Dietary restriction (DR, also known as caloric restriction) is a long-term dietary intervention that consists of a daily 15 to 40% reduction of food intake without malnutrition, and that is well known as a universal anti-aging intervention increasing health-and lifespan (Le Couteur and Simpson, 2018; Lee et al., 2021; Green et al., 2022). DR reduces age-related decline in motor and cognitive function, as well as nervous system pathologies associated with normal aging and age-related neurodegenerative diseases (Mattison et al., 2017; Xie et al., 2020). While the benefits of DR as an anti-aging intervention are well established, the mechanisms by which aging is counteracted are still largely unresolved (Speakman and Mitchell, 2011; Lee et al., 2021). On the one hand, aging itself is a complex multifactorial phenomenon with numerous detrimental processes occurring at cellular (accumulation of damaged macromolecules and intracellular debris), organ (e.g., impaired tissue renewal), as well as whole organism level (changes in hormonal, serological and inflammatory status; Lopez-Otin et al., 2013; Pluvinage and Wyss-Coray, 2020). On the other hand, DR affects numerous intra- and intercellular processes, which include nutrient sensing, metabolic, hormonal and immunoregulatory pathways (Speakman and Mitchell, 2011; Green et al., 2022). Thus, a major challenge in understanding the anti-aging mechanisms of DR lies in connecting the pathways modulated by DR with the variety of aging processes that operate distinctively across tissues and cell types (Speakman and Mitchell, 2011; Ma et al., 2020; Schaum et al., 2020; Vougioukalaki et al., 2022).

We have previously found that DR has a remarkably strong beneficial effect on life-and health span in progeroid Ercc1^Δ/− mice, in which defective DNA repair results in chronic genome instability and numerous features of multimorbidity and accelerated aging (Vermeij et al., 2016a). These data show that DR may alleviate the detrimental effects of impaired DNA repair, and indicate that DR protects against genotoxic stress, which is one of the driving factors of aging (Lopez-Otin et al., 2013). Most notably, DR strongly reduced the dramatic, progressive neurodegeneration (Vermeij et al., 2016a). Since Ercc1^Δ/− mice develop systemic multimorbidity, with severely compromised functions of vital organs, blood vessels and immune system (Vermeij et al., 2016a,b; Vougioukalaki et al., 2022), a key question remained whether DR is neuroprotective indirectly via local and/or systemic components (e.g., by reducing inflammation), or by directly lowering cell-intrinsic DNA-damage-driven neurotoxicity. This question is particularly relevant for the central nervous system in view of increasing evidence favoring a central role of systemic factors in nervous system aging and neurodegeneration (Pluvinage and Wyss-Coray, 2020). Therefore, we examined the effect of DR in a mouse model in which DNA repair deficiency drives accelerated aging only in one single neuronal cell type, i.e., cerebellar Purkinje cells. These neurons are severely affected, but largely protected by DR, in whole-body Ercc1^Δ/− mice (Birkisdottir et al., 2021), and degenerate in human aging (Andersen et al., 2003) as well as in various genome instability disorders (Shiloh, 2020; Kwak et al., 2021).

We generated a Cre-lox-based Purkinje cell-specific ERCC1-deficient (Pcp2-Ercc1^Δ/f) mouse model that carried a floxed Ercc1 allele (Ercc1^f), a C-terminally truncated “delta” Ercc1 allele (Ercc1^Δ; Vermeij et al., 2016b), and a Pcp2-Cre transgene (Mpin line) that drives Cre recombinase expression in postnatal Purkinje cells (Van't Sant et al., 2021). After recombination by Cre, the floxed Ercc1 allele becomes a null allele, resulting in Purkinje cells that have one null and one truncated Ercc1 allele, and that reproduce the severe (but incomplete) reduction in ERCC1/XPF nuclease function of whole body Ercc1^Δ/− mice (Figure 1A), affecting nucleotide excision repair, inter-strand crosslink repair, and single-strand annealing of double-strand breaks (Vermeij et al., 2016b; Mulderrig and Garaycoechea, 2020; Apelt et al., 2021). Pcp2-Ercc1^Δ/f mice had, unlike Ercc1^Δ/− mice, the same weight as control littermates and did not show evidence of neurodegenerative changes outside the cerebellum, consistent with the remainder of the body and nervous system being repair proficient and healthy (Figure 1A; Supplementary Figures S1A,F). Pcp2-Ercc1^Δ/f mice at 12 weeks showed control performance in cerebellar motor tests, but developed progressive deficits thereafter (Supplementary Figures S1B–D). The time of onset and severity of cerebellar motor abnormalities as well as the severity of Purkinje cell degeneration were the same as in Pcp2-Ercc1^Δ/f mice generated with another Pcp2-Cre transgene (Jhdu line), indicating that the Purkinje cell degenerative phenotype is not influenced by the Cre line (Supplementary Figure S1). Since the BAC construct of the Jdhu Pcp2-Cre line may introduce extra copies of genes implicated in neuronal function, splicing and DNA repair (Pnpla6, Stxbp2, Pcp2, and Xab2; Zhang et al., 2004) we used the Pcp2-Cre Mpin line for the DR experiments.

We applied the same 30% DR regimen as used before in full-body Ercc1^Δ/− mice starting with 10% restriction at the age of 7 weeks, 20% at 8 weeks, and 30% thereafter (Vermeij et al., 2016a). DR-treated Pcp2-Ercc1^Δ/f animals showed the same weight loss as DR control mice (Supplementary Figure 2A), and displayed preserved performance in an accelerating rotarod test up to the age of 40 weeks (Figure 1B). In contrast, ad libitum (AL)-fed Pcp2-Ercc1^Δ/f mice showed strongly declining rotarod performance till 35 weeks, after which the descent appears to stabilize (Figure 1B), concomitant with severe Purkinje cell degeneration throughout the cerebellar cortex as revealed by reduced immunohistological staining for calbindin that in the cerebellum is specifically expressed in Purkinje cells (Figures 1C,D; Supplementary Figures S2B). DR-treated Pcp2-Ercc1^Δ/f mice at 40 weeks, instead, showed relatively preserved calbindin staining (Figures 1C,D; Supplementary Figure S2B), indicative of reduced Purkinje cell loss. We quantified Purkinje cells using a stereological approach (Andersen et al., 2003; Birkisdottir et al., 2021), and confirmed that Purkinje cell loss was significantly (p < 0.01) prevented by DR with 1.16 ± 0.12 *10⁵ and 0.47 ± 0.03 *10⁵ residual Purkinje cells in 40-week-old DR-and AL-Pcp2-Ercc1^Δ/f mice, respectively. Additionally, Purkinje cells of DR-treated Pcp2-Ercc1^Δ/f mice showed a lower frequency of morphological abnormalities, such as axonal spheroids (Figure 1D) and displayed less astrocytosis, indicated by GFAP staining, in the molecular layer consistent with reduced Purkinje cell degeneration (Figures 2A,B; de Waard et al., 2010; de Graaf et al., 2013). Next, we analyzed Golgi apparatus morphology of Purkinje cells, following up on our finding that in Ercc1^Δ/− mice DR strongly reduced the frequency of spinal motor neurons with atypical Golgi, putatively representing a proxy of compromised neuronal health (de Waard et al., 2010; Vermeij et al., 2016a). Indeed, AL-fed Pcp2-Ercc1^Δ/f mice showed a spectrum of abnormal Golgi in about 25% of residual Purkinje cells, while this was strongly diminished in DR-treated Pcp2-Ercc1^Δ/f mice (Figures 2C,D). In sum, DR significantly reduces motor dysfunction and Purkinje cell pathology and loss in Pcp2-Ercc1^Δ/f mice at 40 weeks.

To robustly quantify the effect of DR on ERCC1-deficiency in Purkinje cells in time, we focused on stereological analysis of Purkinje cell numbers. At 8 weeks of age at the start of DR treatment, the number of Purkinje cells in Pcp2-Ercc1^Δ/f mice appeared unaltered compared to controls of 8–40 weeks (2.28 ± 0.11 *10⁵ vs. 2.27 ± 0.06 *10⁵, respectively; Supplementary Figure S3). Accordingly, Pcp2-Ercc1^Δ/f mice at 8 weeks showed no evidence of astrocytosis (Supplementary Figure S3D). At the age of 26 weeks, DR and AL mice showed 1.73 ± 0.09 * 10⁵ and 1.36 ± 0.08 *10⁵ remaining Purkinje cells, respectively (Figure 1E) revealing a significant protective effect of DR (p < 0.05). Using the mean value at 8 weeks (2.28 *10⁵) as starting number of Purkinje cells in Pcp2-Ercc1^Δ/f mice (Supplementary Figure S3B), we estimated the number of Purkinje cells that died in AL and DR animals at 26 and 40 weeks (= 2.28 *10⁵- # of surviving Purkinje cells). Subsequently, these values were used to calculate a relative rescue of Purkinje cells by DR with the formula: (average loss in AL animals – loss in DR animal)/average loss in AL animals*100. This analysis indicated that DR attenuated Purkinje cell loss by about 40% at both ages (Figure 3A). This percentage appears somewhat smaller than the 59% estimated relative rescue of Purkinje cell loss by DR for full-body Ercc1^Δ/− mice, examined at the age of 16 weeks, a few weeks before the progeroid mutants become moribund (Figure 3A; Birkisdottir et al., 2021), albeit the difference did not reach statistical significance (p = 0.289 and 0.188 for 26 and 40 weeks respectively). It should be noted, that systemic Ercc1^Δ/− mice do not only differ from Pcp2-Ercc1^Δ/f in regard of ubiquitous versus Purkinje-cell-specific ERCC1-deficiency, but also show earlier onset of ERCC1 cell deficiency (early development versus post-natal week 2), and already exhibit Purkinje cell loss at the onset of DR treatment at 8 weeks (Supplementary Figure S3B).

Hypomorphic Pcp2-Ercc1^Δ/f and full-body Ercc1^Δ/− mice still have some residual repair activity, which may contribute to the protective effect of DR. Therefore, we tested DR in Pcp2-Ercc1^−/f mice that are fully ERCC1-deficient in Purkinje cells (Figure 1A). Pcp2-Ercc1^−/f mice show unaltered numbers of Purkinje cells at 8 weeks (Supplementary Figure S3B), like Pcp2-Ercc1^Δ/f mice and in accord with previous proteomic and neuropathological data (de Graaf et al., 2013). However, they display a larger (p < 0.0001) Purkinje cell loss at 26 weeks with 0.78 ± 0.07 *10⁵ (vs. 1.36 ± 0.04 *10⁵ for Pcp2-Ercc1^Δ/f mice) residual Purkinje cells, and only sparse surviving Purkinje cells at 40 weeks (0.23 ± 0.03 *10⁵; Supplementary Figures S3B,C), indicating a dose–response relationship between degree of repair deficiency and neuronal cell loss. As in Pcp2-Ercc1^Δ/f mice, DR significantly (p < 0.01) reduced Purkinje cell loss in Pcp2-Ercc1^−/f mice at 26 weeks of age (Figures 3B,C), with a relative rescue of about 25% compared to AL animals (Figure 3A). These data indicate that DR also is protective in Purkinje cells with complete loss of ERCC1, that experience more severe genotoxic stress than Ercc1^Δ/− Purkinje cells (Figure 1A).

To get insight into the role of DNA damage we examined P53 as guardian of the genome. Previously, we found a decrease in P53-expressing cells in the neocortex and spinal cord of DR Ercc1^Δ/− mice, supporting that DR reduces DNA damage load (Vermeij et al., 2016a). Here, we also observed an overall reduction [p = 0.024; 2-way ANOVA F(1,16) = 6.019] by DR on P53 positive Purkinje neurons, albeit the difference between each individual AL-DR pair not being statistically significant (Figure 3D). Across all models and cohorts, DR equally reduced the proportion of Purkinje cells with nuclear P53 accumulation by about 50% (Figure 3E).

In this study we show that DR strongly reduces degeneration of ERCC1-deficient Purkinje cells, in mice that are unaffected in other cell types and tissues. The data imply that DR strongly reduces cell-intrinsic DNA-damage driven Purkinje cell degeneration. The data also indicate that reduced Purkinje cell degeneration by DR in whole-body ERCC1-deficient mice primarily results from cell intrinsic mechanisms rather than by restoring function of surrounding glia cells and vital organs or via systemic components. Our data support previous work indicating that DR may both reduce the levels of DNA damage and increase the ability of cells to cope with DNA damage (Heydari et al., 2007), and support the hypothesis that the anti-aging effect of DR in the nervous system, at least in part, is mediated by reducing genotoxic stress in neurons (Vermeij et al., 2016a; Xie et al., 2020).

Previously, we have shown that ERCC1-deficient mice develop multiple age-dependent neurological deficits and progressive neuropathology characterized by a diversity of degenerative changes throughout the nervous affecting both neurons and glia (de Waard et al., 2010; Borgesius et al., 2011; Vegh et al., 2012; Raj et al., 2014; Vermeij et al., 2016a). The nervous system pathology is consistent with a model where stochastic accumulation of DNA damage causes transcription stress, i.e., a variety of yet incompletely defined transcription problems and down-stream effects that can affect cell function in multiple ways, for instance by blocked or deregulated expression of essential genes or problems resulting from persistent transcription stalling (Vermeij et al., 2016b; Niedernhofer et al., 2018; Lans et al., 2019; Mulderrig and Garaycoechea, 2020; Apelt et al., 2021; Jia et al., 2021; Van't Sant et al., 2021). Here, we focused on stereological analysis of residual Purkinje cell numbers as the principal readout, representing an unequivocal quantitative readout that reflects the final outcome of multiple neurodegenerative pathways. In addition, we show that DR has impact on two additional readouts in Purkinje cells: it reduces the frequency of Purkinje cells expressing the genotoxic stress transcription factor P53, and Purkinje cells with abnormal Golgi apparatus morphologies. The varied perturbations in Golgi morphology in more than 20% of ERCC1-deficient Purkinje cells and their strongly reduced frequency in DR animals, are consistent with findings in spinal motor neurons in Ercc1^Δ/− mice (de Waard et al., 2010; Vermeij et al., 2016a), and may reflect a diversity of degenerative and cell stress pathways including DNA damage signaling (Makhoul et al., 2019).

The identity of DNA lesions that accumulate in ERCC1-deficient nervous system and cause neuronal damage are yet poorly understood (Niedernhofer et al., 2018; Lans et al., 2019; Mulderrig et al., 2021). In our previous DR study with Ercc1^Δ/− mice, analyses of liver transcriptomes showed a preferential reduction in expression of long genes in Ercc1^Δ/− liver, which was attenuated by DR (Vermeij et al., 2016a). This is consistent with the notion that long genes have an increased likelihood of accumulating transcription stalling lesions (Lans et al., 2019; Noe Gonzalez et al., 2021), and display reduced damage in DR animals. Hence, analysis of gene-length-dependent transcriptional changes reflect a promising indirect readout for measuring transcription stress and the effect of DR in Ercc1^Δ/− nervous system. Importantly, a bias of reduced expression of long genes also has been found in aged human nervous system (Vermeij et al., 2016a; Niedernhofer et al., 2018; Soheili-Nezhad et al., 2021), indicating that DNA-damage-induced transcription stress is an important factor contributing to nervous system aging and targeted by DR (Vermeij et al., 2016a). Future studies employing high resolution spatial transcriptomic approaches (Asp et al., 2020; Lewis et al., 2021) may uncover the relationships between transcriptional abnormalities and neuronal degeneration in aging and progeroid models.

What changes triggered by DR render Purkinje cells less vulnerable towards DNA repair deficiencies? DR has been shown to promote autophagy and to reduce ribosomal protein S6 phosphorylation in Purkinje cells, indicative of suppression of mTOR signalling (Alirezaei et al., 2010), one of the main pathways linked to DR benefits (Liu and Sabatini, 2020; Green et al., 2022). However, we recently showed that mTOR inhibition via rapamycin or genetic intervention fails to improve lifespan, neurological functioning, and Purkinje cell survival of Ercc1^Δ/− mice (Birkisdottir et al., 2021), hence, pointing to alternative mechanisms. While a spectrum of other potential anti-aging compounds that have been proposed to reproduce aspects of the beneficial effects of DR by influencing nutrient sensing, inflammation, mitochondrial function, or glucose homeostasis, did not have an effect on survival and neurological function in Ercc1^Δ/− mice, we found that two NAD⁺ precursors (nicotinamide riboside and nicotinic acid) induced small benefits in survival and neurological function (Alyodawi et al., 2019; Birkisdottir et al., 2022). Since DR may increase NAD⁺ levels and modulate a variety of NAD⁺-dependent mechanisms in energy and redox metabolism, as well as in DNA damage sensing, protein deacetylation or other enzymatic pathways requiring NAD⁺ (Covarrubias et al., 2020; Green et al., 2022), NAD⁺-dependent pathways may be part of the mechanisms mediating the beneficial effect of DR in ERCC1-deficient Purkinje cells. However, further studies are needed to dissect the mechanisms that underlie the strong beneficial effect of DR in ERCC1-deficient mice. Our Purkinje cell specific ERCC1-deficient mice provides a complementary model to global Ercc1^Δ/− mice in these studies.

The original contributions presented in the study are included in the article/Supplementary material, further inquiries can be directed to the corresponding author/s.

The animal study was reviewed and approved by Animal Welfare Board of the Erasmus MC, ErasmusMC Rotterdam. Written informed consent was obtained from the owners for the participation of their animals in this study.

WV and DJ conceptualized and designed the experiments. MB, LS, WV, and DJ wrote the manuscript and performed statistical analysis of data. JH provided critical advice, and contributed to writing and editing of the manuscript. RB, SB, and WV planned and performed mouse experiments. MB, LS, and DJ performed histopathological assessments. All authors contributed to the article and approved the submitted version.

We acknowledge financial support of the National Institute of Health (NIH)/National Institute of Aging (NIA) (AG17242), European Research Council Advanced Grants Dam2Age (to JH), ONCODE supported by the Dutch Cancer Society, ADPS Longevity Research Award (to WV), Memorabel (ZonMW 733050810), BBoL (NWO-ENW 737.016.015), Deutsche Forschungsgemeinschaft (DFG, German Research Foundation—Project-ID 73111208—SFB 829), Regiodeal Foodvalley (162135), and the European Joint Programme Rare Diseases (TC-NER RD20-113).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.