*Article* **Glucose-Induced Expression of DAPIT in Pancreatic** β**-Cells**

#### **Alberto Leguina-Ruzzi , Anežka Vodiˇcková, Blanka Holendová , Vojt ˇech Pavluch, Jan Tauber, Hana Engstová, Andrea Dlasková and Petr Ježek \***

Department of Mitochondrial Physiology, No.75, Institute of Physiology of the Czech Academy of Sciences, 142 20 Prague, Czech Republic; AlbertoAndres.LeguinaRuzzi@fgu.cas.cz (A.L.-R.); anezka.kahancova@fgu.cas.cz (A.V.); blanka.holendova@fgu.cas.cz (B.H.); vojtech.pavluch@fgu.cas.cz (V.P.); jan.tauber@fgu.cas.cz (J.T.); hana.engstova@fgu.cas.cz (H.E.); andrea.dlaskova@fgu.cas.cz (A.D.)

**\*** Correspondence: jezek@biomed.cas.cz; Tel.: +420-296442760

Received: 20 April 2020; Accepted: 8 July 2020; Published: 10 July 2020

**Abstract:** Transcript levels for selected ATP synthase membrane FO-subunits—including DAPIT—in INS-1E cells were found to be sensitive to lowering glucose down from 11 mM, in which these cells are routinely cultured. Depending on conditions, the diminished mRNA levels recovered when glucose was restored to 11 mM; or were elevated during further 120 min incubations with 20-mM glucose. Asking whether DAPIT expression may be elevated by hyperglycemia in vivo, we studied mice with hyaluronic acid implants delivering glucose for up to 14 days. Such continuous two-week glucose stimulations in mice increased DAPIT mRNA by >5-fold in isolated pancreatic islets (ATP synthase F1α mRNA by 1.5-fold). In INS-1E cells, the glucose-induced ATP increment vanished with DAPIT silencing (6% of ATP rise), likewise a portion of the mtDNA-copy number increment. With 20 and 11-mM glucose the phosphorylating/non-phosphorylating respiration rate ratio diminished to ~70% and 96%, respectively, upon DAPIT silencing, whereas net GSIS rates accounted for 80% and 90% in USMG5/DAPIT-deficient cells. Consequently, the sufficient DAPIT expression and complete ATP synthase assembly is required for maximum ATP synthesis and mitochondrial biogenesis, but not for insulin secretion as such. Elevated DAPIT expression at high glucose further increases the ATP synthesis efficiency.

**Keywords:** mitochondria; USMG5/DAPIT; glucose-stimulated insulin secretion; glucose-induced expression; membrane subunits of ATP synthase; ATP synthase oligomers mitochondrial cristae morphology

#### **1. Introduction**

Pancreatic β-cells sense glucose and respond to its elevated concentration in rich capillaries of the pancreatic islets by exocytosis of insulin granules [1–7]. The increased mitochondrial ATP synthesis due to the elevated oxidative phosphorylation (OXPHOS) represented a consensual canonical mechanism of glucose-stimulated insulin secretion (GSIS). We recently demonstrated that also NADPH-oxidase-4- (NOX4-) mediated H2O<sup>2</sup> release is fundamentally required for GSIS in addition to ATP [8]. The elevated OXPHOS is transformed to the increased ATP/ADP ratio at the peri-plasma membrane loci proximal to the ATP-sensitive K <sup>+</sup> channel. The channel closes when both conditions are fulfilled, i.e., elevated ATP and elevated H2O<sup>2</sup> [8], which results in the depolarization of the plasma membrane. This activates voltage-gated L-type Ca <sup>2</sup><sup>+</sup> channels (CaL). The subsequent Ca <sup>2</sup><sup>+</sup> entry to the cytosol initiates the insulin exocytosis of insulin granules.

To match physiological postprandial glucose concentrations with the sensitivity range of the glucose sensor, numerous factors delicately tune glucose-concentration dependence for the insulin release mechanism. The insulin-independent glucose transporter GLUT-2 (GLUT-1 in humans) rapidly equilibrates blood glucose with the cytosolic glucose in β-cells [9]. The glucokinase (hexokinase IV) is insensitive to the inhibition by glucose-6-phosphate [10]. Moreover, low activities of lactate dehydrogenase and pyruvate dehydrogenase kinase enable a nearly 100% utilization of pyruvate by OXPHOS [1–6].

We found that also the inhibitory factor IF1 of the mitochondrial ATP synthase belongs to the key proteins, ensuring the physiological range of the glucose sensor [11,12]. The IF1 slightly inhibits synthesis of ATP, thus setting the range for elevation of phosphorylating respiration and insulin release above 3-mM glucose [11,13], with half-activation between 3.5 and 4-mM and saturation above 8-mM glucose [11,12]. When such a slight in vivo inhibition was largely cancelled using the silencing of IF1, the elevation of respiration and OXPHOS occurred at very low glucose concentration approaching to zero [9]. Simultaneously, the half-activation for the insulin release dose–response was shifted to the range of 0–2-mM glucose in INS-1E cells [9]. In contrast, overexpression of IF1 substantially blocked GSIS [13].

Another peculiar aspect of the pancreatic β-cell glucose sensor is the steepness of the related dose–response, i.e., insulin release dependence vs. glucose concentration [11,12,14–16]. The revealed narrowing of mitochondrial cristae may hypothetically contribute to this steepness, such as observed in INS-1E cells during GSIS [17] or upon the increased load of cell-permeant Krebs cycle substrate [18,19]. Moreover, the ATP synthase dimers, tetramers [20] or speculatively higher oligomers are associated within the rows along the crista rims [21–26].

The stabilization within the rims may depend on the presence and correct assembly of the membrane embedded subunits *e*, *f*, *g* of the membrane FO-sector [20,26]; or DAPIT, which is the peripheral outmost subunit of the ATP synthase (Figure 1), added lastly upon the assembly [26]. DAPIT is a product of the rat *Usmg*5 gene [27–31]. Similarly, Mic10 was suggested to be inserted between the neighbor dimers or tetramers of the ATP synthase [32,33]. Likewise, crosslinking of two F<sup>1</sup> moieties from the neighbor ATP synthase dimers by the un-phosphorylated dimeric inhibitory factor IF1 can stabilize the rim [20], but at the expense of ATP synthesis, since this interaction is inhibitory [13]. The inner mitochondrial membrane (IMM) bending itself may be ensured by subunits *e* and *g* [20]. In porcine ATP synthase a self-standing subunit *k* belongs to those outmost subunits [14], whereas a yeast subunit *k* was originally annotated as a DAPIT ortholog [26].

DAPIT is bound to the (mtDNA-) mitochondria-encoded subunit *a* (Figure 1). DAPIT may also interact with the second mtDNA-encoded subunit ATP8 [20,26]. However, a tetramer formation may be facilitated by an interaction of DAPIT with the subunit *g* within an interface between all four monomers [20]. Such interface was termed site three among altogether six sites determining the tetramer structure. Since DAPIT was found to be assembled as the last subunit into the FO-sector structure [26], we may speculate that its stoichiometry to the F<sup>O</sup> moiety may be variable.

Moreover, a fraction of Mic10 was found to be associated with the ATP synthase and hypothetically can crosslink the neighbor dimers [32,33]. Mic10 otherwise ensures 90◦ curvature of the IMM at the crista outlets being a part of the MICOS complex [32,33]. Besides Mic10 [32,33], also DAPIT has been suggested to crosslink rows of ATP synthase dimers at crista rims and thus stabilize them [26].

The DAPIT was originally termed diabetes-associated protein in insulin-sensitive tissue. At a complete stoichiometry, the physiologically established dimers and tetramers of the ATP synthase possess two and four FO-sectors, respectively, each containing a single DAPIT subunit, embedded into IMM, actually spanning from its intracristal surface to the matrix IMM surface (Figure 1). DAPIT-knockdown in He–La cells reportedly led to a slightly decreased ATP synthesis activity and slower cell growth, while transcripts for the F1-sector subunits α and β did not change [29]. Moreover, a homozygous splice-site mutation (c.87 + 1G > C) in *Dapit*/*Usmg5* gene reportedly determines suppression of ATP synthase dimers and inhibition of ATP synthesis in patient's fibroblasts [34]. This particular DAPIT mutation causes altered mitochondrial cristae in fibroblasts of Leigh syndrome

patients [35]. Possibly similar or related mechanisms as caused by the DAPIT mutation are involved in sensitizing of pancreatic cancer cells to inhibitors impairing their survival [36].

**Figure 1.** DAPIT location within the ATP synthase structure. Neighbor subunits of DAPIT are shown within the structure of membrane FO-sector of the ATP synthase. DAPIT and the surrounding subunits are depicted in color, the remaining subunits are transparent. A side view represents a transversal section of the crista rim; a "bottom" view is from the intracristal space directly viewing the inner sharpest IMM bend, when one considers the F<sup>1</sup> moiety being positioned at the top. A top view is also the top view for the crista rim and the top of the sharp edge of the IMM bent. The ATP synthase structure was derived from the atomic model for the dimeric F<sup>O</sup> region of mitochondrial ATP synthase published by Guo H. et al. [37], pdb code 6b8 h. The structure was visualized using the PyMOL Molecular Graphics System, Version 1.8 Schrödinger, LLC.

Despite its name pointing to diabetes (alternatively to *upregulated in skeletal muscle gene 5*, *Usmg5* in the rat genome), the DAPIT protein has not been studied in pancreatic β-cells and there is no knowledge concerning its effect on the regulation of the ATP synthesis efficiency and its role in GSIS. Consequently, we attempted to study possible DAPIT role, using the rat pancreatic β-cell line INS-1E.

#### **2. Materials and Methods**

#### *2.1. Model Cells*

#### 2.1.1. Cell Culturing

Rat insulinoma INS-1E cells were kindly provided by Prof. Maechler, University of Geneva or purchased from AddexBio (San Diego, CA, USA; cat. No. C0018009). The results were invariant of the source as documented in previous reports [8,11,12,38,39]. Cells were cultivated with 11-mM glucose in RPMI 1640 medium with l-glutamine, supplemented with 10-mM HEPES, 1-mM pyruvate, 5% (*v*/*v*) fetal calf serum, 50-µmol/L mercaptoethanol, 50 IU/mL penicillin and 50-µg/mL streptomycin [38,39]. Prior to each experiment, cells were preincubated for 30 min (two washes of 15 min each) in cultivation medium containing 3-mM glucose and 2.5% of fetal calf serum.

#### 2.1.2. Silencing

Silencing of *Usmg*5/*Dapit* expression was performed with two database-registered siRNAs (#HA13041841, #HA13041842 by Sigma-Aldrich, St. Louis, MO, USA) at final concentration of 50 nM, transfected with Lipofectamine™ RNAiMAX Reagent (13778150, Thermo Fisher, Life Technologies Waltham, MA, USA), following the manufacturer protocol. Effects of siRNA silencing were tested by individual western blots and qPCR to ensure specificity of the silencing. Scrambled-sequence containing siRNA, nonspecific to any *Rattus norvegicus* gene (Scr siRNA), was used as a control.

#### *2.2. qRT-PCR*

#### 2.2.1. qRT-PCR Protocol

Total RNA was extracted from cells by the acid guanidinium thiocyanate–phenol–chloroform extraction using Trizol reagent (15596-018; ThermoFisher, Waltham, MA, USA). The reverse transcription was performed using QuantiTect Reverse Transcription Kit ( QiaGen, Venlo, The Netherlands; cat. No. 205310), using 1000 ng of RNA previously quantified by NanoDrop 2000 (Thermo Fisher). Real-time polymerase chain reaction was performed with FastStart Essential DNA Green Master Mix (Roche, Basel, Switzerland). The PCR amplification was carried out with initial denaturation at 95 ◦C for 20 s, followed by 45 cycles of 95 ◦C for 3 s and 60 ◦C for 30 s in a LightCycler 96 (Roche). The primers used were as indicated in Table 1. All of them led to a single amplicon product, when the described qRT-PCR procedure was performed. Data were calculated by the 2−∆∆CT method (where CT is cycle threshold and ∆∆CT is CT for the gene of interest minus CT of the internal control), having β-actin ("*Actb*") as an internal control for each experimental condition.




**Table 1.** *Cont.*

#### 2.2.2. Estimation of mtDNA Copy Number

The mtDNA from model β-cells, INS-1E cells, was isolated by phenol–chloroform extraction. This was followed by SYBR Green qPCR amplification with primers annealing on the *Slco2a1* nuclear gene (encoding solute carrier organic anion transporter family member 2A1) and the *Nd5* mitochondrial gene (bp 11,092 to 11,191 according to GenBank sequences from The National Center for Biotechnology Information, Bethesda, MD, USA) [40]. For primers see Table 1. The ratio between *Nd5* amplicon and half of the nuclear amplicon amounts was taken as the mtDNA copy number per cell.

#### *2.3. Protein Separation Methods and Western Blotting*

#### 2.3.1. SDS-Electrophoresis and Western Blotting

To determine relative levels of DAPIT/USMG5, ATP synthase α-subunit, PGC-1α and β-actin proteins, cells were lysed in RIPA buffer (Tris-HCl 20 mM, NaCl 150 mM, Na2EDTA 1 mM, EGTA 1-mM and NP-40 1%) containing protease inhibitors (1-mg/mL aminocaproic acid, 1-mg/mL benzamidine, 0.2-mg/mL SBTI and 3-mM PMSF) and phosphatase inhibitors (0.012-mg/mL sodium orthovanadate, 4.46-mg/mL sodium pyrophosphate and 4.2-mg/mL sodium fluoride). Total protein concentration was determined by the bicinchoninic acid protein assay (Pierce, Waltham, MA, USA).

Proteins (30 µg) from lysates were separated by electrophoresis in 12% SDS–polyacrylamide gel (SDS-PAGE). Proteins were transferred to a 0.45 µm PVDF membrane, which was blocked with 5% non-fat milk in TTBS containing 0.05% Tween-20 at room temperature. Then, the PVDF membrane was incubated overnight at 4 ◦C with the primary antibody anti-ATPA (Abcam, Cambridge, MA, USA, ab110273), PGC1 (ThermoFisher, PA5–38021), DAPIT/USMG5 (Abcam, ab108225) or anti-β-actin mAb (Sigma–Aldrich, A2228) at 1:2000 of dilution, followed by incubation with secondary antibody conjugated to peroxidase at 1:10,000 of dilution (Santa Cruz Biotechnology, Dallas, TX, USA) for one hour at room temperature. Immunoreactive bands were visualized using a chemiluminescent reagent (Western Lightning, PerkinElmer, Waltham, MA, USA) according to the procedure described by the manufacturer. Chemiluminescence was detected by the Chemidoc-IT Imaging System (UVP, LLC) and immunoreactive bands were analyzed by densitometry analysis using the ImageJ software (National Institutes of Health).

#### 2.3.2. Isolation of Mitochondria from INS-1E Cells

Mitochondria were isolated by differential centrifugation in ice-cold medium (180-mM KCl, 5-mM MOPS-K buffer, pH 7.2, 2-mM EGTA and 0.5% bovine serum albumin, BSA) according to a published procedure [41]. Final mitochondrial pellets were washed by resuspension/centrifugation in the isolation medium lacking BSA. The protein content was determined by the BCA method (Sigma).

#### 2.3.3. Blue-Native and Clear-Native Electrophoresis

ATP synthase oligomerization was analyzed by one-dimensional blue native (BN-) and clear native (CN-) polyacrylamide gel electrophoresis (PAGE). Protein complexes were extracted from INS-1E cells or INS-1E cell mitochondria using digitonin (Thermo Fisher; 2.5:1, *w:w*, protein:digitonin ratio). The obtained lysates were separated on bis-Tris 3%–13% gels. Following BN/CN PAGE proteins were transferred by wet electroblotting onto PVDF membranes. ATP synthase and DAPIT were immunodetected with monoclonal antibodies against the ATP synthase F1α-subunit (Abcam, ab110273) and DAPIT/USMG5 (Abcam, ab108225) after acidic membrane-stripping.

#### *2.4. ATP Assay*

Before the experiment, siRNA-transfected cells were preincubated for 2 h in the KRH buffer with 0.1% fatty acid-free BSA and indicated glucose concentrations. Quantification of ATP was performed using the ATP Assay bioluminescence kit HSII (Roche). Cells were mixed with boiling lysis buffer (100-mM Tris, 4-mM EDTA, pH 7.75) and further boiled for 2 min. Samples were centrifuged at 10,000× *g* for 1 min. Diluted supernatants were mixed with luciferase reagent and a Synergy HT luminometer was used to read the bioluminescence. To confirm that the assay procedure does not interfere with the ATP concentration determination, internal ATP standards were added to the samples during the initial experiments.

#### *2.5. High Resolution Respirometry*

Routinely, an oxygraph 2k (Oroboros Instruments GmbH, Innsbruck, Austria) has been used for experiments checking respiration of INS-1E cells as described elsewhere [11,12].

#### *2.6. Transmission Electron Microscopy*

Cells were cultured on poly-l-lysine–coated petri dishes. For transmission electron microscopy (TEM), cells were fixed for 24 h in 0.1-M cacodylate buffer (pH 7.2) containing 2.5% glutaraldehyde and postfixed in 2% OsO<sup>4</sup> in the same buffer. For visualization of conspicuous membranes of mitochondria, post fixation was performed with 2% OsO<sup>4</sup> plus 0.8% K4Fe(CN6) in a PBS buffer. Fixed samples were dehydrated through an ascending ethanol and acetone series and embedded in araldite –Poly/Bed® 812 mixture (ThermoFisher, Waltham, MA, USA). Thin sections were cut on a Reichert-Jung Ultracut E ultramicrotome and stained using uranyl acetate and lead citrate. Sections were examined and photographed using JEOL JEM-1011 electron microscope. Fine structure measurements were performed using a Veleta camera and iTEM 5.1 software (Olympus Soft Imaging Solution GmbH, Hamburg, Germany).

#### *2.7. Experiment with Mice*

#### 2.7.1. Mice

C57BL/6 mice were obtained from The Jackson Laboratory, Bar Harbor, MN. Experiments were approved by the Animal Care and Use Committee (Inst. Molecular Genetics, ASCR) in accordance with the European Union Directive 2010/63/EU for animal experiments, U.K. Animals (Scientific Procedures) Act, 1986 and the Guide for the Care and Use of Laboratory Animals (NIH Publication No. 85–23, revised 1996) and the ARRIVE guidelines (www.nc3rs.org.uk/ARRIVE).

#### 2.7.2. Experiments with Hyaluronic Acid Implants

Our protocol was adopted from Praveen et al. [42,43]. Hyaluronic acid implants were a kind gift from Dr. Semira Kwabi from Skin Clinic UK. A Juvederm Ultra 4 hyaluronic acid filler was mixed with d-glucose to a final concentration of 10 M. Such preparation (either 100 µL or 200 µL) was subcutaneously implanted to mice. Typically, two weeks of treatment with hyaluronic acid implants of the designated volumes were performed. Pancreases were removed from three mice and mRNA was isolated using an RNA isolation kit (Qiagen). This was followed by qRT-PCR as described above. Alternatively pancreatic islets were isolated from three pooled pancreases as described elsewhere [40].

#### *2.8. Statistical Analysis*

Results are presented as mean ± standard deviation (SD) for *N* number of biologic replicates or total number of estimates (*n*). Graphs were plotted and statistical analyses were performed using SigmaPlot 6.0 and SigmaStat 3.1 (Systat Software, San Jose, CA, USA) and ANOVA followed by the Tukey's test on the pre-validated data; or, alternatively using a Prism (GraphPad Software, San Diego, CA, USA) and an unpaired Student's *t*-test (when comparing two groups) or one-way nonparametric ANOVA (Tukey's test) followed by Bonferroni post hoc analysis (for the comparison of more than two groups) or correlation analysis. Statistical significance was set at \*\*\* *p* < 0.001; \*\* *p* < 0.05; \* *p* < 0.1.

#### **3. Results**

#### *3.1. Expression of DAPIT and Other F<sup>O</sup> Subunits Varies according to Metabolic State in INS-1E Cells*

INS-1E cells are routinely cultured with 11-mM glucose, thus keeping optimum autocrine factors and *Ins* gene expression [39,41]. Hence, we attempted to study dependencies on glucose for selected subunits of the membrane F<sup>O</sup> moiety of the ATP synthase, starting with DAPIT. To do this, glucose was first depleted from INS-1E cells, using short prewashing in PBS followed by two 15-min washings in the culturing medium with 3-mM glucose and a half-content of the fetal calf serum. Cells were subsequently incubated at 3 mM, 11-mM and 20-mM glucose in the KRH buffer for up to 120 min. The *Usmg5*/*Dapit* transcript frequently decreased after the two consequent 15 min preincubations with 3-mM glucose (Supplementary File 1; Figure S1a) and for some cell passages we have encountered decreased DAPIT mRNA even after the initial PBS wash. One may regard these conditions as simulating fasting state in vivo, when a slower rate of metabolism is established; nevertheless, for cultured cells this treatment represents a certain form of shock.

Further incubation in KRH containing 11 mM or 20-mM glucose led frequently to restoring the initial levels of DAPIT mRNA already after 30 min (Supplementary File 1; Figure S1a). With further incubations at 20-mM glucose up to 120 min, DAPIT transcript increased up to the ratio to β-actin about 2.5 for some cell passages (Supplementary File 1; Figure S1b). The extent of such elevation was variable, most probably depending on the variability in the initial pretreatment. Nevertheless, specificity of this qRT-PCR assessment was checked in conjunction with pre-transfection by DAPIT siRNA. Indeed, if elevations of DAPIT transcript occurred in control INS-1EScrl cells (transfected with siRNA containing the scrambled sequence), they vanished when cells were silenced for DAPIT (Supplementary File 1; Figure S1b). Moreover, when cells were left in the medium with 11-mM glucose up to 120 min, DAPIT mRNA started to rise around 60 min (Supplementary File 1; Figure S1c). When glucose was initially set to 20 mM, the elevations of DAPIT mRNA levels were higher around and after 30 min (Supplementary File 1; Figure S1c). We may interpret these elevations as responses to the elevated metabolism and beginning at 60 min even some biogenesis could take place. Investigations of these aspects are reported below.

Dealing with the other three selected subunits of the ATP synthase membrane F<sup>O</sup> sector, i.e., subunits *e*, *f* and *g*, their transcripts were less sensitive to pretreatments in the medium containing 3-mM glucose, except of *g* (Supplementary File 1; Figure S2a–c). As expected, DAPIT silencing had no effect to the mRNA recovery of these subunits due to the accelerated metabolism (Supplementary File 1; Figure S2d–f). Their recoveries/elevations were nearly synchronous with those found for DAPIT mRNA.

#### *3.2. Upregulation of DAPIT in Pancreatic* β*-Cells in Mice*

Next, we investigated, whether any glucose-induced upregulation of DAPIT exists in vivo in mice (Figure 2a–e; Figure 3a–h; Figure 4). The transcript of mouse *Usmg*/*Dapit* was estimated in whole pancreases (Figure 2a,b,d), as well as in the isolated pancreatic islets (Figure 2c). Up to 2 weeks of diffusion of glucose was allowed from the hyaluronic acid implants inserted into the omentum of Black6/J mice (Figure 2a). Attained fasting blood glucose levels reached on average to 7.5 and 9-mM for 100 µL and 200 µL volumes of implants, respectively (Figure 2e). The observed increases relatively to β-actin in whole pancreases (Figure 2b) and isolated islets (Figure 2c) support the existence of the glucose-induced expression of DAPIT in vivo. Similarly, an increase in F<sup>1</sup> subunit α mRNA was pronounced in mice having glucose delivery by the implants (Figure 3e), as well as increases in mRNA of the other selected membrane F<sup>O</sup> sector ATP synthase subunits (Figure 3a,b,d,f–h), except of subunit *g* (Figure 3c). In conclusion, two-week glucose delivery induced expression of the ATP synthase subunits of the membrane F<sup>O</sup> sector and subunit α, as a major subunit of the F<sup>1</sup> moiety.

**Figure 2.** Upregulation of DAPIT in pancreas and pancreatic islets in mice. (**a**) typical experiment, (**b**,**d**) DAPIT mRNA in pancreas (**c**) in isolated pancreatic islets after two weeks of treatment with hyaluronic acid implants of the designated volumes. Data were mutually significantly different (*N* = *n* = 5 mice per group): \*\* *p* < 0.05 (except 50 vs. 200 µL) as well as data vs. control in (**c**); (**d**) dependence of DAPIT transcript in pancreas vs. glucose; resulting from fasting glucose levels in mice, calibrated relatively to volume of hyaluronic acid (see panel (**e**); \*\*\* *p* < 0.05). Spearman coefficient for obtained fit is 0.56; *p* = 0.015.

**Figure 3.** Upregulation of transcripts of other ATP synthase subunits (**a**) FOe, (**b**) FOf, (**c**) FOg, (**d**) A6L/ATP8, (**e**) F1α, and (**g**) ATP6 and (**f**) PGC1α and (**h**) and Mic10 in pancreatic islets in mice—after two weeks of treatment with hyaluronic acid implants of the designated volumes. Data except in panel (**c**) are mutually significantly different (*N* = *n* = 5 mice per group): \*\*\* *p* < 0.001; \*\* *p* < 0.05; \* *p* < 0.1 (with exception of controls vs. 100 µl in panels (**a**,**b**,**e**).

Note that the applied procedure did not induce any glucose intolerance as documented by the time courses of insulin release after glucose injection (Figure 4) and concomitantly performed glucose tolerance tests (Figure 4). With higher implant volumes, mice had rather exhausted fasting insulin levels, possibly reflecting starting hypoinsulinemia. Nevertheless, insulin release time course was identical to sham-operated controls (Figure 4).

**Figure 4.** Time course of glycemia and insulin release —i.e., intraperitoneal glucose tolerance test and insulin time course, performed as described in Ref. [8], after a single glucose dose. Mice were treated for two-weeks with hyaluronic-acid implants with no glucose (black symbols represent sham operation); 100 µL (blue symbols) and 200 µL of implants (red symbols). *N* = 3 mice at each time point. Note for rather exhausted fasting insulin levels (at time zero) for treatment with 200 µL of implants.

#### *3.3. Glucose Independent but DAPIT-Dependent Mitochondrial Biogenesis in INS-1E Cells*

Turning attention back to INS-1E cells, we further investigated possible glucose-induced mitochondrial biogenesis, to answer a question whether any glucose-induced replication of mitochondrial DNA (mtDNA) and/or PGC-1α expression exists. Both, when elevated, reflect the initiation of mitochondrial biogenesis. We found that the mtDNA copy number and PGC-1α (mRNA and protein) were not enhanced until 60 min and 120 min, respectively, lasting from the end of pretreatment (from the two 15-min preincubations with 3-mM glucose) (Figure 5a–g). Elevations of mtDNA copy number became significantly ~1.5 higher after 60 min and 3-fold higher after 120 min with all tested glucose concentrations but zero (Figure 5a). For these experiments, INS-1E cells were transfected with siRNA containing the scrambled sequence (i.e., INS-1EScrl cells were used).

**Figure 5.** Glucose-induced expression of mitochondrial biogenesis. Time course (**a**–**c**) and glucose concentration dependence (**d**) for DAPIT-dependent increase in mitochondrial DNA copy number. Panels (**a,b**) show individual data plus their averages, while (**c**) shows the DAPIT-dependent increase in copy number. Changes in transcript of PGC-1α (**e**) and PGC-1α protein (*N* = 3) (**f**,**h**) and F1α protein (**g**) are shown. ANOVA (*N* = 4): (**a**,**b**) vs. zero glucose: \*\* *p* < 0.05; \*\*\* *p* < 0.001. Full size western blot of panel (**g**) and F1α mRNA see Supplementary File 1, Figure S4. For (**e**), thresholds yielded by RT-PCR concerning the PGC-1α amplicon were in triplicates as follows: 24.6; 24.4 and 24.5 for samples of zero glucose, whereas samples treated with 20-mM glucose yielded 22.76; 23.06 and 22.9.

No increase at 60 min, but a smaller increase at 120 min, was recognized for INS-1E cells silenced for DAPIT (Figure 5b; see Supplementary File 1, Figure S3 for silencing efficiency). As a result, only a portion of increases in mtDNA copy number was DAPIT dependent (Figure 5c); however, even this portion did not depend on glucose. Rather permanent mtDNA replication proceeded, as recognized also when no glucose was added while estimating the time course. Thus, the initial effective mtDNA replication was dependent on DAPIT, i.e., on a complete ATP synthase expression/assembly.

PGC-1α mRNA increased apparently only at 120 min (Figure 5e,), similarly to protein levels (Figure 5f,h). In INS-1E silenced for DAPIT, the levels of PGC-1α protein were nearly equal to those in controls for zero and 3-mM glucose but with 20-mM glucose the PGC-1α protein decreased by 25% relative to zero glucose, which was about 50% relative to INS-1E cells transfected with the scrambled siRNA (Figure 5f,h). Levels of PGC-1α were not affected by the pretreatment with 3-mM glucose (not shown). Furthermore, F1α subunit mRNA and protein did not change (Figure 5g; Supplementary File 1; Figure S4).

#### *3.4. ATP Levels Correlate with DAPIT Expression Induced by Glucose in INS-1E Cells*

Seeking for consequences of DAPIT deficiency, we estimated elevations in the total cellular ATP levels during parallel incubations of INS-1E cells with different glucose concentrations and timing, comparing cells transfected with DAPIT siRNA vs. those transfected with siRNA containing scrambled sequence. Cells were again routinely pretreated by two 15 min pre-incubations with 3-mM glucose. The initial increase was apparent during the 120-min time course for absolute total ATP levels (Figure 6a), despite the constant levels of subunits F1α (Figure 5g). Since mitochondrial biogenesis was beginning at around 60 min, as documented above, the initial increases in the total ATP content must originate exclusively from the elevated ATP synthesis by the existing and probably constant number of ATP synthase molecules. This is due to the enhanced metabolism initiated by the added glucose. Such an increase is retarded after 15 min (Figure 6a). Then biogenesis may contribute to the second phase of the observed increase in the total ATP. These two phases are better recognized when the incremental ATP increases were calculated, i.e., when values of the total ATP content was subtracted for samples when no glucose is added (Supplementary File 1; Figure S5).

**Figure 6.** ATP levels in INS-1E cells. (**a**) Time course of the total ATP levels upon glucose induction in INS-1E cells transfected with scrambled siRNA (color coding as in Figure 1) or *Dapit*/*Usmg5* siRNA as indicated; (**b**) DAPIT-dependent part of the total ATP levels (data for *Dapit* siRNA were subtracted from data of scrambled sequence containing siRNA) after glucose induction; (**c**,**d**) Glucose dependencies constructed from the data of panel (**a**). *p* < 0.001 for 11 and 20-mM glucose in (**c**,**d**). Data were significantly different from zero (3 mM) glucose data as indicated (*N* = 3): \*\* *p* < 0.05; \* *p* < 0.1.

When induced with 11 and 20-mM glucose, at least half of the incremental ATP increase was dependent on expression of DAPIT, since only about less than half and ~30% of the incremental ATP increase was reached at 30 min and 120 min, respectively, upon DAPIT silencing (Figure 6a). Thus, when induced with 3-mM glucose, the rise of total ATP with time was independent of DAPIT since data were equal to those upon DAPIT silencing (Figure 6a).

Estimating the total ATP content which would be related to the DAPIT presence, we subtracted the time courses measured in INS-1E cells transfected with DAPIT siRNA from controls transfected with siRNA having scrambled sequence (Figure 6b). Clearly, no ATP increase related to DAPIT exists without the induction by glucose and when induced only with 3-mM glucose. In contrast, when induced with 11 and 20-mM glucose, the revealed biphasic DAPIT-dependent accumulation was similar as the net ATP surplus but reaching about 50% of ATP elevations and about 50% initial rate (Figure 6b). If we anticipate that the ablation of DAPIT using siRNA was incomplete, one may expect even higher inhibition of the ATP surplus at completed DAPIT ablation. The glucose concentration dependence at 15 min and 120 min (Figure 6c,d) again indicated a substantial increase in ATP. In DAPIT-deficient INS-1E cells, absolute ATP levels were insignificantly different with 3-mM glucose, but with 11 and 20-mM glucose the total ATP went drastically down. All these results show that the DAPIT, and therefore the complete ATP synthase, is required for the maximum synthesis of ATP.

#### *3.5. Phosphorylating Respiration of INS-1E Cells at High Glucose Partially Depends on DAPIT*

To estimate independently whether a direct correlation exists between ATP synthesis and DAPIT levels, we evaluated INS-1E cell respiration by standard bioenergetics tests using an oxygraph after 120-min incubations with different glucose concentrations in the KRH buffer, after two previous washings in KRH only (Figure 7a–d). The ratios between phosphorylating vs. non-phosphorylating rates of respiration (*V*3/*V*4) were derived (Figure 7e). In control INS-1EScrl cells, these ratios physiologically increased from 1.1 to 2.6 and 2.9, when glucose was raised from 3 mM (representing the insulin non-stimulating levels) to 11 mM and 20 mM, respectively (Figure 7e). The latter two concentrations of glucose are stimulating the release of insulin (see below). The *V*3/*V*<sup>4</sup> ratios with 11-mM glucose upon DAPIT silencing were statistically insignificantly different from those in controls (siRNA with scrambled sequence) but at 20 mM, the *V*3/*V*<sup>4</sup> ratio was significantly lower by ~30%. This perfectly matches the DAPIT contribution to the ATP accumulation (Figure 6a,b).

The total capacity of the respiratory chain can be expressed as the maximum respiration rate *V*max, evaluated by titration with an uncoupler FCCP, to reach maximum and saturated respiration. With increasing glucose from 3 to 11 mM, values of *V*max respiration increased 1.5-fold and stayed nearly the same at 20-mM glucose, independently of DAPIT silencing (Figure 7d). We can derive an alternative parameter *R*ATP synthesis, expressing the portion of the total respiratory chain capacity required for ATP synthesis, as (*V*<sup>3</sup> − *V*4)/*V*max. The calculated values of *R*ATP synthesis reached only 5% at 3-mM glucose for INS-1EScrl cells or cells transfected with DAPIT siRNA. Independently of DAPIT silencing, the *R*ATP synthesis ratios increased to about 40% with 11-mM and 20-mM glucose (Figure 7f).

**Figure 7.** Respiration of control and DAPIT-silenced INS-1E cells. Glucose concentration dependencies for 120 min induction (*N* = 4, *n* = 6–8), after which respiration of was estimated in INS-1E cells transfected with scrambled sequence containing siRNA (*black*) or *Dapit*/*Usmg*5 siRNA (*gray*) (data, SDs) such as: (**a**) phosphorylating (basal) respiration rates (*V*<sup>3</sup> ); (**b**) non-phosphorylating respiration rates (*V*<sup>4</sup> ); (**c**) ATP synthase-related respiration rates (*V*<sup>3</sup> − *V*<sup>4</sup> ); (**d**) maximum respiration rates *V*max (derived from titration with an uncoupler FCCP); (**e**) Ratios *V*<sup>3</sup> /*V*<sup>4</sup> , i.e., between phosphorylating vs. non-phosphorylating respiration rates; and (**f**) *R*ATP synthesis, i.e., portion of total respiratory chain capacity required for the ATP synthesis calculated as (*V*<sup>3</sup> − *V*<sup>4</sup> )/*V*max. \*\*\* *p* < 0.001, \*\* *p* < 0.05 for significant differences of DAPIT-silenced vs. cells transfected with scrambled siRNA.

#### *3.6. Insulin Release Is Insignificantly A*ff*ected by the Profoundly Decreased DAPIT Expression in INS-1E Cells*

Next, we estimated time courses of secreted insulin during 120-min incubations of INS-1E cells while using different glucose concentrations added initially (Figure 8a,b). At no glucose addition a small basal rate of insulin release was found (Figure 8a). A similar low linearized rate occurred after the initial addition of 3-mM glucose (Figure 8b). When induced with 11-mM (Figure 8b) and 20-mM glucose (Figure 8a), a nearly linear dependence of the released insulin vs. time up to 120 min reflected the established high rate of insulin secretion at these two stimulating concentrations. Differences between these rates and the basal rate at zero glucose addition represent the net GSIS rate. Surprisingly, the net GSIS rate did not vanish upon DAPIT silencing (Figure 8a,b). When induced with 11 and 20-mM glucose, the net GSIS rate in DAPIT-silenced INS-1E cells accounted for 90% and 80%, respectively, of the mean net GSIS rate for control INS-1EScrl cells. The effect after around 100 min may contain a contribution of mitochondrial biogenesis occurring during the time course of this experiment, since the PGC-1α transcript 1.5-fold increased during the same period as well as mtDNA copy number.

**Figure 8.** Insulin release in INS-1EScrl cells vs. DAPIT-silenced cells. Time course of the accumulated insulin with no addition (**a**) and after the addition of 3-mM (**b**), 11-mM (**b**) and 20-mM glucose (**a**). Data originate from *N* = 3 time courses for each condition. Legend: "Scrl", INS-1EScrl cells; "siDAPIT", DAPIT-silenced INS-1E cells.

#### *3.7. Morphology of Mitochondrial Cristae Does Not Change with DAPIT Deficiency*

Next, we evaluated possible changes in morphology of mitochondrial cristae as consequences of the DAPIT deficiency. We were unable to reach substantial changes in cristae morphology, such as those reported elsewhere [35,36]. The resultant transmission electron microscopy (TEM) images showed that the cristae morphology did not significantly change upon the DAPIT silencing (Figure 9a–g).

However, the previously described [17] cristae narrowing upon transition from a low to high glucose proceeded also in DAPIT-silenced cells, albeit in a lower extent, since cristae width was rather smaller within the estimated ensemble (275, 275 and 450 estimates in 20, 11 and 3-mM glucose, respectively), when compared to INS-1E cells transfected with scrambled siRNA. Further studies are required to investigate what stands behind the literally cell memory regarding glucose levels. Indeed, the in vivo glucose-dependent changes persist during the isolation procedure of mitochondria. The fact that the in vivo situation is reflected by certain persisting structural (e.g., of cristae) and/or conformational pattern suggests that this is due to a specific persisting assembly of the ATP synthase or other cristae-shaping proteins. Note also, that at 3-mM glucose ATP synthase tetramers largely disappeared and hexamers vanished (Figure 10a,b), as reported elsewhere for hypoxic HepG2 cells [18,44].

**Figure 9.** Mitochondrial cristae in DAPIT-deficient INS-1E cells—i.e., silenced for DAPIT, when compared to INS-1E cells transfected with scrambled siRNA (*N* = 2). (**a**–**c**) SDS–polyacrylamide of cristae width; (**d**–**g**) exemplar TEM images of sections of mitochondrial tubules (i.e., "mitochondria"). Independent incubations with the designated glucose concentrations were performed twice, while *n* = 275 crista width estimations were made from 15 TEM images for each condition (*n* = 450 for DAPIT siRNA at 3-mM glucose). Values of most frequent crista width are listed with SDs taken as 0.5 of half-width, as derived from Gaussian fits of the data (displayed on histograms).

**Figure 10.** Rough estimations of DAPIT to F1α stoichiometry in ATP synthase dimers. (**a**) Exemplar **BN-PAGE** shows missing hexamers and majority of tetramers at 3-mM glucose; (**b**) **CN-PAGE** immunostained with anti-F1α or anti-DAPIT antibody as indicated; (**c**) **DAPIT**/**F1**α **stoichiometry** was estimated from *N* = 3 CN-PAGEs separating ATP synthase monomers/oligomers after dual western blotting. \*\*\* *p* < 0.001 (*n* = 3), for significant differences of DAPIT-silenced vs. INS-1EScrl cells.

#### *3.8. Rough Estimation of Stoichiometry for the DAPIT Relatively to the ATP synthase F1*α

Finally, we attempted to derive how the DAPIT to F1α stoichiometry may change with the increasing glucose. Moreover, we verified whether such stoichiometry approaches to zero upon DAPIT silencing. Thus, parallel BN-PAGEs (Figure 10a) and parallel CN-PAGEs (Figure 10b) were run, followed by western blotting, and were immunostained either against F1α, to distinguish monomeric, dimeric and tetrameric ATP synthase; or against DAPIT (Supplementary File 1; Figure S6), to identify roughly its bound amount to either monomeric, dimeric and tetrameric ATP synthase. Acid stripping of PVDF membranes allowed to cross-correlate antigen amounts.

Typical results of semiquantification from CN-PAGE are shown in Figure 10c, estimating only situation in dimers, i.e., anti-DAPIT antibody staining density ratios relatively to the immunostaining of ATP synthase dimers by anti-F1α antibodies. Our attempts to evaluate these ratios relatively to tetramers failed due to a rather highly variable fraction of tetramers occurring in CN-PAGEs. Thus, we repeatedly found that the apparent staining ratio of DAPIT to F1α exclusively among the fraction of dimers was increasing at 20-mM glucose (and slightly at 11 mM) relatively to 3-mM glucose (Figure 10c). The expected theoretical maximum saturated stoichiometry should be 2:6 within a single dimer of the ATP synthase. Different antibody efficiency prevents to get absolute stoichiometry values. Note also, that at 3-mM glucose ATP synthase tetramers largely disappeared and hexamers vanished (Figure 10a,b), as reported elsewhere for hypoxic HepG2 cells [18,44].

#### **4. Discussion**

In this work, we demonstrated that when a fraction predominates of vestigial ATP synthase molecules deficient of USMG5/DAPIT, rat pancreatic β-cells (INS-1E cells) almost do not elevate ATP levels as responding to high glucose. Despite these levels were merely elevated only by 6%, such DAPIT deficiency surprisingly did not inhibit glucose-stimulated insulin secretion (GSIS). The latter result represents a remarkable paradox, specific for pancreatic β-cells. The glucose-stimulated insulin secretion (GSIS) was not substantially hampered in DAPIT-deficient cells. This paradox can be explained on the basis of the recently revisited mechanism for the induction of insulin exocytosis. We demonstrated that H2O2, originating from NADPH oxidase-4 (NOX4), is essentially required together with ATP, as a logical sum for insulin secretion stimulated with glucose [8]. Candidate NADPH oxidases, i.e., unidentified isoforms, were implied in GSIS previously [45–48]. Therefore, at virtually no ATP increase, the existing basal ATP levels are sufficient for GSIS, together with NOX4-ensured redox signaling.

Indeed, the basal total ATP in DAPIT-deficient INS-1E cells is sufficient together with elevated H2O<sup>2</sup> (NOX4 was not hampered) to close the ATP-sensitive K<sup>+</sup> channel and so to trigger insulin granule exocytosis. Apparently, the only negligibly increasing ATP at 11 and 20-mM glucose in DAPIT-deficient INS-1E cells and still existing GSIS provide another independent yet indirect support for the existence of the NOX4-dependent mechanism [8]. Note, that constitutively expressed NOX4 does not need to be assembled [49]; on the contrary the deletion of NOX2, an inducible NOX isoform, enhanced GSIS in pancreatic islets via decrease in superoxide production and elevated cAMP concentrations [48]. Actually, the observation, that despite a very little elevation of ATP in DAPIT-deficient cells, there is still virtually unaffected secretion of insulin responding to glucose, is very similar to the effect of oligomycin described in Ref. [7] (Figure 6D therein).

Our demonstrations that the increased DAPIT expression correlates with the elevated incremental ATP levels is similar to the correlation with elevated ATP previously described in He–La cells [29], which were rather of a weak character. In our case, with a prevailing population of vestigial ATP synthase molecules set by DAPIT silencing [26], we recognized almost no elevation in incremental ATP as a response on higher glucose concentration. This finding supports the requirement of completely assembled ATP synthase for efficient and maximum synthesis of ATP.

It is necessary to further study whether any assembly intermediates of the ATP synthase [50] (or vestigial ATP synthase molecules in deletion experiments, [26]) contribute to the synthesis of ATP at all. Note that our silencing left a substantial amount of DAPIT mRNA expressed. Despite this fact, the surplus accumulation of ATP was decreased by 94%, likewise the extent of increase in phosphorylating/non-phosphorylating respiration ratio, calculated as *V*3/*V*<sup>4</sup> − 1, which decreased by 40%. We also observed similar ATP and respiratory decreases upon silencing of the ATP synthase F<sup>O</sup> sector subunit *e* (Leguina Ruzzi, unpublished).

As we demonstrated in INS-1E cells, expression of the DAPIT and other F<sup>O</sup> subunits determined the intensity of ATP synthesis. The latter is indicated by the *V*3/*V*<sup>4</sup> ratio, increases with the increasing glucose from insulin non-stimulating (3 mM) to insulin stimulating concentration (>8 mM) [11,12,14–16]. However, as we now show, this does not proceed in DAPIT-deficient cells. Hence, a special attention should be paid to the relationships of respiration vs. oxidative phosphorylation in DAPIT-deficient cells. With 20-mM glucose, the *V*3/*V*<sup>4</sup> respiration rate ratio diminished only to ~70% upon DAPIT silencing, whereas the net GSIS rate accounted for 80%. The fraction of respiration employed for the synthesis of ATP was only 5% at 3-mM glucose and was constant at ~40% with 11 and 20-mM glucose. This phenomenon should be independent of the DAPIT-deficiency, since DAPIT does not interact with the respiratory chain supercomplexes.

We also revealed that the two-week delivery of glucose from hyaluronic acid implants elevates transcription of USMG5/DAPIT and other subunits of the membrane F<sup>O</sup> sector of the ATP synthase; and, as exemplified by the subunit F1α, one can expect also elevations of other subunits of the F<sup>1</sup> moiety. Currently, we cannot distinguish whether there was a direct glucose-induction of the selected ATP synthase subunits, whether faster metabolism due to permanently higher glucose presence increased transcription rate; or whether the elevated transcription was evoked by the induced mild hypoinsulinemia in mice. Note that in this way, we simulated chronic states, imposing a rather mild hyperglycemia (7.5 and 9-mM blood glucose) by the constant glucose systemic delivery to mice, using hyaluronic acid implants. One may consider that even if the turnover of DAPIT was faster than the turnover of the other F<sup>O</sup> or F<sup>1</sup> subunits, glucose- or metabolism-induced DAPIT expression would ensure the maintenance of complete ATP synthase. Indeed, ubiquitin degradation exists for certain mitochondrial matrix proteins [51] and this may principally differentiate between subunits. In contrast, if autophagy is involved, all subunits would be degraded simultaneously.

In vivo, postprandial and fasting states are alternating [2,5]. For pancreatic β-cells, this means insulin secreting period and basal period. Various secretagogues coming from metabolized meals induce insulin release by different mechanisms during secreting period, while a rather low insulin secretion exists in the basal state [1–6]. Traditionally emphasized prominent role of glucose stems not only from the complex mechanisms of GSIS, but is also supported by demonstrations that glucose also serves in the maintenance of expression of *Ins* gene and other β-cell specific genes [4,6,39,52]; or may act in identity self-checking of β-cells [52]. The revealed upregulation of subunits of the ATP synthase membrane F<sup>O</sup> sector, including USMG5/DAPIT, and upregulation of the subunit F1α, by two-week glucose delivery is thus only one among numerous other complex processes induced by glucose.

One can also speculate that the glucose/metabolism-induced DAPIT expression may be part of physiological regulations in pancreatic β-cells. This would be enabled, if a fraction of vestigial ATP synthase molecules exists within the ensemble of ATP synthase dimers located in rows along the cristae rims, i.e., the sharp cristae edges. In this case, the newly expressed DAPIT molecules may gradually saturate the vacant sites. These vacant sites may hypothetically arise from a preferable DAPIT degradation. Using DAPIT silencing, we demonstrated that if such degradation existed, it would inhibit ATP synthesis. According this hypothetical view, the preferential rapid expression of DAPIT and its binding into the vacant sites would then switch on ATP synthesis. DAPIT would then act reciprocally to the physiological inhibitor of the ATP synthase, the ATPase inhibitory factor IF1 [11,13].

Speculatively, the DAPIT deficiency (a lower stoichiometry relatively to the FO-sector of the ATP synthase) may exist at low glucose (insulin release nonstimulating), such as given by a higher degradation; and this would cause a substantial ATP synthase inhibition. At high glucose (insulin release stimulating), the resulting glucose/metabolism-induced expression of DAPIT could overcome such loss, restoring ATP synthesis. Alternatively, an excessive DAPIT stoichiometry may be established at high glucose, if such an excessive stoichiometry is required to the optimum ATP synthase operation, hence optimum ATP synthesis at its maximum efficiency. We supported these speculations by our

results with rough estimations of possible stoichiometry of DAPIT relatively to the rest of the ATP synthase in dimers. One should take these results with caution, since the provided anti-DAPIT antibody may partly cross-react with some other ATP synthase subunits (Supplementary File 1; Figure S6) and thus distort these estimations. However, since DAPIT silencing substantially reduced immunostaining with these anti-DAPIT antibodies, certain specificity for DAPIT is evident. Our findings of a possible variable DAPIT-ATP synthase stoichiometry thus require further testing.

The above speculations may be plausible if degradation of DAPIT or its glucose- or metabolic induction would be independent of the degradation or expression for the remaining subunits of the ATP synthase. The existence of the vestigial ATP synthase molecules in vivo which lack the USMG5/DAPIT protein would be possible at least for a small fraction, since DAPIT is one of the outmost subunits (Figure 1) that is also lastly added among the FO-sector subunits of the ATP synthase during biogenesis [26].

A complex assembly of the entire human ATP synthase was described recently to consist of separate steps assembling independently the F1-moiety with the c-ring and complex of subunits *b*, *e*, *g* [50]. Since DAPIT binds the mtDNA-encoded subunit *a* [20,26], synchronous processes of translation on ribosomes proximal to mitochondria, adjacent protein import into the mitochondrial matrix and translation by mt ribosomes participate in final steps of the ATP synthase assembly. One may emphasize here that a parallel mtDNA transcription must take place, concerning the mtDNA-encoded subunits *a*/ATP6 and ATP8/A6 L. When this is impaired, DAPIT cannot bind to the ATP synthase and the resulting vestigial macromolecules should contain even less subunits.

In this respect, we observed that the mtDNA replication is largely independent of glucose- or metabolic induced upregulations. However, the optimum mtDNA replication again requires the completely assembled ATP synthase. This represents a feedback loop. Similar results were obtained for induction of PGC-1α, as one of the factors of mitochondrial biogenesis.

In addition, unlike the published reports for fibroblasts [35,36], we showed the absence of major effects of DAPIT deficiency on morphology of mitochondrial cristae in INS-1E cells. The reason of this discrepancy may stem from more stable mitochondrial network and its rich cristae in INS-1E cells relatively to fibroblasts. That is why we can interpret the above discussed respirometry analyses without a major contribution of cristae morphology changes. As we also demonstrated, DAPIT-deficient vestigial ATP synthase dimers may form thinner cristae. This is supported by our evaluated distribution histograms of cristae width (Figure 9a–c). Nevertheless, unlike in reported observations with mutant DAPIT [35,36], general cristae morphology was not disrupted in DAPIT-deficient INS-1E cells and the previously observed cristae narrowing [17] also took place at high glucose, albeit in a smaller extent. Again, we can explain this by the existing rich cristae in INS-1E cells, when compared to fibroblasts. Hypothetically, DAPIT interactions with the subunit *g* may control both ATP synthesis as well as the inner mitochondrial membrane morphology, particularly at the crista rims. This prediction is based on the recently revealed structure of the ATP synthase tetramer, in which interactions contributing to the tetramer formation involve subunits *g* from the neighbor dimers with the DAPIT in proximity to each of them [20].

In this way, the release of DAPIT from the dimeric/tetrameric ATP synthase would destabilize rows of ATP synthase dimers since it would disconnect the two neighbor dimers within tetramers. In an opposite, binding of DAPIT into yet vacant sites in the dimeric/tetrameric ATP synthase would strengthen rows of ATP synthase dimers/tetramers. Thus, longitudinal interactions within the rim hypothetically govern the extent or strength of the rim stabilization [18]. A stabilized rim (row of the ATP synthase dimers) may lead to cristae narrowing; and, in contrast, the destabilized rim may enable wide cristae [18,35]. In addition, prolonged INS-1E cell exposure to 3-mM glucose may fragment mitochondrial network [53].

#### **5. Conclusions**

In conclusion, our results with the DAPIT-deficient rat β-cell line (INS-1E cells) indicate that the vestigial ATP synthase lacking DAPIT exhibits much lower efficiency of ATP synthesis. However, despite the fact that the glucose-induced ATP increment was only around 6% in DAPIT-deficient INS-1E cells, this was still sufficient to initiate GSIS by the mechanism(s) independent of ATP elevation, i.e., independent of the oxidative phosphorylation. The result is compatible with the existence of the recently revealed NOX4-dependent redox signaling essential for GSIS [8]. In general, we demonstrated that the optimum expression of DAPIT is elementary required to reach the maximum ATP synthesis. The elevated expression of DAPIT at high glucose further increases the ATP synthesis efficiency. Moreover, we described rather a compensating response of pancreatic islets cells, i.e., including β-cells, lying in the elevated transcription of the ATP synthase F<sup>O</sup> subunits *e*, *f*, *g* and DAPIT, as induced by the enhanced metabolism upon higher glucose intake.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2218-273X/10/7/1026/s1, Figure S1: Typical changes in DAPIT mRNA levels in INS-1E cells; Figure S2: Downregulation/recovery of transcripts for F<sup>O</sup> subunits *e*, *f* and *g*; Figure S3: Typical results of DAPIT silencing; Figure S4: western blots of F1α; Figure S5: Surplus ATP accumulation; Figure S6: Anti-DAPIT/USMG5 antibodies recognize a complex above ~75 kDa.

**Author Contributions:** Conceptualization, A.L.-R. and P.J.; methodology A.L.-R., A.V., B.H., H.E., V.P., J.T. and A.D.; validation, A.L.-R., B.H., H.E., V.P., J.T.; formal analysis, A.L.-R. and P.J.; investigation, A.L.-R., A.V., B.H., H.E.,V.P., J.T. and A.D.; resources, P.J.; data curation, P.J.; writing—original draft preparation, P.J.; writing—review and editing, P.J.; visualization, P.J.; supervision, P.J.; project administration, P.J.; funding acquisition, P.J. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by the Grant Agency of the Czech Republic (Grantová Agentura Ceske ˇ é Republiky, GACR), Grant Number 20–00408S. ˇ

**Acknowledgments:** We gratefully acknowledge receiving of a Juvederm Ultra 4 hyaluronic acid filler for performing hyaluronic acid implants from Semira Kwabi (Skin Clinic, London, UK) and the excellent technical assistance of Jana Vaicová with insulin release assays and Jitka Smiková with cell culture media preparation and cell cultivation; and Katarína Smolková and Lydie Plecitá-Hlavatá for their advice and fruitful discussion.

**Conflicts of Interest:** The authors declare no conflicts of interest. The funders had no role in the design of the study; in the collection, analyses or interpretation of data; in the writing of the manuscript or in the decision to publish the results.

#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

*Article*

#### **Transport of Ca <sup>2</sup>**<sup>+</sup> **and Ca <sup>2</sup>**+**-Dependent Permeability Transition in the Liver and Heart Mitochondria of Rats with Di**ff**erent Tolerance to Acute Hypoxia**

**Konstantin N. Belosludtsev 1,2, \* , Mikhail V. Dubinin 2 , Eugeny Yu. Talanov 1 , Vlada S. Starinets 2 , Kirill S. Tenkov 2 , Nadezhda M. Zakharova <sup>3</sup> and Natalia V. Belosludtseva 1**


Received: 4 December 2019; Accepted: 7 January 2020; Published: 9 January 2020

**Abstract:** The work examines the kinetic parameters of Ca <sup>2</sup><sup>+</sup> uptake via the mitochondrial calcium uniporter complex (MCUC) and the opening of the Ca <sup>2</sup>+-dependent permeability transition pore (MPT pore) in the liver and heart mitochondria of rats with high resistance (HR) and low resistance (LR) to acute hypoxia. We found that the rate of Ca <sup>2</sup><sup>+</sup> uptake by mitochondria of the liver and heart in HR rats is higher than that in LR rats, which is associated with a higher level of the channel-forming subunit MCU in liver mitochondria of HR rats and a lower content of the dominant-negative channel subunit MCUb in heart mitochondria of HR rats. It was shown that the liver mitochondria of HR rats are more resistant to the induction of the MPT pore than those of LR rats (the calcium retention capacity of liver mitochondria of HR rats was found to be 1.3 times greater than that of LR rats). These data correlate with the fact that the level of F0F1-ATP synthase, a possible structural element of the MPT pore, in the liver mitochondria of HR rats is lower than in LR rats. In heart mitochondria of rats of the two phenotypes, no statistically significant difference in the formation of the MPT pore was revealed. The paper discusses how changes in the expression of the MCUC subunits and the putative components of the MPT pore can affect Ca <sup>2</sup><sup>+</sup> homeostasis of mitochondria in animals with originally different tolerance to hypoxia and in hypoxia-induced tissue injury.

**Keywords:** hypoxia; resistance to hypoxia; mitochondria; mitochondrial calcium transport; mitochondrial calcium uniporter complex; mitochondrial Ca <sup>2</sup>+-induced permeability transition pore; cyclophilin D; ATP synthase

#### **1. Introduction**

Hypoxia/ischemia is a widespread phenomenon that occurs both in conditions of oxygen deficiency in the environment and in various pathologies as a result of a decrease in oxygen delivery to the cell to a level insufficient to maintain its functions and structure. Hypoxic conditions in the body are observed in ischemic and reperfusion injuries of organs, hemodynamic disorders, coronary insufficiency, blood loss, hemorrhagic shock, arterial hypertension, pulmonary insufficiency, systemic inflammatory response syndrome, traumatic shock, exposure to adverse environmental factors, extreme conditions, and other stressful effects.

The main intracellular targets of hypoxia of various etiologies are mitochondria and aerobic energy metabolism [1,2]. Hypoxia alters mitochondrial dynamics and morphology and provokes mitochondrial dysfunction, which can lead to a drop in the ATP synthesis, an increase in the production of reactive oxygen species, and other adverse events, including the disturbance of ionic, mainly Ca2+, homeostasis. Excessive accumulation of Ca2<sup>+</sup> in cytoplasm and mitochondria in hypoxia and subsequent reoxygenation can finally initiate cell injury and death via the opening of the Ca2+-dependent mitochondrial permeability transition pore (MPT pore) and the subsequent release of proapoptotic proteins from organelles [3].

As a major large-capacity buffer system of Ca2<sup>+</sup> ions, mitochondria are involved in the regulation of intracellular Ca2<sup>+</sup> homeostasis. Changes in mitochondrial [Ca2+] modulate key cellular processes, ranging from aerobic metabolism (through Ca2+-sensitive dehydrogenases, and enzymes of the Krebs cycle [4]) to the release of proapoptotic factors [5–8], as well as local modulation of the activity of channels and enzymes [9,10].

In recent studies, it has been found that mitochondria contain several proteins involved in mitochondrial Ca2<sup>+</sup> uptake: mitochondrial Ca2<sup>+</sup> uniporter complex (MCUC), RaM (the rapid mode of Ca2<sup>+</sup> uptake), Letm1 (leucine zipper and EF-hand-containing transmembrane protein 1), mitochondrial ryanodine receptor type 1, and uncoupling proteins [5,11]. Among them, the MCUC is believed to be the main calcium transport system, which mediates the electrophoretic influx of the ion into mitochondria. This multiprotein complex is composed of the pore-forming and calcium-conducting subunit MCU (mitochondrial calcium uniporter subunit), its paralogue MCUb (mitochondrial calcium uniporter dominant negative beta subunit), and regulatory subunits: the small membrane-spanning protein EMRE (essential MCU regulatory subunit), the peripheral membrane proteins acting as gatekeepers MICU1 and MICU2 (mitochondrial calcium uptake proteins 1 and 2), and MCUR1 (MCU regulator protein 1) [5,6]. Variations in the ratio of the subunits of MCUC determine the dynamics of mitochondrial Ca2<sup>+</sup> uptake in different tissues and pathophysiological states. For example, a decrease in the expression of MICU1 in the heart tissue correlates with the lowered activation threshold of Ca2<sup>+</sup> influx and the decreased calcium capacity of heart mitochondria in comparison with those of liver mitochondria [12]. MCUb mRNA is highly expressed in the heart and minimally expressed in skeletal muscle, which can be the cause of variations in the tissue-dependent mitochondrial Ca2<sup>+</sup> uptake: the Ca2<sup>+</sup> influx into skeletal muscle mitochondria is significantly greater than that into cardiac ones [13].

Excessive accumulation of Ca2<sup>+</sup> in mitochondria leads to the opening of the Ca2+-dependent mitochondrial permeability transition pore (MPT pore), which can be a key step in the mechanism of the activation of programmed cell death or, under certain conditions, can serve as an additional nonspecific calcium release pathway. The molecular structure of the MPT pore is as yet not clearly established. The MPT pore complex is considered to be a multiprotein mega-channel, which includes proteins of the inner and outer mitochondrial membranes. Among possible candidates for the role of the channel-forming subunit of the MPT pore are considered a number of proteins of the inner mitochondrial membrane, including ATP synthase, adenine nucleotide translocator, and phosphate carrier. At the same time, the only protein that is currently claimed to be an integral part of the MPT pore is the regulatory protein cyclophilin D, which is able to interact with the above-mentioned proteins. Cyclophilin D is the target of the cyclic undecapeptide cyclosporin A (CsA), which desensitizes the MPT pore complex to calcium ions at nanomolar concentrations [5,14].

Studies on the role of mitochondrial Ca2<sup>+</sup> transport systems in the pathogenesis of hypoxia have been ongoing for a long time. It was found that the pharmacological or genetic modulation of cyclophilin D leads to the inhibition of pore opening in mitochondria and increases the resistance of cell cultures and tissues to the hypoxic state. At the same time, the question of how the structure and operation of the mitochondrial Ca2<sup>+</sup> transport system and the MPT pore can be regulated in hypoxia and hypoxic adaptation remains unclear. The answer to this question should be found when conducting studies on animals with different individual resistance to acute hypoxia [15–17]. It is known that the onset and severity of the acute hypoxia response of rats drastically differ within the population, and two opposite, extreme phenotypes of animals, with high resistance (HR) and low resistance (LR) to hypoxia, have significantly different "functional-metabolic profiles" [18]. Thus, we have shown earlier that the liver mitochondria of HR and LR rats have different resistance to the formation of Ca2+-dependent mitochondrial pores of two types, the MPT pore and the lipid pore induced by saturated long-chain fatty acids [19].

In the present work, we attempted to determine the structural and functional features of the mitochondrial Ca2<sup>+</sup> transport multiprotein systems, the MCUC and the MPT pore, in liver and heart tissues of animals with different tolerance to oxygen shortage. The study demonstrates that: (1) the rate of accumulation of Ca2<sup>+</sup> ions by mitochondria isolated from the liver and heart of HR rats is considerably higher than that of LR rats, which highly correlates changes observed in the content, the ratio, and mRNA level of the subunits of the MCUC. In HR rats, the levels of the pore-forming uniporter subunit, MCU, and the key regulatory uniporter subunit, MICU1, in liver mitochondria are significantly higher than in LR rats. The level of the dominant-negative channel subunit, MCUb, is lower in heart mitochondria of HR rats compared to those of LR rats; (2) the liver mitochondria of HR animals are more resistant to the induction of the MPT pore than those of LR animals. This difference can be associated with the lower level of ATP synthase, a protein considered to be the channel-forming component of the MPT pore, in liver tissue of HR animals. In heart mitochondria of rats of the two phenotypes, no difference in the formation of MPT pore was revealed.

#### **2. Materials and Methods**

#### *2.1. Testing the Tolerance of Rats to Extreme Hypoxia Conditions*

Experiments were conducted on Wistar male rats (220–250 g) with different baseline resistance to hypoxia, low-resistance (LR) and high-resistance (HR) rats. The resistance of rats was tested one month prior to the experiments using a special procedure [19]. Animals were placed into a hypoxic chamber and then tested for the ability to endure a critical life-incompatible O<sup>2</sup> concentration. For this test, the concentration of oxygen in the chamber was reduced to 3.1% by its displacement with nitrogen gas, and the time of the onset of pathological breathing (apnea) (Ta) was determined. This parameter characterizes the viability of animals under extreme hypoxic conditions and reflects the ability to fully mobilize the nonspecific protective functions responsible for the survival of the organism in the sublethal period. After registering Ta, the hypoxic chamber was opened, and the normal posture and locomotor activity of the animals restored within 5–10 min. The T<sup>a</sup> value was 1–2 min for LR rats and more than 10 min for HR rats. Each group of rats amounted to approximately 20% of the total number of animals tested (n = 30). An interval of one month after the procedure of testing the tolerance of rats to hypoxia conditions was chosen on the basis of earlier studies to eliminate the effect of hypoxic exposure and to identify the basal differences between two rat phenotypes [15,16,18]. After this interval, two extreme types of animals with different tolerance to acute oxygen deficiency exhibit characteristic features of the ultrastructure and functional activity of mitochondria of different organs [16,19].

The laboratory animals were treated in accordance with the European Convention for the Protection of Vertebrates used for experimental and other purposes (Strasbourg, 1986) and the principles of the Helsinki Declaration (2000). All the protocols were approved by the Institute of Theoretical and Experimental Biophysics RAS Ethics Committee (Order No. 173/k of 03.10.2011, Protocol No. 03/2019 of 05.03.2019).

#### *2.2. Isolation of Rat Liver and Heart Mitochondria*

Mitochondria were isolated from the liver and heart of Wistar rats by differential centrifugation as described earlier [19]. The homogenization buffer contained 210 mM mannitol, 70 mM sucrose, 1 mM EDTA, and 10 mM Hepes/KOH buffer, pH 7.4. Subsequent centrifugations were performed in the same buffer, except that, instead of EDTA, 100 µM EGTA was used. Final suspensions contained 70–80 mg of mitochondrial protein/mL and 25–35 mg of mitochondrial protein/mL (liver and heart mitochondria, respectively), as determined by the Lowry method [20].

#### *2.3. Ca2*<sup>+</sup> *Uptake by Mitochondria*

The Ca2<sup>+</sup> concentration in the incubation medium was monitored spectrophotometrically with an arsenazo III indicator at 675–685 nm using a plate reader Tecan Spark 10M (Tecan, Männedorf, Switzerland) at 25 ◦C under constant stirring. At a neutral pH, Ca2<sup>+</sup> forms a complex with arsenazo III, which has a blue color. The color intensity is proportional to the concentration of Ca2<sup>+</sup> in a buffer and can be measured spectrophotometrically. Mitochondria (0.4–0.5 mg of mitochondrial protein/mL) were suspended in an incubation medium containing 210 mM mannitol, 70 mM sucrose, 1 mM KH2PO4, 50 µM arsenazo III, 10 µM EGTA, and 10 mM HEPES-KOH (pH 7.4) and energized with 2.5 mM glutamate + 2.5 mM malate. After the addition of 50 µM CaCl2, the rate of Ca2<sup>+</sup> uptake by mitochondria (nmol Ca2<sup>+</sup> × min−<sup>1</sup> × mg−<sup>1</sup> of mitochondrial protein) was determined in the presence of 1 µM CsA. To determine the ability of mitochondria to retain Ca2+, 10 µM CaCl<sup>2</sup> was added into the reaction medium successively, with an interval of ~90 s. After several additions, external [Ca2+] increased, indicating a massive release of the ion from the organelles due to the opening of the MPT pore. The amount of Ca2<sup>+</sup> released upon permeability transition (defined as Ca2<sup>+</sup> retention capacity) was used as a measure of the MPT pore opening probability.

#### *2.4. Electrophoresis and Immunoblotting of Mitochondrial Proteins*

To prepare samples for quantifying the levels of mitochondrial proteins, aliquots of native mitochondria (2 mg/mL) were solubilized in Laemmli buffer in Eppendorf tubes and heated for 3 min at 95 ◦C. Sample aliquots normalized by the protein concentration (10 µg of mitochondrial protein) were applied to the lanes and subjected to electrophoresis followed by Western blot analysis. Mitochondrial samples were separated by 12.5% SDS-PAGE and transferred to a 0.45 µm nitrocellulose membrane (Amersham, Munich, Germany). The proteins of PageRuler Prestained Protein Ladder (Thermo Scientific, Waltham, MA USA) were used as markers. After overnight blocking, the membrane was incubated with the appropriate primary antibody. The monoclonal rabbit anti-MCU (#14997), anti-CBARA/MICU1 (#12524), and anti-ANT2/SLC25A5 (#14671) antibodies were from Cell Signaling Technology Inc. (Danvers, MA, USA). The total OXPHOS Rodent WB Antibody Cocktail (#ab110413) containing the α-subunit of complex V (CV-ATP5A-55 kDa), anti-VDAC1 (#ab154856), and the polyclonal rabbit antibodies Anti-CCDC109B (#ab170715), anti-cyclophilin F (CypD) (#ab64935), and anti-ANT1 (#ab102032) were from Abcam (Cambridge, United Kingdom). The immunoreactivity was detected using the appropriate secondary antibody conjugated to horseradish peroxidase (#7074, Cell Signaling Technology Inc., Danvers, MA, USA). Peroxidase activity was detected with ECL chemiluminescence reagents (Pierce, Rockford, IL, USA). The relative levels of the detected proteins were visualized using an LI-COR system (LI-COR, Lincoln, NE, USA). Optical density measurements were performed using the LI-COR Image Studio software.

#### *2.5. RNA Extraction, Reverse Transcription, and Quantitative Real-Time PCR*

Total RNA was isolated from 100 mg of deep-frozen tissue samples (the liver or the heart) using an ExtractRNA kit (Eurogen, Moscow, Russia) in accordance with the protocol of the manufacturer. The resulting RNA preparation was treated with RNase-free DNase I (Thermo Scientific, Waltham, MA USA). The concentration of total RNA was measured spectrophotometrically using a Nanodrop ND-1000 spectrophotometer (ND Technologies, Fremont, CA, USA). Two micrograms of total RNA was taken for cDNA synthesis; reverse transcription was performed using an oligo(dT)15 primer and MMLV reverse transcriptase (Eurogen, Moscow, Russia) according to the manufacturer's instructions. Real-time PCR was performed using a DTLite5 amplifier (DNA-Technology LLC, Moscow, Russia) using the qPCRmix-HS SYBR reaction mixture (Eurogen, Moscow, Russia). The selection and analysis of gene-specific primers were performed using Primer-BLAST [21] (the oligonucleotide sequences are

presented below; see Table 1). The relative level of the expression of each gene was normalized by the level of mRNA of the cytoskeletal protein beta-actin (Actb), and a comparative C<sup>T</sup> method was used to quantify the results [22].


**Table 1.** List of gene-specific primers for RT-PCR analysis.

#### *2.6. Statistical Analysis*

The data were analyzed using the GraphPad Prism 7 and Excel software and presented as mean (median) ± SEM of 3–5 experiments. Statistical differences between the means were determined by Mann–Whitney U test; *p* < 0.05 was considered to be statistically significant.

#### **3. Results**

#### *3.1. Mitochondrial Ca2*<sup>+</sup> *Uptake and the Features of Subunit Composition of the Mitochondrial Ca2*<sup>+</sup> *Uniporter Complex in the Liver and Heart of Rats with Di*ff*erent Tolerance to Acute Hypoxia*

In this work, we first analyzed the functional and structural features of the system of mitochondrial Ca2<sup>+</sup> transport in the liver and the heart of animals with different tolerance to oxygen shortage, LR and HR rats. Figure 1A shows the kinetics of uptake of Ca2<sup>+</sup> (50 µM) by the liver mitochondria of LR (dotted line) and HR rats (solid line) in the presence of CsA, which was necessary to block the possible opening of the MPT pore. One can see that the rate of Ca2<sup>+</sup> influx into the liver mitochondria of HR rats is 1.3 times higher compared with that of LR rats (Figure 1B). The heart mitochondria of HR rats were found to accumulate Ca2<sup>+</sup> ions also significantly faster than those of LR rats, although the difference is less pronounced than for liver mitochondria (about 10–15%) (Figure 1C).

Mitochondrial Ca2<sup>+</sup> uptake is mediated by an electrogenic uniport, referred to as "Ca2<sup>+</sup> uniporter", a complex of proteins of the inner mitochondrial membrane, including the pore-forming subunit MCU and its dominant-negative form MCUb, and the regulatory subunits MICU1, MICU2, EMRE, and MCUR1. It is considered that the contents of MCU, MCUb, and MICU1 and their stoichiometry can predominantly regulate the mitochondrial Ca2<sup>+</sup> transport in accordance with physiological needs [5,6,23]. Since the mitochondria of HR animals accumulate Ca2<sup>+</sup> faster as compared to the organelles of LR rats, one can assume that the adaptation to hypoxic stress is associated with changes in the relative content of these subunits in the mitochondrial membrane. Therefore, we quantified the uniporter protein constituents and their mRNA level in the liver and cardiac muscle of rats depending on the baseline resistance of animals to hypoxia.

μ μ μ **Figure 1.** Ca2<sup>+</sup> ion uptake by mitochondria of the liver and heart of hypoxia low-resistance (LR) and high-resistance (HR) rats. (**A**) The changes in the external concentration of Ca2<sup>+</sup> ions in the incubation medium during their accumulation by the liver mitochondria of HR (the solid line) and LR (the dotted line) rats. The incubation medium contained 210 mM mannitol, 70 mM sucrose, 2.5 mM malate, 2.5 mM glutamate, 1 mM KH2PO<sup>4</sup> , 10 µM EGTA, 1 µM cyclosporin A, and 10 mM Hepes/KOH buffer (pH 7.4). Additions: rat liver mitochondria (0.4 mg/mL), 50 µM CaCl<sup>2</sup> . The typical traces of five independent experiments are presented. (**B**) The rates of Ca2<sup>+</sup> uptake by liver mitochondria (shaded columns) of HR and LR rats. (**C**) The rates of Ca2<sup>+</sup> uptake by heart mitochondria of HR and LR rats. Values are given as means ± SEM (n = 5). \* The difference between HR and LR animals is statistically significant (*p* < 0.05). μ μ μ

The immunoblotting of the members of the MCUC protein family of the mitochondria isolated from the liver of rats of two phenotypes reveals that in HR rats, the levels of MCU and MICU1 were higher than in LR rats by 1.3 and 1.35 times, respectively (Figure 2A,B), whereas no significant difference in the level of the dominant-negative uniporter subunit MCUb was observed. As for heart tissue, a comparative analysis of the protein level of the components of the MCUC showed that there was a significant decrease in the content of MCUb and a slight tendency to an increase in the content of MCU in the cardiac mitochondria of HR rats relative to those of LR animals (Figure 2C,D). VDAC1 participates in the formation of MAM (mitochondria-associated membranes) contacts, which are the main pathway of Ca2<sup>+</sup> transfer from the endoplasmic reticulum to mitochondria [5,14]. One can see that the amount of VDAC1 in liver mitochondria did not differ. At the same time, the level of VDAC1 in heart mitochondria of HR rats was higher than that of LR rats.

− − − The results of the real-time PCR analysis confirm the data obtained and also indicate that the changes occur at the level of transcription. The mRNA content of MCU in the liver of HR rats is significantly increased in comparison with that of LR rats. At the same time, the expression profile of MICU1 and MCUb in the liver of rats of two phenotypes does not differ (Figure 3A). So, the MCU/MCUb expression ratio grew from 4.6 in the liver of LR animals to 7.0 in the liver of HR rats (2−∆Ct *MCU* = 0.014 ± 0.001 and 0.025 ± 0.001 in the liver tissue of LR and HR rats, respectively; 2 <sup>−</sup>∆Ct *MCUb* = 0. 003 ± 0.0005 and 0.0035 ± 0.00038 in LR and HR rats, respectively). Figure 3B demonstrates that the expression level of MCUb is significantly lower in the heart tissue of HR animals in comparison with that of LR animals. The MCU/MCUb expression ratio grew from 3.5 in the heart of LR animals to 8.8 in the heart of HR rats (2−∆Ct *MCU* = 0.024 ± 0.005 and 0.034 ± 0.008 in the heart tissue of LR and HR rats, respectively; 2−∆Ct *MCUb* = 0.007 ± 0.0007 and 0.0038 ± 0.0005 in LR and HR rats, respectively). − − − −

−

**Figure 2.** Levels of the subunits of MCUC and VDAC1 in liver (**A**,**B**) and heart (**C**,**D**) mitochondria of LR and HR rats. Western blot analysis of the members of the MCU protein family (MCU, MCUb, and MICU1) and VDAC1 in the liver (**A**) and heart (**C**) mitochondria of LR and HR rats. Summarized data of densitometric band analysis of these proteins in the liver (**B**) and heart (**D**) mitochondria of LR and HR rats. Values are given as means ± SEM (n = 3). \* The difference between HR and LR animals is statistically significant (*p* < 0.05).

**Figure 3.** The mRNA levels of the subunits of MCUC in liver (**A**) and heart (**B**) of HR rats normalized to those of LR rats. The median and extreme values are presented (n = 8). The horizontal line is the mean value of the mRNA levels in LR rats. \* The difference between HR and LR animals is statistically significant (*p* < 0.05).

On the basis of the results obtained, one can conclude that an increase in the relative amount of the pore-forming subunit of MCUC, MCU, in liver mitochondria and a decrease in the dominant-negative subunit, MCUb, in heart mitochondria underlie the increased rate of mitochondrial Ca2<sup>+</sup> uptake in HR rats relative to that in LR rats.

#### *3.2. A Comparison of the Resistance of Mitochondria to the Opening of MPT Pore and of the Levels of Its Probable Molecular Components in the Liver and Heart of HR and LR Rats*

The pathophysiological phenomenon of the Ca2+-induced MTP pore opening is well known to be central to mitochondrial vital functions and can play a lethal role in many pathophysiological conditions including hypoxia cell injury. So, the next objective of our work was to examine whether the resistance of mitochondria to the opening of the MPT pore changes depending on the baseline tolerance of an animal to hypoxic condition. The mitochondrial Ca2<sup>+</sup> retention capacity (CRC) is related to the threshold concentration of Ca2<sup>+</sup> necessary for the pore to open. One of the ways to assess this parameter is to introduce Ca2<sup>+</sup> into the suspension of mitochondria in small successive doses, and the number of such additions until the MPT pore is triggered will reflect the CRC of the organelles. Figure 4A shows the results of such experiments. It can be seen that the number of successive Ca2<sup>+</sup> additions (and, therefore, the threshold pore-opening Ca2<sup>+</sup> concentration) in the case of the liver mitochondria of HR rats was greater compared to the organelles of the liver of LR rats. The parameter of CRC of HR rat liver mitochondria increased 1.3-fold (Figure 4B). This implies that their resistance to the opening of the MPT pore would also be higher.

By now, the structure of the MPT pore, which is believed to be a mega-channel penetrating both the inner and outer mitochondrial membranes, has yet to be determined. Three proteins of the inner mitochondrial membrane, namely, adenylate translocator (ANT), ATP synthase, and cyclophilin D, are considered to be essential components of the pore complex [14,24]. To elucidate a possible molecular mechanism of the resistance of HR rat liver mitochondria to the induction of MPT pore opening, we have compared the levels of these proteins in the organelles of the liver of rats of two phenotypes.

Figure 4C,D show immunoblots of cyclophilin D, ANT1, and α-subunit of ATP synthase of mitochondria from the liver of HR and LR rats. One can see that the amount of cyclophilin D and ANT1 in the mitochondria did not differ. The levels of mitochondrial ATP synthase in the liver of HR rats, on the other hand, were reduced, with the reduction being statistically significant. The results of the real-time PCR analysis confirm the data obtained and also indicate that the changes occur at the level of transcription. So, the mRNA content of *Atp5f1a* in the liver of HR rats is significantly decreased in comparison with that of LR rats. At the same time, the expression profile of ANT1 and cyclophilin D in the liver of rats of two phenotypes does not differ (Figure 5A). We suppose that the decrease in the content of ATP synthase can be the cause of the increased resistance of liver mitochondria of HR rats to the MPT pore opening. α

μ **Figure 4.** The parameters of formation and levels of putative protein components of the Ca2+-induced MPT pore in the liver mitochondria of HR and LR rats. (**A**) Changes in the external concentration of Ca2<sup>+</sup> in a suspension of liver mitochondria of HR (the solid line) and LR (the dotted line) rats after sequential additions of 10 µM Ca2+. The incubation medium was the same as in Figure 1, with the exception of cyclosporin A. Additions: rat liver mitochondria (0.4 mg/mL). The figure shows traces of a typical experiment conducted at the same time on the same mitochondrial preparation. Similar results were obtained in five independent experiments. (**B**) Calcium retention capacity of liver mitochondria of LR and HR rats. The values are given as means ± SEM (n = 5). (**C**) Western blot analysis of MPT-related proteins: CypD, ANT1, ATP5A, and VDAC1 in liver mitochondria of LR and HR rats. (**D**) Summarized data on the relative contents of the MPT-related proteins. Values are given as means ± SEM (n = 3). \* The difference between HR and LR animals is statistically significant (*p* < 0.05).

Figure 6 shows the data of the estimation of maximal calcium capacity and Western blot analysis of the MPT-related proteins in mitochondria from the heart of HR and LR animals. In contrast to mitochondria from the liver tissue, cardiac mitochondria from rats of both groups display no statistically significant difference in the CRC and the content of ATP synthase and ANT1. At the same time, the levels of the regulatory protein, cyclophilin D, and VDAC1, which is assumed to form the channel of the MPT pore in the outer mitochondrial membrane, in heart mitochondria of HR rats were higher than those of LR rats. The results of real-time PCR analysis are consistent with the data of immunoblotting (Figure 5B).

**Figure 5.** The mRNA levels of the MPT in liver (**A**) and heart (**B**) of HR rats normalized to those of LR rats. The median and extreme values are presented (n = 8). The horizontal line is the mean value of the mRNA levels in LR rats. \* The difference between HR and LR animals is statistically significant (*p* < 0.05).

μ μ μ μ **Figure 6.** Induction of the Ca2+-induced MPT pore in the heart mitochondria of HR and LR rats. (**A**) Changes in the external concentration of Ca2<sup>+</sup> in a suspension of heart mitochondria of HR (the solid line) and LR (the dotted line) rats after sequential additions of 10 µM Ca2+. The incubation medium was the same as in Figure 1, with the exception of cyclosporin A. Additions: heart mitochondria (0.4 mg/mL); 10 µM CaCl<sup>2</sup> . The figure shows traces of a typical experiment conducted at the same time on the same mitochondrial preparation. Similar results were obtained in five independent experiments. (**B**) Calcium retention capacity of heart mitochondria of LR and HR rats. The values are given as means ± SEM (n = 5). The medium composition and experimental conditions were as indicated in Figure 4A. (**C**) Western blot analysis of MPT-related proteins: CypD, ANT1, ATP5A, and VDAC1, in heart mitochondria of LR and HR rats. (**D**) Summarized data on the relative contents of the MPT-related proteins. Values are given as means ± SEM (n = 3). \* The difference between HR and LR animals is statistically significant (*p* < 0.05).

#### **4. Discussion**

It is generally accepted that mitochondria are one of the main intracellular targets of hypoxia. Structural, biochemical, and functional abnormalities of mitochondria are widely believed to be important pathogenetic factors that underlie hypoxic or ischemic cell injury [1,18]. Apart from disordering mitochondrial ATP synthesis, network dynamics, and redox state, hypoxia also dysregulates the Ca2<sup>+</sup> homeostasis of these organelles and the cell as a whole. The hypoxia-induced dramatic alterations in Ca2<sup>+</sup> transport in mitochondria may not only be related to the decreased oxidative phosphorylation (OXPHOS), a metabolic shift toward glycolysis, and the development of oxidative

stress, but also induce cell death pathways via mitochondrial Ca2<sup>+</sup> overload and the formation of Ca2+-dependent MPT pore in the inner mitochondrial membrane, leading to the release of proapoptotic proteins into the cytosol [3]. Consequently, it has been proposed that mitochondrial Ca2<sup>+</sup> transport pathways might be targets for protective intervention, as well as involved in the formation of the molecular mechanism of cell resistance to hypoxia and ischemia. The purpose of our work was to study in more detail the features of structure and operation of the Ca2<sup>+</sup> transport systems in the mitochondria of the liver and heart of rats with originally different tolerance to acute hypoxia.

The results obtained in this work indicate that mitochondria of HR and LR rats are characterized by fundamentally different genetically programmed rearrangements of the systems regulating Ca2<sup>+</sup> homeostasis. In the case of HR rats, the rearrangements of the Ca2<sup>+</sup> uniporter can increase the efficiency of Ca2<sup>+</sup> accumulation by liver and heart mitochondria. At the same time, the organelles of the liver of HR rats become more resistant to the opening of the Ca2+-dependent MPT pore in comparison with those of LR rats. Along with the well-characterized distinction in the intensity of OXPHOS in mitochondria of the vital organs of LR and HR rats [18,25], these effects might contribute to the development of individual systemic resistance of animals to acute hypoxia.

It is well known that individuals of an animal population differ in their tolerance to oxygen deficiency [15–17,25]. As shown earlier, animals that belong to two opposite, in regard to hypoxia tolerance, types (LR and HR) have essentially different "functional-metabolic" profiles, including the effectiveness of energy support of the organism, the regulation of central nervous cardiovascular systems, neurohumoral regulation, stress-activating and stress-limiting systems, the oxygen-transporting function of blood, and the state of membranes and receptors. LR animals are considered to have a weak type of the nervous system, increased excitability, and emotional reactivity. LR animals respond to hypoxia with agitation and high locomotor activity. In contrast, HR animals have reduced excitability and anxiety, milder aggressiveness, more pronounced internal inhibition, low sensitivity to any provocative factors, and a tendency to social domination [18].

It was previously shown that mitochondria of the vital organs of LR and HR rats differ in both structural and basic functional parameters. Thus, the mitochondria of the cerebral cortex of HR rats are characterized by a denser packing of cristae and a more electron-dense matrix, smaller sizes, and higher concentrations of the respiratory chain complexes and the respiration substrate, succinate (i.e., they are functionally more active compared to mitochondria of LR rats) [16,25]. In contrast, in brain mitochondria of LR rats, the number of mitochondrial cristae is decreased, and this is consistent with the lower content of their respiratory electron carriers compared to mitochondria of HR rats. It was also found that in liver mitochondria of HR animals, the rate of the ATP-dependent potassium transport, which reflects the activity of the mitochondrial ATP-sensitive potassium channel, was increased [17].

Here, we have demonstrated that, in addition to the above structural–functional features, the mitochondria of the liver and heart of rats with originally different tolerance to hypoxia are distinguished in respect to the rate of influx of Ca2<sup>+</sup> ions via MCUC. The liver and heart mitochondria of HR rats are characterized by a significant increase in the efficiency of Ca2<sup>+</sup> uniport compared to the organelles of LR rats, with the difference being especially pronounced for the liver tissue. To elucidate the causes of these functional effects, we have examined structural changes in the Ca2<sup>+</sup> transport system of mitochondria of LR and HR rats and revealed that these changes do occur in the macromolecular complex of the mitochondrial Ca2<sup>+</sup> uniporter in the heart and liver tissues.

The MCUC is considered to consist of two transmembrane channel subunits: MCU and MCUb. MCU forms a highly selective Ca2<sup>+</sup> channel, which transfers the ion across the inner mitochondrial membrane. MCUb is a dominant-negative MCU paralogue, and its overexpression impairs the ion-transporting function of the complex [13]. MCU and MCUb are associated with other subunits: MICU1-2, EMRE, and MCUR1, which act as regulators of the uniporter.

As follows from the data obtained, in HR animals, the contents of the pore-forming subunit MCU and the gate subunit MICU1 are substantially increased in liver mitochondria, whereas the level of the dominant-negative subunit, MCUb, is decreased in heart mitochondria. The data are confirmed by the results of real-time PCR analysis and indicate that these changes occur at the level of transcription. At the same time, there is almost no difference between liver and heart mitochondria of LR and HR animals in the content of the regulatory uniporter subunit EMRE and the expression of its gene.

It is known that the activity of MCUC is regulated by changes in the ratio of the Ca2+-conducting pore subunits MCU and MCUb [5,26,27]. Therefore, both an increase in the level of MCU and a decrease in the content of its paralogue MCUb, which we recorded in mitochondria of the liver and heart of HR animals, can be the cause of the increased rate of Ca2<sup>+</sup> influx into the organelles. Based on the data obtained, we can conclude that the structural changes in the pore-forming subunits of MCUC in HR rats increase the efficiency of Ca2<sup>+</sup> accumulation in the mitochondrial matrix and, as a result, might underlie the fine control of cytoplasmic free Ca2<sup>+</sup> at low levels.

On the other hand, excessive accumulation of Ca2<sup>+</sup> into mitochondria can result in the opening of MPT in the inner mitochondrial membrane. As a result, all transmembrane ionic gradients and the membrane potential collapse, and mitochondria swell, which leads to the rupture of their outer membrane and the release of proapoptotic proteins (cytochrome *c*, apoptosis-inducing factor, and others) from the organelles. This can facilitate the activation of the caspase cascade, DNA damage, and eventually cell death. The opening of the MPT pore is believed to be triggered by hypoxia and subsequent reoxygenation [3,14]. The data obtained in this work show that liver (but not heart) mitochondria of HR rats have a significantly higher Ca2<sup>+</sup> capacity (i.e., are more resistant to the induction of the MPT pore than the organelles of LR rats). Similar results were partially obtained by us earlier [19]. In order to gain insight into the molecular mechanism of tolerance of liver mitochondria of HR rats to MPT, we assayed for levels of MPT-associated mitochondrial proteins: cyclophilin D, ATP synthase (α-subunit), and adenylate translocator ANT1 (Figure 4). Our results indicate that the levels of cyclophilin D and ANT1 in mitochondria of the liver of HR and LR rats are not distinguished. There is, however, a significant decrease in the content of the α-subunit of ATP synthase (a protein that supposedly forms the channel of the MPT pore) in liver mitochondria of HR rats. Thus, the increased tolerance of liver mitochondria of HR rats to MPT, as compared with mitochondria of LR rats, can be related to a lowered expression of the MPT pore-forming protein. This is in agreement with the earlier report that the changes in expression of ATP synthase had significant effects on the probability of MPT induction [28].

However, the question as to which subunit of this multisubunit complex is the central structural component of the MPT pore, or whether it is formed by MPT-specific conformation of dimers of ATP synthase, remains the province of further investigations.

It should be noted that the CRC index of mitochondria isolated from the heart of HR rats was slightly lower compared to that of LR rats, although this difference was not statistically significant. This trend toward a decline in the Ca2<sup>+</sup> buffering capacity is likely explained by the fact that the content of cyclophilin D, a major regulatory protein of the MPT pore [14,29,30], in heart mitochondria of HR rats is increased (Figure 6). It is believed that VDAC1 is also involved in the formation of the MPT pore complex in mitochondria. In addition, this protein participates in the formation of MAM contacts, which are the main pathway of Ca2<sup>+</sup> transfer from the endoplasmic reticulum to mitochondria [5,14]. So, the changes observed in the level of VDAC1 can also contribute to the tissue-specific modulation of mitochondrial Ca2<sup>+</sup> handling and the MTP opening.

Based on the results obtained, one can assume that the increase in the rate of Ca2<sup>+</sup> transport in mitochondria is not accompanied by a higher susceptibility of mitochondria to calcium overload and the induction of the MPT pore. As follows from our data, changes in the levels of the MPT-related proteins can play a crucial role in the regulation of pore formation in mitochondria. At the same time, the high rate of mitochondrial Ca2<sup>+</sup> uptake may provide a more effective removal of excess Ca2<sup>+</sup> ions from the cytosol under physiological conditions and hypoxia.

#### **5. Conclusions**

The results obtained point out that liver and heart mitochondria of HR animals exhibit functional and structural features in Ca2<sup>+</sup> ion transport, which may be essential for proper mitochondrial function in hypoxia and contribute to the formation of adaptive signs to provide the development of cellular response to oxygen shortage. So, apart from the increased activity of the ATP-sensitive potassium channel and respiratory chain complexes, mitochondria of HR animals are characterized by the increased rate of uptake of Ca2<sup>+</sup> ions. At the same time, compared to liver mitochondria of LR animals, the organelles of HR rats are more resistant to the opening of the Ca2+-dependent MPT pore. Taking into account the fact that the MPT pore opening is one of the key pathogenic events in oxygen deficiency, it can be supposed that these changes contribute to the molecular basis underlying cellular sensitivity to hypoxic injury.

**Author Contributions:** Conceptualization, K.N.B.; investigation, K.N.B., N.V.B., E.Y.T., V.S.S., N.M.Z., K.S.T., and M.V.D.; writing—original draft preparation, K.N.B.; writing—review and editing, K.N.B., N.V.B., and M.V.D.; project administration, K.N.B.; funding acquisition, K.N.B. All authors have read and agreed to the published version of the manuscript.

**Funding:** This study was funded by grants from the Russian Foundation for Basic Research (18-315-20011 to Konstantin N. Belosludtsev).

**Acknowledgments:** We thank Ivan V. Semenov from the ICB RAS for technical support in carrying out the experiments.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).
