**2. Results**

Confocal microscopy of a normal Wistar rat cardiac mitochondrial architecture in the vicinity of nuclei revealed a very complicated mitochondrial network organized by tiny branched mitochondrial filaments. Practically all cardiac nuclei were surrounded by a mitochondrial web, deeply penetrating the body of the nucleus (Figure 1, Supplementary Video S1). These visually observed structures were identified with a variety of mitochondrial dyes, at least one of these being potential-dependent tetramethyl rhodamine methyl ester (TMRM). Mitochondrial dye nonyl acridine orange (NAO), apparently interacting with mitochondrial cardiolipin independently of the existence of the membrane potential (Supplementary Video S4), as well as Mitotracker Deep Red (not shown), revealed the same peri(intra) nuclear mitochondrial network suggesting that these nuclear mitochondria are fully functional. To exclude that these structures belong to sarcoplasmic reticulum non-specifically stained with mitochondrial dyes, we used an approach of photo-induced oscillations of the mitochondrial membrane potential [42]. Observed oscillations of a di fferent part of the mitochondrial reticulum including the peri(intra)nuclear part confirmed that these structures were mitochondria with maintained membrane potential (Figure 2, Supplementary Video S2). However, in spite of obvious very deep penetration of mitochondrial fluorescence images into the space occupied by the nucleus, the light microscopic level approach did not allow us to discriminate mitochondria deeply invaginated in the nucleus from those residing in the nucleoplasm. Subsequent electron microscopic study of the normal cardiac myocyte was designed to resolve this question.

**Figure 1.** Variability of the mitochondrial architecture in the vicinity of nuclei in normal rat ventricular cardiac myocyte was stained with the mitochondria-targeted probe tetramethyl rhodamine methyl ester (TMRM; 200 nM). Confocal microscopy. Bright images represent energized mitochondria along myofilaments of the heart cell. We used a pinhole of 150 mμ allowing one to observe tilted mitochondrial chains spanning the cell, thus making an impression of the appearance and disappearance of these chains. In some images, in order to reveal the peri(intra)nuclear mitochondrial network (arrows), the detector gain was artificially enhanced, making the fluorescence of interfibrillar mitochondria saturated. Scale: 5 mμ.

**Figure 2.** Photo-induced oscillations of the membrane potential in mitochondrial clusters within cardiac myocytes loaded with 200 nM TMRM. Colored arrows show some oscillatory elements at different time intervals indicated in the upper-left corner (in seconds) of each confocal scan. The example at the bottom shows periodic changes of the fluorescence intensity of TMRM in the region shown by the red arrow. Scale: 5 mμ. Full-time series of this sample is presented in Supplementary Video S2.

Figure 3A–C represents three consecutive ultrathin sections of cardiac myocytes of a 3-month-old Wistar rat. As shown in Figure 3A, three mitochondria were clearly visible inside the nucleus. The mitochondria were not separated from the contents of the nucleus by the nuclear membrane, that is, they, in fact, were located in the nucleoplasm. In the subsequent sections of this nucleus, the number of mitochondria inside the nucleus was increased. In Figure 3B,C, the contents of the nucleus were in direct contact with a mitochondrial cluster due to the partial absence of the nuclear membrane. It should be noted that in Figure 3A,B, the nuclear area directly surrounding the mitochondria had a fine fibrillar structure differing greatly from the granular structure in the main part of the nucleoplasm. In this case, a fragment of cytoplasm containing mitochondria was supposed to enter the nucleus through the open aperture in the nuclear membrane. Figure 4A,B show direct contact between a mitochondrial cluster and nuclear structures in a cardiomyocyte of a 24-month-old OXYS rat. Furthermore, using heart samples from a 3-month-old Wistar rat, we performed a three-dimensional reconstruction of a part of the nucleus with the mitochondria present inside, showing the architecture of the chromatin and nuclear membrane on the basis of the analysis of a sequential series of ultrathin sections for electron microscopy (Figure 5 and Supplementary Movie 3).

**Figure 3.** (**A**–**C**) Consecutive ultrathin sections from serial sections of a cardiomyocyte nucleus of a 3-month-old Wistar rat. Electron microscopy.

**Figure 4.** (**A**) Direct contact of a mitochondrial cluster with nuclear structures in a cardiomyocyte from a 24-month-old OXYS rat. The area of contact is indicated by an arrow. (**B**) The local fragment of the nucleus indicated by the arrows in Figure 4A. The nuclear membrane was absent, and individual mitochondria were directly adjacent to the nucleoplasm.

**Figure 5.** Serial micrographs of 12 sections over the nucleus of the cardiomyocyte of a 3-month-old Wistar rat with mitochondria embedded in the nucleoplasm.

The main and very important argumen<sup>t</sup> is that this particular examined cell with the intranuclear mitochondria was abnormal. However, this was not confirmed by the ultramicroscopic characteristics of the nucleus and the cytoplasm surrounding this nucleus. We compared the ultrastructure of these cells containing the obvious intranuclear mitochondria with ultrastructure of cells where intranuclear mitochondria were missing and found no significant alterations indicating cell damage (Figure 6).

**Figure 6.** Ultrastructures of nuclei and cytosols of the rat heart tissue with cells, one of which contained intranuclear mitochondria while the others did not. (**A**) Ultrastructure of cardiomyocytes from a 3-month-old Wistar rat with mitochondria in the nucleus (upper cell) and without them (lower cell). White arrows indicate condensed chromatin and black arrows indicate decondensed chromatin. (**B**) Cytosolic ultrastructure of the 24-month-old OXYS rat cell with intra-nuclear mitochondria (cell 1) and adjacent cells (cell 2 and cell 3), connected by intercalated discs (ID). Note that (**B**) is a low zoom of the cardiac tissue containing the region depicted in Figure 4A,B in the manuscript.

In the nuclei of cardiomyocytes with nuclear ruptures, chromatin was preferentially decondensed, and the condensed chromatin was visible at nuclear periphery in close contact with the nuclear envelope, around the nucleoli (perinucleolar chromatin), and inside the nucleoplasm. The blocks of condensed chromatin in the nucleoplasm adjusting the nuclear envelope ruptures had elongated shape, probably due to mechanical tension. The similar localization of condensed chromatin was detected in the nuclei without nuclear envelope ruptures. This apparently mechanical deformation of chromatin complexes was visible near nuclear envelope ruptures, indicating that nuclei were under a strong pulling force action, which potentially could induce these ruptures. Decondensed chromatin was not substantially modified, even in regions that were in direct contact with the mitochondria. Thus, in the nuclear regions adjacent to broken nuclear membrane, the changes in chromatin were minimal, whereas on the nuclear periphery far from these regions, the chromatin configuration was not distinguished in both types of cells (Figure 6A).

As for the cytosolic ultrastructure of the cell with intranuclear mitochondria (Figure 6B), the ultrastructure of myofibrils was conventional with regular position of isotropic and anisotropic regions separated by a Z-line. Myofibrils are longitudinally oriented and densely packed. Sarcomeres have a conventional size of 2–3 microns in length. Intercalated disks are not damaged with a typical structure. The sarcoplasma is not swollen with mitochondria having a normal orthodox structure.

A more striking picture of a direct contact between mitochondria and structures of the nucleus was found in a cardiomyocyte of a 5-year-old naked mole rat (Figure 7A,B). In this case, the sections were made in such a way that the absence of the nuclear membrane was revealed along the entire perimeter of the nucleus. The specific morphology of the nucleus was preserved despite the vast area lacking the nuclear membrane. Figure 7B shows at higher magnification the mitochondrial group indicated by arrows in Figure 7A. It is clearly seen that the mitochondria were in direct contact with the intranuclear structures. The analysis of serial sections of this nucleus (Figure 8) revealed that mitochondria did not form a continuous layer contacting the contents of the nucleus. There were some nuclear areas directly adjacent to myofibrils (indicated by the arrow, Figure 8, section h).

**Figure 7.** (**A**) Direct contact of mitochondria with nuclear structures in a cardiomyocyte from a 5-year-old naked mole-rat. In this section, the nuclear membrane was absent along the entire perimeter of the nucleus. Arrow 1 shows the cytoplasm with organelles including mitochondria was located inside the nuclear invagination. Arrow 2 shows the nuclear membrane of the nucleus invagination. (**B**) A group of mitochondria shown by arrow 3 in Figure 7A at high magnification. Mitochondria were in direct contact with nuclear structures and were submerged in the nucleoplasm.

**Figure 8.** Serial of consecutive sections of the nucleus of a cardiomyocyte (from **a** to **i** with the thickness of each section ~500 Å) from the 5-year-old naked mole-rat presented in Figure 7A. Arrows in sections **a** and **h** point to the same area of the nucleus, proving that mitochondria did not form a continuous layer contacting the nuclear contents. In section **a**, the mitochondria were in direct contact with nuclear structures, whereas in section **h**, there was direct contact of nuclear content with adjacent myofibrils.

The statistical analysis of the occurrence of mitochondria in cardiomyocyte nuclei showed that, on average, the percentage of nuclei with mitochondria was roughly around 1%, and it did not show age and species dependency (see Supplementary Table S1).
