various cell types. Unexpectedly, we find that the effects of CCBs on store-operated Ca²⁺ entry are complex; different CCBs have either no effect on the Ca²⁺ channels, activate the channels indirectly via store depletion, or inhibit the channels. We find that amlodipine is strongly fluorescent in the cytoplasm with an excitation spectrum that overlaps with that of fura-2. Hence, cytosolic Ca²⁺ signals are likely to be heavily masked by the properties of amlodipine itself. Using longer wavelength Ca²⁺-sensitive fluorescent dyes that do not overlap with the excitation spectrum of CCBs, we have been unable to detect direct activation of store-operated Ca²⁺ entry or stimulation of subsequent Ca²⁺-dependent gene expression.

We have performed a systematic review and network meta-analysis of randomized controlled trials (RCTs) that compared CCBs with another drug or placebo and that assessed incident HF after at least 1 mo of active treatment. Evidence from this network meta‐analysis (and underlying RCTs) is not consistent with increased risk of incident HF with CCB therapy and supports the protective effect of other antihypertensive drug classes on development of HF. This is confirmed in a real-world analysis of patients on monotherapy with one of the major antihypertensive drug classes. Our new data show dihydropyridines are associated with a significantly decreased risk of HF compared with other antihypertensive drugs, including nondihydropyridine CCBs. Our results provide reassurance that dihydropyridine CCBs are not associated with increased risk of HF. Abandoning these drugs may result in an increased burden of cardiovascular morbidity and mortality arising from cardiac failure.

Methods and Materials

Cell Culture

HEK293 cells (ATCC, CRL 1573) and RBL2H3 cells (ATCC, CRL 2256) were cultured in Dulbecco’s minimum essential medium (DMEM) supplemented with 10% heat-inactivated fetal bovine serum and 2 mm glutamine and maintained in a humidified 95% air, 5% CO₂ incubator at 37°C. Human aortic smooth muscle cells (HASMC: ATCC, PCS-100-012) were maintained in culture as detailed by ATCC. Briefly, cells were maintained in Vascular Cell Basal Medium (ATCC, PCS-100-030) supplemented with Vascular Smooth Muscle Cell Growth Kit (ATCC, PCS-100-042). After treatment with Trypsin-EDTA (ATCC, PCS-999-003), HASMC cells were seeded in T-75 flasks at a cell density of either 2500 or 5000 cells per cm². The seeded culture flasks were maintained at 37°C with a 5% CO₂ atmosphere and reached confluence in ∼7 d. In general, in preparation for cytosolic Ca²⁺ measurements or confocal microscopy, these cell types were subcultured onto 30-mm round glass coverslips (#1.5 thickness) and maintained in culture for an additional 36-48 h before use in cytosolic Ca²⁺ measurements. In addition, experiments with HEK293 cells, HASMC and bone marrow-derived macrophages (BMDM cells, see below) were performed with a HEPES-buffered salt solution (HBSS: NaCl 120; KCl 5.4; MgCl₂ 0.8; HEPES 20; CaCl₂ 1.8 and glucose 10 mm, with pH 7.4 adjusted by NaOH). For RBL2H3 cells, a slightly modified HEPES-buffered salt solution was used (HBSS: NaCl 145; KCl 2.8; MgCl₂ 2; HEPES 10; CaCl₂ 2 and glucose 10 mm, with pH 7.4 adjusted by NaOH)

Isolation of Bone Marrow-Derived Macrophages Expressing GCaMP6f

Macrophage cells expressing the genetically encoded calcium sensor, GCaMP6f, were derived from murine bone marrow by a method adapted from Meurer et al.¹⁴ Briefly, mice (8-12 wk old) globally expressing GCaMP6f (see below), were euthanized by CO₂ and cervical dislocation, and long bones (femur and tibia) removed whilst being careful to keep the bones intact. Under sterile conditions, both ends of isolated bones were removed with scissors, and each bone flushed with 5 mL of Dulbecco’s Phosphate Buffered Saline (DPBS). Bones were also crushed and flushed further. The flushed medium containing a cell suspension was collected, passed through a 70-µm filter, and then centrifuged for 5 min at 800 rpm. The resulting cell pellet was resuspended in RPMI-based medium (supplemented with l-glutamine, 15% L929-conditioned RPMI medium, 10% FBS, 1% penicillin/streptomycin) and transferred on to 10 cm tissue culture dishes. In culture, BMDM adhere to the dish, whereas nonmacrophage cell types (such as mast cells) are of nonadherent nature and can be removed by replacing the RPMI-medium. After 4 d, cells were washed once with DPBS and fresh RPMI media applied. BMDM cells were maintained in culture for 2 wk to allow cell adhesion and morphology to mature before cryopreservation or use in experiments. In preparation for cytosolic Ca²⁺ measurements, BMDM cells were brought into suspension using Corning Cell Stripper Dissociation Reagent, centrifuged for 5 min at 800 rpm, and resuspended in RPMI medium at a cell density of 5 × 10⁵ cells per mL. BMDM cells were then plated directly onto glass coverslips and maintained in culture for at least 3-5 d before use in calcium experiments. Cryopreserved BMDM cells were also utilized after being thawed and plated directly on to glass coverslips for calcium experiments.

To generate mice expressing whole body GCaMP6f, GCaMP6f and CMV-Cre mice were purchased from Jackson Laboratories, respectively: Jax stock 024105 B6;129S-Gt(ROSA)26Sortm95.1(CAG-GCaMP6f)Hze/J PubMed:25741722 and Jax stock 006054 B6.C-Tg(CMV-cre)1Cgn/J PubMed:8559668. GCaMP6f was activated by crossing homozygous GCaMP6f with CMV-Cre. Resulting offspring ubiquitously express GCaMP6f, and isolated cells can be directly used for cytosolic Ca²⁺ measurements. All animal procedures complied with institutional guidelines and were approved by the NIEHS Animal Care and Use Committee.

Single-Cell Calcium Measurements With UV Ratiometric Calcium Indicator

Fluorescence measurements were made with HEK293 cells loaded with the Ca²⁺-sensitive dye, fura-5F, as described previously.¹⁵ Briefly, cells plated on 30 mm round coverslips and mounted in a Teflon chamber were incubated in DMEM with either 1 μm acetoxymethyl ester of fura-5F (Fura-5F/AM, Setareh Biotech, Eugene, OR, USA) or fura-2 (Fura-2/AM, Molecular Probes, USA) at 37°C in the dark for 25 min. For cytosolic Ca²⁺ measurements, cells were bathed in HEPES-buffered salt solution (HBSS: NaCl 120; KCl 5.4; Mg₂SO₄ 0.8; HEPES 20; CaCl₂ 1.8 and glucose 10 mm, with pH 7.4 adjusted by NaOH) at room temperature. Nominally Ca²⁺-free solutions were HBSS with no added CaCl₂. Fluorescence images of the cells were recorded and analyzed with a digital fluorescence imaging system (InCyt Im2, Intracellular Imaging Inc., Cincinnati, OH, USA). Fura-5F fluorescence was monitored by alternatively exciting the dye at 340 and 380 nm and collecting the emission wavelength at 520 nm. Changes in cytosolic Ca²⁺ are expressed as the “Ratio (F340/F380)” of fura-5F fluorescence due to excitation at 340 and 380 nm (F340/F380). Before starting the experiment, regions of interests identifying cells were created and 25 to 35 cells were monitored per experiment. As needed, ratio values were corrected for contributions by autofluorescence, which was measured after treating cells with 10 μm ionomycin and 20 mm MnCl₂.

Single-Cell Cytosolic Ca²⁺ Measurements With Single-Wavelength Calcium Indicator

Fluorescence measurements were made with HEK293 cells loaded with the single-wavelength calcium sensitive dye, Cal520. Cells plated on 30-mm round coverslips were incubated in DMEM with 1 μm acetoxymethyl ester of Cal520 (Cal520/AM, AAT Bioquest, USA) at 37°C in the dark for 25 min. Fluorescence images of the cells were recorded and analyzed with a digital fluorescence imaging system (InCyt Im1, Intracellular Imaging Inc., Cincinnati, OH, USA). Cal520 fluorescence was monitored by exciting the dye at 488 nm and collecting the emission wavelength at 525 nm. Changes in cytosolic Ca²⁺ are expressed as the “F/Fo” of Cal520 fluorescence, whereby the time course of fluorescence intensities (F) were divided by the initial fluorescence intensity recorded at the start of the experiment (Fo). Before starting the experiment, regions of interests identifying cells were created and typically 25 to 35 cells were monitored per experiment.

Fluorometric Imaging Plate Reader Cytosolic Ca²⁺ Measurements

Cytosolic Ca²⁺ was monitored in Cal520-loaded HEK 293 cells using a fluorometric imaging plate reader (FLIPR^TETRA; Molecular Devices, Inc., Sunnyvale, CA, USA). wtHEK 293 cells were plated 24 h before use on poly-Lysine-coated 96-well plates at 40 000 cells/well. Cells were then loaded with the single visible wavelength indicator Cal520 (4 μm Cal520/AM) in complete DMEM for 45 min at 37°C. After dye loading, cells were washed twice in nominally Ca²⁺-free HBSS and placed on the FLIPR observation stage. The dye-loaded cells were excited at 488 nm with emission fluorescence detected by a cooled charge-coupled device camera via 510-570-nm bandpass filter. Experiments were carried out at room temperature.

“Ca²⁺ Add-Back” Protocol for Monitoring Intracellular Ca²⁺ Release and Store-operated Ca²⁺ Entry

HEK 293 cells were loaded with Cal520/AM and intracellular fluorescence monitored on FLIPR ^TETRA. HEK 293 cells were initially bathed in buffer solutions with no extracellular Ca²⁺ present. Unstimulated HEK 293 cells elicit no detectable change in cytosolic Ca²⁺ on re-addition of extracellular Ca²⁺. Treatment of cells with (2 μm thapsigargin) caused a transient release of Ca²⁺ from intracellular stores, and on “Ca²⁺ re-addition” a rapid and sustained elevation of cytosolic Ca²⁺ was observed, indicative of store-operated Ca²⁺ entry.

Confocal Microscopy

For experiments examining the cellular accumulation of Amlodipine Besylate, HEK293 cells expressing an ER-targeted GFP were prepared for confocal microscopy in a similar way as described for calcium measurements. Cells plated on 30 mm round coverslips and mounted in a Teflon chamber were placed on the stage of a Zeiss LSM 880 confocal microscope equipped with a 40× objective (N.A. 1.2 W), and confocal images were collected with a pinhole set at 1 Airy unit. The observation of ER in HEK 293 cells was achieved by expressing of GFP modified with an ER retention sequence (KDEL). The ER-GFP was observed by excitation with a 488 nm laser, and emission fluorescence observed with 482-553 nm bandpass filter. To observe acidic lysozymes, HEK 293 cells were incubated with LysoTracker Red (50 nm) for 20 min. Accumulated LysoTracker Red was observed with 561 nm laser excitation and emission fluorescence detected with 572-650 nm bandpass filter. After cells were incubated with LysoTracker Red, HEK 293 cells were then exposed to 20 μm amlodipine besylate and intracellular accumulation monitored with 405 nm laser excitation and emission fluorescence observed with 407-481 nm bandpass filter.

Measuring Spectral Characteristics of Amlodipine Besylate and the Fluorescent Calcium Indicators Fura-2 and Fluo-4

Amlodipine besylate (20 and 100 μm final concentration) or free-acid forms of the fluorescent calcium indicators fura-2 and fluo-4 (10 μm final concentration) were dissolved in a Ca²⁺ calibration buffer (Thermo Fisher C3008MP) with either “zero Ca²⁺” ions (Rmin) or 10 mm Ca²⁺’ (Rmax). All spectra were recorded on a SpectraMax M2 microplate reader (Molecular Devices, Sunnyvale, CA, USA). For amlodipine besylate or fura-2, emission spectra were recorded at 510 nm (Supplementary Figure S1) while scanning the sample with excitation wavelengths from 320 to 420 nm.

Gene Reporter Assay

An RBL2H3 cell line stably expressing GFP under an NFAT promoter was generated using a self-inactivating retroviral vector (pSIRV-NFAT-eGFP, Addgene plasmid #118031, kindly deposited by Dr Peter Steinberger). RBL2H3-NFAT-eGFP stable cells were plated in 6-multiwell plates at a cell density of 1 × 10⁶ cells per well and maintained in culture with complete DMEM. On the day of the experiment, cells were treated with either thapsigargin (TG, 100 nm) or various concentrations of amlodipine besylate and incubated for 16 h. The cells were brought into a single-cell suspension with Trypsin/EDTA, washed into Ca²⁺/Mg²⁺-free PBS, and treated with propidium iodide (10 μm). The cells were then sorted and analyzed using a BD LSRFortessa flow cytometer (Becton Dickinson, USA), with a minimum of 20 000 events (cells) recorded per sample. Fluorescence signals were recorded for amlodipine besylate accumulation (ex 405/em 450), NFAT-eGFP expression (eGFP; ex 488/em 530), and propidium iodide (PI) accumulation (PI; ex 561/em 585). The exclusion of PI was used to determine % viable cells. For analysis, the % of viable cells expressing eGFP and accumulating amlodipine besylate was quantified, as well as the mean ± SEM fluorescence intensity of NFAT-eGFP. Data were analyzed using FlowJo software (Becton Dickinson).

Materials

Thapsigargin was purchased from Alexis (San Diego, CA, USA), fura-5F/AM from Setareh Biotech (Eugene, OR, USA), Fura- 2/AM from Molecular Probes, USA, Cal520/AM from ATT Bioquest, CA, USA, Calcium Calibration Buffer Kit #1 (zero and 10 mm CaEGTA, C3008MP) from Thermo Fisher. Amlodipine Besylate and Diltiazem from Selleck Chemicals (Houston, TX, USA), Nifedipine from Sigma Aldrich, Inc., USA, LysoTracker Red DND-99 from Thermo Fisher, USA, 30 mm round #1.5 coverslips from Bioptechs (Butler, PA, USA), BioCoat Poly-D-Lysine 96-well Black/Clear Flat Bottom TC-treated Microplates from Corning, USA.

Statistics

Data are presented as mean ± SEM, and statistical evaluation was carried out using unpaired t-test (GraphPad Prism). Statistical analyses for the meta-analysis and real world study are extensively described in the text and Supplementary Information.

Results

The observation that amlodipine activates CRAC channels without the need for store depletion potentially provides a novel pharmacological tool to study CRAC channels in the presence of functional Ca²⁺ stores. We therefore set out to use amlodipine as a tool to develop our studies on how local Ca²⁺ entry through CRAC channels drives downstream responses.

Amlodipine Besylate Increases Cytosolic Fluorescence in Fura-5F-loaded HEK293 Cells

Amlodipine is prescribed in the form amlodipine besylate, and this was principally used in these studies. Amlodipine Besylate (20 μm) was reported to raise cytosolic Ca²⁺ by activating CRAC channels in fura-2-loaded HEK cells within minutes.⁹ Fura-5F has identical spectral properties to fura-2 but, with a slightly (∼2-fold) lower affinity for Ca²⁺, it has a larger bandwidth for recording cytosolic Ca²⁺ signals. Acute application of 0.5 and 20 μm amlodipine besylate to fura-5F -loaded cells did not increase the fluorescence ratio that is indicative of a cytosolic Ca²⁺ rise. Instead, 20 μm caused a rapid decrease in the fluorescent ratio followed by no further change (Figure 1A). In dissecting the fluorescent ratio signal, we observed two effects of amlodipine besylate on the two excitation wavelengths (340 and 380 nm; Figure 1B). An immediate and sustained step increase in background fluorescence at each wavelength was caused by 20 μm amlodipine besylate. This was observed in cell-free areas of the coverslip (labeled “Bkgd” in Figure 1B) and reflects properties of amlodipine itself. The increase in emission at 380 nm excitation was slightly larger than at 340 nm (Figure 1B). An additional component was observed in fura-5F loaded cells; there was now a further increase in fluorescence at each wavelength that reached a steady-state after about 15 min (Figure 1B). The increase in the 380 nm signal was again slightly larger than the 340 nm signal, resulting in the decrease in the fura-5F ratio observed in Figure 1A. In Figure 1C, the background amlodipine fluorescence observed in cell-free areas of the coverslip was subtracted from the fluorescence signal observed in fura-5F loaded cells. The background corrected signal was ratioed (Figure 1D) and the result shows that 0.5 and 20 μm amlodipine besylate did not increase the fura-5F fluorescence ratio.

Figure 1.

Effects of amlodipine on cytosolic Ca²⁺-signals in fura-5F-loaded HEK 293 cells. (A) Acute application of 0.5 μm amlodipine slightly decreased the ratio of emitted signal for cells excited at 340 and 380 nm. Addition of 20 μm amlodipine resulted in a marked step decrease in the ratio signal. (B) The component 340 and 380 wavelength signal intensities that were used to derive the ratio changes in (A) are shown. While 0.5 μm amlodipine did not clearly change the 340 or 380 nm signal, 20 μm amlodipine caused an initial step increase in background fluorescence followed by a prolonged rise in cellular fluorescence that reached a steady state after 15 min. (C) The fluorescence intensity change observed in background bathing solution was subtracted from the fluorescence signals measured within fura-5F loaded cells. (D) Ratio (F340/F380) signal was derived from ratioing the “background corrected” 340 and 380 signals observed in (C). After “background” correction, 20 μm amlodipine did not raise cytosolic Ca²⁺. The data in (A) and (D) are mean ± SEM for n = 5 experiments.

We repeated these experiments using fura-2 instead. As we observed with fura-5f, amlodipine increased fluorescence substantially in fura-2-loaded cells in a manner identical to that seen when fura-5f was used instead (Supplementary Figure S1).

Amlodipine besylate has been reported to interfere with ratiometric fura indicators directly.¹⁶ This prompted us to carry out an extensive series of control experiments to assess whether UV-excited ratiometric fura dyes are appropriate for measuring cytosolic Ca²⁺ in the presence of amlodipine besylate. In vitro experiments showed that amlodipine besylate had no effect on the fluorescence of fura-2 (free acid) in either Ca²⁺-free or Ca²⁺-containing solution (Supplementary Figure S1).

Since amlodipine besylate did not appear to affect the spectral properties of fura-indicators, we further dissected the effects of the drug on HEK 293 cells that had not been loaded with fura-5F (Figure 2). At a concentration of 1 μm, amlodipine led to a small but gradually developing increase in the 380 nm signal from within cells that was larger than its background fluorescence (Figure 2A). In the presence of 20 μm amlodipine, a much larger but still slowly developing cellular fluorescence signal occurred (Figure 2B). Once accumulated intracellularly, amlodipine washed out very slowly with substantial intracellular fluorescence remaining even 40 min after removal of the drug from the extracellular solution (Figure 2C). Importantly, amlodipine fluorescence could not be quenched by Mn²⁺. Therefore, once inside cells, amlodipine fluorescence will confound fura dye-dependent signals.

Figure 2.

Effects of CCBs on cellular fluorescence signals in non-dye loaded HEK 293 cells. To dissect the effects of amlodipine besylate on HEK 293 cells, experiments were conducted as described in Figure 1, but now on cells that had not been loaded with fura-5F. (A) A coverslip containing HEK 293 cells was exposed to 1 μm amlodipine. Whereas cell-free regions did not show a detectable increase in background fluorescence, a small fluorescent rise occurred over tens of seconds within cells. (B) With 20 μm amlodipine, an initial step increase in background (cell-free) fluorescence occurred and this was followed by a prolonged rise in cellular fluorescence that reached a steady state after ∼15 min, and with a t_1/2 of 5.75 min. (C) Cells that already been incubated with 20 μm amlodipine for 15 min were bathed continuously with amlodipine-free external solution. Amlodipine was removed from the bathing solution at t =  60 sec. The accumulated intracellular fluorescence signal decreased very slowly (estimated t_1/2 of 42 min). The residual amlopidine fluorescence could not be quenched by the addition of ionomycin/Mn²⁺ (10 μm/20 mm, respectively). In panels (A-C), excitation wavelength was 380 nm. (D) Background (buffer) and cellular fluorescence are shown for different amlodipine concentrations at 340 nm excitation. (E) As in panel (D), but intensity of emitted light at 380 nm excitation is shown. The dashed lines in panels (D) and (E) indicate the typical fluorescent signals measured in fura-5F loaded HEK 293 cells at rest (indicated by “fura5F-cell” in F). (F) Intensity of signal following excitation at the wavelengths indicated are compared. (G) Fluorescence to 100 μm of either nifedipine or diltiazem are shown for the conditions indicated. Nifedipine fluorescence increased slightly inside cells compared with cells alone. Unpaired t-test was performed for the comparisons indicated (ns = not significant, ** = P ≤ 0.01). (H) Fluorescence signals were measured in non-dye loaded HEK 293 cells using the excitation and emission settings for Cal520 (ex 488 nm, em 520 nm). Amlodipine besylate (100 μm) did not increase the fluorescence intensity measured in buffer solution (background) or in cells . All data are mean ± SEM for n = 3 experiments.

Although 1 μm amlodipine increased the 380 nm signal only modestly (Figure 2A), concentrations >3 μm increased both 340 and 380 nm wavelength signals significantly (Figure 2D and E). While the changes in the background fluorescence (cell-free areas of the coverslip) exhibited step increases in fluorescence, the cellular fluorescence rose slowly in an amlodipine concentration-dependent manner and reached steady state after about 15 min. The dashed lines in Figure 2D and E show the typical fluorescent signals measured in fura-5F-loaded HEK 293 cells (also included in Figure 2F), indicating that at and above 20 μm, the cellular fluorescence of amlodipine besylate within HEK 293 cells overwhelms fura-5F fluorescence signals and thus dominates the spectral properties of the cells. We tested the effects of other CCBs on cellular fluorescence. Nifedipine and diltiazem exerted small effects even at 100 μm, compared with amlodipine besylate (Figure 2G).

Using confocal microscopy (Supplementary Figure S2), we assessed the intracellular accumulation of amlodipine besylate in HEK 293 cells. While the fluorescence images suggested amlodipine besylate accumulated within intracellular compartments, the intracellular localization was not confined to any specific cellular structure. Instead, the fluorescence signal was rather amorphous, with some indications that the signal associated with the ER, and acidic lysosomes (Supplementary Figure S2), as previously reported.¹⁷ Collectively, these results demonstrate that fura-based Ca²⁺ indicators are not an appropriate choice for measuring cytosolic Ca²⁺ when amlodipine is used.

Cal520 is a Suitable Ca²⁺-Sensitive Fluorescent Dye for Use With CCBs

To identify a more appropriate fluorescent probe, we identified wavelengths that failed to excite amlodipine, either in solution or when accumulated inside cells. At around 490 nm excitation, amlodipine did not excite (Figure 2H). We focused on Cal520 as this can easily be loaded into cells using the acetoxymethyl form.

Dose-dependent Effects of CCBs on Ca²⁺ Signals

In Cal520-loaded HEK 293 cells, stimulation with 0.5 μm amlodipine in the presence of external Ca²⁺ failed to raise cytosolic Ca²⁺ (Figure 3A) and only a very small, transient rise was seen in 20 μm amlodipine. By contrast, a prolonged rise in cytosolic Ca²⁺ was observed when thapsigargin was added instead (Figure 3A; aggregate data summarized in Figure 3B).

Figure 3.

Effects of thapsigargin and amlodipine on Ca²⁺-signals in Cal520-loaded HEK 293 cells. (A) Stimulation with 0.5 μm amlodipine in the presence of external Ca²⁺ failed to raise cytosolic Ca²⁺ and only a very small, transient rise was seen in 20 μm amlodipine. By contrast, thapsigargin (TG; 2 μm) caused a prolonged rise in cytosolic Ca²⁺. (B) Aggregate data of “peak” response are compared. All data are mean ± SEM for n = 3 experiments. Unpaired t-test was performed for each condition against DMSO (ns = not significant, **** = P ≤ 0.0001). Changes in the Ca²⁺signal were monitored using a fluorescence imaging plate reader (FLIPR^TETRA).

We adopted the classical Ca²⁺ add-back protocol for studying store-operated Ca²⁺ entry, namely to apply the stimulus in Ca²⁺-free external solution and then readmit external Ca²⁺ 15 min later. This leads to a cytosolic Ca²⁺ rise as Ca²⁺ permeates the CRAC channels opened by store depletion.

Stimulation of Cal520-loaded HEK 293 cells with thapsigargin in Ca²⁺-free solution led to prominent Ca²⁺ release from the stores. This store depletion resulted in robust Ca²⁺ entry through CRAC channels when external Ca²⁺ was readmitted (Figure 4A). By contrast, stimulation with 0.5 μm amlodipine besylate failed to trigger detectable Ca²⁺ release or subsequent Ca²⁺ entry (above the background Ca²⁺ leak seen in the presence of DMSO, the solvent for thapsigargin and amlodipine). Ca²⁺ release to 20 μm amlodipine was undetectable but now small but resolvable Ca²⁺ entry occurred, which was slightly larger than the Ca²⁺ leak (Figure 4A, middle panel). Increasing amlodipine besylate to 100 μm resulted in prominent Ca²⁺ release from the stores (Figure 4A, lower panel), which was similar in size to that induced by thapsigargin. However, whereas large store-operated Ca²⁺ entry occurred on adding back external Ca²⁺ to thapsigargin-treated cells, Ca²⁺ influx to 100 μm amlodipine was suppressed (Figure 4A, lower panel).

Figure 4.

Effects of CCBs on Ca²⁺ release and Ca²⁺ entry. (A) Different concentrations of amlodipine were applied in Ca²⁺-free solution and then external Ca²⁺ readmitted as indicated. Upper panel shows thapsigargin (2 μm), amlodipine (0.5 μm), and solvent (DMSO) control. Middle and lower panels show 20 and 100 μm amlodipine. (B) As in panel (A), but nifedipine was used instead. (C) As in panel (A), but diltiazem was applied. (D-E), Effects of different concentrations of amlodipine on Ca²⁺ release and Ca²⁺ entry are shown. Changes in the Ca²⁺signal were monitored using a fluorescence imaging plate reader (FLIPR^TETRA) and reported as changes in F/Fo. All data are mean ± SEM for n = 4 experiments.

Amlodipine Besylate Releases Ca²⁺ From the Thapisargin-Sensitive Ca²⁺ Store

In assessing the concentration-dependent effects of amlodipine besylate on Ca²⁺ release and entry in Cal520-loaded HEK 293 cells (Figure 4D-E), concentrations of amlodipine besylate around 50 μm or greater evoked prominent Ca²⁺ release, whereas no discernible Ca²⁺ mobilization from the stores was detected between 1 and 30 μm. We considered two possibilities to explain this. First, there is a threshold amlodipine besylate concentration for Ca²⁺ release and this is >30 μm. Second, lower concentrations of amlodipine release Ca²⁺ but this is slow and the plasma membrane Ca²⁺ ATPase pump effectively exports the Ca²⁺, thereby preventing cytosolic Ca²⁺ from rising. The Ca²⁺ entry signal to 20 μm amlodipine besylate in the Ca²⁺ add-back protocol was slightly larger than that seen when 20 μm amlodipine was applied in the continuous presence of Ca²⁺-containing solution (Figure 3A versus Figure 4D; ΔF/Fo of 0.14 versus 1.05, respectively), suggesting that the Ca²⁺ entry component in the add-back protocol reflected some store depletion at concentrations that did not visibly raise cytosolic Ca²⁺. Consistent with this, the Ca²⁺ rise following Ca²⁺ entry was transient (Figure 4A and D), declining to a steady-state level similar to that seen in Figure 3A.

To test more directly whether amlodipine released Ca²⁺ from internal stores, we carried out two different experiments. First, we measured the Ca²⁺ content of the stores using the Ca²⁺ ionophore ionomycin. Cells were bathed in Ca²⁺-free solution and then 10 μm ionomycin was added.¹⁵ A large transient rise in cytosolic Ca²⁺ occurred, reflecting Ca²⁺ release from the stores (blue trace in Supplementary Figure S3A). Ca²⁺ was exported out of the cells by the plasma membrane Ca²⁺ATPase pump. Application of thapsigargin 20 min after the ionomycin challenge (blue trace) elicited a negligible response. Under these conditions, ionomycin essentially releases all the mobilizable intracellular Ca²⁺ stores.

Exposure to ionomycin 20 min after a DMSO addition (green trace) elicited a Ca²⁺ transient identical to the ionomycin-induced Ca²⁺ release response measured after 2 min (Supplementary Figure S3). This illustrates that intracellular Ca²⁺ stores remain stable over the time course of this experiment, despite being bathed in Ca²⁺-free solution. Application of thapsigargin also caused a robust and transient release of intracellular Ca²⁺, reflecting the depletion of Ca²⁺ principally from ER (black trace). However, application of ionomycin after stimulation with thapsigargin resulted in a very small Ca²⁺ transient (Supplementary Figure S3A; aggregate data in Supplementary Figure S3B), when compared with an ionomycin pulse at the same time but in the absence of thapsigargin pretreatment (green trace, Supplementary Figure S3A). The residual Ca²⁺ release may reflect the release of Ca²⁺ from non-ER stores, most likely the mitochondria. Hence, the vast majority of Ca²⁺ mobilized by ionomycin emanates from the thapsigargin-sensitive Ca²⁺ store. Addition of amlodipine besylate (100 μm) after stimulation with thapsigargin (Supplementary Figure S3A red trace) failed to raise cytosolic Ca²⁺, suggesting that amlodipine besylate mobilizes Ca²⁺ from thapsigargin-sensitive Ca²⁺ stores.

We investigated the relationship between amlodipine besylate concentration and Ca²⁺ release from intracellular stores (Figure 5). Little detectable Ca²⁺ release was seen at amlodipine concentrations below 30 μm (Figure 5A and B), but Ca²⁺ mobilization then rose steeply as amlodipine concentration increased. Using the ionomycin-release protocol to estimate the extent of Ca²⁺ store depletion (Supplementary Figure S3), we found an inverse relationship between amlodipine concentration and remaining store Ca²⁺ content (Figure 5B versus 5C); as amlodipine concentration increased, more Ca²⁺ was released from the stores (Figure 5B) and the size of the ionomycin-accessible Ca²⁺ store subsequently decreased (Figure 5C).

Figure 5.

Effects of amlodipine on intracellular Ca²⁺ pools. (A) Different concentrations of amlodipine were applied in Ca²⁺-free solution (first addition) and then ionomycin was applied (second addition), using a protocol based on data in Supplementary Figure S3. The second addition of ionomycin was used to estimate the extent of Ca²⁺ store depletion. As the concentration of amlodipine increased, the size of the ionomycin-accessible Ca²⁺ store decreased. (B) Aggregate peak responses for the first additions are summarized. (C) Aggregate data showing the size of the ionomycin response (second addition) after different amlodipine concentrations are compared. Included in (A-C) are the responses to thapsigargin (2 μm). All data are mean ± SEM for n = 3 experiments. (D) Ca²⁺ release to the various stimuli indicated, applied in Ca²⁺-free solution, was measured. (E) Aggregate peak responses are compared, and unpaired t-test was performed for each condition against DMSO (ns = not significant, ** = P ≤ 0.01, **** = P ≤ 0.0001). (F) As in panel (D), but 1 mm Gd³⁺ was present to prevent Ca²⁺ extrusion. (G) Aggregate data from experiments as in panel (F) are compared, and unpaired t-test was performed for each condition against DMSO (ns = not significant, ** = P ≤ 0.01, *** = P ≤ 0.001, **** = P ≤ 0.0001). Ca²⁺ measurements were performed on Cal520-loaded HEK 293 cells using a FLIPR^TETRA.

Although 30 μm amlodipine did not transiently raise cytosolic Ca²⁺ to a detectable level, it nevertheless reduced the store Ca²⁺ content (Figure 5C). This suggests that concentrations lower than 30 μm amlodipine might also release Ca²⁺, albeit slowly, so Ca²⁺ extrusion prevents cytosolic Ca²⁺ from rising. To test this, we stimulated cells in Ca²⁺-free solution in either the absence or presence of 1 mm Gd³⁺. Trivalents like Gd³⁺ inhibit plasma membrane Ca²⁺ATPase pumps, thereby preventing Ca²⁺ export out of the cell.^15,18 Application of thapsigargin in Ca²⁺-free solution in the absence of Gd³⁺ resulted in a transient cytosolic Ca²⁺ signal (Figure 5D; aggregate data are in Figure 5E). However, when Gd³⁺ was present, thapsigargin evoked a much larger and more sustained Ca²⁺ rise (Figure 5F; aggregate data are shown in Figure 5G). Stimulation with 100 μm amlodipine caused a rapid Ca²⁺ transient in the absence of Gd³⁺ (Figure 5D, red trace) but this became much larger in size and considerably more prolonged when the trivalent cation was present (Figures 5F, red trace). The small cytosolic Ca²⁺ rise evoked by 40 μm amlodipine was also converted into a much larger and prolonged Ca²⁺ rise in the presence of Gd³⁺ (Figures 5D versus F, blue traces). Strikingly, 20 μm amlodipine besylate, a concentration that failed to elicit a resolvable sustained rise in cytosolic Ca²⁺ under normal conditions, produced a prolonged Ca²⁺ signal when Gd³⁺ was present (Figure 5F). Although this Ca²⁺ rise was small, it was nevertheless significantly increased compared with the DMS0/Gd³⁺ control. A similar trend was seen with 1 μm amlodipine (Figure 5F), although the Ca²⁺ rise in Gd³⁺ was very small (Figure 5G). These results show that even low concentrations of amlodipine release Ca²⁺ from the intracellular store.

Amlodipine Besylate Inhibits Store-Operated Ca²⁺ Entry

Despite substantial store depletion, high concentrations of amlodipine besylate evoked weak store-operated Ca²⁺ entry (Figure 4E). We hypothesized that this was a consequence of amlodipine besylate inhibition of store-operated Ca²⁺ channels. To test this, we applied different concentrations of amlodipine besylate to thapsigargin-treated HEK 293 cells after sustained Ca²⁺ entry was established. Amlodipine led to a concentration-dependent reduction in store-operated Ca²⁺ entry, with maximal inhibition at 40 μm (Supplementary Figure S4).

Effects of Nifedipine and Diltiazem on HEK 293 Cells

Unlike amlodipine besylate, nifedipine (0.5-100 μm) did not evoke Ca²⁺ release or subsequent store-operated Ca²⁺ entry (Figure 4B). Diltiazem also failed to stimulate CRAC channels when applied at sub-micromolar or low-micromolar concentrations (Figure 4C), although higher doses evoked a transient store-operated Ca²⁺ entry signal with no observable Ca²⁺ release.

Effects of CCBs in Immune Cells and Human Vascular Smooth Muscle

CRAC channels are robustly expressed in various cells of the immune system.¹³ Therefore, one would predict prominent store-operated Ca²⁺ entry in these cell types in response to CCBs. Application of 0.5 and 20 μm amlodipine besylate failed to evoke Ca²⁺ entry in RBL-2H3 cells (Supplementary Figure S5A), an immortalized mast cell line that expresses a large store-operated CRAC current.¹⁹ We measured the activity of the NFAT transcription factor, which is activated by CRAC channels, by monitoring expression of GFP under an NFAT promoter. Whereas activation of CRAC channels with a low dose of thapsigargin triggered strong GFP expression, amlodipine (0.2-2 μm) failed to induce reporter gene expression (Supplementary Figure S5B). Higher concentrations of amlodipine (20 μm) led to a significant decrease in cell viability over the time frame required for these measurements. We also failed to observe a detectable Ca²⁺ rise in bone marrow-derived murine macrophages (Supplementary Figure S5C) to 0.5 and 20 μm amlodipine besylate. In human aortic vascular smooth muscle cells, 0.5 μm amlodipine failed to raise cytosolic Ca²⁺ and 20 μm evoked only a small response, when applied in extracellular solution containing 10 mm Ca²⁺ (Supplementary Figure S5D). Robust Ca²⁺ signals were seen with 40 and 100 μm amlodipine (Supplementary Figure S5D). These responses were similar to the initial signal evoked by thapsigargin, suggesting they reflected Ca²⁺ release from the stores. However, whereas the thapsigargin response was relatively sustained due to Ca²⁺ entry through CRAC channels, the response to amlodipine was transient (Supplementary Figure S5D). Although not investigated further, this is consistent with block of CRAC channels by the drug.

Analysis of Effects of CCBs on Cardiovascular Disease From Randomized Clinical Trials

The results of our bibliographic search, study selection, and summary of included studies are presented in theSupplementary section “Systematic Review and Network Meta-Analysis.” From the initial literature search which yielded 4740 publications, we sequentially pruned the list based on inclusion/exclusion criteria and quality of reported data to identify 25 randomized clinical trials, comprising a total of 160 325 participants for inclusion in the meta-analysis (Supplementary Figure S6). A summary of our quality evaluation of included studies is presented in the risk of bias graph (Supplementary Figure S7), showing the majority of studies had low risk of bias.

Supplementary Figures S8 and S9 show the meta-analyses results comparing all CCBs or DHP-CCBs, respectively, against other antihypertensive drug classes or placebo. For the analyses that only included DHP-CCBs, only the analyses involving diuretics and beta-blockers were affected with the exclusion of CONVINCE, INVEST, NORDIL, and VHAS trials. Five studies compared CCBs and placebo and showed a protective effect of CCB on incident HF (pooled relative risk: 0.77, 95% CI [0.65,0.93] [I² = 0%; chi²P = 0.005]). Angiotensin converting enzyme inhibitors (ACEI: 1.15[1.04,1.28] [I² = 0%; P = 0.006]), ARB (1.15[1.01,1.31] [I² = 34%; P = 0.03]), and diuretics (1.29[1.19,1.39] [I² = 33%; P > 0.00001]) showed a slightly lower risk of incident HF compared to CCB. Beta-blockers showed a lower risk of incident HF compared to all CCBs (1.15 [1.02,1.30] [I² = 0%; P = 0.02]), but showed neutral effect when compared to DHP-CCBs (1.06 [0.89,1.26] [I² = 0%; P = 0.54]). Inspection of funnel plots showed no evidence of publication bias (Supplementary Figure S10).

Network Meta-Analysis

The network meta-analysis including all trials comparing DHP-CCB with other antihypertensive drug classes or placebo are presented in Figure 6. The data show that placebo treatment compared to DHP-CCBs was associated with a 28% increased risk of incident HF (1.28 [1.07; 1.53]) and beta-blockers were neutral compared to DHP-CCB in risk of incident HF (0.95 [0.79; 1.13]). ACEI, ARB, and diuretics showed 13%-25% decreased risk of incident HF.

Figure 6.

Results of network meta-analysis. Treatments with direct comparison are linked with a line, the thickness of which is proportional to the number of included trials (single number adjacent to the line). The odds ratio and 95% confidence interval for each pair-wise comparison are presented adjacent to the lines with those in bold representing the network meta-analysis estimate with DHP-CCB as the referrant treatment. ACEI: Angiotensin converting enzyme inhibitors; ARB: Angiotensin receptor blockers; BB: Beta blockers; CCB: Dihydropyridine calcium channel blockers; DI: Thiazide; P: Placebo.

Real-World Analysis of CCBs

We undertook a prospective real-world analysis of patients with no prior history of cardiovascular disease and who had been newly prescribed 1 of the 5 classes of antihypertensive drugs as monotherapy (ACEI/ARB, Beta blocker, DHP-CCB, non-DHP-CCB, Thiazide). The demographics for the overall population and stratified by drug class are presented in Supplementary Tables S5 and S6. We estimated the risk of developing new onset HF in the following 1 yr of follow-up after a minimum 6 mo exposure to the drug. Additionally, we also estimated the risk of other cardiovascular outcomes—Ischemic Heart Disease (IHD), MI, Cerebrovascular Accident (CVA), and all Cardiovascular (CV) events (see Supplementary Table S7). For each outcome, we conducted univariable and multivariable [adjusted for age, sex, and Scottish Index of Multiple Deprivation (SIMD)] logistic regression to estimate the risk of the outcomes for DHP-CCB compared to each drug class. Of the 63 117 patients included in the analyses, the proportions on monotherapy were ACEI/ARB (41.2%), BB (29%), DHP-CCB (15.3%), non-DHP-CCB (3.1%), and Thiazide (11.3%). Multivariable adjusted logistic regression analyses showed DHP-CCB is associated with a significantly decreased risk of HF compared to ACEI/ARB—rate ratio (95% CI): 0.43 (0.33-0.56, P < 0.001), BB—0.47 (0.34-0.65, P < 0.001), and non-DHP-CCB—0.36 (0.23-0.57, P < 0.001), while it had a null effect compared with thiazide—1.16 (0.72-1.85, P = 0.543). Furthermore, we observe a similar pattern of risk for other cardiovascular outcomes (see Supplementary Table S7).

Discussion

The third generation dihydropyridine amlodipine is widely prescribed globally for treating hypertension and angina. It is favored because it preferentially targets CaV1.2 channels in vascular smooth muscle, reducing peripheral vascular resistance with little effect on cardiac output.²⁰ The preferential targeting of vascular smooth muscle is at least partly a consequence of a smooth muscle Cav1.2 splice variant that has higher sensitivity to dihydropyridines.^21–23 Consistent with this, amlodipine was found to be ∼30 times more effective in inhibiting Cav1.2 driven responses in an artery than in cardiac papillary muscle.²⁴

The long half-life of amlodipine allows for dosing once daily, increasing patient compliance, whilst providing sustained blood pressure control. Amlodipine is on the World Health Organization (WHO) list of essential medicines (http://apps.who.int/iris/handle/10665/325771). Amlodipine was the sixth most prescribed drug in the USA⁶ and fourth in the UK (if one discounts paracetamol and vitamin D).⁷ Amlodipine is widely prescribed in Asia; for example, it is the most common single antihypertensive agent in India.⁸ Although prevalent now, the incidence of hypertension is expected to increase further. A World Health Report has projected that by 2025, ∼1.56 billion people will suffer from chronic high blood pressure.²⁵

Recently, the therapeutic benefit of amlodipine and CCBs in general has been questioned by Johnson et al.,⁹ who reported that different types of CCB all activated store-operated Ca²⁺ entry in HEK 293 and rat aortic smooth muscle and stimulated vascular myocyte proliferation. An analysis of ∼20 000 patient records in the Penn State Healthcare system led them to suggest that CCBs increased the risk of HF, and therefore clinical use of this first line drug therapy should be reappraised.

Our data show that CCBs have complex and multimodal actions on store-operated Ca²⁺ entry, and different CCBs have very different effects. At therapeutic concentrations (sub-micromolar), no CCB activated store-operated Ca²⁺ entry. Nifedipine had no effect on cytosolic Ca²⁺ or store-operated Ca²⁺ entry even at high concentrations. By contrast, at pharmacological concentrations, amlodipine released Ca²⁺ from the thapsigargin-sensitive store. Whilst Ca²⁺ release was directly observable at concentrations >30 μm of amlodipine, Ca²⁺ mobilization was detectable even at 1 μm when Ca²⁺ extrusion across the plasma membrane was suppressed. Concentrations of diltiazem in the micromolar range activated store-operated Ca²⁺ entry in the absence of detectable Ca²⁺ release. Despite depleting Ca²⁺ stores, amlodipine evoked very little Ca²⁺ entry, and this was due to inhibition of store-operated Ca²⁺ channel activity. Hence, different CCBs either have no effect on store-operated Ca²⁺ entry (nifedipine), activate the channels (diltiazem), or inhibit them (amlodipine). Activation of store-operated Ca²⁺ entry by amlodipine is most likely a consequence of store depletion because Ca²⁺ influx was observed at concentrations that caused Ca²⁺ release.

One reason for why our data differ from that reported earlier⁹ relates to choice of Ca²⁺-sensitive fluorescent dye to measure cytosolic Ca²⁺. In the previous study,⁹ fura-2 was routinely used to measure Ca²⁺ signals in response to amlodipine and other CCBs. We have found that UV-based fura-2 Ca²⁺ indicator dyes are unsuitable for such experiments for two reasons. First, amlodipine and fura-2 have overlapping excitation spectra, which could be seen by the rise in emission to either 340 and 380 nm excitation when the drug was applied to extracellular solution in the absence of fura-2. Second, and more problematically, amlodipine accumulates within the cytoplasm and fluorescence emission to both excitation wavelengths increase substantially over several minutes of exposure. This dominates the fluorescence signal in fura-2-loaded cells.

We are not the first to find that amlodipine becomes more fluorescent as it builds up within the cytoplasm. Amlodipine was found to slowly accumulate within lysosomes and maneuvers that reduced acidity of these organelles diminished amlodipine fluorescence.¹⁷ It is tempting to speculate that the slow clearance of amlodipine from patients could be due, at least in part, to the intracellular accumulation such that lysosomes act as a reservoir for the drug, prolonging its levels in the plasma. Regardless, we have found that longer wavelength dyes, such as Cal520, are suitable for measuring cytosolic Ca²⁺ when dihydropyridines are used. With Cal520, we find no evidence for direct activation of CRAC channels by dihydropyridines. A recent study in human myometrial cells supports our findings; nifedipine failed to activate CRAC channels despite Orai1, 2, and 3 and STIM proteins all being expressed.²⁶

Our findings that pharmacological concentrations of amlodipine inhibit store-operated Ca²⁺ entry is in good agreement with prior literature. For example, in the low micromolar range, dihydropyridines inhibited store-operated Ca²⁺ entry in U937 cells,²⁷ in human uterine smooth muscle cells,²⁸ in vascular smooth muscle cells from human mammary artery,²⁹ in Jurkat T lymphocytes³⁰ and in leukemic HL-60 cells.³¹ Amlodipine has also been found to partially deplete the thapsigargin-sensitive Ca²⁺ pool in vascular smooth muscle,³² consistent with our results that show amlodipine reduced the thapsigargin- and ionomycin-mobilizable Ca²⁺ store.

Effects of amlodipine on vascular smooth muscle proliferation remain unclear. Amlodipine was reported to increase vascular myocyte proliferation,⁹ but other studies showed dihydropyridines, including amlodipine, applied over a similar concentration range, inhibited vascular myocyte proliferation.^33–36 The Ca²⁺-dependent transcription factor, NFAT, is linked to expression of genes associated with proliferation. However, we were unable to detect activation of NFAT-dependent gene expression in response to amlodipine, over a range of relatively low concentrations.

A critical question is whether the concentrations of amlodipine used in vitro to affect CRAC channels are of any relevance to the therapeutic levels seen in the serum of patients. The serum levels of amlodipine have been measured in patients and comatose-fatal levels of amlodipine are in the range 0.245-0.490 μm (0.1-0.2 mg/L).³⁷ Toxic levels are seen at 0.215 μm and therapeutic levels are in the range 0.73-36 nm.³⁷ The mean plasma concentration-time profile of amlodipine has also been measured in human subjects in response to an oral therapeutic dose (2.5 mg) and was ∼3 nm.³⁸ A plasma concentration of 0.5 μm amlodipine or more would therefore be fatal to a patient and is of no clinical relevance. Dihydropyridines block Cav1.2 channels in vascular smooth muscle with an IC₅₀ in the low nm range at a holding potential close to the resting membrane potential.²³ These values are well within the range of serum levels measured in patients. Therefore, at therapeutic levels seen in patients, amlodipine will inhibit Cav1.2 channels, leaving CRAC channels unaffected.

There is considerable evidence from systematic review of RCTs that show effective antihypertensive therapy with ACEI, ARB, diuretics, and CCBs reduce the risk of incident HF.^39,40 The relative efficacy of each drug class in terms of magnitude of protection against new onset HF put ACEI, ARBs, and diuretics as the most effective. Two meta-analyses in 2009 and 2011 show CCBs have a protective effect against incident HF.^39,40 Our updated meta-analysis and network meta-analysis confirms and extends previous findings that DHP-CCBs are protective against incident HF. We follow this up with a real-world analysis of over 63 000 patients without prior cardiovascular disease commenced on monotherapy with 1 of the 5 antihypertensive drug classes (Supplementary Figure S7). We find that DHP-CCBs are associated with significantly decreased risk of incident HF, ischemic heart disease, and myocardial infarction over a 1-yr follow-up period compared to ACEI/ARB, beta-blockers, and non-DHP-CCBs. The difference between the real-world analyses and the meta-analysis of RCTs may reflect the population characteristics where the real-world analysis included patients who were on monotherapy and without pre-existing cardiovascular disease, thus representing a low-risk cohort. While confounding by indication cannot be ruled out in the real-world analysis, as CCB monotherapy is generally offered as first line antihypertensive therapy to low-risk patients, the unequivocally reduced risk of incident HF compared to ACEI/ARB and BB in the real-world study coupled with the unequivocal protective effect of DHP-CCB against placebo in the meta-analysis of RCTs provide reassurance that DHP-CCBs do not increase risk of HF. We note DHP-CCBs are also protective against incident HF compared with non-DHP CCBs.

The conclusions from our meta-analysis of a large number of random controlled trials along with our real world analysis, where HF diagnosis was adjudicated by an independent panel of experts, contrast with that reached in an earlier study.⁹ In addition to cohort size, several other factors could contribute to the discrepancy. Those authors used a cross-sectional study simply comparing co-prescription of CCB with HF versus HF with non-CCB drugs.⁹ This is not the preferred method to infer a relationship between any drug and an outcome as this is subject to wide-ranging confounding and biases. No differentiation was made between CCBs (verapamil and diltiazem) that have greater cardio-depressant effect and DHPs. The former are not prescribed in patients with HF or at high risk of HF compared with the latter. The two groups in the previous study were not randomized, nor blinded in terms of the drug intervention and thus any observed differences are likely to be confounded.⁹ Furthermore, there was no adjustment for known confounders, including age, ethnicity, smoking, prevalent cardiovascular disease, prevalent diabetes, or renal disease.⁹ The control group in the previous study that was not on CCBs might have been on drugs that have a protective effect against HF, such as ACE inhibitors, angiotensin receptor blockers, or beta-blockers.⁹

Our data suggest that removal of dihydropyridines from the arsenal of drugs available for treating hypertension is not advisable. Cessation of DHP-CCBs, widely used as a first line of treatment, would have a negative impact on blood pressure control, increasing the risk of secondary morbidities, including myocardial infarction, HF, stroke, retinopathy, and kidney disease.

Funding

This work was supported by the US National Institutes of Health Intramural Research Program; the US National Institute of Environmental Health Sciences (NIEHS; ZIA ES103353-03 to A.B.P.). S.P. is supported by the British Heart Foundation Centre of Excellence Award (RE/18/6/34217) and the UKRI Strength in Places Fund (SIPF00007/1).

Conflict of Interest

A.B.P. holds the position of Executive Editor for Function and is blinded from reviewing or making decisions for the manuscript.

Data Availability

The data underlying this article will be shared on reasonable request to the corresponding authors.