Animal ethics statement

The study conformed to the National Institute of Health Guide for the Care and Use of Laboratory Animals (8th Edition, 2011) and was approved by the Institutional Animal Care and Use Committee of the University of Utah (Protocol number 17–08005). Animal euthanasia was performed using inhalant isoflurane.

NRVM preparation and adenoviral infection

Coverslips (no. 0) were manually cut and placed in 48 well tissue culture plates (Genesee Scientific, 25–108) followed by sterilization in 70% ethanol overnight. The next day, the culture plate containing coverslips was thoroughly washed with sterile water before treating for 2 hours at room temperature with 50 μg/mL fibronectin (Sigma F1141-5mg) diluted in PBS. The fibronectin solution was removed and coverslips were allowed to dry for an additional 2 hours before cell plating.

Neonatal rat ventricular myocyte monolayers (NRVMs) were isolated and plated using reagents and a modified protocol from Worthington Biochemical (Worthington, LK003300) (http://worthington-biochem.com/NCIS/default.html). Briefly, hearts from 0–1 day old Sprague Dawley SAS400 (Charles River) rat pups were excised and washed. The ventricles were separated from the rest of the heart and used for the remainder of the protocol. The tissue was minced into fragments approximately < 1 mm³, and then incubated in trypsin solution overnight at 4 °C. The next day, a trypsin inhibitor solution was added and the tissue was warmed to 37 °C. Following this, a collagenase solution was applied to the preparation and allowed to incubate at 37 °C with light shaking for approximately 30 minutes. Gentle dissociation using a serological pipet, followed by mild centrifugation in enzyme-free culture media was used to further separate myocytes and remove the majority of red blood cells. This step was repeated twice to ensure complete removal of the collagenase enzyme. The preparation was pre-plated in a tissue culture incubator for two hours at 37 °C / 6% CO₂ to reduce the fibroblast population. Following this pre-plating period, the myocyte rich suspension was collected and counted on a hemocytometer to determine total yield, then plated at 100K—300K confluency in DMEM (Corning 10-013-CV) complete medium containing 10% FBS (Sigma F0926), 1% non-essential amino acids (Sigma, M7145), and 1% antibiotic solution (Corning 30-004-CI). The NRVM cultures were incubated undisturbed at 37 °C and 6% CO₂ for 24 hours to allow cellular adhesion to the coverslips. After the initial 24 hour tissue culture period, the media was removed, cultures were washed twice in PBS, followed by replacement of culture media. Cells in wells destined to express genetically encoded mitochondrial Ca²⁺ indicator were infected with an adenoviral construct (Vector Biolabs) containing the LAR-GECO1.2 plasmid obtained from Addgene (www.addgene.org) and developed by Wu et al. [10]. Following 24 hours of infection, the cultures were double washed again with PBS and fresh culture media was added. The genetically encoded indicator was expressed for an additional 24 hours before experiments commenced. All experiments were completed between 3–4 days after plating.

Anoxia/Reoxygenation protocol

Prepared NRVM coverslips were placed into a custom perfusion bath and allowed to equilibrate for at least 20 minutes (baseline). The perfusion solution for both baseline and “reperfusion” contained 126 mM NaCl, 4.4 mM KCl, 1 mM MgCl₂, 11 mM glucose, 24 mM HEPES and was titrated with NaOH to pH = 7.4. In the simulated “ischemic” solution, [K⁺] was increased from 4.4 to 8.8 mM, pH was decreased from 7.4 to 6.5, and glucose was replaced with 2-deoxyglucose (11 mM). All solutions contained 0.5 mM Probenecid to aid in the retention of AM dyes. The “ischemic” cocktail was gassed with nitrogen for at least 1 hour to deplete oxygen. Several minutes before the anoxic period, sodium-hydrosulfate (1 mM) was added to eliminate any remaining oxygen. Upon completion of the baseline period, “ischemia” commenced. During ischemic episodes we monitored fluorescence images and waited until we observed the loss of Ca transient, a decrease of ΔΨ_m-sensitive fluorescence to below 50% of the average baseline level, and myocyte shrinking and blebbing. These events occurred in different NRVM monolayers between 15 and 30 min of “ischemia”. Once the observable signs of anoxic effects were complete, coverslips underwent “reperfusion” for at least 1 hour. Throughout the entire protocol, temperature was maintained at 37 ± 1°C. Cells were paced at 0.5–1 Hz beginning at baseline, and only coverslips which retained Ca²⁺ transients for the full baseline period were accepted to move forward with the simulated I/R protocol.

Intervention groups

Fluorescent indicators

To track [Ca²⁺]_Cy in the Control, Rya-Thap, Zero-Ca and CsA groups, NRVMs were incubated with Fluo-4 AM (9.1 μM, Thermo Fisher Scientific) in Optimem for 25 mins at room temperature previous to placement in the imaging perfusion chamber. In 3 special experiments in which cells were stained simultaneously with Fluo-4 AM and the cell-impermeable nucleic acid dye TO-PRO3, the loss of Fluo-4 was coincident with the uptake of TO-PRO3 (see S1 Fig), hence the loss of Fluo-4 could be used as the indicator of “canonical” SP during I/R. All NRVMs in these groups expressed the genetically encoded indicator, LAR-GECO1.2, to track [Ca²⁺]_Mi. NRVMs were also stained with a ΔΨ_m indicator MitoView633 (10 nM, Biotium) added to the perfusate for 8 min prior to baseline imaging, and then again upon “reperfusion” for additional 6 min to replenish dye lost due to ΔΨ_m depolarization during the simulated ischemic period. This combination of indicators allowed for the simultaneous monitoring of [Ca²⁺]_Cy, [Ca²⁺]_Mi, ΔΨ_m, and the onset of SP (as the cellular loss of Fluo-4).

In the Zinc group, the cytoplasmic concentration of Zn²⁺ ([Zn²⁺]_Cy) was monitored with a zinc-specific dye, Newport Green DCF (7.5 μM, Thermo Fisher Scientific). Rhod-2 AM (2.5 μM, Cayman Chemical) was employed to detect [Ca²⁺]_Cy and SP instead of Fluo-4 AM, to avoid overlap with the emission spectrum of Newport Green DCF. Both dyes were loaded by incubation in Optimem for 35 mins at room temperature, and MitoView633 was used identically to the other groups. The Zinc group was not infected with the LAR-GECO1.2 indicator, as it would overlap with the emission spectrum of Rhod-2. Thus, in the Zinc group, four parameters were also tracked, but [Ca²⁺]_Mi was replaced with [Zn²⁺]_Cy. In some experiments, the SP indicator TO-PRO3 (167 nM, Life Technologies) was used as an additional method to detect SP. In spite of its overlapping spectra with MitoView633, the two indicators are spatially distinct (MitoView633 in the mitochondria, TO-PRO3 in the nucleus) and in practice TO-PRO3 uptake almost always occurred after MitoView633 loss. TO-PRO3 is normally cell impermeable and similar to the commonly used SP indicator propidium-iodide, in that it binds to nucleic acids upon entering permeable cells which amplifies the fluorescent intensity over 100-fold.

Confocal imaging

NRVMs were placed in the perfusion bath and imaged on a Leica SP8 confocal microscope. An oil immersion 20x lens was employed for the collection of all images. Ca²⁺ transients were captured as a continuous time-series every 100 ms for 80 frames, equaling a recording total of 8 secs. The field of view (FOV) was set to a 128 x 128 pixel resolution and a 4.5x magnification with the 20x objective, along with an imaging speed of 1400 to achieve the stated continuous time resolution. For tracking all parameters throughout the full experimental period, time-series were utilized again with a frame rate of either 30 secs or 1 min depending on the experiment. These frame rates were chosen as the best balance between better time resolution and avoiding phototoxicity and bleaching due to excessive laser exposure induced by higher frame rates, such as every 15 sec. The spatial resolution of 512 x 512 pixels at a magnification of 1.3x yielded a 450 x 450 μm FOV. During a time-series, if there appeared to be a slight drift in imaging plane then the series was stopped and the plane was quickly refocused to minimize the time between the acquired series.

All long time-series images were obtained in three channels sequentially. In the first channel, either Fluo-4 or Newport Green was excited with a 488 nm wavelength laser and the emitted signal was filtered through the Leica SP8 prism. Light between 491 and 551 nm was collected with a HyD detector. The second channel, also collected with a HyD detector, imaged either LAR-GECO1.2 or Rhod-2 and was excited with a 561 nm laser and band passed between 566 and 630 nm. In the third channel MitoView633, and sometimes additionally TO-PRO3, were excited at 633 nm and long pass filtered to a PMT detector at 643 nm. Using the Leica SP8 prism technology and imaging each channel sequentially, as opposed to simultaneously, avoided possible crosstalk between signals and ensured quality intensity curves upon analysis.

Data analysis methods and criteria

We randomly selected 5 to 7 NRVMs per plate which met the criteria of (1) being a myocyte (exhibiting rhythmic Ca²⁺ transients at baseline); (2) having clearly demarcated cell boundaries to allow for accurate cell segmentation; (3) having overt signs of “ischemic” effects including at least 50% loss in the ΔΨ_m and cell shrinking, and (3) surviving within 6 minutes of “reperfusion”, as demonstrated by at least 50% recovery in the ΔΨ_m during MitoView633 restaining combined with the lack of SP (i.e., retainment of Fluo-4 inside cells). These criteria were imposed due to the fact that these characteristics are observed in the whole heart model [7], and the aim here was to replicate whole heart events so the findings could be applied in entirety and not as a model specific phenomenon. Once acceptable myocytes were identified, they were manually segmented across all images of a time-series, and the area and mean fluorescence for each selected cell in each frame was calculated using ImageJ software.

The area and mean fluorescence were then multiplied together to determine the total fluorescence per cell. In Control, Rya-Thap and CsA groups the signals were normalized to the range between minimum and maximum values that occurred during the “reperfusion” phase. Specifically, the fluorescence values during reperfusion were computed as (F-F_Min)/(F_max-F_min), where F is the actual level of fluorescence reported by the microscope; F_Min is the lowest fluorescence level during reperfusion; and F_Max is the highest fluorescence level during reperfusion. We defined the time of a critical increase in [Ca²⁺]_Cy (T_CaCy) and a critical increase in [Ca²⁺]_Mi (T_CaMi) as the time when the respective signals (Fluo-4 or LAR-GECO1.2 fluorescence) increased above the 50% level between the minimum and maximum. We defined the time of MPT (T_MPT) as the decrease of MitoView633 fluorescence below the 50% level. Finally, we defined the time of SP (T_SP) as the time when Fluo-4 fluorescence decreased below the 50% level. We interpreted the loss of the Fluo-4 signal during “reperfusion” as evidence of dye leakage from cells through sarcolemmal pores. This interpretation was confirmed in separate experiments, in which the loss of Fluo-4 signal during “reperfusion” correlated well with the uptake of the normally cell-impermeable dye, TO-PRO3 (see S1 Fig).

For the Zinc group, we also defined T_CaCy and T_SP, but additionally we defined T_ZnCy as the time of abnormal uptake of Zn²⁺. However, here we used a slightly different approach. First, we used Rhod-2 to detect abnormal uptake of Ca²⁺ into cells instead of Fluo-4, for spectral compatibility with the other two fluorophores, Newport Green and TO-PRO3. Unlike Fluo-4, a significant fraction of Rhod-2 may redistribute into mitochondria [13] although the cytoplasmic fraction is always large [14]. However, from Control experiments we determined that the mitochondrial Ca²⁺ uptake, as indicated by an increase in LAR-GECO1.2 fluorescence, always lags behind the cytoplasmic Ca²⁺ uptake as indicated by Fluo-4 (see Results). Hence, the earliest detectable increase in Rhod-2 fluorescence should reflect mostly [Ca²⁺]_Cy, but even if it reflects some weighed combination of [Ca²⁺]_Cy and [Ca²⁺]_Mi, this still would indicate a net Ca²⁺ influx into the cell, which was the main event of interest. Also, in the Zinc group we first searched for the minimum of Rhod-2 signal during “reperfusion”, and normalized the signal to the range between that minimum and the maximum value during “reperfusion”. Lastly, for all the events (T_CaCy, T_ZnCy and T_SP), we set the threshold of detection at 10% level of the normalized signal, as opposed to 50% level used in other series. The low threshold was chosen to facilitate early detection of Zn²⁺ and TO-PRO3 uptake, which developed in time rather slowly, and by visual inspection was deemed to be robust enough not to produce false-positive detections.

Statistics

All values within groups are presented as the mean ± standard error of means. Quantification plots portray a variety of data including 1) the detected values according to the analysis criteria, 2) box and whiskers plots which portray median values, the upper and lower quartiles, and the highest and lowest observations excluding outliers, and 3) mean diamond plots which incorporate the mean values and 95% confidence intervals. These plots were generated with JMP trial software (www.jmp.com). Statistical tests included Student t-tests and one-way ANOVA with Newman-Keuls post-hoc test for pairwise comparisons (XLSTAT by Addinsoft, www.xlstat.com), as indicated in specific figure legends. Differences with p < 0.05 were considered statistically significant.

Validation of LAR-GECO1.2 localization and the viability of infected NRVMs

Before beginning experiments with the simulated I/R, the infected NRVM model was tested to ensure the [Ca²⁺]_Mi indicator, LAR-GECO1.2, was properly expressed and localized to the mitochondria. To ensure appropriate localization, NRVMs expressing LAR-GECO1.2 were stained with the ΔΨ_m dye, MitoView633, and co-localization of the two indicators was determined using high resolution imaging. Fig 1 confirms that LAR-GECO1.2 is indeed expressed homogeneously throughout all mitochondria within the imaged cell, and exhibits a signal strong enough for experimental purposes.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Evidence for localization of LAR-GECO1.2 to mitochondria in transfected NRVMs.

A-C, high resolution images of the ΔΨ_m dye MitoView633 (A), the genetically encoded mitochondrial [Ca²⁺] indicator LAR-GECO1.2 (B), and their merged image (C). The images support predominantly mitochondrial localization of LAR-GECO1.2. Note the abundance of yellow (and lack of red) areas in C, suggesting that LAR-GECO1.2 is expressed homogeneously within mitochondria in infected cells. Green areas here indicate mitochondria that are expressing LAR-GECO1.2, but undergo ΔΨ_m, flickering, possibly resulting from laser exposure due to imaging. This further implies that in spite of ΔΨ_m depolarization, LAR-GECO1.2 is still retained in the mitochondria.

https://doi.org/10.1371/journal.pone.0212076.g001

LAR-GECO1.2 appears to be highly co-localized with MitoView633, as seen in the yellow areas of Panel C, and thus is expressed exclusively in mitochondria without other offhand targets within the cell. The limited green areas in Panel C that are devoid of MitoView633 staining represent temporarily depolarized mitochondria, and reflect mitochondrial depolarization-repolarization cycles (flicker) sometimes observed at baseline. In sham experiments, the cell-averaged level of LAR-GECO1.2 fluorescence remained relatively constant over a full 2-hour test period (S2 Fig). A known problem of GECO family of indicators, which includes LAR-GECO1.2, is a significant pH sensitivity, such that the fluorescence decreases at lower pH [15] (see also S3 Fig). Our use of LAR-GECO1.2 to monitor [Ca²⁺]_Mi during “reperfusion” was based on the assumption that after recovery of pH in the reperfusion solution to 7.4, any further significant increase in LAR-GECO1.2 fluorescence should be mostly due to an increase [Ca²⁺]_Mi rather than an increase in pH. Increase in intra-mitochondrial pH (alkalization) is expected when mitochondria hyperpolarize and increase the pH gradient across the inner mitochondrial membrane. However, during “reperfusion” we observed a large increase in LAR-GECO1.2 signal just before the loss of ΔΨ_m, and this increase was sustained after the loss of ΔΨ_m when the mitochondrial matrix is expected to become more acidic. Hence, it seems that under the tested conditions, changes in LAR-GECO1.2 fluorescence reflected a true [Ca²⁺]_Mi increase, even if attenuated due to expected acidification of depolarized mitochondria.

The viability of infected NRVMs was also examined to guarantee that they functioned similarly to non-infected NRVMs and preserved normal Ca²⁺ handling. Coverslips were incubated with Fluo-4 AM to detect [Ca²⁺]_Cy, and subsequently placed in the imaging perfusion chamber. Cells were paced at 1 Hz and continuously imaged every 100 ms. The resulting Ca²⁺ transients can be seen in Fig 2C, with 1:1 pacing capture and robust transient upstrokes. The Fluo-4 signal during the peak and trough of the Ca²⁺ transient is shown in Fig 2A and 2B respectively. In sham experiments, the infected NRVMs maintained Ca²⁺ transients for the entirety of a 2-hour experimental duration, and had not undergone any detectable SP (see S2 Fig). Due to these positive functional and viability results, NRVMs expressing LAR-GECO1.2 were deemed adequate for use in the simulated I/R experiments.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Paced [Ca²⁺] transients in infected NRVMs.

Images from infected NRVMS loaded with the [Ca²⁺]_Cy indicator, Fluo-4 AM, during the peak (A) and the trough (B) of the [Ca²⁺] transient. Monolayers were paced at 1 Hz and imaged continuously every 100 msec. The resulting [Ca²⁺] transients (Panel C) show 1:1 capture, verifying that infected cells are viable, appear to have normal [Ca²⁺] cycling, and are able to sustain pacing.

https://doi.org/10.1371/journal.pone.0212076.g002

Consistent chronology of critical events in NRVMs during “reperfusion”

In Control experiments, the NRVMs transfected with LAR-GECO1.2 were also loaded with Fluo-4 and MitoView633 and were subjected to 15–30 min of simulated ischemia, followed by 60 min of “reperfusion”. During simulated ischemia, NRVMs exhibited various degrees of ΔΨ_m dissipation and cellular shrinkage, but the vast majority of cells survived until reperfusion. However, the vast majority of cells surviving “ischemia” underwent critical transitions including full ΔΨ_m loss and SP later in “reperfusion”. Importantly, the dynamics of the critical events varied between myocytes which underscored the necessity to track these events on a cell-by-cell basis.

Representative curves from four different Control NRVMs are shown in Fig 3. (The decrease in LAR-GECO1.2 signal during simulated ischemia is attributed to imposed acidic conditions and is ignored). Note the considerable variability in [Ca²⁺]_Cy dynamics during simulated ischemia and during the first 5–7 min of “reperfusion”, which was present even among NRVMs from the same monolayer. An initial [Ca²⁺]_Cy rise could occur during ischemia (Fig 3B), or immediately upon reperfusion (Fig 3A, 3C and 3D). It is of interest, however, that regardless of the presence or the absence of these early [Ca²⁺]_Cy events, they were followed by a period of an apparent tranquility until the secondary [Ca²⁺]_Cy increase invariably occurred, heralding the onset of cellular transition to death. It can be seen that despite variability in the dynamics of specific signals among the presented cells, there was a consistent sequence of critical events, in the order of T_CaCy < T_CaMi < T_MPT < T_SP. This sequence was conserved in all but one analyzed Control cell (see Fig 4; asterisk denotes the exception). [Ca²⁺]_Cy overload occurred at an average of 22.7 ± 4.9 mins of “reperfusion”, followed by [Ca²⁺]_Mi overload approximately 3.4 ± 1.7 mins later, then MPT at 32.5 ± 5.5 mins, and finally SP at 38.1 ± 6.3 mins. T_CaMi, T_MPT, and T_SP were found to be significantly separated from their respective previous event (all p values < 0.05 by paired t-test). The application of CsA (0.2–0.4 μM) (CsA group), an MPT inhibitor which decreases the MPT pore sensitivity to Ca²⁺, did not change the timing or order of the critical events (S4 Fig).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Chronology of post-reperfusion events in Control cells.

Data from 4 separate representative Control cells (A-D) imaged throughout the full experimental duration at a rate of 1 frame per min. Yellow, green, and red curves show total per-cell fluorescence of Fluo-4 ([Ca²⁺]_Cy), LAR-GECO1.2 ([Ca²⁺]_Mi) and MitoView633 (ΔΨ_m), respectively. All curves are normalized to their respective minimum and maximum intensity values during the “reperfusion” period. The anoxic phase is depicted by the grey box, while the thin horizontal black line during “reperfusion” indicates the 50% level used to determine the associated time points of events (colored open circles, as labeled in Panel B). The time points of T_CaCy and T_CaMi were estimated at the 50% rise in the fluorescence of Fluo-4 and LAR-GECO1.2, respectively, whereas T_MPT and T_SP were estimated as the 50% decrease of MitoView633 and Fluo-4 fluorescence, respectively (the latter indicating the dye efflux from the cell). Notice that in all 4 cells, this sequence of events remains the same regardless of when in reoxygenation they begin to occur.

https://doi.org/10.1371/journal.pone.0212076.g003

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. Quantification of critical events timing in the Control group.

A, the timing of detected events (T_CaCy, T_CaMi, T_MPT, T_SP) with respect to the onset of “reperfusion” for all analyzed Control cells. The style is such that each color indicates the time interval between the specific event (as indicated) and the previous event. For T_CaCy the previous event is the onset of “reperfusion”. In all cells [Ca²⁺]_Cy increase was the first observed event, followed by [Ca²⁺]_Mi increase, then MPT, and finally SP (as detected by loss of Fluo-4). This sequence was preserved in every cell in the Control group, with the exception of cell #29, where MPT and SP occurred at the same time (hence the absence of a grey bar), and also cell #1 (indicated by an asterisk) where both MPT and SP occurred very quickly and, according to the detection criteria, SP occurred 1 min ahead of MPT. B, the spread of detected time points for each event (black dots) overlaid with the associated box and whiskers plot (red) and mean diamond plot (blue). The average time of T_CaMi, T_MPT, and T_SP each were statistically different from the respective previous event (all p < 0.05).

https://doi.org/10.1371/journal.pone.0212076.g004

Summarizing, in this model we detected a [Ca²⁺]_Cy overload event which is separate from the [Ca²⁺]_Cy fluctuations occurring immediately upon reperfusion, and which invariably leads to [Ca²⁺]_Mi overload, MPT and SP. We further tried to sort out whether the observed increases in [Ca²⁺]_Cy and [Ca²⁺]_Mi are due to Ca²⁺ release from SR, or due to a Ca²⁺ flux across the cell membrane.

Sarcoplasmic reticulum Ca²⁺ depletion accelerates the timing of critical events in reperfusion

Previously, a Ca²⁺ release from SR during reperfusion/reoxygenation was implicated in mitochondrial Ca²⁺ loading, MPT, and cell death [4]. To test the possible role of a large Ca²⁺ release from the SR as the trigger of the observed sequence of critical events, the SR was depleted with continuous perfusion of ryanodine (5 μM) and thapsigargin (1 μM), keeping ryanodine receptor channels in a continually open state while also inhibiting Ca²⁺ uptake into the SR by blocking SERCA2a (Rya-Thap group). Panel A in Fig 5 shows normalized intensity curves from a representative Rya-Thap cell during reperfusion, analyzed identically to cells in the Control group. Most notably, the major sequence of events remained the same as observed in Control, with [Ca²⁺]_Cy overload occurring first, followed by [Ca²⁺]_Mi overload, then MPT and SP. When quantified and compared to Control (Panel B), the sequence of critical events indeed remained the same as seen in Control, with T_CaMi, T_MPT, and T_SP occurring significantly later than the respective previous event (indicated by an asterisk, p<0.05). Interestingly, in spite of the retention of the event sequence, all critical events in the Rya-Thap group occurred significantly earlier than their Control counterpart (indicated by a §, p<0.05). The cause and/or meaning of this observation will be deliberated in the Discussion section.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5. Effect of sarcoplasmic reticulum depletion by ryanodine/thapsigargin.

A, fluorescence intensity curves during “reperfusion” from a representative cell in the Rya-Thap group. All labeling and notations are the same as in Fig 3, but only “reperfusion” phase is shown. B, when quantified, the sequence of critical events in the Rya-Thap group (grey bars) is the same as in Control group (black bars), however, each event in the Rya-Thap group occurred at a significantly earlier time point of “reperfusion” than in Control group. *, p < 0.05 as compared to the timing of previous event in the same group (paired t-test); §, p < 0.05 as compared to the timing of the same event in different group (unpaired t-test).

https://doi.org/10.1371/journal.pone.0212076.g005

Cytoplasmic and mitochondrial Ca²⁺ overload during reperfusion is due to Ca influx across the sarcolemma

Since the depletion of SR did not affect the dynamics of [Ca²⁺]_Cy and [Ca²⁺]_Mi increase during reperfusion, we hypothesized that the source of these increases is an influx of Ca²⁺ ions across the sarcolemma. To prove this point, we applied 2–3 min pulses of nominally zero Ca²⁺ solution (0 Ca²⁺) after the initial phase of the [Ca²⁺]_Cy overload during “reperfusion” became evident during real-time monitoring in pre-selected regions of interest. Fig 6 shows representative data from a single NRVM with two 0 Ca²⁺ pulses applied after the steep rise in [Ca²⁺]_Cy. As is apparent, both [Ca²⁺]_Cy and [Ca²⁺]_Mi drop immediately upon the application of 0 Ca²⁺, and return directly back upon the reapplication of normal [Ca²⁺]. Note the almost symmetric response of [Ca²⁺]_Cy to decreases and increases of [Ca²⁺]_o, suggesting equal permeability for Ca²⁺ in both transmembrane directions.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 6. [Ca²⁺]_Cy and [Ca²⁺]_Mi directly follow changes in extracellular [Ca²⁺] after the start of [Ca²⁺]_Cy overload.

A representative example of responses to pulses of nominally 0 [Ca²⁺] applied after the start of noticeable Ca²⁺ overload during “reperfusion” (Zero-Ca group). Blue shades indicate the time of perfusion with nominally 0 [Ca²⁺]. Other notations are the same as in Fig 3. Note that both [Ca²⁺]_Cy and [Ca²⁺]_Mi strictly follow the changes in extracellular [Ca²⁺], with apparently symmetric fluxes in both inward and outward directions. The fact that after the second pulse of 0 Ca²⁺ the [Ca²⁺]_Cy signal does not return to overloaded levels (arrow), indicates that at this point the sarcolemma becomes permeable to [Ca²⁺]_Cy indicator, Fluo-4, which leaks out of the cell preventing further ability to track [Ca²⁺]_Cy.

https://doi.org/10.1371/journal.pone.0212076.g006

The symmetrical inward and outward flux of Ca²⁺ observed during “reperfusion” could be mediated Na⁺-Ca²⁺ exchanger (NCX) [16]. To investigate possible role of NCX in the [Ca²⁺]_Cy overload we applied 5 mM Ni²⁺ to the perfusate in the Nickel group. At this concentration, Ni²⁺ blocks NCX by 80% [11]. Ni²⁺ affected the events during reperfusion in multiple ways and caused astounding variability in the outcomes. Most importantly, it blunted [Ca²⁺]_Cy dynamics in such a way that in the majority of cells the criterion of T_CaCy as the time when [Ca²⁺]_Cy curve crossed the 50% level of the ascending phase of the dynamic range during reperfusion (see Fig 3) was not met, precluding proper comparison with Control. Fig 7 shows examples of diversity observed in the Nickel group.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 7. Representative examples of reperfusion events in cells subjected to 5 mM Ni²⁺.

All labeling and notations are the same as in Figure, unless indicated otherwise. Note overall blunted dynamics of Fluor-4 signal during reperfusion, including the case when no apparent [Ca²⁺]_Cy overload ever occurred (B). Arrow in C indicate the time when TO-PRO3 was added to perfusate, immediately revealing that the sarcolemmal permeabilization had already occurred.

https://doi.org/10.1371/journal.pone.0212076.g007

Fig 7A shows a case when the events occurred similarly to the Control group. However, the post-reperfusion recovery of fluor-4 signal reflecting [Ca²⁺]_Cy was very poor, the minimum level barely scratching the 50% of the dynamic range. Also, the recovery of LAR-GECO1.2 signal, reflecting the combination of changes in [Ca²⁺]_Mi and pH was extremely slow as compared to Control cases (see Fig 3). This means that either [Ca²⁺]_Mi, or pH, or both were affected by Ni²⁺. Fig 7B shows a case when post-reperfusion [Ca²⁺]_Cy overload did not occur at all. Yet, there was a gradual decrease in MitoView633 signal reflecting a slow loss of ΔΨ_m. It was followed by fast loss of MitoView633 and Fluo-4. Almost simultaneously, there was a sharp increase in LAR-GECO1.2 signal, reflecting an abrupt increase in [Ca²⁺]_Mi. These three events are consistent with an almost simultaneous SP and MPT, reminiscent of our observations in whole adult hearts [7]. Lastly, Fig 7C shows a case when there was a moderate increase in Fluo-4 signal during “ischemia”, followed by a poor recovery during”reperfusion” (just touching 50% level), followed by a blunted secondary increase, followed by a very slow decrease in Fluo-4 signal to the level of full disappearance. The Fluo-4 signal loss was so slow that we even doubted that it reflected true event of SP, and we applied TO-PRO3 at the moment indicated by the arrow in Fig 7C. The immediate uptake of TO-PRO3 proved that the cell was indeed permeabilized by that time. Typical for cells subjected to Ni²⁺, there was a phase of slow loss of ΔΨ_m followed by a phase of fast ΔΨ_m loss. Increase in [Ca²⁺]_Mi followed increase in [Ca²⁺]_Cy and preceded the phase of fast ΔΨ_m loss. Thus, qualitatively the sequence of critical events is similar to that in Control, but all events are somewhat blunted.

To provide a quantitative support for the notion that Ni²⁺ blunted [Ca²⁺]_Cy increase during reperfusion, we calculated the maximum level of Fluo-4 signal during “reperfusion” as the percent of the maximum baseline signal. (Recall that at baseline the maximal level of Fluo-4 reflected a possibly underestimated amplitude of the normal [Ca²⁺]_Cy transient). In analyzed Control cells maximal post-reperfusion level of Fluo-4 signal reached 154±6% of the maximal baseline level, whereas in analyzed Nickel cells it was 100±14% of the maximal baseline level (p < 0.0001 by unpaired two-tail t-test).

Evolving sarcolemmal permeability observed in the presence of Zn²⁺

Whereas the effects of Ni²⁺ described above nominally supported the role of NCX as the conduit of post-reperfusion [Ca²⁺]_Cy overload, we had our doubts (explained in Discussion) and we tested an alternative hypothesis that the putative channel may be non-selective. We tested the possibility that the channel responsible for the critical increase in [Ca²⁺]_Cy is also permeable to ions of Zn²⁺. We chose Zn²⁺ because it is normally present in cardiac myocytes at extremely low (single nanomolar) concentrations. Also, under normal conditions, the exposure of cardiomyocytes to extracellular Zn²⁺ at 10 to 100 μM concentrations does not cause a detectable increase in [Zn²⁺]_Cy, and does not adversely affect cardiomyocytes apart from a slight inhibition of I_Ca,L [12]. In addition, Zn²⁺ is not transported by NCX [17] and does not affect Ca transport through NCX [18]. Hence, a significant increase in Zn²⁺ uptake during reperfusion would indicate abnormal cell permeability independent of NCX. Experiments were designed to determine the relative timing of permeability to Ca²⁺, Zn²⁺ and the normally cell-impermeable nucleic acid indicator, TO-PRO3, during “reperfusion” (Zinc group). ZnCl₂ (10 μM) was present in the perfusate at least from the beginning of “reperfusion”. Rhod-2 AM and Newport Green DCF were used to indicate [Ca²⁺]_Cy and [Zn²⁺]_Cy, respectively. TO-PRO3 fluorescence was collected in the same channel as MitoView633, since the signals could be easily separated spatially and temporally (see Methods).

Fig 8A–8C portrays data from three different representative cells from the Zinc group. Events were detected at a 10% increase in each signal (thin black line) in the range between minimum and maximum measured during “reperfusion”. In each cell, the onset of [Ca²⁺]_Cy rise was detected the earliest, followed by the onset of Zn²⁺ uptake, and finally uptake of TO-PRO3 (colored open circles in Fig 8). Note that the time course of the red curve reflects two different events, where a rapid drop indicates a decrease in MitoView633 fluorescence due to ΔΨ_m loss (solid red line), whereas the final and sustained increase (dashed red line) indicates uptake of TO-PRO3 signifying “canonical” SP. Overall, the timing of the three permeability events varied between cells, but it was always the case that Zn²⁺ uptake started after Ca²⁺ uptake and before, or simultaneously with, TO-PRO3 uptake (T_CaCy < T_ZnCy ≤ T_SP, see Fig 8D). Note that this order might reflect the increasing size of the penetrating species (hydrated radius of Ca²⁺, 4.1 Å; Zn²⁺, 4.3 Å; TO-PRO3, Stokes radius ~5 Å).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 8. Three stages of permeability detected in the presence of Zn²⁺.

A-C, the time course of fluorescence emitted by Rhod-2 (yellow), Newport Green (green), MitoView 633 (solid red) and TO-PRO3 (dashed red) from 3 representative cells in the Zinc group during “reperfusion”. The levels of fluorescence are normalized the same way as in Fig 3. Rhod-2 fluorescence mostly reflects [Ca²⁺]_Cy, but also to some extent [Ca²⁺]_Mi. Newport Green fluorescence reflects [Zn²⁺]_Cy. MitoView633 and TO-PRO3 track ΔΨ_m and “canonical” SP, respectively. Although MitoView633 and TO-PRO3 are imaged in the same channel, the loss of ΔΨ_m (presumably indicating MPT) and the uptake of the normally impermeable TO-PRO3 can be distinguished because they are separated in time. The detection criteria for uptake of Ca²⁺, Zn²⁺, and TO-PRO3 was set at the 10% level of the dynamic range during “reperfusion”. Note that Ca²⁺ uptake occurs first, followed by Zn²⁺ uptake, and followed lastly by SP, with the zinc and SP events occurring at the same time in the last curve (colored open circles). In B, there is a ‘step’ in the [Ca²⁺]_Cy signal that was never observed in Control experiments (beginning of second phase of [Ca²⁺]_Cy indicated by vertical dashed line). D, overall, the onset of Zn²⁺ permeability occurred between the onset of permeability to Ca²⁺ and the onset of permeability to TO-PRO3. *, p < 0.0001 by one-way ANOVA with Newman-Keuls (SNK) post-hoc test for pairwise comparisons.

https://doi.org/10.1371/journal.pone.0212076.g008

When scrutinized more closely, it became apparent that in the Zinc group, there exists a phenomenon in some, but not all cells, that is not present in Control experiments. This phenomenon is a ‘step’ or a ‘gap’ in the rising phase of the [Ca²⁺]_Cy curve during “reperfusion”. This ‘step’ can be detected by visual inspection and quantified as the time difference between the 20 and 80% levels of the rising phase of [Ca²⁺]_Cy. Fig 9 shows four different cells from the Zinc group with varying ‘step’ durations, including a cell without a step (similar to those observed in Control cases), cells with a median and long step, and one with a “gap” (two phases of [Ca²⁺]_Cy increase separated by a period of [Ca²⁺]_Cy decrease). The thin black lines indicate the 20 and 80% levels of [Ca²⁺]_Cy rise and the colored open circles are the detected time points (denoted T20_Ca and T80_Ca, respectively). The time difference between the points demonstrates how the step size may vary between different myocytes.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 9. Zn²⁺-induced changes in [Ca²⁺]_Cy uptake dynamics.

In the presence of zinc, [Ca²⁺]_Cy dynamics change and the analyzed cells can be divided into two groups, those with a [Ca²⁺]_Cy ‘step’ and those without. A-D, data from representative cells in the Zinc group portraying varying durations of the step ranging from no step (similar to Control cases), to a median and long step, and then a ‘gap’ between two phases of [Ca²⁺]_Cy rise. The step duration is calculated between the points where the [Ca²⁺]_Cy curve crosses the 20 and 80% threshold, indicated by the horizontal thin black lines and colored open circles (T20_Ca and T80_Ca, respectively). This demonstrates great variability of Zn²⁺ interference in Ca²⁺ influx between different cells.

https://doi.org/10.1371/journal.pone.0212076.g009

Due to the great variability of Zn²⁺ effects, the time interval between T20_Ca and T80_Ca was not statistically different between the Control and Zinc groups. However, when the Zinc group was divided into two subgroups with and without a “step” in [Ca²⁺]_Cy rise, statistical analysis showed that the subset of Zinc cells with a “step” had a significantly longer interval between T20_Ca and T80_Ca as compared to Control cells and Zinc cells without the “step” (Fig 10).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 10. Quantification of the interval between 20 and 80% level of [Ca²⁺]_Cy increase during “reperfusion” (T80_Ca—T20_Ca) in the presence and in the absence of Zn²⁺ in the perfusate.

Zinc cells without observable “steps” in [Ca²⁺]_Cy rise are similar to Control cells, while Zinc cells with “steps” have a significantly longer T80_Ca—T20_Ca interval than either Control or Zinc without “steps” (one-way ANOVA with Newman-Keuls (SNK) post-hoc test for pairwise comparisons). Black dots indicate individual values of T80_Ca—T20_Ca interval in each group, and are overlaid with both the box and whiskers plot (red) and the mean diamonds plot (blue).

https://doi.org/10.1371/journal.pone.0212076.g010

Summarizing, experiments using Zn²⁺ in the perfusate suggest progressive sarcolemmal permeabilization, allowing for the intake of ions and molecules of increasing size as reperfusion advances. The peculiarities of mutual dynamics of Ca2⁺, Zn2⁺, and TO-PRO3 uptake suggest that all of these species permeate through the same transmembrane conduit, as discussed below.

Chronology of critical events leading to cardiomyocyte death during I/R

The mechanistic understanding of I/R injury in the heart has evolved into an exceedingly complicated network of interwoven pathways which conceptually includes necrosis, necroptosis, pyroptosis, apoptosis, and autophagy (to find topical reviews, see these recent meta-reviews [19, 20]). Whether these pathways interact and converge to produce a single critical event which decides the cell fate, or there is a combinatorial explosion of death phenotypes due to variable manifestations of specific mechanisms co-occurring in the same heart, remains poorly understood. For a long time, various groups identified the opening of the MPT pore as the pivotal event in cardiomyocyte death. The MPT pore is a non-selective large channel in the outer mitochondrial membrane (with still disputed molecular composition [21]), which is triggered by a combination of increased mitochondrial [Ca²⁺], reactive oxygen species (ROS), and inorganic phosphate [22]. Opening of the MPT pore causes collapse of the mitochondrial membrane potential (ΔΨ_m) and uncoupling of oxidative phosphorylation leading to cellular ATP depletion. In addition, it causes mitochondrial matrix swelling and rupture of the outer mitochondrial membrane resulting in the release of pro-apoptotic factors such as cytochrome c into the cytosol, thereby initiating the apoptotic pathway. Crompton and colleagues could be the first to suggest that MPT pore opening is a potential factor of acute myocardial I/R injury [22] and were the first to show that inhibition of MPT pore opening by CsA protected isolated ventricular myocytes from cell death caused by simulated I/R [23]. CsA is the most prevalent MPT pore inhibitor in experimental studies and is also the drug used to prevent reperfusion injury in clinical trials. Despite the fact that the majority (but not all [19]) experimental studies reported reduction of infarct size in CsA treated hearts/animals, the clinical trials overall show a lack of benefit when CsA was administered in patients undergoing percutaneous coronary intervention (CIRCUS trial [8]).

Different reasons were debated to explain the disappointing outcomes of the CIRCUS trial, but one important consideration here is the fact that CsA is a not a blocker, but rather a modulator, of the MPT pore. By blocking cyclophilin D, CsA reduces the sensitivity of the pore to [Ca²⁺], but cannot prevent MPT pore opening when a sufficiently high [Ca²⁺] in the vicinity of the pore is achieved [24]. In fact, MPT still occurs even in the complete absence of cyclophilin D (reviewed in [21])). Hence, pending discovery of true blockers of the MPT pore, the problem of MPT in reperfusion is reducible to the problem of a high enough level of mitochondrial Ca²⁺ overload during I/R.

Before the advent of the MPT pore theory, there was a considerable body of work supporting the notion that cytoplasmic Ca²⁺ overload during reperfusion causes cell death via excessive cardiomyocyte contraction (“hypercontracture”), leading to excessive membrane tension and mechanical rupture of the sarcolemma [25]. Clearly, sarcolemmal rupture is expected to allow a massive influx of Ca²⁺ into the cell, which could create a vicious circle further promoting hypercontracture, but also could trigger MPT. On the other hand, it was speculated long ago that the event of MPT leading to dissipation of ΔΨ_m, can be a cause of hypercontracture and sarcolemmal rupture, due to a critical depletion of ATP and a release of Ca²⁺ from the mitochondrial matrix into the cytoplasm [26]. A study supporting this sequence was later published [3].

From the above one can see that the known or speculative events during reperfusion could form a double vicious circle coupled through [Ca²⁺]_Cy. However, any vicious circle has to start from some event outside the feedback loop(s). There has to be an external trigger either for MPT, or [Ca²⁺]_Cy elevation, or both. To that effect, a canonical view has been developed which emphasized events occurring at the moment of restoration of oxygen and ionic imbalances at the onset of reperfusion. It is generally believed that in this setting a net Ca²⁺ influx takes place due to a coupled action of the sodium-proton exchanger and NCX in response to abrupt restoration of extracellular pH upon reperfusion (for review, see [27]). There is a theory that this excessive Ca²⁺ influx leads to SR Ca²⁺ overload and that spontaneous Ca releases from the SR facilitate mitochondrial Ca²⁺ overload through a direct Ca²⁺ cross-talk between SR and mitochondria, leading to MPT [4]. Another important acute post-reperfusion event is a burst of reactive oxygen species (ROS) occurring due to a rapid oxidation of NADH [28] or succinate [29] by a possibly damaged [30] electron transport chain. ROS elevation per se is sufficient to induce MPT. Importantly, these events are expected to occur within minutes of reperfusion [30]. Traditionally, it is also believed that MPT should occur within minutes of reperfusion, mostly based on a single line of work demonstrating that the largest increase in mitochondrial uptake of deoxyglucose, which is normally unable to cross the inner mitochondrial membrane, occurs between 2 and 5 minutes of reperfusion [31]. However, confocal imaging studies in whole hearts, using an abrupt and irreversible loss of ΔΨ_m-sensitive fluorophores in discrete myocytes as an indicator of MPT, showed that these events spread over a period of time from tens of minutes to hours [7, 9, 32]. At least some of those events cannot be explained by Ca²⁺ overload and/or ROS occurring upon restoration of perfusion and re-introduction of oxygen.

The findings of the current study are in line with prior observations in whole hearts [7, 9, 32] that the critical cellular events are not strictly coupled to the moment of reperfusion. While there were cells which simply did not survive the period of simulated ischemia, the majority of cells underwent critical transitions 15–25 min after the restoration of normal perfusion. Although we observed transient [Ca²⁺]_Cy elevations consistent with an activation of NCX in the first minutes of “reperfusion” (see Fig 3), those did not directly lead to other critical events. Regardless of [Ca²⁺]_Cy dynamics at the end of “ischemia” and beginning of “reperfusion”, which varied among cells, it was the “secondary” increase in [Ca²⁺]_Cy occurring after a latent period, which led to a consistent sequence of critical events. This sequence affirms the notion that an increase in [Ca²⁺]_Cy is the primary event of cell damage during reperfusion, and that it is a necessary and sufficient trigger of MPT. This sequence agrees with the sequence of events found in mouse hearts subjected to anoxia/reoxygenation [9], but is different from our prior observations in whole rabbit hearts subjected to global ischemia/reperfusion [7]. In our previous study the event of MPT was virtually simultaneous with the event of SP, and administration of CsA led to SP being ahead of MPT [7]. In our current study MPT clearly preceded SP, with no effect of CsA on the sequence of events. Hence, whereas in our previous study we could implicate the occurrence of large non-selective pores in the sarcolemma in cellular Ca²⁺ overload leading to MPT, in the current study we cannot, knowing that the critical increase in [Ca²⁺]_Cy occurred several minutes before the “canonical” membrane permeabilization (as indicated by cellular uptake of TO-PRO3 or/or cellular loss of a Ca²⁺ indicator). We will now discuss the possible source of the [Ca²⁺]_Cy increase in reperfusion, and whether findings of our current study can be reconciled with earlier observations, including our own.

The direction of pathological Ca²⁺ fluxes during reperfusion

As mentioned above, different prior studies have supported different directions of Ca²⁺ shifts between various cellular compartments during reperfusion. It is universally accepted that a net influx of Ca²⁺ into cells upon reperfusion is a critical component of I/R injury. However, the sources, the sinks, and the conduits for abnormal Ca²⁺ transport during reperfusion are not well established. The canonical view implicated reverse-mode NCX as a major source of Ca²⁺ influx into the cell during reperfusion [27]. Ca²⁺ ions entering cells should redistribute between three major compartments, the cytoplasm, mitochondria, and SR, assuming for simplicity that the Ca²⁺ absorbing capacity of other organelles, such as lysosomes and the nucleus, are small compared to SR and mitochondria. [Ca²⁺]_Cy overload is the prerequisite for hypercontracture, whereas [Ca²⁺]_Mi is the determinant of MPT. Upon SR Ca²⁺ release, RyR channels situated in close apposition to mitochondria may create a much higher [Ca²⁺] near the outer opening of the mitochondrial Ca²⁺ uniporter channel than the bulk [Ca²⁺]_Cy, creating a SR-mitochondrial “Ca²⁺ cross-talk” which was shown to significantly contribute to the regulation of [Ca²⁺]_Mi [33]. This cross-talk was implicated in facilitating MPT upon reperfusion, so that mitochondrial calcein release, used as an indicator of MPT, was partially blocked by the application of ryanodine/thapsigargin[4]. It was also postulated that upon the event of MPT, mitochondrial Ca²⁺ may redistribute to the cytoplasm, which was implicated in hypercontracture[3] or a regenerative spread of MPT [2].

In our experiments Ni²⁺, the most reliable blocker of NCX, overall blunted [Ca²⁺]_Cy dynamics and limited the maximum level of [Ca²⁺]_Cy observed during “reperfusion”. This can be considered as the evidence that NCX is a significant source of Ca influx triggering the death sequence in our model. However, there are several arguments against the possibility that reverse-mode NCX operation is the actual mechanism for the massive post-reperfusion increase in [Ca²⁺]_Cy observed in our model. First, this event occurred at variable delays after the restoration of normal extracellular milieu. The activation of reverse-mode NCX during reperfusion is linked to the abrupt restoration of the extracellular pH, which triggers influx of Na⁺ through sodium-proton exchanger, which in turn activates reverse-mode NCX extruding Na⁺ and bringing Ca²⁺ in. This sequence of events is expected to occur within the first minute of “reperfusion”. It is not clear why it would occur 15–20 minutes after reintroduction of oxygen and normalization of pH. Second, Zn²⁺ is not transported by NCX and does not influence transport of Ca²⁺ by NCX, and yet in our experiments Zn²⁺ influx was correlated with Ca²⁺ influx, and also interfered with Ca²⁺ influx. Alternative explanation of Ni²⁺ effect may be related to the competitive hinderance by Ni²⁺ of a putative pore which presumably also passes Ca²⁺, Zn²⁺ and TO-PRO3. At the high concentration necessary to block NCX (5 mM) and having the hydrated radius (3.11 Å [34]) smaller than that of Ca2+ (4.1 Å), Ni²⁺ ions could hinder and slow down Ca²⁺ influx through an aqueous pore. In addition, Ni²⁺ intake through the putative pore could have unintended effects inside the cells, possibly contributing to altered dynamics of ΔΨ_m and [Ca²⁺]_Mi observed in Nickel group (compare Figs 3 and 7).

Considering the role of SR, depletion of SR with ryanodine/thapsigargin in our experiments did not affect the sequence of critical events. In fact, depletion of SR accelerated the timing of all critical events (see Fig 5). Speculatively, this could be due to faster cytoplasmic and mitochondrial Ca²⁺ loading in the absence of a major Ca²⁺-absorbing entity.

Lastly, our data does not support the flux of Ca²⁺ from mitochondria to the cytoplasm at any time of reperfusion studied. In all cases, [Ca²⁺]_Mi elevation followed [Ca²⁺]_Cy elevation at some delay, and after MPT (as indicated by a precipitous loss of ΔΨ_m) [Ca²⁺]_Mi stayed constant or further increased, with the exception of occasional fluctuations (see Fig 3). Moreover, after the onset of the “terminal” [Ca²⁺]_Cy rise, abrupt decreases and increases in extracellular [Ca²⁺] immediately propagated to the cytoplasm and mitochondria, symmetrically in inward and outward directions (see Fig 6). Overall, our data suggest that the transition to cell death starts with an abnormal elevation of [Ca²⁺]_Cy which occurs via a non-canonical transmembrane conduit, but not due to SR release, and requires some hidden intracellular events to take place after the onset of reperfusion. [Ca²⁺]_Mi elevation follows [Ca²⁺]_Cy elevation after a ~5 min delay, which probably reflects a limited ability of mitochondria to actively resist Ca²⁺ overload (possibly due to Ca²⁺ buffering by intra-mitochondrial polyphosphates [35]). However, after this ability is saturated or broken, Ca²⁺ ions appear to move freely between extracellular space, cytoplasm, and mitochondrial matrix.

Sarcolemmal pore expanding during reperfusion—A hypothesis

Whereas for a number of years the focus has been on MPT as the pivotal event in myocardial I/R injury, our findings suggest that MPT is an epi-phenomenon, and an unavoidable consequence, of a critical increase in [Ca²⁺]_Cy. Since treatment attempting to prevent MPT has not been effective [8], prevention of the [Ca²⁺]_Cy elevation may be an effective strategy to salvage cardiomyocytes in the wake of I/R insult, provided we understand the mechanism of this elevation. We believe our data suggests that the critical Ca²⁺ influx occurs through a non-selective sarcolemmal pore expanding with time. A revealing observation here is the relationship between uptake of Ca²⁺, Zn²⁺, and TO-PRO3. The onset of detectable uptake of these species correlated with their estimated size when dissolved in water. The hydrated radii of Ca²⁺ and Zn²⁺ are 4.1 and 4.3 Å, respectively. Stokes radius of TO-PRO3 was not published but the estimated stokes radius of YO-PRO1, a close analog of TO-PRO3, is ~5 Å [36], and we determined in special experiments that TO-PRO3 and YO-PRO1 uptake during “reperfusion” is virtually simultaneous (S5 Fig). It cannot be excluded that the influx of the three species occurred through different channels; but then it would be hard to explain coordination of the presumably independent channels in time. More importantly, there appeared to be specific interactions between the influxes of the three species tested. First, the presence of Zn²⁺ changed the dynamics of Ca²⁺ influx in 57% of analyzed myocytes, introducing a “stairstep” or a “gap” between two phases of [Ca²⁺]_Cy rise (see Fig 8). Within the same NRVM monolayer exposed to Zn²⁺ in the perfusate, different cells exhibited variable durations and shapes of these Ca²⁺ influx discontinuities, but those were never observed in Control experiments. Also note that in some cells exhibiting the stairstep, the second phase of Ca²⁺ uptake was correlated with the onset of Zn²⁺ uptake (e.g., Fig 7B and 7C). While there isn’t direct evidence as to why this occurred, speculation could envision that Zn²⁺ might have the ability to briefly interfere with the developing pores and to hinder passage of Ca²⁺, if the pores were expanding slowly and were not yet large enough for Zn²⁺ to pass through. As the pores continue to expand, there could be a specific time point when both Ca²⁺ and Zn²⁺ could enter the cell. In cases where this step does not occur, it’s possible that the pores developed more quickly, such that there wasn’t a detectable window of opportunity for Zn²⁺ interference before it could flow into the cell. Lastly, it is worth mentioning that in some cells there was a remarkably similar profile of Zn²⁺ uptake and TO-PRO3 uptake (e.g., Fig 6C). It seems to be improbable that such cases were due to coincidence. Taken together, these different observations could be explained by variations in the dynamics of the pore expansion, which could occur in three stages. If the pore expands really quickly, then the uptake of all three species is very tightly coupled (an indeed we have examples of such cases, see S6 Fig). Otherwise, the relative timing of detectable Ca2+, Zn²⁺, and TO-PRO3 uptake could reflect the time spent by the pore in different stages of expansion.

The expanding pore hypothesis could reconcile discrepancies between findings of Davidson et al. [9], our own previous study [7], and the results of the current study. Amid the pore expansion, the moment of MPT would be determined by the time when the amount of Ca²⁺ absorbed by mitochondria exceeds the threshold for MPT. Depending on the interplay between the pore dynamics and the Ca²⁺ threshold for MPT in different models, the “canonical” SP event detected by uptake of small fluorophores could occur either before [7] or after [9] detectable MPT. Using a genetically encoded cytoplasmic Ca²⁺ indicator in mouse hearts, Davidson et al. detected peculiar Ca²⁺ “waves” which predicted MPT in the same cells. They could not explain the nature of those Ca²⁺ waves. We speculate that the Ca²⁺ waves they saw are equivalent to the critical [Ca²⁺]_Cy increases observed in the current study, and may be a result of opening of non-selective pores in the sarcolemma. We also predict that testing the uptake of Zn²⁺ in the whole heart I/R model would show that Zn²⁺ uptake events spatially correlate with Ca²⁺ “waves” and possibly precedes the event of MPT. However, these predictions need to be confirmed in future studies.

Whereas NCX operating in reverse-mode does not seem a be a plausible substrate of the putative expanding pore, there are a few membrane channels that can provide large sarcolemma permeability, and that can become conductive after metabolic inhibition (ischemia/reperfusion). These channels can be formed by proteins identified as connexins or pannexins.

Connexin channels are known to be present at the membrane of many cell types, including cardiac myocytes, and forming structures known as hemichannels [37]. These hemichannels are highly unselective in size and charge [38] and could allow the fast permeation of molecules close to 1KDa, especially during dephosphorylation [39]. Ventricular myocytes express mostly connexin43 (Cx43) [40–42]. Importantly, Cx43 is a phosphoprotein [43], and the conductive properties of the channels they form can be increased during dephosphorylation [44]. Cx43 hemichannels have been shown to open during metabolic inhibition in astrocytes [45, 46] [47]. There is evidence that blockade of Cx43 hemichannels by selective peptide inhibitors is protective in myocardial I/R [48], although the effect is modest [49]. In our experiments requiring continuous perfusion of the imaged cells, the use of Cx43 hemichannels blockers such as gap19 or gap26 was cost-prohibitive; thus, the possibility that Cx43 hemichannels underlie the putative expanding pore remains open. We are working on a miniaturized cell perfusion system which would enable to radically reduce the volume of perfusate and enable application of “gapXX” blockers in our continuous imaging protocol.

Pannexin channels are as complex as connexins; they form unopposed channels in the sarcolemma [50] These channels have been shown also to be highly conductive allowing passage of large molecules, including YO-PR01 [51], and become open during specific metabolic events [52]. They are found associated with ATP-gated ion channels known as P2X receptors which permits an additional conductance in the membrane [53]. A recent study provided evidence that pannexins are expressed in cardiac tissue and that their opening is induced by Trypanosoma cruzi infection, which mediate Chagas disease [54]. However, that effect, observed in NRVMs similar to those used in our study, was blocked by 400 μM probenecid. Since in our experiments probenecid was present in all solutions at the concentration of 500 μM, the role of pannexin channels can be ruled out.

Conclusions

In this study, we established a methodological framework which enabled us to reveal a consistent chronology of critical cellular events during simulated I/R in NRVMs. Despite a highly complex and multi-facetted mechanistic concept of myocardial I/R injury, in our NRVM model cell death upon reperfusion seemed to follow a quite specific path which begins with a pathological [Ca²⁺]_Cy increase due to Ca²⁺ influx through a possibly expanding non-selective transmembrane pore. Once a sufficient [Ca²⁺]_Cy increase occurred, the subsequent [Ca²⁺]_Mi increase followed by MPT are logical and probably unavoidable consequences. The nature of the putative pore remains unknown, but the fact that the abnormal Ca²⁺ uptake route can be hindered by extracellular Zn²⁺, and that Ca²⁺, Zn²⁺, and TO-PRO3 may pass through the same route, should provide initial guidance for steps towards establishing the molecular identity of the putative pore/channel. In future studies it would be important to reproduce the chronology of critical cellular events observed here in a whole heart model of I/R. If the sequence is confirmed, then the putative sarcolemmal pore mediating the abnormal Ca²⁺ influx during reperfusion has to be identified and explored as a target for therapies aimed at salvaging myocardium in the wake of an I/R episode.