Quantification of Human Cardiac Ion Channels by Parallel Reaction Monitoring and In-Sample Calibration Authors: Soroush Torkamannejad, Hoi-ying Yip, Bingyun Sun Published: 2026 ABSTRACT Cardiac ion channels are responsible for action potentials in cardiomyocytes. Dysfunction of ion channels in human hearts by abnormal expression levels causes electrical remodeling in the heart and arrhythmia. Direct quantification of ion channels is scarce because these proteins are large, multipass transmembrane proteins. No targeted proteomic approaches have been carried out to study the copy numbers of these ion channels in human hearts. Following our previous effort to use extended enzymatic digestion on the sodium deoxycholate (SDC) insoluble fractions of the human ventricle to obtain a more comprehensive proteome of sarcomere proteins, we report here the ion channel subunits quantified in copy numbers in the same fraction by a targeted parallel reaction monitoring (PRM) proteomic approach. Using Heavy-Match-Light (HML) in-sample calibration, we successfully quantified three major endogenous ion channel subunits, Nav1.5, hERG, and Cav1.2, and the selected auxiliary subunits, Cavβ1 and Cavα2δ1, in the SDC insoluble fraction of the human ventricle. Our approach showed higher sensitivity than global proteomics for identifying these large membrane proteins. We hope that the developed targeted proteomic method can allow future systematic quantification of these functionally important membrane proteins. We also hope that the absolute quantity obtained from the copy numbers of these proteins allows more comparisons across studies and biological systems, as we demonstrated here for better biological insights. Introduction Cardiovascular diseases (CVDs) are the leading cause of death globally. In 2019 alone, 17.9 million global deaths were attributable to CVDs, and by 2030 this number could rise to 22.2 million. , Almost all heart diseases result in an irregular rate, rhythm, or cessation of beating at the end stage of heart failure or sudden cardiac arrest. Although tremendous advances have been made in cardiac therapy, heart diseases remain the leading cause of death worldwide. Cardiac muscle contraction is regulated by the electrical system through excitation-contraction coupling. The key regulatory factor, i.e., the cardiac action potential, is generated by the coordinated opening and closing of more than a dozen major types of ion channels in the heart. Protter illustrations depicting the structure of some selected ion channel subunits, including Nav1.5, hERG, Cav1.2, Cavβ1 and Cavα2δ1, with respect to the plasma membrane are shown in Figure . Illustration of the transmembrane topology of the studied ion channel subunits by Protter and the location of the targeted peptides in each of these proteins with zoomed-in view of the amino acid composition. (A) Nav1.5 protein, SCN5A gene with UniProt entry of Q14524; (B) hERG protein, KCNH2 gene with UniProt entry of Q12809; (C) Cav1.2 protein, CACNA1C gene with UniProt entry of Q13936; (D) Cavβ1 protein, CACNB1 gene with UniProt entry of Q02641; (E) Cavα2δ1 protein, CACNA2D1 gene with UniProt entry of P54289. Maintaining stable and normal expression of these ion channels is critical for normal heart function. For example, Nav1.5 is responsible for the fast upstroke in phase 0 of the cardiac action potential. Nav1.5 is more than 220 kDa, has 24 transmembrane passes, and forms four homologous domains, , as shown in Figure A. Malfunction of Nav1.5 can cause many heart diseases including longQT3 and Brugada1 syndromes. HERG participates in phase 3 of the cardiac action potential, in which the hERG produces an inward rectifying current acting along with several other K+ channels such as KvLQT1, Kir 2.1/2.2 and Kir 6.2 to repolarize the cardiac membrane potential. With a molecular weight greater than 120 kDa and 6 transmembrane passes as shown in Figure B, hERG forms tetrameric K+ channels, and is a frequent target responsible for cardiac drug toxicity induced arrhythmia. Malfunction of the hERG also contributes to longQT and shortQT syndromes. Cav1.2 is the major L-type voltage-gated Ca2+ channel in the myocardium, which is responsible for the depolarization in phase 2 of the cardiac action potential and indirectly contributes to excitation-contraction coupling through Ca2+ triggered Ca2+ release. Like Nav1.5, Cav1.2 is over 240 kDa, also has 24 transmembrane passes and forms four homodomains as shown in Figure C. Over or under expression of Cav1.2 causes longQT and Brugada syndromes, respectively, along with other cardiac diseases such as hypertrophic cardiomyopathy. Cav1.2 also contains well-studied and critical auxiliary Cavβ (Figure D) and Cavα2δ (Figure E) subunits in the heart to facilitate its membrane expression and activity. We detected two important isoforms, Cavβ1 (Figure D) and Cavα2δ1 (Figure E) in the heart. Malfunction of these proteins can cause shortQT and Brugada syndromes. Accurate and direct quantitation of these large ion channel proteins is limited. Currently, most quantifications of ion channels in cardiac tissue involve qPCR on the corresponding transcripts. − Direct detection of expressed proteins often involves antibody-based immunofluorescence imaging or Western blotting. , These measurements provide relative quantitation to the corresponding controls and often serve as a validation of the results from transcripts. , Because mRNA and protein are expressed at different levels, only the trend of increase, decrease or no change can be used for validation, a relative quantity even though it is meaningful only to the control is sufficient for these studies. However, owing to the large number of possible controls available for different laboratories, ranging from human cardiac tissues, isolated human cardiomyocytes and human induced pluripotent stem cell-derived cardiomyocytes (hiPSC-CMs) to cells and tissues from a variety of animal models such as pigs, dogs, rabbits, and rodents, to heterologous expression systems, the controls used are often different in studies carried out by different laboratories, and the obtained trends are sometimes conflicting. A powerful technique that can characterize ion channel protein function on the cell membrane is the use of patch clamp technique. , Analysis is typically performed on individual cells and flexible for characterizing a single channel or the sum of one type of channel from an entire cell. The obtained values have been the gold standard for the construction of various mathematical cardiac models. However, each electrophysical measurement can target only one of a dozen major types of ion channels in cardiomyocytes. In addition, the isolation of individual cardiomyocytes eliminates valuable cell–cell interactions, which can be crucial to the functions of certain channels, such as Nav1.5, that colocalize with desmosomes in the intercalated disk, which only forms in cell–cell junctions. We intend to develop a quantitative mass spectrometry (MS)-based assay and to accurately and directly determine the absolute quantity, i.e., the copy number, of ion channel proteins in human cardiac tissues, which maintains the native cell–cell interaction states, does not need antibodies, and can be compared with similar results obtained from any cardiac system across different laboratories similar to those from patch clamp. Our assay uses nanoflow liquid chromatography (LC) coupled with tandem MS (MS/MS) and parallel reaction monitoring (PRM) as the data acquisition method for quantitation. In PRM quantitation, a calibration curve is established for every fragment of each peptide from all target proteins using stable isotope incorporated reference peptides with known concentrations. Furthermore, multiple fragments from the same peptide, and multiple peptides from the same protein can be simultaneously assayed for accurate and sensitive measurements of absolute quantity. From the calibration curve, several necessary performance parameters of the assay, such as the sensitivity, the lower limit of quantitation (LLOQ), and the linear range can be assessed for the quantitation reliability. To date, an external calibration curve is often used and built in a surrogate sample matrix, because the authentic sample matrix has endogenous analytes that cannot provide the background signal that is required to assess the LLODs. Alternatively, surrogate analytes have been used to be spiked into the authentic matrix for calibration, in which stable isotope incorporated heavy peptides have identical chemical properties as the target analytes; therefore, they are the best choices. In such cases, an internal standard is also needed to normalize sample loading. To meet such requirements, a second stable isotopologue is typically employed or a chemical analogue is used as a compromise. , The matrix effect is among the main factors hampering the sensitivity and accuracy of MS-based absolute quantitation. To minimize the impact of the matrix effect on the quantitation, the calibration curve needs to be constructed in the same matrix as that of the samples. However, obtaining a surrogate sample matching the matrix of the authentic sample is technically challenging. − Standard addition as an in-sample calibration (internal calibration curve) method is widely used to address the matrix effect, but it cannot evaluate LLOD owing to the presence of endogenous analytes. The use of alternative analytes is also rare as it requires two heavy isotopologues or one heavy isotopologue and one chemical analogue with compromised performance. To address this challenge, we developed a new in-sample calibration strategy differing from Standard Addition. We termed our method Heavy-Match-Light (HML) method, in which we used one heavy stable isotopologue as the “calibrator” and used the endogenous peptide as the “internal standard” to construct the calibration curve. This method is better than the standard addition approach as it allows the evaluation of the true matrix effects and the LLOD values for each peptide analyte of interest. Our approach is simpler and more economical than the alternative analyte approach because it uses only one stable isotopologue. We have applied this method to quantify the voltage-gated K+ channel hERG in a heterologous expression system and compared the results with those obtained from the standard addition method as well as values estimated from single-cell patch clamp. Overall, the results agreed well. In the current study, we applied our HML method to study ion channels in human cardiac tissue. This is the first application of a targeted MS-based quantitation strategy to human cardiac tissue. We discuss the challenges encountered, the solutions we took, and our results compared with known quantities of these ion channels measured by global proteomics approaches. We hope our study will provide an effective tool to help advance the knowledge on cardiac electrophysiology and cardiac diseases. Materials and Methods Materials: Sodium deoxycholate (SDC), tris­(2-carboxyethyl)­phosphine (TCEP), dithiothreitol (DTT), iodoacetamide (IAA), ethyl acetate, bovine serum albumin (BSA), ovalbumin from hen egg white, and protease inhibitor cocktail (PIC) were obtained from Sigma-Aldrich (St. Louis, MO). RapiZyme trypsin, MS grade (cat. No. 186010106), and MCX cartridges (cat. No. 186000252) were obtained from Waters (Milford, MA). Heavy [13C6, 15N2] lysine- or [13C6, 15N4] arginine-incorporated peptides as summarized in Table were customized by JPT Peptide Technologies (Berlin, Germany), and were selected from 5 ion channel proteins. Additional heavy peptides were selected from two reference proteins, i.e., bovine serum albumin, and hen ovalbumin to assess the recovery and quantitation accuracy of the assay. MS grade acetonitrile, formic acid (FA), and HPLC grade water were purchased from Thermo Fisher Scientific (Mississauga, ON, Canada). All the other chemicals without specification were obtained from Fisher Scientific (Pittsburgh, PA). Snap frozen human left ventricular myocardium from the biopsy of a post-mortem donor deceased from noncardiac-related causes was purchased from AMSBIO LLC. (Cambridge MA, USA). 1Parameters for Peptides of Target Proteins Including the Normalized Collision Energy (NCE), the monitored transitions, their Corresponding Charges (z), Mass to Charge Ratios (m/z), and Lower Limit of Quantitation (LLOQ) Values m/z m/z LLOQ (fmol)proteingenepeptideNCEtransition Z (Light)(Heavy)suppelNav1.5SCN5AYFFSPTLFR27precursor2589.3057594.30980.30.1 precursor [M+1]2589.8072594.81140.030.2 precursor [M+2]2590.3086595.312820.2 y811014.5411024.54900 y71867.4723877.480600 y61720.4039730.412200 y51633.3719643.380100 b21311.139311.13900 b31458.2074458.207400.1 b41545.2395545.2395 0.4 y52317.1896322.193700YQGFIFDIVTK22precursor2665.8557669.8628 0.2 precursor [M+1]2666.3572670.364450.02 precursor [M+2]2666.8587670.865852 y911039.5821047.597 0.08 y81982.5608990.57500 y71835.4924843.506 0 y61722.4083730.422500 y31347.2289355.2431 0 y21248.1605256.1747 1 b21292.1292292.129200 b31349.1506349.1506 0DSDSGLLPR22precursor2480.2433485.247410.04 precursor [M+1]2480.7447485.74890.90.02 precursor [M+2]2481.2459486.2501 2 y71757.4203767.428500.02 y61642.3933652.401600 y51555.3613565.369600 y41498.3398508.348100 y31385.2558395.26400 y11175.119185.1272 0.1 y21272.1717282.180.20 b21203.0662203.06620.20GSTTLQESR20precursor2489.7462494.75030.10.06 precursor [M+1]2490.2476495.2518 0.03 y61733.3839743.392200.03 y51632.3362642.34450.020 y41519.2522529.260400 y31391.1936401.201800 y72417.7194422.72360.90 y32196.1004201.10460 b51460.2402460.240210 b2273.03473.0340 hERGKCNH2ALVGPGSPPR22precursor2475.7745480.77870.10.01 precursor [M+1]2476.276481.28020.20.2 y81766.4206776.42890.10 y71677.3522677.36051 × 10–6 0 y61610.3307620.33900 y31369.2245379.232700 y32185.1159190.1200 b31284.1969284.196900 b41341.2183341.21830.020 SAPGQLPSPR32precursor2505.2749510.279 0.2 precursor [M+1]2505.7763510.780521 y81581.4734861.481600 y71754.4206764.428900 y61697.3991707.407400 y51569.3406579.348800 y41456.2565466.264800 y21272.1717282.180.20 y11175.119185.12720.80.08 b21159.0764159.0764 0 GHVAPQNTFLDTIIR30precursor3561.3055564.64161 precursor [M+1]3561.6398564.97590.03 y51617.3617627.370 y41502.3348512.3430.05 y11175.119185.12810.3 b41365.1932365.19320.04 b61590.3045590.30560 b42183.1002183.1002 0 b62295.6559295.655900Cav1.2CACNA1CADGANALGGK19precursor2437.2249441.2320.30.4 precursor [M+1]2437.7263441.73341E10.7 precursor [M+2]2438.2275442.2346 2E1 y81687.3784695.392600.07 y61559.3198567.33400.03 y51445.2769453.291100 y31261.1557269.169900 b21187.0713187.07130.20.2 b31422.0928244.09280.10 AGSSQGDTESPSHEK19precursor3506.2219508.89334 precursor [M+1]3506.5561509.22750.2 y61684.3311629.34530 y51597.2991605.31330.05 y142723.3106727.31770.06 y132694.7999698.8070.1 y102543.7385547.74560.07 b21129.0659129.06590.001 Cavβ1CACNB1HLNVQIAASEK28precursor [M+1]2605.8344609.84153E15.7 y1011072.61080.61402 y91959.5156967.52980 y81845.4727853.48690 y31363.1874371.20164 y11147.1128155.1271E13 b21251.153251.15031E14 ALFDFLK22precursor2427.2445431.2516 3 precursor [M+1]2427.7461431.7532 0.5 precursor [M+2]2428.2475432.2546 1 y51669.3606677.3748 1 y41522.2922530.3064 0.6 y31407.2653415.2795 0.3 y21260.1969268.2111 0.1 y11147.1128155.127 0.3 b21185.1285185.1285 0.7Cavα2δ1CACNA2D1FVVTDGGITR19precursor2532.7904537.7945 0.2 precursor [M+1]2533.2918538.296 0.2 precursor [M+2]2533.7932538.7974 0.2 y81818.4367828.44490.40.4 y71719.3682729.3765 0.3 y61618.3206628.32880.30.3 y51503.2936513.30190 b21247.1441247.14410.080.6 b31346.2125346.212500.1 VFTFSVGQHNYDR19precursor3523.9197527.25582E13 precursor [M+1]3524.254527.590178 precursor [M+2]3524.5882527.92433E13 y71889.3911899.399400 y122735.8417740.845830.4 y112662.3075667.31161E15 y102611.7836616.787872 y92538.2494543.253560.8 b21247.1441247.14412E10 YQDLYTVEPNNAR25precursor2791.8784796.88264E1 precursor [M+1]2792.3799797.38419E2 precursor [M+2]2792.8812797.88546E2 y911063.5171073.5250.9 y81900.4534910.46160.4 y61700.3373710.34550 y51571.2947581.3030.5 b21292.1292292.12920 b31407.1561407.15610 Parameters for Peptides of Target Proteins Including the Normalized Collision Energy (NCE), the monitored transitions, their Corresponding Charges (z), Mass to Charge Ratios (m/z), and Lower Limit of Quantitation (LLOQ) Values Protein extraction from myocardium. This study was approved by the Institutional Research Ethics Committee at Simon Fraser University (No. 30001371). The procedure was based on published protocols ,, with minor modifications. In detail, frozen left ventricular myocardium from a female post-mortem donor was removed from −80 freezer and immediately shattered in liquid nitrogen held by a ceramic mortar. The selected tissue pieces were transferred to an Eppendorf vial, weighed, and cut into smaller sizes before being rinsed with chilled hypotonic buffer (10 mM Tris, 10 mM EGTA, 10 mM KCl, 1.5 mL MgCl2) with 1:100 diluted protease inhibitor cocktail (PIC) on ice to remove blood. The cleaned tissue was submerged in 500 μL hypotonic buffer and homogenized by a Precellys homogenizer­(Bertin Technologies, Montigny-le-Bretonneux, France), a commercial beat beater, at 4 °C and 6500 rpm for 2 × 20 s cycles with ceramic beads. The homogenized sample was centrifuged at 14,000 rpm for 5 min to separate the supernatant (soluble fraction) from the membrane enriched pellet. The membrane fraction was dissolved in 20% SDC by being boiled at 100 °C for 10 min. The protein content of the soluble fraction and the supernatant of the membrane fraction were quantified by bicinchoninic acid (BCA) assay (Thermo Fisher Scientific, Mississauga, ON, Canada). Proteolysis of the insoluble membrane fraction. The pellet resistant to 20% SDC was dispersed in 20% SDC, 10 mM EDTA, 10 mM TCEP, and 40 mM Tris at pH 8 and further boiled at 100 °C for 10 min. The sample suspension was allowed to cool to room temperature before urea powder was added to the sample for a final concentration of 8 M. The sample vial was incubated at 37 °C for 30 min with end-to-end rotation, before IAA was introduced for a final concentration of 10 mM to alkylate the free thiols in the dark at room temperature for another 30 min with end-to-end rotation. DTT (10 mM) was then introduced to quench the reaction for 15 min at room temperature with rotation. The suspension was diluted to 0.5 mL with 40 mM Tris at pH 8, and a final concentration of 10 mM CaCl2 was introduced to facilitate the following proteolysis. RapiZyme trypsin (2 μg) was introduced in the suspension prior to the incubation of the sample vial at 37 °C overnight with end-to-end rotation. The next day, an additional 2 μg of trypsin was introduced, before the third dose of 2 μg trypsin was introduced after the second day of digestion. The digestion was stopped by acidifying the sample to pH 2–3 with 1 M HCl after 3 consecutive days of trypsin proteolysis at 37 °C with end-to-end rotation. Sample purification. The sample was spun at 10,000 rpm for 10 min, and the supernatant (SDC supernatant fraction) was separated from the pellet that contained acid-insoluble SDC. The pellet was resuspended in ammonium hydroxide to recover potential sample peptides precipitated with SDC, and the SDC in this sample was removed by three repeated phase extractions with ethyl acetate at a sample to extractant volume ratio of 1:6. After phase extraction, the aqueous solution from the sample was dried by a SpeedVac (Model SPD121P-230 from Thermo Fisher Scientific, Mississauga, ON, Canada) and resuspended in 0.1% TFA (SDC pellet fraction). Both the SDC supernatant and the pellet fractions were further purified with MCX cartridges and dried in a SpeedVac prior to nanoLC-MS/MS analysis. LC-MS/MS Analysis The purified and dried peptides were reconstituted in MS loading buffer (0.1% FA), and various amounts of 13C- and 15N-labeled reference peptides were spiked into the sample (final concentrations ranged of 4 ng/mL to 1.4 μg/mL) before LC-MS/MS analysis by an EASY nLC 1000 coupled with a Q-Exactive HF Hybrid Quadrupole-Orbitrap mass spectrometer through a nano-EASY-Spray source (all obtained from ThermoFisher Scientific, Mississauga, ON, Canada). A control sample not spiked with any heavy peptides was also prepared for identical LC-MS/MS analysis. The separation was carried out using a 2 cm trap column and a 10 cm analytical column (both packed with Magic C18aq from Michrom Bioresources, Inc.) of 75 μm ID and 3 μm resin with 100 Å pore size for a 60 min gradient of 2–35% buffer B (0.1% FA in acetonitrile). A voltage of 2.3 kV was used for electrospray, and the ion transfer tube was heated to 320 °C. For Q-Exactive HF MS/MS analysis, a parallel reaction monitoring mode was used to target an inclusion list of 32 peptide precursors for both the isotopically labeled heavy reference and the light peptides from 5 ion channel proteins as well as BSA and ovalbumin. MS2 spectra were collected after optimized higher-energy collisional dissociation (HCD) fragmentation, which ranged from 19% to 35% normalized collision energy (NCE). MS1 precursors scans were acquired in the m/z range of 400–1200 with a mass resolution of 120,000, and the MS2 resolution was 30,000 with automatic m/z ranges. The automatic gain control (AGC) values for the MS1 and MS2 scans were 3E6 and 2E5, respectively, and the ion injection (IT) times for MS1 and MS2 were 200 and 100 ms, respectively. The mass isolation window for MS2 was 2.0 m/z offset by 0.5 m/z to include more isotopic peaks of the target peptides. Skyline Data analysis. Like a previously reported procedure, the area under the curve of both heavy and light peptides was reported from Skyline v21.1, after manually selecting the peak boundaries. The selection of the transition peaks followed the criteria we reported previously. Our criteria were that (1) all transitions including both the precursors and the fragments from the same peptide sequence including both the light and the heavy form should have the same retention time; (2) the retention time of the targeted peptides identified in Skyline should match those reported from the searches carried out by Sequest against the human UniProt database via Discoverer. To take advantage of the complete MS2 spectra recorded by the Q-Exactive HF, a maximum of ten transitions from each peptide precursor were selected for analysis. The ratios of the peak areas between light and heavy peptides with the concentration of the spiked heavy peptides in each dilution were used to generate the calibration curve. Each calibration curve was constructed on five heavy peptide concentrations including the sample-alone blank (without spiking with heavy peptides) via linear regression for each transition. The reproducibility of the assay was evaluated by up to four repeated runs on different days, and for each run, a different set of calibration curves was generated for all the transitions. All of the R 2 values of the calibration curves were assessed, and only reliable results with R2 values greater than or equal to 0.96 were chosen for quantification. The lower limit of quantitation (LLOQ) was computed as LLOQ = 10 s/m, , in which s is the standard deviation of the background noise, i.e., the response in the heavy peptide-free sample, and m is the slope of the calibration curve. This calibration curve allows us to determine the absolute quantity of the target proteins in the sample using the Heavy Match Light (HML) method. In detail, because the light and heavy peptides have identical sequences and coelute at the same retention time experiencing the same ionization efficiency, when the area ratio between the light and heavy peptides was 1, the concentration of the heavy peptide matched that of the light peptide. Therefore, the endogenous light peptide concentration was calculated by obtaining the corresponding x axis value (heavy peptide concentration) when the y axis value (area ratio between light and heavy peptide) was “1”. Results and Discussion Ion channels are specialized transmembrane proteins responsible for regulating the flow of ions across the cell membrane. Many ion channels are expressed in the human heart and can be activated through a variety of mechanisms including ligand-gated, lipid-gated, and voltage-gated channels. There are three classes of voltage-gated ion channels, i.e., sodium, potassium, and calcium ion channels, that function critically in the heart. Each ion channel is a large molecular complex with several subunits, including at least one pore-forming α subunit and one or several auxiliary subunits. For example, cardiac Ca2+ channels typically have one β subunit and one α2δ subunit as auxiliary subunits shown in Figure . These auxiliary subunits can largely increase the surface presentation of the α subunit. Each subunit also has several isoforms, for example the voltage gated sodium channel has 9 isoforms, Nav1.1-Nav1.9. As a polit study, we selectively characterized some major α subunit isoforms in each of the three major cardiac ion channels mentioned above. Specifically, we studied the human Nav1.5 protein encoded by the SCN5A gene and responsible for the INa current in the voltage-gated Na+ channels, the human ERG (hERG) protein encoded by the KCNH2 gene and responsible for the IKr current in the voltage-gated K+ channel, and the human Cav1.2 protein encoded by the CACNA1C gene and responsible for the ICa,L current in the voltage-gated Ca2+ channels. We also targeted the Cavβ1 (CACNB1) and Cavα2δ1 (CACAN2D1) subunits in voltage-gated Ca2+ channels for measurement. Table summarizes the targeted proteins, their corresponding genes, our detected peptides, and their optimized collision energy, as well as the corresponding m/z values of the monitored fragments for both the endogenous light form and the stable isotope incorporated heavy form. The MS2 spectra of the newly selected peptides excluding those previously published by us ,, are summarized in Figure . The m/z values of the selected transitions for both light and heavy peptides are included in Table . MS2 spectra of the targeted peptides, in which color denotes the type of ions matched to the spectral library. Purple marks b ions, blue marks y ions, and yellow marks c ions. Highlighted fragments are chosen transitions with their specific m/z values listed in Table . Ion channel proteins are rarely quantified in cardiac tissues, because most α-subunit proteins are large membrane proteins as shown in Figure . In cardiomyocytes, the major cell type in the heart, the dominant myofiber structure and the tenacious intercalated discs that are responsible for holding beating cardiomyocytes together throughout the entire lifespan of humans are difficult to solubilize in aqueous solution for quantitative measurements. In conventional proteomic studies of human hearts, only the soluble fraction was analyzed, and the insoluble fraction was discarded. Recently, we reported that the typically discarded insoluble fraction included a substantial number of proteins after 3 days of extended trypsin digestion in SDC containing buffer. Because of the large size and membrane residing nature of ion channels, we suspect that this fraction may contain a large number of ion channels. Therefore, we carried out targeted PRM quantification on this insoluble fraction of human heart tissue homogenate. Because we used SDC to solubilize the membrane fraction of the heart lysate, we named this fraction the SDC insoluble fraction. In the past, we discovered that in the detergent (SDC) removal step, substantial amounts of peptides remained in the SDC pellet after acid precipitation. , We therefore quantified channel proteins in both the supernatant after SDC removal and the SDC pellet. For each fraction, 4 repeated measurements were carried out on different days to assess the reproducibility of the assay. The extracted ion chromatograms from Skyline on the selected transitions from both the heavy and light peptides of all the proteins are shown in Figure . Because the matrix effect can largely affect the quantitation accuracy, we performed a previously developed HML in-sample calibration to best evaluate the matrix effect in the authentic matrix. In our approach, various concentrations of heavy peptides were spiked into the sample solution as calibrators. The stable-isotope dilution approach was used, and the endogenous light peptides were treated as internal standards to obtain the Heavy/Light ratio against the concentrations of the Heavy calibrators for the calibration curve as shown in Figure . Nonspiked samples from four replicates without calibrators can be used to accurately assess background noise and LLOQ. Because two fractions, the SDC supernatant and SDC pellet, were separately analyzed and each with four technical replicates, the fraction specific LLOQ values are summarized in Table . The LLOQ results indicated that the SDC pellet fractions more frequently presented lower detection limits than did the supernatant fraction, which agreed with our previous observations on porcine heart tissues. In our global proteomic analyses of the two fractions, the supernatant fraction contained one fold more proteins and peptides than did the pellet, suggesting that the matrix effect was greater in the supernatant fraction, which in part explained the better LLOQs in the pellet. Extracted ion chromatograms of reference proteins and measured ion channel subunits. Furthermore, in our HML approach, the use of endogenous peptides as the internal standards allowed us to eliminate the use of second heavy isotopologue or chemical analog as internal standards. The advantages of such treatment include reducing the synthesis cost of additional isotopologues that are expensive, simplifying the spectra, and preventing the compromise of choosing chemical analogs instead. The challenge for such treatment is that our target endogenous analytes are used to build the calibration curve itself, which is not available to obtain its concentration by the typical calibration approach, in which the unknown target analyte signal should be an independent signal from those included in the calibration curve. To solve this problem, we implemented a mathematical trick, in which we selected a special Y value of “1”, when the heavy and light peptide concentrations were equal, and the corresponding Heavy calibrator concentration was the Light endogenous peptide concentration. Because the preparation procedure was lengthy with extensive digestion and purification, we also spiked two reference proteins, BSA and ovalbumin, with known concentrations at the beginning of the 3-day digestion to evaluate sample recovery. Table summarizes these proteins, their peptides, the optimized collision energy, and the transition details of both heavy and light peptides with corresponding LLOQs. The MS2 spectra of the selected peptides have been previously detailed. We summarized the selected extracted ion chromatograms of their transitions and the calibration curves in human heart samples in Figure . To eliminate transitions suffering from severe interference, we used R 2 > 0.96 of the calibration curve to remove poor transitions. 2Parameters for Peptides of Reference Proteins m/z m/z LLOQ (fmol) proteingenepeptideNCEtransition Z (light)(heavy)suppelBSAALBAEFVEVTK22precursor2461.7477465.7548 0.05 precursor [M+1]2462.2492466.2563 0.1 precursor [M+2]2462.7505466.7576 0.7 y62361.7078365.7149 0.05 b21201.087201.087 0.04 y21248.1605256.17470.10 y41476.2715484.285700 y51575.3399583.354100 y61722.4083730.4225 0.05 b31348.1554348.15540.50 b51576.2664576.26640.060OvalbuminSERPINB14GGLEPINFQTAADQAR22precursor2844.4236849.42775E23E1 precursor [M+1]2844.925849.9292 1 precursor [M+2]2845.4263850.4305 2E2 y122666.3388671.3429 0 b31228.1343228.1343 0 b41357.1769357.176930 y1011121.5331131.54200.04 y1211331.671341.679 0 y911007.491017.49900 y81860.4221870.430300 Parameters for Peptides of Reference Proteins The differential quantitation of all targeted peptides from both the SDC-supernatant and the SDC-pellet fraction is shown in Figure . The results suggested that the majority of peptides were detected in both fractions. Both our study and others reported that the peptides in the SDC pellet were more hydrophobic. We analyzed the Gravy score (https://www.gravy-calculator.de/index.php, accessed in 2024–2025) detailed in our previous study of the current peptides and labeled them on top of each peptide quantity value. A further analysis of the distribution of peptides between the two fractions was carried out to determine the peptide quantity ratio between the SDC pellet and SDC supernatant over the gravy score. The results are shown in Figure , in which a positive correlation can be observed with a correlation of determination of 0.48 for linear correlation. These results largely agreed with previous reports that hydrophobic peptides can partially be coprecipitated with SDC such that phase extraction can be used to recover them. Some reports have suggested the use of phase extraction over acid precipitation for SDC removal to minimize peptide loss in the SDC pellet. Our studies suggested that even though it is tedious to perform both acid precipitation and phase extraction, the resolved pellet can function as one additional fraction with less peptide complexity and matrix effects for better LLOQs. In addition, a further analysis of the coefficient of variation (CV) suggested that the quantitation in pellet fraction has less variations (CV = 0.2) than that of the supernatant (CV = 0.3), which can also be explained by less matrix effect in the pellet fraction. Quantity of targeted peptides from each ion channel subunit in SDC pellet and SDC supernatant. The GRAVY score is on top of each peptide. Quantity ratio of commonly detected peptides between SDC pellet and SDC supernatant fractions as a function of their Gravy score. To allow our results to be comparable to future results, we performed absolute quantitation and reported the quantity of the measured ion channel proteins as copy number per cardiomyocytes. To obtain the quantity of peptides, all the transitions of the same peptide were further averaged. The corresponding quantity of proteins was inferred from the average of the surrogate peptides. The copy number of each ion channel protein was calculated by dividing the quantity of ion channel subunits over the number of cardiomyocytes in the tissue. The number of cardiomyocytes in our sample was estimated based the on 24.8 million/g of female human left ventricle derived from the studies of Bergmann et al., which matched the sex and heart chamber of our sample. Such proximity has certain errors, yet for the very first report of heart ion channels in copy numbers, it can serve as the foundation for more accurate measurements in the future. All the copy numbers obtained from the selected proteins and their corresponding peptides are summarized in Figure A. In addition, the percentage of recovery for the two reference proteins is also summarized in Figure B, in which BSA showed satisfactory recovery (∼100%), whereas the recovery of ovalbumin was poor (<50%). Such recovery was consistent with our results obtained from porcine heart tissue analysis, suggesting robust peptide-dependent losses instead of sample dependence, i.e., human vs pig hearts. These findings also indicate the robustness of the assay. (A) Summary of our measured copy numbers of ion channels by PRM and those reported by Doll et al. from human left ventricle and Linscheid et al. from mouse sinus notes. (B) The quantified recovery of our reference proteins. Finally, we compared the copy numbers we obtained from previous global proteomic studies also shown in Figure A. We found two studies that used global proteomic characterization to study cardiac tissue and reported detected proteins in copy numbers including some ion channels. One study analyzed isolated left ventricle cardiomyocytes sorted by flow cytometry of the human heart by Doll et al., and the other studied sinus node cardiomyocytes in mouse heart tissue by Linscheid et al. Both studies used intensity-based label free quantitation and global scaling for copy numbers instead of using a targeted approach for each individual protein and their peptides as in our study. We demonstrated that the intensity of peptides is highly susceptible to local matrix effects (LLODs in Table ). Even though both studies performed prefractionation on samples before LC-MS/MS analysis, ruling out the differential matrix effects and ion suppression of all the detected peptides, which could compromise the obtained copy numbers, is difficult. Nevertheless, their report in terms of copy numbers as a measure of protein quantity allows a comparison between our study and these previous studies, as shown in Figure . First, among the 5 channel subunits we analyzed, hERG, an important K+ ion channel responsible for most drug induced cardiac toxicity, was absent from both of the other studies. Doll et al. detected only three subunits, Nav1.5, Cav1.2 and Cavα2δ1, whereas Linscheid et al. detected an additional subunit, Cavβ1. Second for the commonly detected ion channels, Nav1.5 showed similar copy numbers between our study (49k copies) and Doll et al. (52k copies), whereas Cav subunits showed large differences (Figure A). Linscheid et al. had a lower quantity for all the detected proteins, likely due in part to interspecies differences and smaller sinus node cells than ventricular cardiomyocytes. The consistently greater quantity of the three subunits of Cav, i.e., Cav1.2, Cavβ1 and Cavα2δ1, than Nav1.5 and Kv11.1 (the homotetramer of the hERG) in our results agrees with the known stoichiometry of the Cav channel. The particularly high quantity in our results might suggest that the insoluble fraction of the heart homogenate had enriched in more of these large membrane complexes (Cav1.2, 249 kDa + Cavβ1, 66kD + Cavα2δ1, 125 kDa = 440 kDa), and our proteomic approach was more effective at recovering and quantifying these proteins than others. Conclusions In summary, we present here a targeted approach to successfully quantify the absolute copy number of ion channels in human cardiac tissue using Heavy-Match-Light (HML) in-sample calibration method. We also demonstrated that the commonly discarded insoluble fraction contained a large number of ion channel molecules, and that extended digestion was able to solubilize these proteins for quantitative analysis. Finally, we have illustrated the power of reporting ion channels in copy numbers by comparing our results with those of two previous reports carried out by different laboratories using separate biological systems: one used isolated individual cardiomyocytes, and the other used mouse sinus nodes. Although these values are mostly different, suggesting variations and complexity in biological systems, this provides a framework to integrate more similar values in the future for identifying true biological insights from noisy data. Although we demonstrated the use of endogenous analytes as internal standards and HML for protein quantitation, this method can be equally applied to the quantification of metabolites, for which the synthesis of structural analogs or stable isotope-labeled molecules as internal standards is challenging. The accuracy of the measurement is ensured by performing calibration for each individual sample. When a large number of samples need to be analyzed and the assay linearity has been well characterized previously, the number of points in the calibration curve can be reduced to two, as discussed in our previous study, to save analysis time and sample consumption.