Primary adult HSFs from healthy controls (n = 5) and SZ patients (n = 5) were obtained from the Research Centre for Medical Genetics. A description of the five healthy subjects and the five SZ patients, who were the primary donors of skin cells, is presented in Table 1. The examination was conducted in the round-the-clock inpatient units of N.A. Alekseev Clinical Psychiatric Hospital No. 1 (Moscow, Russia), where the patients received standard psychiatric care. The skin samples were obtained from SZ cases with a long-term continuous form of schizophrenia. Each patient had been taking standard antipsychotics for several years (e.g., haloperidol, aminazin, risperidone, seroquel, olanzapin). Four of the five patients had nearest relatives with the same disease.

The control skin sample donors were staff members of the Research Centre for Medical 100 Genetics. They were apparently healthy, took no pharmaceuticals, and had no mental problems or relatives with mental disorders.

The cells were maintained in Dulbecco’s Modified Eagle Medium supplemented with 10% fetal bovine serum (HyClone Laboratories, Logan, UT, USA), 2 mM glutamine, 50 U/mL penicillin, and 50 µg/mL streptomycin at 37 °C in a 5% CO${}_2$ incubator. All of the experiments were performed using cells from passage 4. Cultures with less than 2% of cells in the S phase were considered confluent. In general, cells reached confluence after 3–4 days of plating. This was marked as Day 1 of the experiment.

The study was carried out in accordance with the latest version of the Declaration of Helsinki and was approved by the Independent Interdisciplinary Ethics Committee on Ethical Review for Clinical Studies (Protocol No. 4 dated March 15, 2019 for the scientific minimally interventional study “Molecular and neurophysiological markers of endogenous human psychoses”). All participants provided written informed consent to participate in the study after the procedures had been completely explained.

To isolate the DNA, we used a standard method as previously described [51]. Briefly, 2 mL of the solution (2% sodium lauryl sarcosylate, 0.04 M EDTA, and 150 µg/mL RNAse A; Sigma, St. Louis, MO, USA) was added to the cell mass for 45 min (37 °C) and then treated with proteinase K (200 µg/mL; Promega, Madison, WI, USA) for 24 h at 37 °C. The lysates were extracted with an equal volume of phenol, phenol/chloroform/isoamyl alcohol (25:24:1) and chloroform/isoamyl alcohol (24:1). DNA was precipitated by adding 1/10 the volume of 3 M sodium acetate (pH 5.2) and a 2.5-fold volume of ice-cold ethanol. DNA was collected by centrifugation (10,000 $\times$g for 15 min at 4 °C), washed with 70% ethanol (v/v), and dissolved in water. DNA concentration was measured in two steps, as described earlier [51]. The final DNA quantification was performed fluorometrically using PicoGreen dsDNA Quantification Reagent (Invitrogen, Carlsbad, CA, USA). The DNA concentration in a sample was calculated according to a standard curve for DNA. We used the Enspire Multimode Microplate Reader (PerkinElmer, Turku, Finland) at $\lambda$${}_{\text{ex}}$ = 488 nm and $\lambda$${}_{\text{em}}$ = 528 nm.

The nonradioactive quantitative hybridization (NQH) method is based on the complementary hybridization of DNA immobilized on a filter with a biotin-labeled DNA probe. NQH was used for the quantification of f-SatIII, TR, and rDNA repeats, as previously detailed without modification [52, 53] and Supplements of [39]. Five reference DNA probes with a known repeat content were applied to the same filter. After hybridization and biotin detection using the streptavidin-alkaline phosphatase conjugate (BCIP and nitro blue tetrazolium substrates), the filter was scanned and the spot signal intensity was measured using a customized software pack. A calibration curve was constructed, reflecting the dependence of the signal intensity on the genome’s repeat copy number. The relative standard error for NQH was 5 $\pm$ 2%. The major overall error of the experiment was contributed to by the step of isolating DNA from the cells. The total standard error was 11 $\pm$ 7%. The DNA probe for SatIII (1) quantification was a 1.77-kb cloned EcoRI fragment of human satellite DNA [1] labeled with biotin-11-dUTP using nick translation. Dr. Cook (Medical Research Council, Edinburgh, UK) kindly supplied the human chromosome lql2-specific repetitive satellite DNA probe pUC1.77. The DNA probe for rDNA quantification contains rDNA sequences (5836 bp) cloned into an EcoRI site of the pBR322 vector. The rDNA fragment covered positions from –515 to 5321 of the human rDNA (GenBank accession No. U13369). DNA Probe used for the detection of the human TR was biotin-(TTAGGG)${}_7$. Syntol (Moscow, Russia) performed the synthesis and biotin labeling of the oligo probe.

Cells were analyzed using the CytoFLEX S flow cytometer (Beckman Coulter, Brea, CA, USA). HSFs were grown in 60 mm dishes. Prior to flow cytometry analysis (FCA), cells were washed with Versene solution and treated with 0.25% trypsin under light microscope observation. Then the cells were transferred to Eppendorf tubes, washed with culture medium, centrifuged, and resuspended in phosphate-buffered saline (PBS). Staining the cells with antibodies was performed as previously described [54]. Briefly, to fix the cells, paraformaldehyde (PFA; Sigma, Kawasaki, Japan) was added (3% at 37 °C for 10 min). Then the cells were washed three times with 0.5% bovine serum albumin-PBS and permeabilized with 0.1% Triton X-100 (Sigma) in PBS. Cells (~50 $\times$ 10${}^3$) were stained with the following antibodies: NRF2-FITC (bs1074r-fitc; Bioss Inc., Boston, MA, USA), 8-oxodG-PE (sc-393871 PE; Santa Cruz Biotechnology, Inc., Santa Cruz, CA, USA), NOX4-PC5.5 (bs-1091r-cy5-5; Bioss Inc.), $\gamma$H2AX-pb450 (nb100-384AF405; Novus Biologicals LLC, Centennial, CO, USA), SOD1-PE (sc-11407 and rabbit IgG-PE sc-3753; Santa Cruz), BAX-PE (rabbit Nb120-7977; Novus Biologicals), secondary mouse anti-rabbit IgG-PE (sc-3753; Santa Cruz), A350-BCL2 (bs-15533r-a350; Bioss Antibodies Inc. Woburn, USA), LC3DyLight 488 (NB100-2220G; Novus Biologicals). To quantify the background fluorescence, we stained a portion of the cells with secondary FITC (PE, PC5.5, pb450, DAPI)-conjugated antibodies only. Primary data are presented as median signal intensities minus signal background values. For a more convenient presentation of the data, the values of the medians of FL-signal were normalized to the maximum FL-signal value in the sample (n = 10). The relative standard error of the FCA was 4 $\pm$ 2%.

The fluorescent assay was applied for ROS quantification in HSFs in the format of a 96-well tablet reader at $\lambda$${}_{\text{ex}}$ = 488 nm and $\lambda$${}_{\text{em}}$ = 528 nm (Enspire, PerkinElmer, Turku, Finland). The medium was replaced with 5 µm H2DCFH-DA (Molecular Probes/Invitrogen, Thermo Fisher Scientific, Waltham, MA, USA) in PBS, and a fluorescence intensity was measured at 37 °C. Eight (4 $\times$ 2) measurements were made for each DNA sample and 16 for the control. The slope of the line (k), which reflects the dependence of the fluorescence intensity on time, characterizes the ROS content in HSFs. This method allows quantification of the total ROS amount in HSFs (intracellular) and in the extracellular medium. For the intracellular ROS assay (i.e., only in the cells), FCA was applied. The medium was replaced with 5 µm H2DCFH-DA in PBS for 1 h at 37 °C. Prior to FCA, cells were washed with Versene solution and treated with 0.25% trypsin under light microscope observation. Then the cells were transferred to Eppendorf tubes, washed with PBS, centrifuged, and resuspended in PBS.

Isolation of the RNA was performed using the RNeasy Mini kit (Qiagen, Hilden, Germany). RNA was quantified using the Quant-iTRiboGreen RNA Reagent (MoBiTec, Göttingen, Germany) on a multimode plate reader (EnSpire; PerkinElmer) at $\lambda$${}_{\text{em}}$ = 487 nm, $\lambda$${}_{\text{fl}}$ = 524 nm. The RNA samples were further treated with DNase I to remove possible trace amounts of DNA. After DNAse I treatment, RNA samples were reverse transcribed using the Reverse Transcriptase Kit (Sileks, Moscow, Russia). PCR was conducted with specific primers and SYBR Green intercalating dye on the StepOnePlus Real-Time Polymerase Chain Reaction (PCR) System (Applied Biosystems, Foster City, CA, USA). The primers were selected and synthesized by Evrogen (Moscow, Russia). The internal standard was the TATA box-binding protein gene. For the RNASATIII assay, primers described in [34] were applied. In each series of experiments, RNA samples treated with DNase I were used as a negative control. The PCR reaction mixture in a volume of 25 µL consisted of 2.5 µL PCR buffer (700 mmol/L Tris-HCl, pH 8.6); 166 mmol/L ammonium sulfate, 35 mmol/L MgCl${}_2$, 2 µL of 1.5 mmol/L dNTP solution; and 1 µL of 30 pmol/L primer solution, cDNA. PCR conditions were chosen individually for each primer pair. After denaturation for 4 min at 95 °C, 40 amplification cycles were performed in the following order: 94 °C for 20 s, 56–62 °C for 30 s, 72 °C for 30 s, and 72 °C for 5 min. The data were processed using a calibration plot with a resultant error of 2%.

The Axio Scope.A1 microscope (Carl Zeiss, Oberkochen, Germany) was used for fluorescence microscopy. For $\gamma$H2AX detection, cells were washed three times with PBS, fixed for 20 min in 3% PFA at 4 °C, washed with PBS, and incubated for 4 h at 4 °C with $\gamma$H2AX-FITC antibodies. After washing with 0.01% Triton X-100 in PBS, HSFs were washed with PBS and stained with 2 µg/mL DAPI.

For fluorescence in situ hybridization (FISH), the HSFs in slide flasks were washed with PBS. The slides were removed and placed for 10 min into a cold fixation solution. After the procedure was repeated three times, the slides were dried and subjected to FISH. Before hybridization, the slides were treated with RNAse A (100 µg/mL). Hybridization was conducted using the protocol and solutions from Abbott Laboratories (Abbott Park, IL, USA) in the Statspin ThermoBrite Slide Denaturation and Hybridization System (Abbot Molecular Inc., Des Plaines, IL, USA) at 42°. Nuclei were stained with propidium iodide. The f-SatIII FISH probe was a 1.77-kb-cloned EcoRI fragment of human satellite DNA [54]. Labeling of plasmid pUC1.77 was performed using nick translation with the CGH Nick Translation Kit (Abbott Molecular) according to the manufacturer’s protocol.

Each test was repeated in triplicate. In FCA, the medians of the signal intensities were analyzed. The significance of the observed differences was analyzed using the nonparametric Mann–Whitney test (p) and Kolmogorov–Smirnov test (D and $\alpha$). Spearman analysis (Rs, p) was used to analyze correlations between the two parameters. p 0.01 and $\alpha$ 0.04 were considered statistically significant. The data were analyzed with Microsoft Excel and Office (Microsoft, Redmond, WA, USA), StatPlus2007 Professional software (http://www.analystsoft.com accessed on March 19, 2022), and StatGraphics (Statgraphics Technologies, The Plains, VA, USA).

Fig. 1 shows the content data of the three tandem repeats (SatIII (1), TR, and rDNA) in DNA isolated from cells that had reached confluence (control). For reference, the contents of the repeats in actively proliferating cell culture 2 days before the confluent state (–2 days) were also given.

In proliferating HSFs, the SatIII (1) content in DNA varied from 11 pg/ng DNA (3SZ) up to 16 pg/ng (1SZ). Proliferating sz-HSFs and hc-HSFs did not significantly differ by SatIII (1) content in DNA (Fig. 1A,a1). The SatIII (1) content in the DNA of confluent cells varied from 11 pg/ng (4K) up to 36 pg/ng (1SZ) (Fig. 1A,a1). The DNA isolated from confluent sz-HSFs contained more SatIII (1) repeats than DNA isolated from hc-HSFs (D = 0.8, $\alpha$ = 0.036; p 0.01).

The TR content in DNA varied from 71 pg/µg DNA (1SZ) up to 335 pg/µg (4K) (Fig. 1A,a2), corresponding with the average LTL, approximately 2 to 7 kb. Hc-HSFs carried more copies of TR (D = 0.8, $\alpha$ = 0.036; p 0.01) in their genomes than sz-HSFs, suggesting longer telomeres in hc-HSFs. A comparison of confluent cell DNA (control) to DNA isolated from actively proliferating cells (–2 days) showed a slight decrease of TR contents in DNA of confluent sz-HSFs. The TR content in DNA of the confluent HSFs negatively correlated (Fig. 1A,a4) with the SatIII (1) content (Rs = –0.88, p 0.001, n = 10).

The numbers of rDNA copies in HSFs varied from 386 (3SZ) up to 454(5K). Sz-HSFs and hc-HSFs did not differ by rDNA abundance in the cellular DNA (Fig. 1A,a3). This index did not change during cell cultivation. Thus, in the process of culturing HSFs to a confluent state, the SatIII (1) content in the isolated DNA increases. One of the possible reasons for the SatIII (1) content increase may be an increase in the amount of RNA SATIII (1).

In confluent cells, SatIII (1) repeats were transcribed, as evidenced by the detected amounts of RNASATIII (1). SatIII (1) transcription intensity in sz-HSFs was higher (D = 0.8, $\alpha$ = 0.036; p 0.01) than that in hc-HSFs (Fig. 1B,b1). There was a positive correlation between the level of RNASATIII and the amount of SatIII (1) DNA (Fig. 1B,b3). For comparison, we quantified the transcription level of another satellite III, which is localized in the pericentromeric region of chromosome 9. In contrast to RNASATIII (1), the RNASATIII (9) content in sz-HSFs was lower (D = 0.8, $\alpha$ = 0.036; p 0.05) than in hc-HSFs (Fig. 1B,b2). In confluent sz-HSFs, the RNASATIII (1)/RNASATIII (9) ratio was 8-fold higher than in hc-HSFs (Fig. 1B,b4).

Thus, it was shown that the content of the SatIII (1) repeat in DNA isolated from confluent cells correlated with the content of RNASATIII in RNA isolated from the same cells. Stress is an inducer of SatIII (1) transcription [34, 35]. To determine the level of endogenous stress that could induce satellite transcription and subsequent satellite DNA amplification, we determined some characteristics, which indicate the level of oxidative stress in the HSF pool.

To estimate the double-strand DNA breaks rate in nuclear DNA, antibodies to 301 phosphorylated form of histone Н2АХ ($\gamma$H2AX) [43] were applied (Fig. 2A,a1). For the reader’s convenience, a median value of cell signal for each HSF cell line was normalized to the maximum median value of the signal in the sample (parameter $\gamma$H2AX = m[FL-$\gamma$H2AX]i/m[FL-$\gamma$H2AX]max). The amounts of $\gamma$H2AX in sz-HSFs were higher than those in hc-HSFs (D = 0.8, $\alpha$ = 0.036; p 0.01). The maximum damage degree was observed for SZ1, whereas the minimum one was observed for K4 (Fig. 2A,a2).

An analysis of the dependence of $\gamma$H2AX on FSC parameter, which indicates the cell size, allowed us to distinguish the cell subset R1 (Fig. 2C,c1–c3 and 2E,e1) with large sizes and relatively high, compared to the other cells, level of the DNA damage marker. This fraction varied in the confluent HSFs from 6% to 16% and correlated positively with SatIII (1) repeat content in the DNA (Rs = +0.82, p = 0.0038, n = 10), while correlated negatively (Rs = –0.84, p = 0.0022, n = 10) with the TR abundance (Fig. 2D,d1,d2). The rDNA abundance in the genome was not associated with the size of R1 fraction (Fig. 2D,d3).

FISH was conducted to analyze the state of SatIII (1) repeat (Fig. 2E,e2). SatIII (1) is localized in two homologous sites of the nucleus. In cells of normal size, they are visualized as two compact spots. In large cells, the SatIII (1) signal area is much larger, suggesting both chromatin decondensation and an increased total copy number of the repeat in such cells.

The DNA oxidation degree was tested using antibodies against an oxidation marker 8-oxodG (Fig. 2B,b1). The 8-oxodG rate (parameter 8-oxodG = m[FL-8-oxodG]i/m[FL-8-oxodG]max) was higher in sz-HSFs than in hc-HSFs (D = 0.8, $\alpha$ = 0.036; p 0.01). The maximum DNA oxidation degree was observed for SZ1, and the minimum one was observed for K4 (Fig. 2B).

Thus, the oxidative modification seems to be the major cause of DNA breaks, resulting from the high content of endogenous ROS in HSF cell pools. The ROS amount depends on the balance of pro- and antioxidant enzyme activity. We assayed the ROS amount and the contents of three proteins (NOX4, SOD1, and NRF2) that regulate ROS abundance in the cell.

The ROS were quantified using H${}_2$DCFH-DA chemical reagent and a tablet reader (Fig. 3A). To estimate the ROS amount, the synthesis rate constant DCF (k) was determined in confluent sz-HSFs after the reagent had been added. The k value for sz-HSFs was considerably higher than that for hc-HSFs, suggesting a higher ROS level in the cells derived from SZ patients (Fig. 3A,a1,a2).

The distribution of unfixed cells by ROS amount was estimated using FCA (Fig. 3A,a3). The HSF cell pool appeared to be heterogeneous by this parameter. Three fractions could be distinguished. R1 were large cells with an elevated ROS content. R2 and R3 were cells of approximately the same normal size, but with different ROS levels.

One way to determine an increase in ROS is NOX4 gene activity (Fig. 3B). The amount of RNANOX4 was on the average more in sz-HSFs than in hc-HSFs (p 0.05). 2K line cells were an exclusion showing a higher RNANOX4 level than the other hc-HSFs (Fig. 4B,b1). NOX4 protein content as determined using FCA (NOX4 index) was positively correlated with RNANOX4 content (Rs = +0.75, p = 0.012, n = 10). The NOX4 index was higher for szHSFs than for hc-HSFs (D = 0.8, $\alpha$ = 0.036; p 0.01).

The amounts of RNA transcribed from the NFE2L2 gene encoding NRF2 transcription factor were equal in sz-HSFs and hc-HSFs (p $>$ 0.1; Fig. 3C,c1). However, NRF2 protein content (NRF2 index) was slightly increased (differences were not significant) in sz-HSFs compared to hc-HSFs (Fig. 3C,c2,c3). The NRF2 index did not correlate with RNANFE2L2 amount.

The mRNA and protein levels of SOD1 (SOD1 index) were the same in sz-HSFs and hc-HSFs (p $>$ 0.1; Fig. 3D). К5 line cells contained the maximum level of the protein. The SOD1 index did not correlate with RNASOD1 amount.

We found no significant differences in the RNA and protein levels of BAX and BCL2 between sz-HSFs and hc-HSFs (Fig. 4A–C). The maximum BAX index was observed for SZ1, whereas the minimum one was observed for K3 and K4 (Fig. 4A,a3). A BAX/BCL2 ratio, which represents the apoptosis intensity [55], was higher for sz-HSFs than for hc-HSFs (p 0.015), with the exception of the К5 cell line. For К5, high values of RNABAX/RNABCL2 and BAX/BCL2 indices were observed (Fig. 4C).

In lines SZ1–SZ4, a considerable increase in ATG16L1 RNA was found (Fig. 4D,d1). The ATG16L1 gene governs the process of autophagy. We determined the amount of protein LC3, which indicates the autophagy intensity in the cells (Fig. 4D,d2,d3). For sz-HSFs, the LC3 index was higher than that for hc-HSFs (D = 0.8, $\alpha$ = 0.036; p 0.01).

The large size cell fraction (R1) was enriched with cells with a high level of LC3 protein (Fig. 4E,e1,e2). A joint analysis of levels of $\gamma$H2AX and LC3 in the same cells showed that the process of autophagy was activated in cells with a high DNA damage degree, which were part of fraction R1 (Fig. 4E,e3).

Fig. 5 shows the dependence of SatIII (1) content on the abovementioned characteristics of the cell lines. The SatIII (1) repeat abundance was positively correlated with the levels of DNA damage markers ($\gamma$H2AX and 8-oxodG), proteins that modulate ROS amount (NOX4 and NRF2), and indices that represent the apoptosis and autophagy intensity (BAX and LC3). The largest SatIII (1) copy number, as measured for DNA derived from SZ1 cells, corresponds to the maximum values of the parameters that display the level of oxidative stress. A positive correlation between the stress indicators and the content of the SatIII (1) repeat in DNA was also preserved in the individual groups (Fig. 5). Thus, we showed that a high SatIII (1) content in cell DNA is associated with exaggerated endogenous oxidative stress in the cell population.

A high level of oxidative stress and DNA damage markers in HSFs was associated with increased SatIII (1) repeat content in the DNA isolated from the confluent cells (Fig. 5). An analysis of histograms of cell distribution by $\gamma$H2AX marker content showed that many cells of sz-HSFs had a high marker content, but especially increased $\gamma$H2AX was observed in a small subset of large-size cells. The fraction of such cells was correlated with the SatIII (1) content in the total DNA isolated from the cells (Fig. 2). The data obtained using FISH suggested that in these cells, the SatIII (1) repeat copy numbers were increased and/or chromatin decondensation occurred at the site of hybridization with a probe for SatIII (1).

A high SatIII (1) content in total DNA was correlated with a low TR content (Fig. 1). Earlier, opposite changes of TR and SatIII (1) contents in the cellular DNA were revealed when studying variation of these tandem repeats during replicative senescence of five HSF lines [39], in various brain regions of a SZ patient [62] and in the research of the human blood cell response for an affective stress [63]. In every case, increased SatIII (1) content in the cellular DNA was linked to a reduced TR count.

For the SatIII (1) repeat region, DNA decondensation and demethylation were previously found at late passages, accompanied by an exaggerated transcription of SatIII (1) DNA [34]. We measured the RNASATIII (1) levels in HSFs (Fig. 1). The transcript amount in sz-HSFs exceeded that in hc-HSFs, correlating with a higher SatIII (1) repeat content in sz-HSFs. An increased level of satellite transcription under conditions of endogenous oxidative stress may be one of the reasons for the increase in the content of the SatIII (1) repeat in some cells of the population.

One report [34] also showed that senescence was not accompanied by the transcription of SatIII (9) repeats localized in the pericentromeric region of chromosome 9, in contrast to SatIII (1). An assay of RNASATIII (9) showed that the amount of the transcript was decreased in sz-HSFs compared to hc-HSFs. The activation of SatIII (1) transcription was accompanied by a decrease in SatIII (9) transcript level (Fig. 1B,b4). We previously observed a similar phenomenon when studying the effects of cell exposure to ionizing radiation [35].

In total, our findings suggest that under the stress caused by endogenous factors, an increase in SatIII (1) copy numbers occurs in cells with a low count of TRs. Seemingly, these cells undergo the biggest ROS impact. Telomere shortening is associated with heterochromatin decondensation and activation of transcription from these regions. A mechanism previously proposed by Bersani et al. [38] may be involved in the increase of SatIII (1) cellular content under the action of stress factors. By means of reverse transcriptase, a DNA chain is synthesized from RNASATIII, and the DNA–RNA hybrids then forms, inserts into the transcribed region, and undergoes the action of DNA repair enzymes. The region 1q12 under exploration is known to be one of the most unstable chromatin sites [63]. In this chromatin region, the double-strand break repair processes are suppressed, thus increasing the chances of successful integration of the DNA–RNA hybrids into DNA [64].

The insertion of additional SatIII (1) fragments into the heterochromatin regions in some cells cannot be detected by analyzing the size of the chromatin block on metaphase chromosomes. We previously showed that cells with an elevated SatIII (1) content are unable to respond to various impacts that require genome architecture rearrangements necessary for changing the gene expression profiles in response to stress. In particular, these cells do not respond to proliferative stimuli (i.e., cannot enter the mitosis phase). In addition, such cells fail to launch an adaptive response for an exposure to small doses of ionizing radiation [35] and perish as a result of even a weak toxic impact.

The correlation of SatIII (1) copy number with stress markers and sufficiently high SatIII (1) transcription level in HSFs suggests that even relatively young cell pools (at early passages) can accrue cells with high SatIII (1) content during cultivation, which do not divide and are not resistant to stronger types of stressor impacts. To prove this assumption, we analyzed the changes of SatIII (1) content caused by two types of stress.

Cultivating cells for a long time without medium replacement is accompanied by stress resulting from a decline of nutrients in the medium and cellular debris accumulation due to apoptosis, which can induce extra ROS synthesis in living cells [65]. Four days after HSFs had reached confluency, the SatIII (1) copy numbers significantly decreased in DNA samples isolated from sz-HSFs (Fig. 6A,a1,a3). A slight decrease of SatIII (1) content also occurred in four DNA samples isolated from hc-HSFs. Simultaneously with the decrease of SatIII (1) content, an increase in TR abundance was observed in DNA isolated from HSFs (Fig. 6B,b2). The copy numbers of ribosomal repeat remained virtually unchanged in the isolated DNA (Fig. 6B,b1). The observed changes in the contents of the two repeats in DNA isolated from the cells after long-term growth without medium replacement suggest that this stress type induces death of cells with high SatIII (1) content and relatively low TR abundance in the pools of HSFs. These findings corroborate the reports of hypersensibility of cells with high SatIII (1) contents to damaging impacts due to a blockage of the adaptive response in these cells [35].

The cell’s response to heat shock considerably differs from the its response to prolonged cultivation without medium replacement by the RNASATIII (1)/RNASATIII (9) ratio change patterns. Heat shock induced SatIII (9) transcription, as previously described [14], while the resultant amount of SatIII (1) transcripts was considerably reduced (Fig. 6C–E).

Despite the difference in the changes of transcription of different SatIII fragments under heat shock and cultivation without medium replacement, the patterns of changes in SatIII (1) and TR copy numbers in the cellular DNA were similar: while SatIII (1) repeats were reduced, the TR arrays were enlarged (Fig. 6B). These data suggest a stress-induced process of death in cells with a high abundance of SatIII (1), and simultaneously, a low content of TR.

An integrated analysis to summarize the previous reports [35, 39] and the data obtained in this study suggested a scheme of satellite III (1q12) CNV in cultured cells with different levels of endogenous ROS production (Fig. 7).

The cell pool of HSFs is heterogeneous by the copy numbers of satellite III (1q12) repeats in the cell genomes. As early as the first passages, they contain a small fraction of cells with an elevated SatIII (1) content [39]. The clones with high SatIII (1) content consist of a few cells, because the proliferative capacity of these cells is considerably abated as compared to cells harboring low SatIII (1) abundance in the genome. The amounts of TR in cells that had formed clones (dividing cell) with high and with low SatIII (1) contents did not differ [39].

The cell pool of HSFs is also heterogeneous by ROS level (Fig. 3A). In confluent HSFs, cells with high ROS content formed a subset, which was characterized by a large cell size and high DNA damage degree. The size of this cell fraction correlated with increased SatIII (1) content and decreased TR content in the DNA isolated from the whole bulk of cells. FISH data suggested that SatIII (1) content was increased, for the most part, in the cells that belonged to this fraction (Fig. 2). As previously shown, cells with the same characteristics accrued during replicative senescence [39].

An extra stress impact induces death of the cells with a high SatIII (1) content and a low TR count. Such cells are not capable of development of the adaptive response and have hypersensibility to DNA damaging agents [35]. The abovementioned process occurs during various stress types – cultivation without medium replacement and heat shock (our study’s findings), as well as exposure to ionization radiation [35]. As a result, the after-stress DNA isolated from HSFs had a decreased SatIII (1) content, and to a lesser degree, an increased TR content (Fig. 6). Perhaps, the cells with high SatIII (1) content and low TR content dying under exposure to small radiation doses, are a source of extracellular DNA fragments, which induce a bystander effect and an adaptive response in the other (surviving) cells of the cell pool. Existence of such a subset in every cell pool was hypothesized earlier [66].

In vivo, the body inevitably undergo the environmental stress impact, sooner or later. Thus, the SatIII (1) content in the total DNA passes a cycle of an increase and a decrease, and the TR content passes a similar, but phase-shifted cycle, as shown in Fig. 7. The figure shows one oscillation, but provided that cell pools are regularly renewed in a living body and the stresses are also repeated from time to time all over the lifetime, one can hypothesize that the oscillations are recurrent in vivo. This speculation had become the basis for a ‘pendulum’ model of SatIII (1) dynamics we proposed early [53] and corroborated in the present study.

The stress evokes an increase of RNASATIII (1) amount in HSFs during normal growth (Figs. 1,6). Transcription of the two fractions of satellite III, localized on chromosome 1 and chromosome 9, depends on the stress type. A high ROS level during cultivation induces an augmentation of RNASATIII (1) transcripts and a decrease of RNASATIII (9) transcripts. By contrast, a heat shock induces an augmentation of RNASATIII (9) amount and a decrease of RNASATIII (1) amount by several times (Fig. 6). As shown earlier, decondensation and transcription of heterochromatin during replicative senescence (when oxidative stress is increased) were observed for SatIII (1) fragment only, but not for SatIII (9) fragment [34].

The different patterns of stress response found in the two satellite III fragments seem to be associated with different mechanisms of transcription activation in these chromatin regions. For SatIII (9), a transcription inducer is heat shock factor [31, 32, 33]. A factor that triggers SatIII (1) transcription under oxidative stress and blocks transcription under heat shock is still unknown.

A bulk of data obtained in recent decades has shown that SatIII (1) repeat CNV in the same cell pool is a process typical for each type of cultivated and primary cells. The augmentation of SatIII (1) copy numbers after a stress impact of medium intensity and the decline after a stronger impact was earlier observed for cultivated human blood lymphocytes, mesenchymal stem cells, and skin fibroblasts [35, 39]. Similar changes were observed in healthy human blood leucocytes under an affective stress [52]. On the peak of stress development, SatIII (1) copy number increased and TR copy number decreased in the leucocyte DNA. After the stress reverse changes were observed in the contents of the two types of DNA repeats. The same regularity was observed in the cells from various human brain regions – a negative correlation between SatIII (1) and TR contents. A chronic stress favors elimination of the cells with high SatIII (1) content from blood in vivo. Perhaps that is why blood leucocyte DNA isolated from SZ patients contained low copy numbers of SatIII (1) [62].

We believe that the increase in SatIII (1) content in some cells of the pool is a manifestation of a general mechanism that is aimed at division arrest in cells with damaged chromatin. A high content of SatIII (1) in the cell is associated with blocking changes in the chromatin structure in response to proliferative and other stimuli. Such cells are subject to being eliminated from the pool; the more actively, the higher the stress in the body. At a relatively low level of stress, these cells accumulate in the pool during senescence. These cells are harmful ballast, as they can disrupt the functional activity of the entire cell population. This is especially true for cells of the nervous tissue that are involved in signal transmission along the cell chain.

It remains unknown what causes the arrest of chromatin mobility under stress conditions, and we do not consider the large size of SatIII (1) as the major cause of arrest. Other satellite chromatin fragments seem to be also involved in a similar scheme of CNVs. This mechanism warrants further research. First, answers to the following questions should be obtained.

(1) How does stress increase the size of the SatIII (1) repeat block in some cells of a normal (non-cancer) cell pool? The transcription from the satellite repeats is launched in most cells of the pool, but just a small fraction of the cells demonstrates an increased abundance of the repeat. One can assume that these cells may have an elevated level of reverse transcriptase, or their 1q12 region contains an increased number of chromatin break sites, where the hybrid molecules can be inserted.

(2) Are there any CNVs of other repeats, such as SatIII (9), during senescence and stress?

(3) What is the difference between the transcription activation mechanisms of the two fragments of satellite III, which are localized on chromosomes 1 and 9? Why is SatIII (1) transcription activation linked to SatIII (9) transcription suppression?