**4. Discussion**

Matrix free [Ca<sup>2</sup>+] ([Ca<sup>2</sup>+]m) plays two important roles: (i) Activation of Ca2+-dependent dehydrogenases for oxidative phosphorylation at low concentrations [46]; and (ii) regulation of cytosolic Ca2+ by sequestration of excess Ca2+ at high concentrations [47]. Excessive accumulation of free [Ca<sup>2</sup>+]m is a leading factor in inducing mPTP opening. It is well established that repetitive mitochondrial Ca2+ loading triggers a gradual increase in [Ca<sup>2</sup>+]m, leading to a loss of IMM integrity that results in dissipation of ΔΨm and release of Ca2+. CsA is known to delay pore opening, in part, by inhibiting the PPIase activity of Cyp-D [31]. Whether CsA-mediated delay in mPTP opening involves regulation of [Ca<sup>2</sup>+]m by Pi-induced matrix Ca2+ buffering has not been addressed before. In this study, we investigated the effects of CsA on [Ca<sup>2</sup>+]m regulation during repeated Ca2+ loading and its functional significance in mPTP opening. Additionally, we determined if changes in [Ca<sup>2</sup>+]m induced by CsA correlated with changes in mitochondrial bioenergetics under identical experimental conditions and if matrix Pi was required for the observed CsA effects.

Since the key postulate was that CsA contributes to mitochondrial Ca2+ buffering, all experiments were performed in Na<sup>+</sup>-free condition to completely block NCLX as a route for efflux of excess matrix Ca2+. This allowed us to directly assess mitochondrial Ca2+ buffering capacity under different treatments. Our major findings during repetitive CaCl2 bolus challenges are: (i) CsA maintained basal ss[Ca<sup>2</sup>+]m owing to increased mitochondrial Ca2+ buffering capacity; (ii) the effectiveness of CsA to maintain basal ss[Ca<sup>2</sup>+]m correlates well with preserved mitochondrial bioenergetics; (iii) the buffering effect of CsA in a Pi-replete buffer was more pronounced than the known buffering effect of OMN+ADP; (iv) CsA-induced buffering was abolished in Pi-depleted mitochondria and Pi-free experimental medium. We conclude that the CsA-mediated delay in mPTP opening could, in large part, be attributed to CsA-induced activation of a Pi-dependent mitochondrial Ca2+ buffering system (MCBS), which maintains a low free [Ca<sup>2</sup>+]m and preserves mitochondrial bioenergetics.

#### *4.1. CsA-Mediated Inhibition of mPTP Opening Relates to the ss[Ca<sup>2</sup>*+*]m*

Using the two protocols (Figure 1A,B), we examined the changes in [Ca<sup>2</sup>+]e and [Ca<sup>2</sup>+]m in response to boluses of CaCl2 in the presence of vehicle (DMSO), CsA, ADP, OMN, or OMN+ADP over time. Our experimental approaches allowed us to define the contribution of CsA in the regulation of [Ca<sup>2</sup>+]m when CsA was given before the CaCl2 boluses (Protocol A) and at the threshold for pore opening under condition of increased free [Ca<sup>2</sup>+]m accumulation (Protocol B). Our results clearly indicate that the effect of CsA on delaying mPTP opening is due largely to its efficacy in maintaining free ss[Ca<sup>2</sup>+]m by activating the MCBS in a Pi-dependent manner, and thereby preclude early mitochondrial Ca2+ overload and delay induction of mPTP opening. Sustained low ss[Ca<sup>2</sup>+]e in the CsA-treated group indicated increasing mitochondrial Ca2+ uptake driven by the enhanced sequestration of free [Ca<sup>2</sup>+]m to maintain a transmembrane Ca2+ gradient and a charged ΔΨm that facilitated additional Ca2+ uptake (Figure 2). Unlike previous studies [14,33,34], NCLX was blocked under our experimental conditions, to prevent Ca2+ efflux during the repetitive CaCl2 additions; therefore, the net free ss[Ca<sup>2</sup>+]m in our study was determined by the balance between Ca2+ uptake and Ca2+ sequestration.

Notably, the CsA-induced buffering of mitochondrial Ca2+ resulted in greater Ca2+ uptake to attain a steady-state, as shown by the gradual decrease in ss[Ca<sup>2</sup>+]m with each added CaCl2 pulse (Figure 3). Insofar as Ca2+–Pi precipitation is a major mechanism for mitochondrial Ca2+ buffering, the sustained ss[Ca<sup>2</sup>+]m after each CaCl2 bolus indicated matrix Ca2+ storage, likely in the form of various inorganic Ca–Pi complexes [14]. The low and maintained ss[Ca<sup>2</sup>+]m during continuous matrix Ca2+ uptake is consistent with formation of these complexes. Although our study did not provide direct experimental evidence for CsA-induced matrix Ca–Pi complex formation, the continuous rise in estimated bound Ca2+:free Ca2+ ratio with each CaCl2 bolus as well as the ten-fold increase in <sup>m</sup>βCa clearly reflects a CsA effect on [Ca<sup>2</sup>+]m buffering capacity (Figure 3).

The protective effect of CsA in delaying mPTP opening has long been reported [28,31,32]. Our findings; however, provide the first direct evidence for a novel effect of CsA to enhance the capacity of mitochondria to sequester Ca2+ by which it obviates Ca2+-induced mPTP formation. Moreover, the effect of CsA in mediating greater matrix Ca2+ buffering explains the sustained free [Ca<sup>2</sup>+]m reported by Chalmers and Nicholls [14] and the CsA-induced inhibition of mitochondrial Ca2+ efflux observed in other prior studies [33,34].

#### *4.2. Underlying Mechanism of the CsA-Mediated [Ca<sup>2</sup>*+*]m Regulation*

It is well established that mitochondria are able to sequester large amounts of Ca2+, while maintaining free [Ca<sup>2</sup>+]m over a range of 0.1 and 10 μM depending on the Ca2+ load [14]; however, the mechanism and kinetics for this are unclear. Matrix Ca2+ buffering capacity is determined by: i) The quantity of Ca2+ that can be retained, and ii) the Ca2+ threshold level for release when Ca2+ exchangers are blocked or maximally operated [48]. The role of Pi as a physiological buffer in regulation of [Ca<sup>2</sup>+]m has been extensively studied [14,44,45,49]. The major mechanism of Pi-mediated Ca2+ sequestration in mitochondria is believed to be achieved by formation of amorphous Ca2+–Pi complexes in the matrix [48,50,51], which in turn maintain the free [Ca<sup>2</sup>+]m at a low level. Hence, sustained [Ca<sup>2</sup>+]m cyclically promotes more Ca2+ uptake via the MCU due to better preservation of both the Ca2+ gradient and ΔΨm.

Though Pi plays an essential role in matrix Ca2+ buffering, Pi has also been suggested to induce mPTP opening [52]. A recent study associated Ca2+–Pi precipitation with complex I inhibition and reduced ATP synthase rate during Ca2+ overload [53]. Another report demonstrated that increasing [Pi] decreased the mitochondrial Ca2+ loading capacity [14]. It was suggested that the mPTP-sensitizing effects of Pi was likely due to its effect in decreasing matrix-free Mg<sup>2</sup>+, an mPTP inhibitor [20]. In addition, formation of polyphosphate, a known inducer of mPTP, could be a factor in regulating the Ca2+ threshold for mPTP activation [54,55]. Interestingly, two prior studies [56,57] indicated that Pi is necessary for the inhibitory effect of CsA on mPTP opening. However, two other studies reported

that CsA inhibits mPTP opening even in the absence of Pi [58,59]. Conversely, in our study, the CsA-induced enhancement of matrix Ca2+ buffering was completely annulled when both mitochondria and the experimental medium were depleted of Pi (Figure 7). This loss of Ca2+ sequestration by CsA was reinstated when exogenous Pi was added just before activation of the mPTP (Figure 8). These observations provide the essential explanation for the requirement of Pi in the CsA-mediated MCBS and delay in mPTP opening.

The importance of mitochondrial matrix Ca2+ buffering via Pi is underscored by the studies of Wei et al. [44,45]. They reported that Pi modulates the total amount of Ca2+ uptake with smaller CaCl2 boluses, whereas Pi modulates Ca2+ buffering capacity with larger CaCl2 boluses. Since we had Pi in our experimental medium and the mitochondria were replete with exogenous Pi, the observation that CsA induced low ss[Ca<sup>2</sup>+]e and ss[Ca<sup>2</sup>+]m could be explained by the following: (i) CsA activates Pi-dependent matrix Ca2+ buffering potentially by maintaining the rate of Ca2+–Pi complex formation; and (ii) CsA may activate Pi transport processes (via H+/Pi transporter and/or phosphate carrier) that help to maintain both the IMM pHm and ΔΨm gradients. These processes would limit the increase in free [Ca<sup>2</sup>+]m, which in turn would contribute to more Ca2+ uptake and retention by increasing the electrochemical driving force for Ca2+ influx.

#### *4.3. CsA vs. ADP; As a Regulator of [Ca<sup>2</sup>*+*]m*

AdN are implicated as one of the multiple matrix factors responsible for sequestering Ca2+ by mitochondria [29,60–62]. AdN can potentiate mitochondrial Ca2+ buffering by maintaining high matrix Pi concentrations that can facilitate precipitation of AdN-Ca–Pi complexes, including, ATP-Mg<sup>2</sup>−/Pi<sup>2</sup>− and HADP2−/Pi<sup>2</sup><sup>−</sup>, and thereby increase the Ca2+ threshold for mPTP opening [60,63]. In a study by Carafoli et al. [60], it was reported that mitochondrial Ca2+-buffering is proportional to mitochondrial ADP uptake. In our Pi-replete study, OMN+ADP had a relatively small effect on Ca2+ buffering compared to CsA, but it had a significantly larger effect than ADP or OMN alone (Figures 2, 3 and 5). A reasonable explanation could be that OMN, an ATP synthase (Complex V) inhibitor [64], could contribute towards augmenting the AdN pool and thus enhance matrix Ca2+ buffering. Consistent with our findings, a previous study also showed a greater CRC with a low-concentration of ADP with OMN compared to 10-fold larger concentration of ADP alone [30]. Thus, in agreemen<sup>t</sup> with Sokolova et al. [30], the observed high buffering capacity and expanded CRC with OMN+ADP is largely attributed to the ADP component of the matrix AdN pool. However, a previous study [62] reported that AdN also prevent mitochondrial Ca2+ influx by directly chelating Ca2+ by a Ca–ATP complexation [61]. Contrary to this observation, in our study, the direct effect of ADP on binding free Ca2+ was negligible, as assessed by adding ADP and CaCl2 together in mitochondria-free experimental buffer (Figure S4). Additionally, carboxyatractyloside-mediated inhibition of ADP uptake via adenine nucleotide translocase precluded matrix Ca2+ buffering and blunted the CRC by OMN+ADP or ADP alone (Figure S5). In this case, the extra-matrix ADP that accumulated did not chelate the Ca2+ added to the buffer. Altogether, these observations indicate that a direct sequestration of Ca2+ outside the mitochondria does not explain the effect of ADP alone or OMN+ADP on the enhanced CRC in our study.

A previous study [42] from our group proposed that the MCBS relies on at least two classes of Ca2+ buffers. The first class could represent classical Ca2+ buffers, including mostly metabolites (ATP, ADP, and Pi) and mobile proteins that bind a single Ca2+ ion at a single binding site. A second class of buffers could be associated with the formation of amorphous Ca2+ phosphates, which may be capable of binding multiple Ca2+ ions at a single site in a cooperative fashion [35,38,39,42]. Genge et al. [65] showed, in an in vitro study, that annexins, a diverse class of proteins, are required for Ca2<sup>+</sup>-phosphate nucleation. Additionally, many studies have suggested an AdN-dependent Ca2+-binding property of annexins [66]. Interestingly, mitochondria exposed to ADP alone or OMN+ADP retained their ability to maintain low ss[Ca<sup>2</sup>+]e and ss[Ca<sup>2</sup>+]m for an extended period of cumulative CaCl2 additions, and showed a higher Ca2+ threshold for mPTP opening without Pi compared to with Pi (Figure 7). This

extended delay in mPTP opening in the Pi-depleted state compared to the Pi-replete state reflects the ability of Pi to induce early mPTP opening under certain conditions [52]. In this case, the presence of Pi appears to counteract the ADP delay effect and induce a much earlier pore opening compared to the Pi-depleted state. The mechanisms for this AdN-mediated massive matrix Ca2+ loading capacity in the absence of exogenous Pi is unclear and needs to be further investigated. A plausible hypothesis could be that, in the absence of Pi, a significant Ca2+ loading capacity of AdN might be mediated via direct interaction with annexins. CsA, on the other hand, might function as a mediator that activates a Pi-dependent Ca2+ buffering system. Another possibility is that Cyp D, as a PPIase, reduces free phosphate levels in the matrix or blocks the Ca2+ binding property of annexins; this then would be relieved by CsA's effect to block Cyp D.

#### *4.4. Implication of CsA-Mediated Ca2*+ *Bu*ff*ering on Mitochondrial Bioenergetics*

Elevated [Ca<sup>2</sup>+]m over the nanomolar range is reported to increase NADH generation in part by stimulating Ca2<sup>+</sup>-sensitive dehydrogenases of the TCA cycle [67,68] and activating the F0F1-ATP synthase [69], thereby accelerating oxidative phosphorylation (OXPHOS). However, excess mitochondrial free Ca2+ can dissipate ΔΨm and impede OXPHOS. The IMM ΔΨm is the key factor in generating the proton motive force across the IMM; it is also one of the primary driving forces for Ca2+ uptake via the MCU [70] and triggers Ca2+ efflux via the NCLX [71,72]. Therefore, if mitochondria continue to take up Ca2+ under increased extra-matrix Ca2+ exposure, the Ca2+ would have to be buffered or ejected to prevent excess free [Ca<sup>2</sup>+]m accumulation that could dissipate ΔΨm and increase oxidation of NADH.

The stability of Ca2+–Pi precipitates inside the mitochondrial matrix largely depends on pHm [50]. It is also proposed that the matrix [Pi] depends on the pH gradient (e.g., a change in pH from 7 to 8 has been estimated to increase [Pi] by a factor of 1000 [14,50]). Thus, matrix alkaline conditions could facilitate Ca2+–Pi precipitation, whereas matrix acidification could lead to a destabilization of the Ca2+–Pi precipitate and so enhance matrix free Ca2+ levels [14]. Consequently, we correlated the changes in [Ca<sup>2</sup>+]m with indices of mitochondrial bioenergetics (ΔΨm, NADH, and pH) (Figure 5) to have a better understanding of the CsA-mediated MCBS. Mitochondria exposed to CsA before the repetitive CaCl2 boluses, exhibited robust mitochondrial Ca2+ uptake and rapid [Ca<sup>2</sup>+]m buffering while maintaining basal ΔΨm, NADH, and an alkalinized pHm until mPTP opened (Figure 5). Maintaining ΔΨm during excess Ca2+ uptake in the absence of functioning NCLX suggests a strong matrix buffering effect that is induced by CsA.

In Protocol B, when CsA was added just before the onset of pore opening, NADH and ΔΨm levels transiently increased but immediately returned to baseline with each added CaCl2 bolus. This transient depolarization and NADH oxidation with each addition of CaCl2 was not observed in Protocol A. The reason for this is unclear. Nonetheless, the observed transient oxidation of NADH helped to restore ΔΨm after Ca2+ induced transient depolarization before the next CaCl2 bolus (Figure 5D,E). The transient redox oxidation and ΔΨm depolarization sugges<sup>t</sup> that the CsA added at the point just before mPTP opening activated MCBS more slowly compared to Protocol A. In addition, CsA maintained the pHm gradient during prolonged Ca2+ pulse challenges (Figure 5C). This finding also likely excludes a contribution of the mitochondrial calcium–hydrogen exchange (mCHE) to the Ca2+ extrusion in the absence of NaCl. We have recently reported that CsA obviates mCHE activity at low extra-matrix pH [19]. However, based on our current results, it is likely that CsA triggered an enhancement of mitochondrial Ca2+ buffering so that the resulting low [Ca<sup>2</sup>+]m and maintained ΔΨm and pHm accounted for the inactivity of mCHE.
