Cisplatin-induced peripheral neuropathy is associated with neuronal senescence-like response

Abstract

Background

Cisplatin-induced peripheral neuropathy (CIPN) is a frequent serious dose-dependent adverse event that can determine dosage limitations for cancer treatment. CIPN severity correlates with the amount of platinum detected in sensory neurons of the dorsal root ganglia (DRG). However, the exact pathophysiology of CIPN is poorly understood, so the chance of developing neuroprotective treatment is reduced. The aim of this study was to determine the exact mechanisms involved in CIPN development.

Methods

By single-cell RNA-sequencing (scRNAseq), we have studied the transcriptomic profile of DRG sensory neurons from a well-characterized neurophysiological mouse model of CIPN.

Results

Gene Ontology analysis of the scRNAseq data indicated that cisplatin treatment induces the upregulation of biological pathways related to DNA damage response (DDR) in the DRG neuronal population. Moreover, DRG neurons also upregulated the Cdkn1a gene, confirmed later by the measurement of its protein product p21. While apoptosis activation pathways were not observed in DRG sensory neurons of cisplatin-treated mice, these neurons did express several senescence hallmarks, including senescence-associated β-galactosidase, phospho-H2AX, and nuclear factor kappa B (Nfkb)–p65 proteins.

Conclusions

In this study, we determined that after cisplatin-induced DNA damage, p21 appears as the most relevant downstream factor of the DDR in DRG sensory neurons in vivo, which survive in a nonfunctional senescence-like state.

Key Points.

1. Cisplatin induces DDR activation and p21 upregulation in mouse DRG sensory neurons.

2. Cisplatin-induced DDR activation and p21 upregulation do not drive apoptosis but a senescence-like phenotype in mouse DRG sensory neurons.

Importance of the Study.

This is the first study demonstrating that the DNA damage induced by cisplatin treatment does not induce apoptosis but a senescence-like response in DRG sensory neurons in vivo. It proposes a new mechanism of CIPN development that may lead to new approaches in the prevention or even treatment of this clinical issue.

Chemotherapy-induced peripheral neuropathy (CIPN) is a well-known and unresolved adverse event of drugs widely used to treat prevalent cancers.¹ The development of CIPN is dose limiting and can lead to treatment withdrawal. As a consequence, CIPN has a direct impact on patients’ survival and quality of life due to its long-lasting nature.² Moreover, the management of patients with CIPN imposes a relevant economic burden for health systems.³

Among neurotoxic cytostatic drugs causing CIPN, platinum compounds are among the most widely used. Despite the successful emergence of new approaches to cancer treatment, platinum drugs are still the cornerstone treatment in several clinical settings for highly prevalent tumors, both in children and in adulthood.⁴ Cisplatin, the most common platinum drug, induces a progressive pure sensory neuropathy in a typical sock-and-glove distribution. Symptomatology ranges from mild-moderate, with decreased vibratory sensitivity, numbness, paresthesia, and hearing loss, to more severe symptoms like ataxic gait or fine sensory-motor impairment that severely impacts on self-care and daily working capabilities.^4,5

Prevention or treatment of platinum-induced peripheral neuropathy is still an unmet clinical need.^5–7 A major reason for this lack of effective treatment is the incomplete knowledge of the pathogenesis of this peripheral neuropathy. Data from neoplastic cells indicate that cisplatin-induced genotoxic lesions (formation of platinum DNA adducts) and increased levels of oxidative stress are the main mechanisms in inhibiting cell division.⁸ On the other hand, initial observations of the peripheral nervous system showed a correlation between the severity of neuropathy and the amount of platinum in the patients’ dorsal root ganglia (DRG) cells.⁹ Since then, preclinical studies have attempted to elucidate the mechanisms underlying this primary sensory neuropathy. A constellation of multiple effector molecular pathways leading to DRG neuron death has been studied. Defects in nuclear and mitochondrial DNA platinum adduct repair as well as the increase of oxidative cellular stress leading to apoptosis are the main factors reported in in vitro experiments. However, mechanisms of neuronal death by apoptosis in animal models of cisplatin-induced neuropathy are poorly supported.¹⁰ In any case, an integrative functional network encompassing the more relevant findings is still missing. This fact and the incomplete knowledge of pathogenic factors may explain the unsuccessful transfer from preclinical results to therapeutic clinical trials.⁶

Therefore, we sought to further decipher the neuropathogenic mechanisms underlying cisplatin neurotoxicity in a well-characterized mouse model of cisplatin-induced neuropathy that closely mimics the clinical features observed in patients. By single-cell RNA sequencing (scRNAseq) of DRG cells we found that cisplatin upregulated the cyclin-dependent kinase inhibitor 1a (Cdkn1a) gene in sensory neurons. The product of this gene, the p21 protein, has multiple divergent roles, including apoptosis modulation or senescence induction.¹¹ Therefore, we evaluated morphological and molecular hallmarks of these two pathways.¹² Altogether, our data indicate that neurons exposed to cisplatin, in response to the suffered DNA damage, activate p21 and survive in a senescence-like phenotype with an impaired functional status.

Materials and Methods

Animals and Treatment Schedule

Ten-week-old female BALB/cAnNCrl (BALB/c) mice (19–22 g on arrival; Janvier) were intraperitoneally (i.p.) injected with cisplatin (Selleckchem) once a week until they reached a total cumulated dose of 42 mg/kg. This dose is equivalent to human doses in which neuropathy starts developing, according to published allometric dose translations.^5,13 The cisplatin administration schedule was planned to give 7 mg/kg cisplatin once a week until reaching the total cumulated dose. However, mice treated with such cisplatin dosage experienced a weight loss ≥10% after the second cisplatin administration. Therefore, the dose was reduced to a half (3.5 mg/kg) once a week from weeks 2–8 (Figure 1A). After each administration, mice received a subcutaneous (s.c.) injection of 1 mL saline to prevent cisplatin-induced nephrotoxicity. As a control group, mice received an i.p. injection of saline solution once a week for 10 weeks. All the animal studies were approved by the ethics committee of the Autonomous University of Barcelona.

Fig. 1.

Characterization of the CIPN mouse model. (A) Experimental scheme for the development of the CIPN model. Cisplatin (CDDP) was administred i.p. once a week for 10 weeks (neuropathy induction time). After this time, animals were still evaluated to assess the coasting effect (coasting-effect evaluation time). (B–F) Functional tests. SNAP (B) and caudal CNAP (C and D) decreased along time in the cisplatin group, whereas CMAP was not affected (E). Mice treated with cisplatin developed mechanical allodynia along time (F). n = 20–25 mice per group, repeated measures two-way ANOVA test with Bonferroni post-hoc test for multiple comparisons. *P < 0.05 vs Control; $P < 0.05 vs Basal. (G–I) Histological studies (n = 4–6 mice per group). The number of myelinated axons in the sciatic nerve (G) and in the tibial nerve (H) and the intra-epidermal nerve fiber density (IENFD) of PGP+ fibers was not affected by cisplatin. One-way ANOVA. Data are represented as group mean ± SD.

Functional Tests

Functional tests were performed during all the induction time (from 0 to 9 week) and up to 6 weeks after the last dose of cisplatin administration (from 10 to 16 week) in order to evaluate the “coasting effect,” the maintenance or worsening of the neuropathy after withdrawal of treatment (Figure 1A).

Nerve conduction studies

To assess sensory and motor nerve conduction, the sciatic nerve was stimulated percutaneously through a pair of needle electrodes placed at the sciatic notch and at the ankle.¹⁴ The compound muscle action potential (CMAP) was recorded by microneedle electrodes placed at the plantar muscle. The sensory nerve action potential (SNAP) was recorded from the fourth toe near the digital nerve. The caudal compound nerve action potential (caudal CNAP) was also recorded by placing a pair of recording needle electrodes at the base of the tail and a couple of stimulating needle electrodes at 3.5 cm or at 5 cm distally to the recording points. Electrophysiological tests were performed before starting the treatment (baseline) and every 2 weeks during all the follow-up time.

Algesimetry test

Mechanical allodynia was tested by using an electronic Von Frey test device (IITC Life Science) at baseline and every 3 weeks. The mean withdrawal pressure of 3 applications to the left foot was calculated for each animal, and group means were calculated. Detailed protocol is available in the Supplementary Methods.

Single-Cell RNA-Sequencing

At 10 weeks of the study, 5 control and 5 cisplatin-treated mice were sacrificed and their DRGs processed for scRNAseq analysis. Only those cisplatin-treated mice that had a decrease >15% in their SNAP or CNAP were used for the scRNAseq. A detailed protocol of the procedure and data analysis can be found in the Supplementary Methods.

Single-cell sorting of the DRG

DRG cells from control and cisplatin-treated mice were labeled against Tyrosine receptor kinase (Trk) and single-cell sorted into 96-well plates.

RNA-isolation and library construction

Full-length scRNAseq libraries of each isolated cell were prepared using the Smart-seq2 protocol¹⁵ with minor modifications.

Single-cell RNAseq analysis, cell type clustering, and transcriptome analysis

Single-cell RNAseq data were analyzed using the Seurat pipeline¹⁶ implemented in the R language¹⁷ by the Seurat package for single-cell analysis.

Histological Methods

Histological studies were performed at 10 and 16 weeks of study to further describe the animal model and to corroborate the data obtained with scRNAseq.

Myelinated axons density

A segment of the sciatic nerve at mid-thigh and the distal part of the tibial nerve at the ankle were processed for microthin sectioning as described in the Supplementary Methods. To estimate the number of myelinated fibers in the sciatic and tibial nerves, axons were counted in images taken on a light microscope. At least 30% of the nerve cross-section area was analyzed.

Intra-epidermal nerve fiber density

Cryotome distal plantar pads sections were immunostained with anti-protein gene product 9.5 (PGP9.5) or rabbit anti-calcitonin gene related peptide (CGRP) antibodies as detailed in the Supplementary Methods. The number and density of nerve fibers present in the epidermis of the paw pads was quantified in an epifluorescence microscope (Olympus BX51).

Transmission electron microscopy of the DRG

DRG cells were processed for their analysis by transmission electron microscopy as detailed in the Supplementary Methods. DRG ultrathin sections were observed with a Joel 1400 transmission electron microscope equipped with a Gatan Ultrascan ES1000 charge-coupled camera.

DRG immunofluorescence

DRG paraffin sections were immunostained with primary antibodies: anti-p21 (1:200, Abcam), rabbit anti–H2AX-pSer139 (1:200, Cell Signaling), rabbit anti–nuclear factor kappa B (Nfkb)–p65 (1:200, Cell Signaling), and mouse anti–β-III tubulin (1:500, Hybridoma), as detailed in the Supplementary Methods. The percentages of neuronal nuclei positive for p21, p-H2AX, and Nfkb-p65 were analyzed in 80–100 neurons for each animal by ImageJ software.

Senescence-associated β-galactosidase activity assay

The senescence-associated β-galactosidase activity was determined in DRG slices following the protocol instructions of the Senescence Detection Kit (Abcam) and detailed in the Supplementary Methods. The intensity of the senescence-associated β-galactosidase reaction (seen as blue precipitate) was quantified in a total of 50 neurons for each animal using ImageJ software.

Western Blot Analysis

Thirty micrograms of protein were loaded in 15% sodium dodecyl sulfate–polyacrylamide gels. After blockade, membranes were incubated overnight at 4°C with primary antibodies. Horseradish peroxidase–coupled secondary antibody incubation was performed for 90 minutes at room temperature and membranes visualized using the enhanced chemiluminescence method with the Clarity Western ECL Substrate (Bio-Rad). Images were collected using a chemidoc apparatus. Western blots were then analyzed using the Lane and band plug-in from Image Lab software (Bio-Rad). Further details are explained in the Supplementary Methods.

Statistical Analysis

The results of functional tests are expressed as a percentage with respect to baseline values for each mouse, and statistical analysis was performed by using a repeated measures two-way ANOVA test. For the other comparisons, a one-way ANOVA test was used. The Bonferroni post hoc test was applied when needed. GraphPad Prism v8.4.0 software was used for statistical inference analysis and graphically represents the data, which are expressed as group mean ± SD. Differences among groups or time points were considered significant at P < 0.05.

Results

Cisplatin-Treated Mice Develop a Painful Peripheral Sensory Neuropathy

Mice treated with cisplatin experienced a significant progressive decrease in the amplitudes of SNAPs recorded in the digital nerves (Figure 1B). Similarly, the mixed sensory-motor CNAP recorded in the proximal and distal caudal nerves also experienced a significant progressive decrease over time in the treated group (Figure 1C, D). The decrease in the digital and caudal CNAPs started at 10 weeks of study and gradually worsened over time, with a maximum negative peak at 16 weeks of follow-up. In contrast, the amplitude of the CMAP recorded at the plantar muscle in cisplatin-treated mice was maintained equivalent to control mice (Figure 1E). In the algesimetry test, cisplatin-treated animals had significantly reduced withdrawal thresholds to mechanical stimulation applied to the plantar paw, indicative of hyperalgesia. This increased response to mechanical stimuli reached a peak at 9 weeks and tended to recover to basal values during the coasting effect follow-up period (Figure 1F). These functional findings are in accordance with those reported in patients with cisplatin-induced neuropathy.⁵

The number and density of myelinated axons in semithin cross sections of the sciatic and tibial nerves did not differ between control and cisplatin-treated mice at the different time points evaluated (Figure 1G, H). Likewise, the intra-epidermal nerve fiber density in the hindpaw pad did not show differences between control and cisplatin-treated animals when labeling fibers with the pan-neuronal marker PGP9.5 (Figure 1I). However, labeling against the peptidergic neuronal marker CGRP indicated a significant reduction of positive intra-epidermal fibers at the end of treatment (Supplementary Figure 1).

Isolation and Identification of DRG Cell Populations by Single-Cell RNA Sequencing

To study the intrinsic response of sensory neurons to cisplatin exposure, we performed scRNAseq of DRG cells previously isolated by single-cell sorting from control and cisplatin-treated mice (Figure 2A, B). The analysis was performed one week after the end of cisplatin treatment. A total of 182 control and 179 cisplatin-treated cells were sequenced. Principal component analysis of expression magnitudes across all cells and genes revealed 7 distinct cell clusters. These clusters are represented in a t-distributed stochastic neighbor embedding (t-SNE) plot to simplify their visualization (Figure 2C). In order to identify cluster-specific marker genes, the difference in expression of each gene between one cluster and the average in the resting clusters was calculated (Figure 2D, Supplementary File 1). Identification of each cell cluster was then determined by comparing the cluster-specific marker genes with the mouse nervous system scRNAseq database published by Linnarsson Lab.¹⁸ The different cell clusters were compatible with satellite glia cells, glial-like cells, neurons, endothelial/pericytes, perivascular macrophages, and vascular smooth muscle cells. The identity of one of the clusters was unclear, presenting mixed markers for different cell types, and therefore it was noted as undefined. Overlaying cells in the t-SNE plot with classical markers for neurons (Tubb3, Eno2)¹⁹ and satellite glial cells (Cdh19, Fabp7)^20,21 corroborate the identity of these 2 main cell populations of the DRG cells (Figure 2E).

Fig. 2.

Specific DRG cell populations were obtained by scRNAseq with previous cell isolation by single-cell sorting. (A) Experimental design of the study. Individual DRG cells of control and treated mice were isolated in 96-well plates by single-cell sorting. Each sorted cell was then sequenced by scRNAseq (smartSeq2) and differences between the transcriptome of control and cisplatin-treated cell populations were analyzed. (B) Process followed to isolate DRG cells by single cell sorting. (C) t-SNE plot representing the data of the principal component analysis of mRNA transcriptome in the individual sorted DRG cells (each point refers to an individual cell). Cells were tagged for a specific cell type according to their different expression patterns of marker genes. (D) Heatmap showing the top 10 genes defining each cellular cluster. Unique cell clusters were not altered by cisplatin treatment. (E) Relative expression level of DRG satellite glia (Cdh19 and Fabp7) and sensory neurons (Tubb3 and Eno2) markers are mapped to each cell in the t-SNE plot. Color key represents normalized gene expression with the highest expression marked purple and the lowest marked gray.

Cisplatin Treatment Induces Upregulation of Cdkn1a Transcription DRG Primary Neurons

Once the neuronal population was identified, the mRNA transcriptome profiles of 15 control and 43 cisplatin-treated sensory neurons were compared. A total of 122 differentially expressed genes (DEGs) (P < 0.05) were detected between both experimental conditions (Supplementary File 2). Gene Ontology analysis of DEG with P < 0.01 revealed that most of the biological processes upregulated in the cisplatin group are related to DNA integrity checkpoints as well as DNA damage responses (Supplementary File 3). However, to reduce the risk of false significant results after multiple comparisons, P-values were corrected and adjusted using the false discovery rate method, known as adjusted P-value (P-val-adj).²² In our experiment, among all the 122 DEGs, only the Cdkn1a gene had a P-val-adj < 0.05, so it was considered as a reliable DEG between control and cisplatin conditions (P-val-adj = 0.0018; log_FC = 0.99) (Supplementary File 2).

To corroborate this significant finding, we first performed a western blot analysis of whole DRG lysates from control and cisplatin-treated animals to check the levels of p21 protein, which is the product of the Cdkn1a gene. We observed a progressive increase of p21 protein levels during the assessment (Figure 3A). In addition, we wanted to identify the specific cell type where p21 was overexpressed. Immunofluorescence analysis of DRG cross sections also revealed an increase of neuronal nuclei positive for p21 immunoreactivity in the cisplatin group, which reached statistical significance at 16 weeks (Figure 3B).

Fig. 3.

Cisplatin treatment induces molecular senescence hallmarks in mice DRG neurons without evidence of apoptosis induction. (A) Up: quantification of p21, Cleaved Caspase-3, Bcl-2, p-H2AX and Nfkb-p65 protein levels after western blot analysis of DRG from control and cisplatin treated mice at 10 and 16 weeks, n = 3–6 mice per group, one-way ANOVA with Bonferroni post-hoc test for multiple comparisons. *P < 0.05 vs Control. Data are represented as group mean ± SD. Below: representative cropped blots of the corresponding protein. +Control refers to SH-ST5Y cells treated with 1 µM staurosporine for 6 hours. (B) Up: representative images of p21, p-H2AX and Nfkb-p65 immunofluorescence (red) in the DRG of control and cisplatin-treated mice. Neurons were stained with the pan-neuronal marker β-III tubulin (β-III-tub, green) and nuclei were counterstained with DAPI (blue). White arrows show neuronal nuclei positive for p21, p-H2AX, or Nfkb-p65. Scale bar: 50 µm. Down: quantification of the percentage of neuronal nuclei positive for p21, p-H2AX, and Nfkb-p65 proteins, n = 4–7 mice per group. One-way ANOVA with Bonferroni post-hoc test for multiple comparisons. *P < 0.05 vs Control. Data are represented as group mean ± SD.

P21: Apoptosis vs Senescence Pathways

According to our scRNAseq results, p21 emerges as the most relevant downstream checkpoint of the cisplatin-induced DNA damage response in DRG sensory neurons. However, in nonquiescent cells, p21 can be involved in 2 different and opposite adaptive cellular responses to stressors: apoptosis and senescence.¹¹ Up until now, apoptosis has been extensively described as the main molecular mechanism underlying cisplatin neurotoxicity.^23–25 To study the activation of apoptosis in our animal model, we checked the levels of cleaved caspase-3, which is necessary for apoptosis execution.²⁶ Western blot analysis in the whole DRG showed that there was no active fragment of caspase-3 (15–17 kDa) at any of the time points evaluated in mice treated with cisplatin (Figure 3A). Moreover, levels of B-cell lymphoma 2 (Bcl-2) protein were not different among control and cisplatin groups (Figure 3A). Similarly, no nuclear apoptotic changes were observed (Supplementary Figure 2). These data point out that the increased levels of p21 do not lead to apoptosis activation. Therefore, we wondered whether cisplatin could be inducing a senescence-like phenotype in DRG neurons in our animal model.

Cisplatin Treatment Triggers a Senescence-Like Phenotype in DRG Sensory Neurons

Senescent cells show a lack of uniform definition and phenotype, especially postmitotic cells. Thus, a combination of multiple biomarkers and morphological features has been used to define this anti-apoptotic and non-proliferative cellular state.²⁷

To corroborate the hypothesis of a cisplatin-induced senescence-like phenotype in DRG sensory neurons, we first checked the levels of phosphorylated H2AX protein (p-H2AX). This is an early molecular hallmark of DNA damage that has been linked with the occurrence of cellular senescence and final downstream p21 upregulation.²⁸ By western blot, we observed an upregulation of this protein in DRG of the cisplatin-treated mice starting at 10 weeks of treatment (Figure 3A). Immunofluorescence analysis of DRG slices showed that the increase of p-H2AX protein occurs in neuronal nuclei (Figure 3B).

Another feature of senescent cells is the development of a senescence-associated secretory phenotype (SASP) that consists of a concerted hypersecretion of pro-inflammatory factors and extracellular matrix proteases.^29,30 Thus, we checked the levels of Nfkb-p65 protein, an important inductor of the SASP phenotype.³¹ Results of western blot showed that cisplatin induces an upregulation of the Nfkb-p65 protein at 16 weeks of the study (Figure 3A). We observed a progressive significant expression of Nfkb-p65 protein in the nucleus of DRG neurons by immunofluorescence analysis in cisplatin-treated mice (Figure 3B).

To further corroborate the senescence-like phenotype in sensory neurons after cisplatin exposure, we performed the senescence-associated beta-galactosidase (SA-β-Gal) assay in DRG slices, a widely used molecular marker of cell senescence.³² Mice treated with cisplatin showed a significantly higher intensity of SA-β-Gal staining at 16 weeks compared with controls (Figure 4).

Fig. 4.

SA-β-Gal staining in DRG of control and cisplatin-treated mice. (A) SA-β-Gal positive neurons are seen in DRG of control and cisplatin-treated mice at 10 weeks and 16 weeks of study. Images below correspond to a magnification of the box-delimited areas of the upper images. Scale: 100 µm. (B) Quantification of the SA-β-Gal staining intensity in control and cisplatin treated mice n = 4–7 mice per group, one-way ANOVA test. Bonferroni post-hoc test was used for multiple comparisons. *P < 0.05. Data are represented as group mean ± SD.

Finally, we evaluated the structural changes that DRG sensory neurons suffered after cisplatin treatment by transmission electron microscope. In comparison to controls, neurons of cisplatin-treated animals presented larger mitochondria, with frequent fusion/fission-like phenomena. Most of these neurons also presented an enlarged endoplasmic reticulum with autophagosome-like vesicles in the periplasmic membrane space. All these changes were qualitatively more pronounced at 16 than at 10 weeks (Figure 5).

Fig. 5.

Transmission electron microscope images of DRG neurons from control and cisplatin-treated mice at 10 weeks and 16 weeks. (A–C) General view of the neuronal cytoplasm of control (A) and cisplatin-treated mice at 10 weeks (B) and 16 weeks (C). Mitochondria (Mit) and endoplasmic reticulum (ER) appear dilated in cisplatin-treated animals at both 10 weeks and 16 weeks compared with control. Moreover, at 16 weeks, neurons from the cisplatin condition present lysosome vesicles (Ly) and lipofuscin granules (Lip). (D–I) magnified views of the neuronal cytoplasm of control (D) and cisplatin-treated mice (E–I) . In the cisplatin conditions, it is frequent to see fission/fusion mitochondrial phenomena (indicated with an asterisk *) (E–F) , which are rare in the control condition. Lipofuscin granules (Lip), lysosomes vesicles, and autophagosome-like vesicles (Aut) are seen in DRG sensory neurons from cisplatin-treated mice at 16 weeks. At this time point, neuronal nuclei (N) have normal morphology with no alterations in the nucleoli (n) nor in the chromatin.

In addition, we observed accumulation of lipofuscin granules, another senescence hallmark³³ in the neuronal cytoplasm of cisplatin-treated animals at 16 weeks. These granules were not seen in controls or in treated animals at 10 weeks (Figure 5).

Discussion

The findings of this study suggest that cisplatin administration induces a senescence-like phenotype in the sensory neurons of the mouse DRG. This cellular process is already activated 10 weeks after the first cisplatin administration, as indicated by the increased levels of Cdnk1a gene expression, p-H2AX, and a concomitant reduction of sensory nerve potential amplitudes. The senescence-like phenotype progresses and is well established by 16 weeks, when multiple molecular and morphological events were observed in the DRG neurons. At this point in time, even when cisplatin has been withdrawn for 6 weeks, a more marked reduction of the amplitude of nerve action potentials is observed, according to the coasting effect.

After DNA damage, cells activate a complex and tightly regulated network of signaling pathways that constitute the DNA damage response (DDR) intended to safeguard genome integrity.³⁴ When the DDR cannot deal with this damage, cells undergo apoptosis or activate senescence pathways in order to preserve their function or minimize tissue damage.¹² While apoptotic programs are well defined,³⁵ cell senescence is a collective phenotype of multiple effector programs, mostly described in replicative cells.²⁷ In fact, the most accepted starting point definition of cell senescence state is provided by the stress-induced arrest and resistance to mitogenic stimuli.^36,37 However, this definition is difficult to fit for postmitotic differentiated cells. Interestingly, molecular markers and effector pathways resembling the senescence phenotype processes have also been identified in nonreplicative cells like Purkinje and cortical brain neurons.^28,38 In fact, in the context of the neural tissue, with low or restricted regenerative capacity in response to cellular insults, predominance of the senescence response seems more adaptive than activation of pro-apoptotic programs. In this study, we have demonstrated that cisplatin treatment induces a senescence-like phenotype in DRG sensory neurons in a well-established animal model of peripheral neuropathy. Since senescence is a collective phenotype of multiple effector programs, the induced senescence-phenotype by cisplatin has been corroborated by using a variety of accepted markers implicated in this response.

Single-cell RNAseq of sorted neurons from DRG of cisplatin-treated mice showed an upregulation of Gene Ontology terms involved in DDRs and increased of Cdkn1a levels. The progressive increased expression of Cdkn1a product p21 and the early phosphorylation of H2AX as a fast responder to DNA damage corroborated the initial findings obtained in the scRNAseq analysis. Reinforcing our data, previous studies pointed to platinum-induced inter- and intra-strand DNA crosslink damage as the main cause of platinum neurotoxicity.^25,39 Moreover, there were neither cleaved caspase-3 nor nuclear apoptotic morphological changes observed in DRG neurons of cisplatin-treated mice, indicating that the DDR did not lead to neuronal death by apoptosis. On the other hand, we have found a preservation of myelinated nerve fibers in our model, as described by others,^14,40 with significant reduction in the amplitude of the compound nerve action potentials. These seemingly discrepant results are compatible with neuronal survival but with a dysfunctional phenotype.

Most studies regarding cisplatin neurotoxicity are focused on activation of the apoptotic pathway triggered by inefficient DNA repair mechanisms. Evidence from in vitro neuronal cultures demonstrated involvement of apoptosis.^24,25 However, the evidence in animal models is scarce or even absent.¹⁰ It has been well established that the cellular response to DNA insults varies depending on the degree and duration of damage. Mild or low prolonged DNA damage results in increased levels of scaffold/matrix attachment region 1 (SMAR1), activation of p21, and the direct transcriptional inhibition of BAX (Bcl-2-associated X protein) and PUMA (p53 upregulated modulator of apoptosis).⁴¹ In contrast, severe DNA damage induces the sequestration of SMAR1 and the activation of apoptotic programs.⁴² Therefore, a reliable experimental model that closely mimics the clinical situation, with similar equivalent cumulated dose and clinical features, is crucial to identify the etiopathogenic mechanisms of this neuropathy, especially when adequate patient samples are not available.

Supporting our results, a previous report on microarray analysis of whole DRG from cisplatin treated rats supports our results, as it also found an increased expression of Cdkn1a and Mmp9.²³ The authors related these changes with apoptosis, although the overexpression of these genes can also be involved in senescence. In addition, our hypothesis of cisplatin-induced senescence-like phenotype was further supported by the activation of other senescent markers, like increased activity of SA-β-Gal, a maker of lysosome activity,⁴³ and overexpression of Nfkb-p65, a key factor in SASP induction.³¹ Nfkb-p65 regulates many cellular processes, including proliferation, apoptosis, and survival. When inactivated, Nfkb-p65 stays in the cytoplasm by binding to inhibitor of kappa B (Ikb). After activation, Ikb is degraded through the proteasome pathway and Nfkb-p65 is phosphorylated for its translocation to the nucleus.⁴⁴ Thus, the localization of p65 into the nucleus is a proxy for Nfkb activation.⁴⁵

These molecular features are reinforced by the morphological changes that we have found in DRG sensory neurons of cisplatin-treated mice, also reported to be characteristic of cellular senescence, including mitochondrial disturbances, reticulum enlargements, and accumulation of intracytoplasmic granules.^27,46

Most of the senescence markers and the cellular changes observed in our model were more marked 16 weeks after the onset of treatment including 6 weeks withdrawal, while neuropathy continues progressing in the coasting effect, which is widely described in patients.⁴⁷ In contrast, p-H2AX increase was already significant at 10 weeks. Phosphorylation of H2AX in Ser139 is a fast response in front DNA damage⁴⁸ that has been involved with the development of senescence.²⁸ Therefore, it is to be expected in front of a situation of active DNA damage, like the one caused by the presence of cisplatin in the cells.

The senescence-like phenotype that sensory neurons develop as a response to cisplatin-induced DNA damage (demonstrated in this study) helps to fit some of the previous findings reported in other models of cisplatin-induced neuropathy. In addition, it proposes a new etiopathogenic mechanism of platinum-induced peripheral neuropathy. The modulation of the senescence inductor p21, its downstream pathways, or even other pathways involved in the senescence program may lead to new approaches in the prevention or even treatment of this important clinical condition that has seen a lack of relevant advances during past decades.

Funding

The authors’ research was supported by funds from CIBERNED and TERCEL networks to XN, and by a PI1501303 grant to JB from the Instituto de Salud Carlos III of Spain, co-funded by European Union (ERDF/ESF, “Investing in your future”). JB has received support from grant number SLT008/18/00028 from the Department of Health of the Government of Catalonia, CERCA Program. AC was recipient of a predoctoral fellowship from the Secretaria d’Universitats i Recerca of the Catalan Government and the European Social Fund (FI fellowship).