Postmitotic neurons develop a p21-dependent senescence-like phenotype driven by a DNA damage response

Abstract

In senescent cells, a DNA damage response drives not only irreversible loss of replicative capacity but also production and secretion of reactive oxygen species (ROS) and bioactive peptides including pro-inflammatory cytokines. This makes senescent cells a potential cause of tissue functional decline in aging. To our knowledge, we show here for the first time evidence suggesting that DNA damage induces a senescence-like state in mature postmitotic neurons in vivo. About 40-80% of Purkinje neurons and 20-40% of cortical, hippocampal and peripheral neurons in the myenteric plexus from old C57Bl/6 mice showed severe DNA damage, activated p38MAPkinase, high ROS production and oxidative damage, interleukin IL-6 production, heterochromatinization and senescence-associated β-galactosidase activity. Frequencies of these senescence-like neurons increased with age. Short-term caloric restriction tended to decrease frequencies of positive cells. The phenotype was aggravated in brains of late-generation TERC-/- mice with dysfunctional telomeres. It was fully rescued by loss of p21(CDKN1A) function in late-generation TERC-/-CDKN1A-/- mice, indicating p21 as the necessary signal transducer between DNA damage response and senescence-like phenotype in neurons, as in senescing fibroblasts and other proliferation-competent cells. We conclude that a senescence-like phenotype is possibly not restricted to proliferation-competent cells. Rather, dysfunctional telomeres and/or accumulated DNA damage can induce a DNA damage response leading to a phenotype in postmitotic neurons that resembles cell senescence in multiple features. Senescence-like neurons might be a source of oxidative and inflammatory stress and a contributor to brain aging.

Fig. 1

Purkinje neurons in old (32 months) but not in young (4 months) mice are positive for multiple markers of the senescent phenotype. Representative images are shown. Size marker bars indicate 20 μm. (A–E) Cerebellar sections of mice at the indicated ages were stained with DAPI (blue, nuclei), the neuronal marker calbindin (purple, omitted in E for clarity) and the antibody of interest visualized by IgG-FITC (green). Shown are the antibody of interest (left) and the merged image (right). (A) γH2A.X, (B) activated p38MAPK, (C) 4-hydroxynonenal (4-HNE, a marker for lipid peroxidation), (D) IL-6, (E) mH2A. (F) Autofluorescence on unstained sections. (G) sen-β-Gal activity. Positive cells show blue cytoplasmic staining.

Fig. 2

Cortical neurons in old mice are positive for multiple markers of the senescent phenotype. Representative images from 4 to 32 months old mice are shown. The size marker bars indicate 20 μm. (A–D) Cortical sections were stained with DAPI (blue), the neuronal marker calbindin (purple, omitted in D for clarity) and the antibody of interest (green). Shown are the antibody of interest (left) and the merged image (right). (A) γH2A.X, (B) activated p38MAPK, (C) IL-6, (D) mH2A. (E) Autofluorescence on unstained sections. (F) Sen-β-Gal activity (blue).

Fig. 3

Myenteric ganglia in old mice are positive for multiple markers of the senescent phenotype. Intestinal strip preparations from young (5 months) and old (22–24 months) mice were stained for the following markers: (A) HuC/D and wheat germ agglutinin (WGA) to discriminate between neurons (HuC/D, red) and glial cells (WGA, green) (done in old mice only). Parallel preparations were immunostained for γH2AX by immunofluorescence (B), blue: DAPI, red: γH2AX foci) and immunohistochemistry (C), or activated p38MAPK (D). (E) Fresh preparations were incubated with dihydrorhodamine 123 (DHR-123, a marker for cellular peroxide production) or were stained for sen-β-Gal activity (F). Representative images are shown. The size marker bars indicate 20 μm.

Fig. 4

Aging increases the frequency of neurons with a senescence-like phenotype. (A) Frequencies of Purkinje neurons positive for the indicated marker (in %) at the indicated ages as measured by immunofluorescence and immunohistochemistry. (B) Purkinje neuron autofluorescence intensity (in arbitrary units) at the indicated ages. (C) Frequencies of cortical neurons positive for the indicated marker (in %). (D) Cortical neuron autofluorescence intensity (in arbitrary units). All data are mean ± SEM, from at least three animals per age group. * and # indicate significant differences to 4 and 8 months old animals, respectively (anova with post hoc Tukey, P < 0.05). (E) Frequencies of gut neurons positive for γH2A.X (in % of all neurons per ganglion) and (F) intensity of neuronal DHR-123 staining (arbitrary units) in young (5 months) and old (22–24 months) mice. Data are from > 15 ganglia from 2 or 3 mice per age group.*P < 0.01, t-test). (G) Frequencies of Purkinje neurons positive for the indicated marker (in %) and (H) autofluorescence intensity (in arbitrary units) in 17 months old ad libitum fed (AL) animals and in mice kept under dietary restriction (DR) from 14 to 17 months of age. Data are mean ± SEM, from three animals per age group. * indicate significant differences (P < 0.05, t-test).

Fig. 5

Purkinje cells are positive for multiple markers of the senescent phenotype as shown by double immunofluorescence for γH2A.X and 4-HNE (A), phospho-p38MAPK and 4-HNE (B) and Il-6 and 4-HNE (C) in cerebellar sections from 32 months old mice. Nuclei are stained with DAPI, which is shown on the merged images (bottom). Red arrows mark Purkinje neurons strongly positive for both antigens, white arrows indicate Purkinje cells that express both tested markers weakly or not at all. Bars indicate 20 μm.

Fig. 6

p21(CDKN1A) is necessary for the senescent phenotype in mouse brain neurons. (A) Cerebellar sections from mice with long telomeres (first and second row) either positive (TERC+/−CDKN1A+/+) or negative (TERC+/−CDKN1A−/−) for p21 expression and late-generation telomerase knockout mice with short telomeres (third and fourth row) positive for p21 expression (F4TERC−/−CDKN1A+/+, third row) or with additional p21 knockout (F4TERC−/−CDKN1A−/−, bottom row) were stained for γH2A.X, phopho-p38, 4-HNE and IL-6. Representative images are shown. Positive cells appear red/brown. (B) Quantitative evaluation for Purkinje cells. Data are percentage of positive cells, mean ± SEM. Numbers of mice per group were 3 (F4CDKN1A+/+), 4 (CONTROL and F4CDKN1A−/−) and 5 (TERC+/−CDKN1A−/−), respectively. * and # indicate significant differences to control and F4TERC−/−CDKN1A−/−, respectively (anova with post hoc Holm–Sidak test, P < 0.05). (C) Quantitative evaluation for cortical neurons. Symbols and statistics as in B).