Microdomain-Specific Modulation of L-Type Calcium Channels Leads to Triggered Ventricular Arrhythmia in Heart Failure

Supplemental Digital Content is available in the text.

Ventricular cardiomyocyte isolation was done as previously described. 1 Briefly, Sprague-Dawley rats (150-250 g) were anesthetized with 5% isoflurane-95% O 2  Cardiomyocytes were plated on dishes coated with laminin and left to stick to bottom for at least 45 minutes before experiments. Cardiomyocytes were used on the same day as isolation.

Patients groups and human cardiomyocyte isolation
To control for factors of regional-dependent heterogeneity, the presence of acute ischemia, and ischemic injury, the current study was conducted on dilated end-stage cardiomyopathic human hearts (n=6, patients, average age 48±5 years, two females and fourth males, Online Table I Failing ventricular cardiomyocytes were isolated from the apical section of the posteriorlateral left ventricle free wall by enzymatic digestion as previously described. 2 Briefly, individual specimens were transferred to ice-cold calcium free Krebs-Ringer saline solution consisting of (in g/L): 7 Sigma-Aldrich) under gentle agitation. The partially digested tissue was transferred to 10 ml of Krebs-Ringer saline supplemented with collagenase type XIV (1mg/ml Sigma-Aldrich).
The tissue was incubated thrice with this solution for 10 min each at 37 o C with gentle agitation. Usually, cardiomyocytes were visible by phase contrast light microscopy after the first incubation step, with the biggest amount of cells after the second incubation step. After each incubation step, the supernatants were transferred to a tube and centrifuged at 600 rpm for 3 min. The pellets were re-suspended in 2-3 mL of Krebs-Ringer solution. After isolation, human cardiomyocytes were plated following the same protocol as rat cardiomyocytes. Nonfailing control human ventricular cardiomyocytes were isolated from the left ventricle papillary muscles.

T-tubule labelling and analysis
Di-8-ANEPPS was excited at a wavelength of 488 nm and confocal images were taken at 63x magnification using a Zeiss LSM-780 inverted confocal microscope. The 40x5 microns size area inside the sarcolemma that did not coincide with the cell nuclei was chosen, automatically thresholded into binarised, and the TT density was determined as ratio of black pixels versus white pixels. The TT regularity was calculated through a single dimension Fourier transformation using a custom-written macro for Matlab (The MathWorks, Inc., Natick, MA, USA) as described before. 6 For the visualization of the TT network structure in combination with SICM, cardiomyocytes were stained with the fluorescent dye Di-8-ANEPPS (10µM) for 1 min. Di-8-ANEPPS was excited at a wavelength of 473 nm with a StradusTM473 laser (Vortran) and confocal images were taken at 100x magnification using a Photomultiplier Detection System (PTI).

Super-resolution scanning patch-clamp with pipette clipping modification
This technique combines scanning ion conductance microscopy (SICM) and patch-clamp electrophysiology with a pipette clipping modification to increase the throughput of recording ion channel activity from cell surface microdomains. SICM uses a sharp pipette (~100 nm inner diameter, ID, 100 MΩ resistance) as a scanning probe to generate high resolution topography images of live cells. 7 SICM is based on the principle that the ion current through the electrolyte-filled micropipette is partially occluded when the pipette approaches the surface of a cell. 8 Therefore, the position of the tip of the pipette relative to the cell surface strongly influences the ion current through the pipette. This ion current is digitized and fed into the feedback and scan control system which provides the feedback signal to control the vertical position of the pipette keeping the pipette-sample separation constant. The pipette raster scans the sample and generates a 3D topography image of the cell surface.
Next, to clip the pipette tip to generate a wider pipette necessary to increase the chances of getting an ion channel, the pipette is positioned above a free surface away from cells.
Then, moving down with high velocity the pipette is allowed to impact onto the surface and the tip of the pipette clips in a controlled manner. 7 This is possible because at a high enough velocity of approaching the surface the feedback control of the scanning system cannot generate enough upward movement of the piezo-drive quickly enough to avoid contact with the surface. With this method at 500 nm/ms the pipette tip is clipped to approximately ~350 nm ID, 30 MΩ resistance. The pipette is then positioned back onto the cellular microdomain of choice (in this case T-tubule or crest) taking help of the coordinates from the image acquired before clipping. The feedback is turned off; pipette is lowered until it touches the cell surface and a very gentle suction is applied to form a gigaseal (Online Figure II). Cellattached single channel recordings are then performed by the conventional patch-clamp technique.

5.a. Cell attached LTCC
Cell-attached patch-clamp recordings of single LTCC currents were performed at room temperature using the following solutions; external solution containing in (mmol/L): 120 Kgluconate, 25 KCl, 2 MgCl2, 1 CaCl 2 , 2 EGTA, 10 Glucose, 10 HEPES, pH 7.4 with NaOH,  (I-V) relationship, the membrane under the patch was held at a voltage of -80 mV and voltage pulses were applied from -30 to +30 mV in the incremental steps of 10 mV. Analysis was performed as previously described. 7 Single channels were sampled at 10 kHz and filtered at 2 kHz (-3 dB, 8-pole Bessel). Single channel data were analysed using Clampfit version 10.2.
A liquid junction potential, calculated to be -16.7 mV, was corrected from the data shown.
Every time a channel is recorded a minimum of 100 sweeps at -6.7 mV (after liquid junction potential correction) and 3 full protocols at different voltage steps were done. All the sweeps are checked for the presence of more than one channel. When no second level of spikes is observed in any of the sweeps the recording is classified as one channel. Overall, from 89 recordings showing activity in this work, 70 (79%) of them were classified as one channel, 13 (14%) show two channels, 5 (6%) triple channels, and only in one occasion (1%) the recording showed 4 channels, and this is similar between TT and Crest regions. From 22 recordings of human cells 64% were classified as one channel versus rat cells where from 67 recordings 84% were classified as one channel. Seal data is showed in detail on Online Table   III.
Occurrence of LTCCs was calculated as the percentage of recording showing activity versus the total number of recordings, higher occurrence can be interpreted as a higher density of channels on that specific location or group.
The open probability (Po) was averaged from 10-20 sweeps at -6.7 mV for each cell.
Each cell was recorded only once, and only one value of Po per cell was used. The total number of channels in the recording was input into pClamp software to calculate the Po and the peak current (calculated as the overall average current) of one single channel.

5.b. Action potential recordings
APs were recorded from isolated LV rat cardiomyocytes using the current-clamp configuration of the patch-clamp technique ( Figure S12 The bath was connected to the ground via an Ag-AgCl pellet. Data were sampled at 10 kHz.
All recordings were performed at 33-35°C. APs were elicited with 5-ms current pulses at 1Hz. AP characteristics were analysed using pClamp software.
Both "f" and "fca" gates have fast and slow gating modes. CaMKII unphosphorylated channels inactivate in both fast and slow gating modes. Inactivation occurs via the slow mode only for channels phosphorylated by CaMKII.

b. Stochastic Sweeps using Gillespie Exact Algorithm
To evolve the channel gating in response to a 1-s voltage change step to -6.7 mV (from resting state, marked *), we used the Gillespie Exact Algorithm, implemented in Matlab (The Mathworks Inc.). The built-in pseudo-random number generator function, "rand", was seeded to the system clock and used twice at each state change step, as required by the algorithm. Example single channel sweeps are shown in Online Figure VIII.

c. Reverting to Hodgkin-Huxley Formalism
Once single channel current simulation results were generated and validated with experimental P o measurements, the I Ca,L model was reverted back to an HH formulation. The

b. Sub-sarcolemmal Volume
In the Grandi et al human cell model, 13 13,14 As was done by Grandi et al and Shannon et al, we set the flux rate between SL and myoplasm to be 0.2213 times that between dyad and SL. We assumed that the geometry of the dyad/SL interface was identical to the original O'Hara-Rudy dyad/myoplasm interface.

c. LTCC Current Density at TT and Crest Locations
The In failing human cells, our experimental findings (data not shown) indicated that LTCC single channel amplitude was slightly, but statistically significantly greater in TTs (1.2 ± 0.07 -fold greater) and slightly but statistically significantly lower at crest sites (0.85 ± 0.06 times) when compared to control TT amplitude, the later being our reference amplitude.
Thus, the LTCC % occurrence (as a fraction of total LTCC occurrence), (manuscript Figure   2A To arrive at a value for fPCa failing,TT , an additional factor accounting for the degree of TT loss was incorporated. This was necessary because TT loss reduces whole cell density of LTCC TT channels, which lowers fPCa failing,TT . Loss of TTs in failing human myocytes was the average of fractional TT density (failing/control = 0.51 ± 0.08) and fractional Z-groove ratio (failing/control = 0.41 ± 0.06, average of 0.54). Thus, fPCa failing,TT = 0.75*1.2*0.54 = 0.486.
Finally, experimental evidence from human myocytes has indicated (see for example 15 ) that HF ion channel remodeling does not have an effect on peak I CaL . Therefore, we adjusted the whole cell LTCC peak current in failing myocytes to match that of control.

d. Na + /Ca 2+ Exchanger (NCX)
The original O'Hara-Rudy model assumes that 20% of NCX channels reside on the surface membrane. In the present simulations, the percent of NCX residing at crest sites (i.e. non-TT-residing NCX) was increased in accordance with the degree of TT loss, assuming that NCX migrates to the Crest in a way similar to LTCCs migration following TT loss. Since the value for the fraction of intact TTs was 0.54 (see above), 46% (100%-54%) of the original 80% of NCX residing in TTs were relocated to the Crest in failing myocytes. The importance of accounting for NCX re-distribution following TT degradation arises from the fact that it affects concentrations in the SL volume.

f. Heart Failure Remodeling
A recent publication provided a comprehensive literature review of heart failure remodeling data in the human myocyte and a summary of how the data have been used in previous computational models. 15  of the nanopipette is widened to ~350 nm by controlled clipping to increase the area of attachment. First, the pipette was navigated to an area free of cardiomyocytes, the fall rate was increased and pipette was allowed to impact on surface. As a result the pipette breaks its tip and increases its diameter because of the conical shape of the pipette. Pipette tip breaking resulted in stepwise increases of the pipette current as its resistance dropped. The breaking was automatically stopped by returning the fall rate to baseline (60 nm/ms) once the pipette current reached a desired level. After clipping, the pipette is lowered to a specific location (TT or crest) and a gigaseal is established. Single ion channels are recorded in cell-attached mode. Insets show the pipette tip size before and after clipping. inactivation ("f CaMK " and "f Ca,CaMK " in purple and pink are slower than "f" and "f Ca " in red and orange, left). When CaMKII phosphorylated, inactivation followed "slow" rather than "fast" paths (indicated by thick "slow" and thin "fast" connections in purple and pink).