Increased apoptosis, p53 up-regulation, andcerebellar neuronal degeneration in repair-deficientCockayne syndrome miceR. R. Laposa*, E. J. Huang†, and J. E. Cleaver*‡

Departments of *Dermatology and Cancer Center and †Pathology, University of California, San Francisco, CA 94143-0808

Contributed by J. E. Cleaver, November 30, 2006 (sent for review August 1, 2006)

Cockayne syndrome (CS) is a rare recessive childhood-onset neu-rodegenerative disease, characterized by a deficiency in the DNArepair pathway of transcription-coupled nucleotide excision repair.Mice with a targeted deletion of the CSB gene (Csb�/�) exhibit amuch milder ataxic phenotype than human patients. Csb�/� micethat are also deficient in global genomic repair [Csb�/�/xerodermapigmentosum C (Xpc)�/�] are more profoundly affected, exhibitingwhole-body wasting, ataxia, and neural loss by postnatal day 21.Cerebellar granule cells demonstrated high TUNEL staining indic-ative of apoptosis. Purkinje cells, identified by the marker calbi-ndin, were severely depleted and, although not TUNEL-positive,displayed strong immunoreactivity for p53, indicating cellularstress. A subset of animals heterozygous for Csb and Xpc deficien-cies was more mildly affected, demonstrating ataxia and Purkinjecell loss at 3 months of age. Mouse, Csb�/�, and Xpc�/� embryonicfibroblasts each exhibited increased sensitivity to UV light, whichgenerates bulky DNA damage that is a substrate for excision repair.Whereas Csb�/�/Xpc�/� fibroblasts were more UV-sensitive thaneither single knockout, double-heterozygote fibroblasts had nor-mal UV sensitivity. Csb�/� mice crossed with a strain defective inbase excision repair (Ogg1) demonstrated no enhanced neurode-generative phenotype. Complete deficiency in nucleotide excisionrepair therefore renders the brain profoundly sensitive to neuro-degeneration in specific cell types of the cerebellum, possiblybecause of unrepaired endogenous DNA damage that is a sub-strate for nucleotide but not base excision repair.

DNA damage � nucleotide excision � base excision �xeroderma pigmentosum � Purkinje cells

Cockayne syndrome (CS) is a segmental progeroid syndromewith cachexia and wasting and progressive neurological

deficits, including demyelination, ataxia, cerebellar atrophy, andcalcification in the brain, particularly in the basal ganglia, butwith no increased risk of cancer (1, 2). Symptoms may be presentat birth or during childhood, and the average lifespan is 12 years(1, 2). CS individuals are deficient in a subpathway of nucleotideexcision repair (NER) known as transcription-coupled repair(TCR) (3–5). The neurological symptoms of CS patients havebeen ascribed to defective repair in the brain of endogenousoxidative damage that blocks transcription (6–11). CSA and CSBgenes have subsequently been ascribed other functions, such asubiquitylation, transcription regulation, and chromatin remod-eling, that could themselves be proximate causes of the TCR andother cellular defects (4, 5, 12, 13).

Mouse knockouts of the Csa and Csb genes present a weakerneurological phenotype than the corresponding human disorder(14–16), but detailed pathological investigations of the neuro-degenerative disease have not yet been done. CS mice, unlikehumans, are predisposed to UVB- and chemically induced skincancer (15, 17–19). Mouse embryo fibroblasts (MEFs) derivedfrom Csb�/� mice are sensitive to x-rays and oxidative damage(10, 20), but the mice do not accumulate 8-oxo-guanine (8-OH-G) lesions spontaneously with age, as would be expected ifthey did not repair 8-OH-G (10). Csb and 8-oxyguanine-DNA

glycosylase (Ogg) knockout cells both showed small reductionsin global repair of 8-OH-G, but a Csb�/�/Ogg1�/� strain wascompletely defective, indicating that the two pathways operateindependently (10). Ogg1 is a DNA repair enzyme that functionsin the base excision repair pathway to remove oxidative DNAlesions such as 8-hydroxy-2�-deoxyguanosine and formamidopy-rimine that can accumulate in vivo (21, 22). Both human CSAand CSB cells are sensitive to UV light and oxidative damage(23); in the mouse, however, only Csb cells are sensitive tooxidative damage, whereas Csa cells are resistant (20). Simplecorrelations of CS neurodegeneration to deficient repair ofoxidative damage, therefore, are difficult.

Because the neural phenotype of Csb�/� mice is much lesssevere than the human CS patient syndrome (1, 2, 16, 24), wedecided to determine the consequence of reducing the repaircapacity of Csb mice further for either base excision repair(BER) (Ogg1�/�) or global NER [xeroderma pigmentosum C(Xpc)�/�]. Xpc is the major DNA damage-binding protein fornontranscribed DNA (global genome repair) (3), and XPCpatients, as well as Xpc knockout mice, have no neurologicalsymptoms (25). We cross-bred Csb�/� mice with Xpc and Ogg1mice to identify the relative importance of single and doubledeficiencies in NER and BER in the pathology of the nervoussystem.

ResultsPathology of Compound Homozygous Animals. A subset of Csb�/�

mice were reported to exhibit mild ataxia at 12 months of age, whichwe also observed (14). Eight litters of mice were weighed duringearly postnatal life. The weight of Csb�/�/Xpc�/� mice was signif-icantly lower than controls at birth, and the Csb�/�/Xpc�/� micedemonstrated a slower growth rate than controls during earlypostnatal life, weighing significantly less than littermate controls atall time points assessed (P � 0.001) (Fig. 1B). By postnatal day(P)21, they were dramatically smaller than their littermate controls(Fig. 1A). Csb�/�/Xpc�/� mice survived only 1 or 2 days past ageP21. At age P21, Csb�/�/Xpc�/� mice moved more slowly andshowed poor coordination compared with littermate controls.Ataxia in Csb�/�/Xpc�/� mice was assessed by rotarod testing incomparison with control mice. Control mice were able to stay on therotarod 140% longer than compound heterozygotes, but this dif-ference was not statistically significant (P � 0.10). The distancebetween footsteps of Csb�/�/Xpc�/� was approximately half that of

Author contributions: R.R.L. and J.E.C. designed research; R.R.L. and E.J.H. performedresearch; E.J.H. contributed new reagents/analytic tools; R.R.L., E.J.H., and J.E.C. analyzeddata; and R.R.L., E.J.H., and J.E.C. wrote the paper.

The authors declare no conflict of interest.

Abbreviations: CS, Cockayne syndrome; NER, nucleotide excision repair; TCR, transcription-coupled repair; 8-OH-G, 8-oxo-guanine; Pn, postnatal day n; MEF, mouse embryonicfibroblast; Ogg, 8-oxyguanine-DNA glycosylase.

‡To whom correspondence should be addressed. E-mail: jcleaver@cc.ucsf.edu.

This article contains supporting information online at www.pnas.org/cgi/content/full/0610619104/DC1.

© 2007 by The National Academy of Sciences of the USA

www.pnas.org�cgi�doi�10.1073�pnas.0610619104 PNAS � January 23, 2007 � vol. 104 � no. 4 � 1389–1394

NEU

ROSC

IEN

CE

control mice. Compound heterozygous Csb�/�/Xpc�/� mice wereviable, and a subset of these mice showed an intermediate neuro-logical phenotype of ataxia at age 3 months. At this age, thecomplexity of dendritic arborization and the intensity of calbindin-D28K was reduced in Csb�/�/Xpc�/� mice as compared withlittermate controls (Fig. 2), indicating a reduction in Purkinje cellhealth. We concentrated on analysis of cerebella functions and haveyet to examine retinal morphology, which is known to degeneratein human CS patients.

To investigate the contribution of oxidative DNA lesions tothe phenotype of Csb�/� mice, Csb�/�/Ogg�/� mice were gen-erated. Ogg�/� mice have only a subtle phenotype (26), and weobserved no obvious phenotype in double-knockout Csb�/�/Ogg�/� mice when followed to 1 year of age (data not shown).

H&E staining demonstrated a progressive loss of Purkinjecells in Csb�/�/Xpc�/� mice from P8 to P21 (Fig. 1 D–I). Losses

ranged from a few missing cells at P14 (Fig. 1H) to long tractsof missing cells at P21 (Fig. 1I). Adjacent sections were stainedfor calbindin. At age P21, the age when differences in calbindinstaining appeared the strongest, the frequency of calbindin-positive cells in Csb�/�/Xpc�/� mice was 86% that of controls(Fig. 1C). This difference was not statistically significant (P �0.18). The Csb�/�/Xpc�/� mice had areas of focal Purkinje cellloss (Fig. 1 G–I) compared with controls (Fig. 1 D–F), but at thelevel of the whole cerebellum, this loss did not statistically reducethe overall number of Purkinje cells.

Apoptosis and Neurogenesis in the Cerebellum of Csb�/�/Xpc�/� Mice.To determine whether the progressive loss of Purkinje cells wascaused by apoptosis, brain sections were labeled with TUNEL(Fig. 3 A–D). At age P21, TUNEL was much more extensive inthe external granule layer of the cerebellum of Csb�/�/Xpc�/�

mice (Fig. 3D) relative to littermate controls (Fig. 3C) and wasobserved rarely in the Purkinje cell layer. In Csb�/�/Xpc�/� mice,TUNEL was more extensive at age P21 (Fig. 3D) than at age P14(Fig. 3B). At age P21, the age when differences in TUNELstaining appeared the strongest, there were 6.8 times moreTUNEL-positive cells per unit area of cerebellar folia in theCsb�/�/Xpc�/� mice compared with controls (P � 0.001) (Fig.3E). To determine whether there were alterations in neurogen-esis in Csb�/�/Xpc�/� mice, mice were injected with BrdU tolabel newly dividing cells, and brain sections were stained forBrdU (Fig. 4). At age P14, no obvious differences in BrdUlabeling were observed between Csb�/�/Xpc�/� (Fig. 4B) miceand control littermates (Fig. 4A).

p53 Immunohistochemistry. We examined the patterns and inten-sity of staining with p53 to identify cells under genotoxic stress.

012345678910

0 10 20 0

0.5

1

1.5

2

2.5

P8

control

Csb-/-/Xpc-/-P14

Csb-/-/Xpc-/-

control

P21A

D

B

H & E

Csb-/-/Xpc-/-control

P21

P21

Pu

rkin

je c

ells

(X 1

0^

3)/

nm

2

(m

ean

+ S

.E.M

.)

wei

gh

t (g

, mea

n +

S.E

.M.)

Age (days)

C

E F

G H I

Fig. 1. Overview of runting and cerebellar morphology in Csb�/�/Xpc�/� mice.(A) Relative size of Csb�/�/Xpc�/� mice and a littermate control at P21. (B) Meanweight of Csb�/�/Xpc�/� mice compared with littermate controls. (C) Frequencyof Purkinje cells per unit area of cerebellum in Csb�/�/Xpc�/� mice compared withlittermate controls. (D–I) Histological sections of cerebellum of control (D–F) anddouble-knockout Csb�/�/Xpc�/� (G–I) mice stained with H&E demonstrating aprogressive loss of Purkinje cells in double-knockout Csb�/� Xpc�/� mice. Micewere examined at P8 (D and G), P14 (E and H), and P21 (F and I). [Scale bars: D, 50�m (applies to D and G); E, 50 �m (applies to E, F, H, and I).]

Fig. 2. Histological sections of mouse cerebella-stained with calbindin-D28kto identify Purkinje cells, showing lack of degeneration in wild type (A) andshowing degeneration (indicated by arrows in B and C) in both Csb�/� andcompound heterozygous Csb�/�/Xpc�/� mice at 3 months of age.

TUNEL

control

control P14

P21

A B

C D

TUN

EL-p

osi

tive

cel

ls /

nm

2

(mea

n +

S.E

.M.)

0

100

200

300

400

500

600control

Csb-/-/Xpc-/- * P21E

Fig. 3. Apoptosis in double-knockout Csb�/� Xpc�/� mice. Sections of cere-bellum of control (A and B) and double-knockout Csb�/�/Xpc�/� (C and D) micelabeled with TUNEL at ages P14 (A and C) and P21 (B and D) showing apoptosisin the external granule cell layer (arrowheads, D) and occasionally in thePurkinje cell layer. [Scale bar: A, 50 �m (applies to A–D).] (E) Frequency ofTUNEL-positive cells per unit area of cerebellar folia in the Csb�/�/Xpc�/� micecompared with controls.

1390 � www.pnas.org�cgi�doi�10.1073�pnas.0610619104 Laposa et al.

Purkinje cells in Csb�/�/Xpc�/� mice showed strong p53 staining,which was absent in littermate controls (Fig. 5). In Csb�/�/Xpc�/� mice, the pattern of p53 staining was in bright focal areas,perhaps indicating nucleolar localization (Fig. 5B). p53 stainingwas absent elsewhere in the cerebella of Csb�/�/Xpc�/� mice(Fig. 5B). p53 staining was observed exclusively in Purkinje cellsthat were also positive for calbindin-D in Csb�/�/Xpc�/� mice(Fig. 5).

Xpc Immunohistochemistry. Immunohistochemical staining for theXpc protein revealed that this damage-binding protein is local-ized to Purkinje cells in wild-type mice with little expressionelsewhere in the cerebellum and absent from Xpc mice (Fig. 6).Xpc�/� mice showed normal cerebellar histology (Fig. 6A).

UV Survival in Embryonic Fibroblasts. We established the functionalconsequences of Csb and Xpc deficiency in MEFs (Fig. 7) usingUV damage, which generates transcription-blocking DNA le-sions that are substrates for global NER and TCR. The Csb�/�

cells were UV-sensitive, Xpc�/� was intermediate, and heterozy-gous Csb�/� and Csb�/�/Xpc�/� and wild type were UV-resistant. Addition of the Ogg�/� genotype to the Csb�/� geno-type did not change the sensitivity. Csb�/�/Xpc�/� MEFs weremostly UV-sensitive, representing a synergistic effect of thecompound deficiency in Csb and Xpc.

Microarray Results. The majority of gene targets in the cerebellumon an expression microarray were unaltered between Csb�/�

mice and wild-type littermate controls; only a small number ofgenes relevant to neurobiology, transcription, and oxidativestress showed significantly altered expression [supporting infor-mation (SI) Table 1]. Underexpressed genes included Purkinjecell protein 4 (pcp4), glutathione peroxidase 1, and RNA-binding motif protein 4. Overexpressed genes included DNAprimase, p58 subunit, and polymerase �2 (mitochondrial). Over-expressed neural genes included myelin basic protein, Synapto-gyrin 3 (which plays a role in learning and memory), neogenin(which plays a role in axonal guidance in the nervous system),

and the potassium voltage-gated channel, shaker-related sub-family, � member 1. Previous results have also shown increasedlevels of the p21 cell cycle regulatory protein and underexpres-sion of collagen 15a1 in human fibroblasts from CS patients (27).

DiscussionPathology of CS in Mice and Humans. Csb�/�/Xpc�/� mice wererunted relative to their littermate controls, being notable at ageP14 and dramatic at age P21, and survived only a few days longer.This phenotype corresponds to major criteria for clinical diag-nosis of CS, which is height below fifth percentile for age and sexwith a mean age at death of 12 years (1, 2). Most CS patientsappear normal at birth (type I or ‘‘classic’’ CS) and exhibitprogressive cachectic dwarfism early in life (28); a subset of CSpatients of type II or early-onset CS present with symptoms atbirth or soon thereafter (1). Progressive growth retardation andearly lethality are also observed in other mouse models of CS:Csb�/�/Xpa�/� (29), Xpg�/� (30), and Xpg�/�/Xpa�/� (31).

The range of neurological phenotypes of our Csb�/�/Xpc�/�,Csb�/�/Xpc�/�, and Csb�/� mice resemble, to a first approxi-mation, the range of CS disease previously classified (1);Csb�/�/Xpc�/� would correspond to the most severe form thatresults in neonatal death, cerebro-oculo-facial-skeletal syn-drome (32).

The cerebellum contains two main cell types, granule andPurkinje cells, and is the primary center of motor coordinationin the CNS. Granule cells are the most abundant and provideinput to the Purkinje cells, whose axons are the sole output fromthe cerebellar cortex providing input to multiple target cells. Thedevelopment of the human cerebellum begins in utero andcontinues for several years postnatally. Protracted development

p53/Calbindin

control Csb-/-/Xpc-/-P21

A B

Fig. 5. p53 and calbindin immunohistochemistry in control littermate (A) andCsb�/�/Xpc�/� (B) mice at age P21. [Scale bar: A, 50 �m (applies to A and B).]

wild typeXpc-/- P21

A BXpc

Fig. 6. Immunohistochemical staining for the Xpc protein in cerebellarsections of wild-type (A) mice at age P21 and Xpc�/� controls. (B) Demonstra-tion of the presence of the Xpc protein in Purkinje cells (arrowheads, B). [Scalebar: A, 50 �m (applies to A and B).]

Fig. 7. Survival of MEFs after UV irradiation. Data are shown for wild type(solid square), Csb�/� (open circle), Csb�/� (closed triangle), Xpc�/� (soliddiamond), compound heterozygous Csb�/�/Xpc�/� (open square), double-knockout Csb�/�/Xpc�/� (closed circle), and double-knockout Csb�/�/Ogg�/�

(closed inverted triangle).

AA BrdU BrdU

control P14

BB

Csb-/-/Xpc-/- P14

Fig. 4. BrdU staining indicating cell proliferation in the external granule celllayer of a control littermate (A) and Csb�/�/Xpc�/� (B) at age P14. [Scale bar:A, 50 �m (applies to A and B).]

Laposa et al. PNAS � January 23, 2007 � vol. 104 � no. 4 � 1391

NEU

ROSC

IEN

CE

puts the cerebellum at risk for damage over a long period fromagents acting in the perinatal or early postnatal period (33).Cerebellar granule cells are vulnerable to a variety of toxins thatdecrease glutathione levels and make the cells more vulnerableto DNA and other cellular damage from reactive oxygen species(34, 35). Purkinje cells are susceptible to ischemic death becauseof their reduced capacity to sequester glutamate and reducedability to generate energy during anoxia (36). Some tissues thatdegenerate in CS appear to be sensitive to oxygen levels,including the Purkinje cells, retina, and oligodendrocytes (37–40). Retinal degeneration is linked to that of Purkinje cells inboth CS and the nervous (nr) mutant mouse (41).

We observed loss of both Purkinje cells and granule cells inCsb�/�/Xpc�/� mice. Stunted dendritic trees of Purkinje cellswere observed in compound knockout Csb�/�/Xpa�/� mice,along with decreased thickness of the molecular layer andinternal granule layer (29). A central neuropathological findingin CS patients is extensive cerebellar atrophy, including a markedreduction of granule cells and the loss of Purkinje cells (2, 28).These observations suggest that the cerebellum is particularlysusceptible to DNA damage, and that DNA repair deficiencieshave a profound effect on cerebellar pathology. Although Csbexpression may be ubiquitous in the brain (P. J. Brooks, personalcommunication), Xpc protein was uniquely abundant in Pur-kinje cells (Fig. 6), which may explain why Purkinje neurons wereparticularly affected in Csb�/�/Xpc�/� mice. There was noobvious pathology of Purkinje cells in Xpc�/� mice, but TCR inhuman XPC cells results in completion of repair at a reduced ratein nonproliferating cells (42). Deficiencies in Xpc may thereforebe unimportant in the brain, being compensated by Csb activity,but are important when associated with TCR deficiency inCsb�/�/Xpc�/� mice.

Neurodegeneration in Repair-Deficient and Other Mouse KnockoutStrains. A dramatic increase in apoptosis was identified in theexternal granule cells of the cerebella of Csb�/�/Xpc�/� mice,suggesting that the loss of both Csb and Xpc rendered these cellsvulnerable to cell death. We saw no stimulation of cells into cellcycle, which would have resulted in the incorporation of BrdU,as suggested by some models of neurodegeneration (43). Loss ofPurkinje cells in Csb�/�/Xpc�/� mice was not associated with thetypical TUNEL-positive staining indicative of classical apopto-sis. The difference in these two cell types suggests that themechanism contributing to loss of Purkinje cells may vary inproliferating precursors and in postmitotic neurons. Consistentwith this notion, Purkinje cells of Csb�/�/Xpc�/� mice stainedstrongly with an antibody to total p53, suggesting that these cellsare under significant stress and have an alternative cell deathpathway. Cell death in the Purkinje layer may be due toendogenous activation of p53 or loss of maintenance signals,such as cytokines, through the horizontal fibers. Both intrinsicand extrinsic components contribute to the neurodegenerationof the cerebellum as a result of Csb and Xpc deficiency.

Reducing repair capacity by crossing with Xpa�/� mice thathave no NER can also increase the severity of neurodegenerativephenotype. These include Csb�/� mice crossed with Xpa�/� mice(29), trichothiodystrophy (TTD) mice crossed with Xpa�/� (24),and Xpg�/� crossed with Xpa�/� (31). Crossing Xpc�/� withXpa�/� did not, however, enhance the cancer susceptibility ordevelopment of neurological symptoms (16). Unlike humanXPA patients who are usually very severely affected, Xpa�/�

mice have no neurological deficit and have only a slight reductionin lifespan (29, 44). The Xpc�/�/Xpa�/� cross therefore repre-sents a combination in which neither possesses a neurologicalphenotype and is uninformative as to mechanisms of neurode-generation.

Other repair-deficient mouse strains exhibit increased celldeath and/or increased p53 staining in the brain. Excision repair

cross-complementing group 1 (ERCC1) in association with XPF,forms the structure-specific 5� endonuclease of NER (3);Ercc1�/� mice show increased p53 levels in the brain and liver(45). Mice deficient in DNA ligase IV, which functions in thenonhomologous end-joining pathway of DNA repair, exhibitapoptosis (TUNEL staining) in the developing nervous system(46) and increased p53 (47). In Xpg�/� mice, pyknotic figures andreduced calbindin staining were observed in the Purkinje layer,whereas TUNEL staining was strongest in the granule layer andrare in the Purkinje layer (48). In Csb�/�/Xpa�/� mice, TUNELstaining was increased in the external granule layer at P8 (29).Oxidative stress has been identified as a mechanism for the lossof Purkinje cells in the repair-deficient Atm mouse (37, 39, 49).TUNEL staining and oxidative stress have been similarly ob-served in cerebellar granule cells from clinical samples of CS andXPA patients (50–53).

Purkinje cells are capable of dying by apoptosis in response toDNA damage under some circumstances. In organotypic slicecultures of mouse cerebellum, bleomycin, which damages DNA byoxidative stress, increased the number of TUNEL- and p53-positiveneurons in the internal granule layer and Purkinje cell layer (54).These responses were not observed in p53�/� mice, demonstratingthat p53 is required for TUNEL-positive apoptosis in both thegranule and Purkinje cell layers, at least in culture (54). In othersituations, such as trophic factor deprivation, Purkinje cells die bya form of programmed cell death distinct from the apoptotic deathof neighboring granule neurons (55).

Are DNA Lesions Important for Neurodegeneration in CS? The mostobvious source of endogenous DNA damage would be the highlevel of oxidative metabolism in the brain (56), combined withits relative deficiency of antioxidative protective mechanisms(57), which is increased in many instances of neurodegeneration(58, 59). DNA lesions are repaired more slowly in neuronsrelative to dividing cells, suggesting they are likely to accumulatein neurons over time (59). Therefore, DNA damage responsesand repair pathways are likely relevant to a number of neuro-degenerative diseases, even if they do not arise directly from amutation in a DNA repair enzyme. Mutations in the XPDcomponent of the transcription factor TFIIH, with which CSproteins interact, also result in many transcriptional defects,including reduced responses to nuclear receptors (60, 61), dys-regulation of peroxisome proliferator-activated target genes(62), vitamin D receptor-responsive genes (63), and thyroidhormone-responsive genes leading to hypomyelination (64).Both CSA and CSB cells also exhibit down-regulation of collagen15a1 expression and dysregulation of p21 (27), and CSB plays arole in chromatin structure (13).

We observed that the sensitivity of MEFs to UV damageincreased in the order: WT; Csb�/�/Xpc�/�; Xpc�/�; Csb�/�;Csb�/�/Xpc�/� (Fig. 7). No increase in sensitivity of Csb�/�

MEFs occurred by adding Ogg1 deficiency (Fig. 7). With theexception of the double heterozygote that was phenotypicallyvariable, this order approximates the pathological severity. Be-cause UV sensitivity is a measure of NER capacity, theseobservations would be consistent with a role for NER in theclinical phenotype.

Several reports demonstrate that human and mouse CS cellsare sensitive to oxidative damage and do not repair the oxidativelesion 8-OH-G (6–11, 23). The CSB protein also interacts withPARP-1, a sensor of DNA breaks from oxidative damage (65,66). No differences were observed in 8-OH-G between CS andcontrol autopsy material (51), despite the higher amounts ofprotein oxidation and lipid oxidation in the brains of CS patients(50). The more severe phenotype of Csb�/�/Xpc�/� mice com-pared with Csb�/�/Ogg�/� mice suggests that lesions that aresubstrates for global NER and block transcription are moreimportant than smaller oxidative DNA lesions for progressive

1392 � www.pnas.org�cgi�doi�10.1073�pnas.0610619104 Laposa et al.

neurodegeneration, or that the Csb�/� genotype already resultsin defective repair of 8-OH-G (5).

Some DNA lesions, such as O6-alkylguanine (67) and thymineglycol (68), are substrates for either base excision repair or NER,and Ogg1 acts on several oxidative lesions. Oxidative stress canalso generate oxidative purine products (5�,8-purine cyclode-oxynucleosides) that could be formed in the brain and requireNER for their repair and block transcription (69, 70). Lipidoxidation products produce malondialdehyde-deoxyguanosineadducts in DNA that block transcription (71). Thus, NER maybe of more importance in repairing oxidative damage than wasthought previously, although its precise mechanism in terminallydifferentiated neurons may be subtly different from that individing cells (72).

We envision a scenario in which TCR (mediated by CSB)continually repairs endogenous oxidative bulky oxidative DNAlesions in terminally differentiated neurons, and in which aninability to carry out this process results in neurodegeneration.The relative importance of backup DNA repair pathways forthese lesions (mediated by XPC) must differ between humansand mice. In mice, where a strong neuronal phenotype wasobserved in double-knockout Csb�/�/Xpc�/� animals, the role ofXpc-mediated backup repair of such lesions must play a largerrole than in humans.

The loss of autophagy would seem a very nonspecific defectunrelated to CS, but it results in a pattern of cell death in thecerebellum very similar to Csb mice. TUNEL-positive cells wereobserved in the granular layer and a marked reduction inPurkinje cell number without concomitant TUNEL staining (73,74). The nervous (nr) mutant involves overexpression of theplasminogen activator pathway and undergoes degeneration ofboth Purkinje and retinal cells (41). Xpc and Csb are bothcomponents of the endogenous proteolytic system, Xpc throughthe proteasome (75) and Csb and Csa as cofactors of an E3ubiquitylation ligase (12). Consequently, CS, like many otherneurodegenerative disorders, could be caused by defective clear-ance of denatured proteins (76). This view is supported by thereport of a mildly UV-sensitive patient with minimal neurode-generative symptoms, in which there was negligible CSB proteinbecause of a truncation mutation close to the 5� terminus (77).Defective TCR in CS would then be a consequence of anupstream primary defect in protein processing that leads toneurodegeneration, rather a proximate cause.

Materials and MethodsMouse Strains. Csb�/� mice were a gift from C. van der Horst andJ. H. J. Hoeijmakers (Erasmus University Medical Center,Rotterdam, The Netherlands) (14). Genotyping was performedby PCR analysis of DNA isolated from tail tips (19) or com-mercially (Transnetyx, Memphis, TN). Xpc�/� male mice (25)were purchased from Taconic Farms (Germantown, NY) andmated to the heterozygous Csb�/� mice. Genotyping for the Xpcalleles was performed by Southern blotting, using a plasmid

probe provided by Taconic Farms (25). Mice deficient in Ogg1(26) were obtained from C. A. Walter (University of TexasHealth Science Center, San Antonio, TX) with permission ofThomas Lindahl (Cancer Research UK, London, U.K.); geno-typing was performed by PCR, and they were mated with Csb�/�

mice. Further crosses were made to generate double-heterozygous and homozygous strains (see SI Text).

MEFs and UV Survival. MEFs were derived from single embryonicday (E)13–E14 embryos (without head, liver, and blood organs)of each genotype and kept in liquid nitrogen until genotyping wascompleted. The yolk sac was kept for genotyping. For cellsurvival, a rapid well assay was used involving measuring thegrowth of cells for 7 days in 12-well plates (78). Cells (5 � 104)were seeded in each well and irradiated with graded doses of UV(254 nm) light, and cell numbers remaining attached werecounted at 7 days.

Histology and Immunohistochemistry. Mice were anesthetized withpentobarbital (200 mg/kg) and perfusion-fixed with ice-coldfreshly prepared 4% paraformaldehyde (Sigma, St. Louis, MO)in PBS. Brains were removed, fixed in ice-cold freshly prepared4% paraformaldehyde in PBS, and processed for paraffin sec-tions. Slides were incubated with antibodies against calbindin-D28K. Apoptosis was determined in sections by using TUNELstaining. DNA replication was measured by using BrdU staining.Mice were injected i.p. with BrdU (Roche Applied Science,Indianapolis, IN) dissolved in PBS (50 mg/kg) 1 h before deathand processed with an In Situ Cell Proliferation Kit (FLUOSRoche Applied Science). Xpc was detected by using a mousemonoclonal antibody (no. GTX70294; Genetex, Boston, MA).Slides were counterstained with hematoxylin (see SI Text forcomplete details).

Microarrays. RNA was extracted from cerebellar homogenates of6-month-old Csb-null or wild-type mice by using affinity resinchromatography (RNEasy; Qiagen, Valencia, CA). ReferenceRNA was extracted as a large batch from pooled mouse embryosby using standard TRIzol extraction methods. Hybridization wasconducted with RNA from one mouse hybridized against referenceRNA, and five arrays for each genotype were hybridized onoligonucleotide microarrays (see SI Text for complete details).

We thank Amy Tang for performing the cryosectioning and Luz Feeneyfor assisting with MEF culture. We thank the Luke O’Brien Foundation,whose continued support and encouragement were instrumental infurthering this work on CS. The work described here was also supportedin part by the National Institute of Environmental Health Sciences[Grant 1 R01 ES 8061 (to J.E.C.)], the National Institute of NeurologicalDisorders and Stroke [Grant 1R01NS052781 (to J.E.C.)], the NationalInstitutes of Health [Grants NS44223 and NS48393 (to E.J.H.)], and theCancer Center Support [Grant P30 CA82103 (Principal Investigator, F.McCormick), and by a National Cancer Institute Ruth KirstensenPostdoctoral Fellowship [1 F32 CA099499-01A1 (to R.R.L.)].

1. Nance MA, Berry SA (1992) Am J Med Genet 42:68–84.2. Leech RW, Brumback RA, Miller RH, Otsuka F, Tarone RE, Robbins JH

(1985) J Neuropathol Exp Neurol 44:507–519.3. Wood RD, Mitchell M, Sgouros J, Lindahl T (2001) Science 291:

1284–1289.4. Bohr VA, Sander M, Kraemer KH (2004) DNA Repair (Amsterdam) 4:293–

302.5. Licht CL, Stevnser T, Bohr VA (2003) Am J Hum Genet 73:1217–1239.6. Cooper PK, Nouspikel T, Clarkson SG, Leadon SA (1997) Science 275:990–993.7. LePage F, Kuroh EE, Aurutskaya A, Gentil A, Leadon SA, Sarasin A, Cooper

PK (2000) Cell 101:159–171.8. Check E (2005) Nature 435:1015.9. Tuo J, Jaruga P, Rodriguez H, Bohr VA, Dizdaroglu M (2003) FASEB 17:668–674.

10. Osterod M, Larsen E, Le Page F, Hengstler JG, Van Der Horst GT, BoiteuxS, Klungland A, Epe B (2002) Oncogene 21:8232–8239.

11. de Waard H, de Wit J, Gorgels TG, van den Aardweg G, Andressoo JO,Vermeij M, van Steeg H, Hoeijmakers JH, van der Horst GT (2003) DNARepair (Amsterdam) 2:13–25.

12. Groisman R, Polanowski J, Kuraoka I, Sawada, J-I, Saijo M, Drapkin R,Kisselev AF, Tanaka K, Nakatani Y (2003) Cell 113:357–367.

13. Newman JC, Bailey AD, Weiner AM (2006) Proc Natl Acad Sci USA 103:9613–9618.14. van der Horst GT, van Steeg H, Berg RJ, van Gool AJ, de Wit J, Weeda G,

Morreau H, Beems RB, van Kreijl CF, de Gruijl FR, et al. (1997) Cell89:425–435.

15. van der Horst GT, Meira L, Gorgels TG, de Wit J, Velasco-Miguel S,Richardson JA, Kamp Y, Vreeswijk MP, Smit B, Bootsma D, et al. (2002) DNARepair 1:143–157.

16. Friedberg EC, Meira LB, Cheo DL (1998) Mutat Res 407:217–226.17. Berg RJ, Rebel H, van der Horst GT, van Kranen HJ, Mullenders LH, van

Vloten WA, de Gruijl FR (2000) Cancer Res 60:2858–2863.

Laposa et al. PNAS � January 23, 2007 � vol. 104 � no. 4 � 1393

NEU

ROSC

IEN

CE

18. Wijnhoven SW, Kool HJ, van Oostrom CT, Beems RB, Mullenders LH, vanZeeland AA, van der Horst GT, Vrieling H, van Steeg H (2000) Cancer Res60:5681–5687.

19. Wijnhoven SW, Kool HJ, Mullenders LH, Slater R, van Zeeland AA, VrielingH (2001) Carcinogenesis 1099–1106.

20. de Waard H, de Wit J, Andressoo JO, van Oostrom CT, Riis B, Weimann A,Poulsen HE, cvan Steeg H, Hoejmakers JH, van der Horst GH (2004) Mol CellBiol 24:7941–7948.

21. Ide H (2001) Prog Nucleic Acid Res Mol Biol 68:207–221.22. Laposa RR, Henderson JT, Wells PG (2003) FASEB J 17:1343–1345.23. Spivak G, Hanawalt PC (2006) DNA Repair (Amsterdam) 5:13–22.24. de Boer J, Andressoo JO, de Wit J, Huijmans J, Beems RB, van Steeg H, Weeda

G, van der Horst GTJ, van Leeuwan W, Themmen APN, et al. (2002) Science296:1276–1279.

25. Sands AT, Abuin A, Sanchez A, Conti CJ, Bradley A (1995) Nature 377:162–165.

26. Klungland A, Rosewell I, Hollenbach S, Larsen E, Daly G, Epe B, Seeberg E,Lindahl T, Barnes DE (1999) Proc Natl Acad Sci USA 96:13300–13305.

27. Cleaver JE, Hefner E, Laposa RR, Karentz D, Marti T (2006) Neuroscience,in press.

28. Otsuka F, Robbins JH (1985) Am J Dermatopathol 7:387–392.29. Murai M, Enokido Y, Inamura N, Yoshino M, Nakatsu Y, van der Horst GT,

Hoeijmakers JH, Tanaka K, Hatanaka H (2001) Proc Natl Acad Sci USA98:13379–13384.

30. Harada YN, Shiomi N, Koike M, Ikawa M, Okabe M, Hirota S, Kitamura Y,Kitagawa M, Matsunaga T, Nikaido O, Shiomi T (1999) Mol Cell Biol19:2366–2372.

31. Shiomi N, Mori M, Kito S, Harada Y-N, Tanaka K, Shiomi T (2005) DNARepair (Amsterdam) 4:351–357.

32. Graham JMJ, Anyane-Yeboa K, Raams A, Appeldoorn E, Kleijer WJ,Garritsen VH, Busch D, Edersheim, TG, Jaspers NG (2001) Am J Hum Genet69:291–300.

33. ten Donkelaar HJ, Lammens M, Wesseling P, Thijssen HO, Renier WO (2003)J Neurol 250:1025–1036.

34. Fonnum F, Lock EA (2000) Toxicol Lett 112- 113:9–16.35. Fonnum F, Lock EA (2004) J Neurosci 88:513–531.36. Welsh JP, Yuen G, Placantonakis DG, Vu TQ, Haiss F, O’Hearn E, Molliver

ME, Aicher SA (2002) Adv Neurol 89:331–359.37. Chen P, Peng C, Luff J, Spring K, Watters D, Bottle S, Furuya S, Lavin MF

(2003) J Neurosci Res 23:11453–11460.38. Liu K, Akula JD, Falk C, Hansen RM, Fulton AB (2006) Invest Ophthalmol

Visual Sci 47:2639–2647.39. Barlow C, Dennery PA, Shigenaga MK, Smith MA, Morrow JD, Roberts LJ

n, Wynshaw-Boris A, Levine RL (1999) Proc Natl Acad Sci USA 96:9915–9919.40. Goldbaum O, Richter-Landsberg C (2004) J Neurosci 24:5748–5757.41. Li J, Ma Y, Teng YD, Zheng K, Vartanian TK, Snyder EY, Sidman RL (2006)

Proc Natl Acad Sci USA 103:7847–7852.42. Maher VM, Dorney DJ, Mendrake AL, Konze-Thomas B, McCormick JJ

(1979) Mutat Res 62:311–323.43. Herrup K, Neve R, Ackerman SL, Copani A (2004) J Neurosci 24:9232–9239.44. Dolle ME, Busuttil RA, Garcia AM, Wijnhoven S, van Drunen E, Niedern-

hofer LJ, van der Horst G, Hoeijmakers JH, van Steeg H, Vijg J (2006) MutatRes 596:22–35.

45. McWhir J, Selfridge J, Harrison DJ, Squires S, Melton D (1993) Nat Genet5:217–224.

46. Barnes DE, Stamp G, Rosewell I, Denzel A, Lindahl T (1998) Curr Biol8:1395–1398.

47. Lee Y, Barnes DE, Lindahl T, McKinnon PJ (2000) Genes Dev 14:2576–2580.48. Sun XZ, Harada YN, Takahashi S, Shiomi N, Shiomi T (2001) J Neurosci Res

64:348–354.49. Barlow C, Hirotsune S, Paylor R, Liyanage M, Eckhaus M, Collins F, Shiloh

Y, Crawley JN, Ried T, Tagle D, Wynshaw-Boris A (1966) Cell 86:159–171.50. Hayashi M, Itoh M, Araki S, Kumada S, Shioda K, Tamagawa K, Mizutani T,

Morimatsu Y, Minagawa M, Oda M (2001) J Neuropathol Exp Neurol 60:350–356.

51. Hayashi M, Araki S, Kohyama J, Shioda K, Fukatsu R (2005) Brain Dev27:34–38.

52. Kohji T, Hayashi M, Shioda K, Minagawa M, Morimatsu Y, Tamagawa K, OdaM (1998) Neurosci Lett 243:133–138.

53. Hayashi M (1999) No To Hattatsu 31:146–152.54. Inamura N, Araki T, Enokido Y, Nishio C, Aizawa S, Hatanaka H (2000)

J Neurosci Res 60:450–457.55. Florez-McClure ML, Linseman DA, Chu CT, Barker PA, Bouchard RJ, Le SS,

Laessig TA, Heidenreich KA (2004) J Neurosci 24:4498–4509.56. Lindahl T (1993) Nature 362:709–715.57. Olanow CW (1992) Ann Neurol 32:S2–S9.58. Kruman I (2004) Cell Cycle 3:769–773.59. McMurray CT (2005) Mutat Res 577:260–274.60. Bastien J, Adam-Stitah S, Riedl T, Egly JM, Chambon P, Rochette-Egly C

(2000) J Biol Chem 275:21896–21904.61. Keriel A, Stary A, Sarasin A, Rochette-Egly C, Egly JM (2002) Cell 109:125–

135.62. Compe E, Drane P, Laurent C, Diderich K, Braun C, Hoeijmakers JH, Egly

JM (2005) Mol Cell Biol 25:6065–6076.63. Drane P, Compe E, Catez P, Chymkowitch P, Egly JM (2004) Mol Cell

16:187–197.64. Compe E, Malerbe M, Soler L, Marescaux J, Borrelli E, Egly JM (2007) Cell,

in press.65. Flohr C, Burkle A, Radicella JP, Epe B (2003) Nucleic Acids Res 31:5332–5337.66. Thorsland T, von Kobbe C, Harrigan JA, Indig FE, Christiansen M, Stevsner

T, Bohr VA (2005) Mol Cell Biol 25:7625–7636.67. Voigt JM, Van Houten B, Sancar A, Topal MD (1989) J Biol Chem 264:5172–

5176.68. Kow YW, Wallace SS, Van Houten B (1990) Mutat Res 235:147–156.69. Kuraoka I, Robins P, Masutani C, Hanaoka F, Gasparutto D, Cadet J, Wood

RD, Lindahl T (2001) J Biol Chem 276:49283–49288.70. Brooks PJ, Wise DS, Berry DA, Kosmoski JV, Smerdon MJ, Somers RL,

Mackie H, Spoonde AY, Ackerman EJ, Coleman K, et al. (2000) J Biol Chem275:22355–22362.

71. Cline SD, Riggins JN, Tornaletti S, Marnett LJ, Hanawalt PC (2004) Proc NatlAcad Sci USA 101:7275–7280.

72. Nouspikel T, Hanawalt P (2000) Mol Cell Biol 20:1562–1570.73. Komatsu M, Waguri S, Chiba T, Murata S, Iwata J, Tanida I, Ueno T, Koike

M, Uchiyama Y, Kominami E, Tanaka K (2006) Nature 441:880–886.74. Hara T, Nakamura K, Matsui M, Yamamoto A, Nakahara Y, Suzuki-

Migishima R, Yokoyama M, Mishima K, Saito I, Okano H, Mizushima N(2006) Nature 441:885–889.

75. El-Mahdy MA, Zhu Q, Wang QE, Wani G, Praetorius-Ibba M, Wani AA(2006) J Biol Chem 281:13404–13411.

76. Ciechanover A, Brundin P (2003) Neuron 40:427–446.77. Horibata K, Iwamoto Y, Kuraoka I, Jaspers NG, Kurimasa A, Oshimura M,

Ichihashi M, Tanaka K (2004) Proc Natl Acad Sci USA 101:15410–15415.78. Cleaver JE, Thomas GH (1988) J Invest Dermatol 90:467–471.

1394 � www.pnas.org�cgi�doi�10.1073�pnas.0610619104 Laposa et al.