Ciprofloxacin causes oxidative stress that leads to aging-like changes in human skin cells grown in the lab

To investigate the effects of the selected fluoroquinolones on cellular senescence, we used 200 μM CIP, 200 μM MOX, and 500 μM LEV. These concentrations were selected on the basis of the resazurin reduction assay after 7 days of exposure in primary human fibroblasts (Supplementary Figure S1). CIP, but not MOX or LEV, significantly increased SA-β-Gal activity per cell, the senescence index (SI), which reflects cellular autofluorescence and cell size, and lysosomal content (Figure 1a–d). MOX and LEV produced a modest reduction in relative SA-β-Gal activity but did not affect SI or lysosomal content (Figure 1a–d). We have additionally tested 25, 50, 100, 200 and 400 μM CIP on human fibroblasts and found that only 200 and 400 μM increased SA-β-Gal activity (Supplementary Figure S2).

We next examined whether CIP similarly affects SA-β-Gal activity in immortalized cell lines used in senescence research. CIP increased SA-β-Gal activity in the human endothelial EA.hy926 cell line () and in H9c2 rat cardiomyoblasts (). Supplementary Figure S3a,b Supplementary Figure S3c,d

We then analyzed mRNA expression of senescence-associated stress-response and cell-cycle-regulatory genes in human fibroblasts, including ANKRD1, TRIM21, CDKN2A/p16, CDKN1A/p21, and TP53/p53, together with cytokine- and stress-related genes GDF15, IL1B, IL6, TNF, CXCL2, and CXCL8/IL8 (Figure 2). For IL1B and TNF, RT-qPCR Ct values exceeded 33 cycles, precluding reliable quantification of transcript levels.

Among the cell-cycle-regulatory markers, CIP significantly increased CDKN1A/p21 expression by approximately 3-fold (p < 0.05), whereas MOX and LEV had no detectable effect on this gene (Figure 2b). CDKN2A/p16 expression showed a moderate upward trend after CIP and MOX treatment, but neither change reached statistical significance (Figure 2a). TP53/p53 expression was not increased by any of the tested fluoroquinolones and instead tended to decrease after treatment (Figure 2c).

The largest transcriptional changes occurred in stress-response genes. CIP increased ANKRD1 expression by approximately 2.5–3-fold, whereas MOX produced an approximately 6-fold increase; LEV had no significant effect (Figure 2d). TRIM21 expression also increased, most prominently after CIP treatment, which elevated TRIM21 mRNA by approximately 1.7-fold (p < 0.001). MOX induced a smaller but significant increase, whereas LEV had little or no effect (Figure 2e). GDF15 expression increased after treatment with all three fluoroquinolones, reaching approximately 2–3 times the control level (Figure 2h).

Fluoroquinolone treatment did not upregulate expression of the analyzed pro-inflammatory cytokine genes. IL6 expression was significantly reduced after MOX and LEV exposure (Figure 2f), while CXCL2 expression decreased after CIP and LEV treatment, with a similar downward trend after MOX exposure (Figure 2g). Thus, although CIP induced several stress- and senescence-associated genes, including CDKN1A/p21, ANKRD1, TRIM21, and GDF15, none of the tested fluoroquinolones elicited a broad pro-inflammatory SASP-like transcriptional response under these conditions (Figure 2).

We next determined whether the transcriptional changes were accompanied by increased secretion of SASP-associated cytokines. After 7 days of exposure to the selected antibiotics, secretion of IL-1β, IL-6, and CXCL8/IL-8 remained largely unchanged (Figure 3a). After 14 days of CIP treatment, however, IL-1β secretion increased modestly by approximately 1.3-fold (p < 0.05) (Figure 3b), indicating delayed induction of at least one SASP component.

Because fluoroquinolones can inhibit cellular topoisomerases in vitro and induce DNA damage [20], we assessed phosphorylation of histone H2AX at Ser139 (γH2AX), a widely used marker of the DNA damage response, particularly to DNA double-strand breaks. CIP significantly increased both the proportion of γH2AX-positive cells and the mean number of γH2AX foci per nucleus (p < 0.01 for both comparisons), whereas MOX and LEV produced no significant effects (Figure 4).

We next evaluated the temporal development of oxidative stress in CIP-treated human fibroblasts using the redox-sensitive probes dihydroethidium (DHE) and 2′,7′-dichlorodihydrofluorescein diacetate (DCFH-DA), together with the MitoCLox fluorescence ratio (520/590 nm) as a measure of mitochondrial lipid peroxidation. A preliminary time-course experiment showed that the DHE signal increased after 72 h of CIP exposure (Figure 5a), whereas the DCF signal increased earlier, after 6 h of incubation (Figure 5b). The MitoCLox fluorescence ratio remained unchanged (Figure 5c). Based on these kinetics, we selected 72 h for DHE measurements and 24 h for DCF measurements in subsequent experiments.

We examined whether antioxidant co-treatment could attenuate CIP-induced DNA damage response. After 7 days, CIP increased the proportion of γH2AX-positive nuclei from approximately 2% in control cells to approximately 20–22% (Figure 6a,b). Co-treatment with 80 nM SkQ3 or 100 μM Trolox reduced γH2AX positivity to approximately 10–12%, indicating attenuation of the CIP-induced γH2AX response. Consistently, CIP markedly increased the mean number of γH2AX foci per nucleus, whereas both SkQ3 and Trolox significantly reduced this increase (Figure 6c). Thus, both the mitochondria-targeted antioxidant SkQ3 and the water-soluble antioxidant Trolox attenuated CIP-induced γH2AX accumulation in human fibroblasts.

We next examined the effects of antioxidant co-treatment on CIP-induced oxidative stress (Figure 7). After 72 h, CIP increased DHE fluorescence by approximately 1.5-fold, consistent with increased superoxide-associated oxidative stress (Figure 7a). Co-treatment with either Trolox (100 μM) or the mitochondria-targeted antioxidant SkQ3 (80 nM) reduced DHE fluorescence by approximately 2-fold relative to CIP-treated cells (p < 0.05), resulting in values slightly below the untreated control level. DCF fluorescence also increased after 24 h of CIP exposure, but the two antioxidants had different effects on this signal. Trolox reduced the CIP-induced increase in DCF fluorescence to approximately the control level, whereas SkQ3 further increased DCF fluorescence (Figure 7b).

We also assessed mitochondrial lipid oxidation using MitoCLox. CIP did not significantly alter the MitoCLox signal after either 24 or 72 h of exposure (Figure 5c), indicating that the oxidative stress detected by DHE and DCFH-DA was not accompanied by a measurable increase in mitochondrial lipid peroxidation under these conditions. Both SkQ3 and Trolox, however, reduced the basal MitoCLox signal, consistent with lower steady-state levels of oxidized mitochondrial lipids (Figure 7c).

We next tested whether antioxidant co-treatment could attenuate CIP-induced senescence-like changes. Both DOX and CIP increased SA-β-Gal staining, whereas co-treatment with SkQ3 or Trolox reduced SA-β-Gal staining in CIP-treated cells (Figure 8a,b). CIP also markedly increased the senescence index, and both antioxidants reduced this parameter to values close to the untreated control level (Figure 8c). LumiTracker Lyso Red fluorescence increased significantly after CIP treatment, while co-treatment with SkQ3 or Trolox produced a downward trend (Figure 8d). Collectively, these data indicate that antioxidant co-treatment attenuates multiple features of CIP-induced senescence-like remodeling in human fibroblasts.

It should be noted that the data shown in Figure 8 were derived from an independent experimental series (different donor fibroblast preparation and imaging session) from those presented in Figure 1. Accordingly, absolute values of integrated density and senescence index differ between the two figures, reflecting inter-experiment biological variability; however, the relative pattern—CIP and DOX increasing SA-β-Gal activity and SI compared to untreated control—is consistent across both datasets.

Western blotting was used to determine whether CIP-induced senescence-like changes were accompanied by alterations in canonical senescence-associated cell-cycle regulators (Figure 9). CIP did not increase p53, p16, or p21 protein levels relative to untreated control cells. Trolox co-treatment produced a tendency toward higher p53, p16, and p21 levels, whereas SkQ3 did not show a similar effect. DOX, used as a positive control, increased all these proteins. In contrast, RB phosphorylation at Ser795 decreased after CIP treatment and was further reduced in cells treated with CIP plus SkQ3 as well as in cells treated with DOX alone. These data indicate that CIP-induced senescence-like remodeling occurs without a robust increase in p53, p16, or p21 protein abundance but could be associated with reduced RB phosphorylation.

To determine whether the effect of prolonged CIP exposure persisted after drug withdrawal, we performed washout experiments. The fibroblasts for this experiment were initially seeded at ~50% confluency. Continuous CIP exposure for 3 weeks reduced resazurin fluorescence to approximately 50% of the untreated control (Figure 10). In contrast, cells exposed to CIP for 1 week followed by a 2-week drug-free period, or for 2 weeks followed by a 1-week drug-free period, exhibited resazurin signals comparable to those of untreated cells. Consistent with these measurements, routine bright-field microscopy revealed a visibly lower cell density after continuous 3-week CIP exposure, whereas cells cultured in CIP-free medium after drug withdrawal progressively repopulated the culture surface and reached an approximately confluent monolayer by the end of the recovery period. Together, these observations suggest that the growth arrest induced by prolonged CIP exposure is not maintained after CIP removal.

MitoCLox, a ratiometric fluorescent probe sensitive to lipid peroxyl radicals, was synthesized from the succinimidyl ester of C11-BODIPY581/591 and (5-[(4-aminobutyl)amino]-5-oxopentyl)triphenylphosphonium bromide as described previously [40]. SkQ3 was kindly provided by LLC “Institute of Mitoengineering, MSU”. CM-H2DCFDA, and dihydroethidium (DHE) were purchased from Invitrogen Life Technologies (Waltham, MA, USA). LumiTracker^® Lyso Red was from Lumiprobe Corporation (Hannover, Germany). Unless otherwise indicated, all other reagents were obtained from Sigma-Aldrich (St. Louis, MO, USA).

Rat H9c2 cardiomyoblasts, a spontaneously immortalized cell line (EcACC Cat. No. 88092904), and the human endothelial cell line EA.hy926 were from ATCC collection (Cat. No. CRL-2922), and primary human skin fibroblasts were from three healthy male volunteers obtained from the Common Use Center “Biobank” (Research Centre for Medical Genetics, Moscow, Russia). The donors were healthy adults aged 28–35 years with no known chronic diseases. The fibroblasts were cultured in Dulbecco’s modified Eagle’s medium (DMEM, 4.5 g/L D-glucose, 25 mM; Gibco, Thermo Fisher Scientific, Waltham, MA, USA) supplemented with 2 mM glutamine and 10% fetal bovine serum (FBS; HyClone, Logan, UT, USA). Experiments were performed at passages comprised between 2 and 7. Cells were seeded in experimental plates at approximately 40% confluence and allowed to reach 50–60% confluence before treatment, unless indicated otherwise. Antibiotics and antioxidants were then added for the indicated exposure periods and were not replenished during incubation. For the positive-control condition, DOX was removed after 24 h by replacing the medium with fresh DOX-free medium. Cell confluence reached approximately 100% by the end of the treatment period. In the 14-day experiments, cells were not passaged, and the culture medium was not replaced after seeding. Ciprofloxacin, moxifloxacin, and levofloxacin were obtained as commercial solutions for intravenous infusion from PJSC KrasPharma (Krasnoyarsk, Russia). The ciprofloxacin infusion solution (2 mg/mL) contained sodium chloride (9 mg/mL), lactic acid (0.65 mg/mL), and disodium edetate dihydrate (0.20 mg/mL) as excipients (pH 3.5–4.6). The moxifloxacin infusion solution (1.6 mg/mL) contained sodium chloride (8 mg/mL) as the sole excipient (pH 4.1–4.6). The levofloxacin infusion solution (5 mg/mL) contained sodium chloride (9 mg/mL) as the sole excipient (pH 4.3–5.3). At the working concentrations used in this study (200 μM CIP, 200 μM MOX, 500 μM LEV), the final concentrations of excipients in the culture medium were: NaCl, 5.1–6.9 mM (added to ~110 mM already present in DMEM); lactic acid, 0.24 mM (CIP only); and disodium edetate, 0.018 mM (CIP only). The pH of the culture medium was not altered by the addition of the infusion solutions at these dilutions, as expected from the ~20–30-fold dilution in bicarbonate-buffered DMEM and confirmed by visual inspection of phenol red color. Cells with no treatment were used as control cells.

SA-β-Gal activity was measured using a histochemical staining kit (Servicebio, Wuhan, China, Cat. No. G1073-100T) according to the manufacturer’s instructions. Cells were fixed and incubated with X-Gal staining solution at pH 6.0 at 37 °C for 4–12 h (the presence or absence of staining was regularly monitored using microscopy). Incubation time was the same for each condition within biological replicate. Bright-field images of SA-β-Gal-stained cells and corresponding Hoechst 33342 nuclear images were captured using an Olympus IX83 microscope (Olympus Corporation, Tokyo, Japan). For each condition, 3 (for 4× magnification) or 10 (for 20× magnification) randomly selected fields were imaged per biological replicate. ImageJ (NIH) was used to measure the integrated density of the SA-β-Gal signal in each field, and the nuclear count was determined from the Hoechst image of the same field. SA-β-Gal activity was expressed as integrated density per cell (integrated density/nuclear count). Multiple fields from each sample were averaged to obtain one value per biological replicate. The same quantification procedure and unit notation were applied to all cell types (primary human fibroblasts, EA.hy926, and H9c2) across all figures.

Cell viability/cytotoxicity was assessed by the resazurin reduction assay using CellTiter-Blue^® reagent (Promega, Madison, WI, USA) according to the manufacturer’s protocol. Fluorescence was measured with a Fluoroskan Ascent FL microplate reader (Thermo Labsystems, Waltham, MA, USA).

Flow cytometric measurements were performed using an Amnis^® FlowSight^® Imaging Flow Cytometer (Luminex, Austin, TX, USA) provided by the Moscow State University Development Program PNR5. Cells were detached with trypsin/Versene solution (1:1), centrifuged at 1000× g for 5 min at 4 °C, and resuspended in 50 µL PBS. Mean fluorescence intensity (MFI) was measured on a per-cell basis for all probes. Data were analyzed using IDEAS software version 6.3 (Luminex). The gating cascade was applied in the following order: (1) single cells were selected using a plot of bright-field area versus aspect ratio to exclude doublets and debris; (2) focused cells were selected from the single-cell population using the gradient RMS feature of the bright-field image, with a threshold of 50 to exclude out-of-focus events (gradient RMS is a texture-based sharpness metric computed as the root mean square of the local intensity gradient across image pixels); and (3) the final gated population, defined as the intersection of the single-cell and focus gates, was used for all downstream fluorescence measurements. The same gating template and thresholds were applied identically to all experimental groups within each experiment. Gates were defined on the control (untreated) population and applied uniformly to all samples. The representative plots for the imaging flow cytometry analyses are provided in Supplementary Figure S4.

The senescence index was determined by flow cytometry from the increases in cell size and lipofuscin-associated autofluorescence characteristic of senescent cells, as described previously [19]. SI is an empirical composite index calculated from normalized cell diameter (nD) and normalized autofluorescence (nAF), relative to proliferating control cells, using the formula: SI = ((nAF − 1) + 5 × (nD − 1))/2. The weighting factor of 5 for the diameter reflects the empirical observation that senescent cells exhibit much smaller increase in diameter (~10%) rather than increase in autofluorescence (~50%), so the 5× multiplier is introduced to equalize the contribution of the two parameters. Dividing to a coefficient of 2 is necessary to balance the contributions of the two parameters.

To label lysosomes, primary human fibroblasts were incubated with 50 μM LumiTracker^® Lyso Red for 15 min before flow cytometric analysis.

For oxidative stress measurements, primary human fibroblasts were incubated with 2 μM DCFH-DA or 2 μM DHE for 30 min before flow cytometric analysis. The mitochondria-targeted ratiometric fluorescent probe MitoCLox, which is sensitive to lipid peroxyl radicals, was used as described previously [40]. For all probes, unstained cells (without probe) were analyzed in parallel to establish baseline autofluorescence, and this background was subtracted from probe-treated samples. Cells treated with the probes but without fluoroquinolone served as the untreated control. For MitoCLox, which is a ratiometric probe (520/590 nm fluorescence ratio), the ratio inherently corrects for variable probe loading, as both the oxidized and reduced forms of the probe are measured simultaneously in the same cell.

After incubation with the indicated compounds, cells were lysed in hot sample buffer (62.5 mM Tris-HCl, pH 6.8; 2% SDS; 10% glycerol; 50 mM DTT; 0.01% bromophenol blue) for 5 min at 94 °C. Proteins were separated by SDS-PAGE and transferred to a PVDF membrane (Bio-Rad, Hercules, CA, USA). Nonspecific binding sites were blocked with 5% BSA. Membranes were incubated overnight at 4 °C with primary antibodies diluted in 5% BSA in TBST (25 mM Tris, pH 7.4; 0.15 M NaCl; 0.1% Tween 20), followed by HRP-conjugated secondary antibodies in 5% BSA/TBST for 1 h at room temperature. Chemiluminescence was developed using West Dura Extended Duration Substrate (Thermo Fisher Scientific, USA), and images were acquired with a ChemiDoc™ gel imaging system (Bio-Rad, USA). Densitometric analysis was performed using Image Lab software v.6.0.1. Signals for proteins of interest were normalized to total protein content measured by 2,2,2-trichloroethanol fluorescence as described in [41,42].

Total cellular RNA was isolated using the Quick-RNA MiniPrep kit (Zymo Research, Irvine, CA, USA) according to the manufacturer’s protocol. RNA concentration was determined with a NanoDrop ND-1000 spectrophotometer (Thermo Fisher Scientific, USA), and RNA quality was verified by electrophoresis. Reverse transcription was performed using the REVERTA-L reagent kit (AmpliSens, Moscow, Russia) according to the manufacturer’s protocol. The resulting cDNA was analyzed by real-time PCR using EvaGreen I intercalating dye (Syntol, Moscow, Russia). The following primers were used: IL-6 (forward: 5′-GACAACTCATCTCATTCTG-3′; reverse: 5′-CCATTAACAACAACAATCTG-3′); CXCL2 (forward: 5′-AACTGCGCTGCCAGTGCT-3′; reverse: 5′-CCCATTCTTGAGTGTGGCTA-3′); GDF15 (forward: 5′-GTTCCTGAGACACCCGATT-3′; reverse: 5′-AATACACAGTTCCATCAGACCA-3′); ANKRD1 (forward: 5′-AGTAGAGGAACTGGTCACTGG-3′; reverse: 5′-TGGGCTAGAAGTGTCTTCAGAT-3′); TRIM21 (forward: 5′-ACCCCCTAAAGGTCTCCACA-3′; reverse: 5′-ATCATTGTCAAGCGTGCTGC-3′); p16 (CDKN2A) (forward: 5′-GGGGGCACCAGAGGCAGT-3′; reverse: 5′-GGTTGTGGCGGGGGCAGTT-3′); p21 (CDKN1A) (forward: 5′-GACACCACTGGAGGGTGACT-3′; reverse: 5′-CAGGTCCACATGGTCTTCCT-3′); p53 (TP53) (forward: 5′-ACCACCATCCACTACAACTACAT-3′; reverse: 5′-ACAAACACGCACCTCAAAGC-3′); GAPDH (forward: 5′-GATGACATCAAGAAGGTGGTG-3′; reverse: 5′-GCTGTAGCCAAATTCGTTGTC-3′); ACTB (forward: 5′-CATCGAGCACGGCATCGTCA-3′; reverse: 5′-TAGCACAGCCTGGATAGCAAC-3′); RPL32 (forward: 5′-CATCTCCTTCTCGGCATCA-3′; and reverse: 5′-AACCCTGTTGTCAATGCCTC-3′). Each sample was analyzed in triplicate (technical replicates) and the values were averaged before normalization, and a no-template control was included in each run. PCR efficiency (E) was determined from standard curves using the equation E = 10^(−1/slope). Target gene mRNA levels were normalized to three reference genes: RPL32, ACTB, and GAPDH according to [43].

Cells were seeded on glass coverslips in 6-well plates at 100,000 cells per well. After washing with DMEM, cells were fixed with 2% paraformaldehyde in DMEM for 10 min at 37 °C and permeabilized with 0.2% Triton X-100 in PBS for 5 min at room temperature. Cells were incubated overnight at 4 °C with a monoclonal anti-phospho-Histone H2AX (Ser139) antibody (sc-517348; Santa Cruz Biotechnology, Dallas, TX, USA). Secondary staining was performed for 1 h at room temperature using cross-adsorbed goat anti-mouse IgG (H+L) conjugated to Fluor647 (S0014; Affinity Biosciences, Changzhou, China) together with Hoechst 33342 (Biotium, Fremont, CA, USA). After washing with PBS, coverslips were mounted using Aqua-Poly/Mount (Polysciences, Warrington, PA, USA). Fluorescence microscopy was performed using an Olympus IX83 microscope (Olympus Corporation, Tokyo, Japan). Quantification of γH2AX foci was performed using Fiji/ImageJ2 software Version: 2.16.0/1.54p. Cell nuclei were segmented based on Hoechst 33342 fluorescence. Background fluorescence was subtracted from the γH2AX channel, followed by application of an identical intensity threshold to all experimental groups. γH2AX-positive cells were identified by visual scoring of nuclei with γH2AX staining clearly above background. Individual γH2AX-positive foci located within the boundaries of each nucleus were identified and counted using particle analysis. The number of γH2AX foci was determined separately for each nucleus, and the mean number of foci per nucleus was calculated for each experimental group. For each experimental group, γH2AX foci were quantified in 10 randomly selected fields of view in each of the three independent experiments. Images from all groups were acquired and analyzed using identical acquisition and image-processing parameters.

Relative levels of IL-1β, IL-6, and IL-8/CXCL8 in culture supernatants from primary human fibroblasts were quantified using commercial ELISA kits (Vector-Best, Koltsovo, Russia) according to the manufacturer’s instructions. Before analysis, supernatants were centrifuged at 3000× g for 10 min.

Unless otherwise indicated, n refers to the number of independent biological replicates (i.e., independent experiments performed with cells from different donors or independent cell preparations). For each experiment, cells from each donor were seeded independently and treated in parallel. No experiment was repeated using cells from the same donor at different passages to generate additional “biological replicates” unless otherwise indicated. Technical replicates (e.g., triplicate qPCR reactions, multiple microscopy fields per sample) were averaged to yield a single value per biological replicate before statistical analysis. For qPCR, ELISA, and flow cytometry the number of technical replicates is 3. For Western immunoblotting the number of technical replicates is 2. In microscopy, unless the opposite mentioned, the number of technical replicates (different fields of view within a single biological replicate) varies depending on the magnification at which the image was captured: for 4× magnification, the number of fields of view was 3; for 20× magnification, it was at least 10. The number of biological replicates (n) and the statistical tests used for each experiment are specified in the corresponding figure legends. Also, unless otherwise indicated, data are presented as mean ± SEM. Statistical analyses and graph preparation were performed using GraphPad Prism 9 (GraphPad Software, San Diego, CA, USA). Given the small sample size (n = 3 per group), formal normality and variance homogeneity testing have insufficient statistical power to be informative. Parametric one-way ANOVA was used based on the conventional assumption of approximately normal distributions for in vitro quantitative data and on the robustness of ANOVA to moderate deviations from normality and variance heterogeneity in equal group sizes. One-way ANOVA was followed by Dunnett’s post hoc test for comparisons of each treatment group against the untreated control, and Bonferroni-corrected pairwise comparisons for comparisons between CIP-treated and antioxidant co-treated groups. A p value < 0.05 was considered statistically significant. Because each biological replicate corresponds to a different donor, the reported variability (SEM) reflects inter-donor variability. Formal modeling of donor-to-donor variability (e.g., mixed-effects analysis) was not performed due to the limited sample size.

The original contributions presented in this study are included in the article/. Further inquiries can be directed to the corresponding author. Supplementary Materials