A New Locus for Spinocerebellar Ataxia (SCA21) Maps to Chromosome 7p21.3-p15.1 Isabelle Vuillaume, MD,1 David Devos, MD,3 Susanna Schraen-Maschke, PharmD, PhD,1,2 Christian Dina, PhD,4 Arnaud Lemainque, PhD,5 Francis Vasseur, PhD,4 Guy Bocquillon, BS,1 Patrick Devos, BS,1 Carole Kocinski, BS,1 Christiane Marzys, BS,1 Alain Destée, MD, PhD3 and Bernard Sablonnière, MD, PhD1,2 We investigated a French family with a new type of autosomal dominant spinocerebellar ataxia that was excluded from all previously identified genes and loci. The patients exhibited a slowly progressive gait and limb ataxia variably associated with akinesia, rigidity, tremor, and hyporeflexia. A mild cognitive impairment also was observed in some cases. We performed a genomewide search and found significant evidence for linkage to chromosome 7p21.3-p15.1. Analysis of key recombinants and haplotype reconstruction traced this novel spinocerebellar ataxia locus to a 24cM interval flanked by D7S2464 and D7S516. Ann Neurol 2002;52:666 – 670 The autosomal dominant cerebellar ataxias (ADCAs) constitute a group of neurodegenerative diseases that can be classified clinically into three categories.1 All ADCAs are characterized by ataxia variably associated with other neurological features. Clinical features are variable among ADCAs, even in the same family, and genetic studies have shown a great heterogeneity in these diseases.2 Thus, classification based on the responsible genes or genetic loci is becoming prevalent. Molecular genetic studies have shown that ADCAs are genetically heterogeneous, and 15 different loci already have been reported to date.2– 4 Among these loci, causative genes have been further identified as expansions of trinucleotide (CAG) repeats for SCA1 (MIM 164400), SCA2 (MIM 183090), SCA3 (MIM 109150), SCA6 (MIM 183086), SCA7 (MIM 164500), and more recently for SCA17 (MIM 600075),5 whereas untranslated CTG or CAG tracts are identified in SCA8 (MIM 603680) or SCA12 (MIM 604326). In SCA10 (MIM 603516), the responsible gene has been identified as a large expansion of a ATTCT repeat tract.6 However, the genes or loci have not been identified in approximately 40% of French ADCA families,4 implying the presence of other unidentified responsible genes. We recently have identified a four-generation French family that segregates a new form of ADCA.7 All known ADCA genes and loci were excluded by direct mutation or geneticlinkage analysis. Using a genomewide linkage analysis, we mapped the locus for this family to chromosome 7p21.3-p15.1. This novel locus was registered as SCA21, with approval from the Human Genome Nomenclature Committee. Subjects and Methods Subjects The subjects were recruited from a four-generation French family in which a new form of ADCA was ascertained, clinically characterized by gait and limb ataxia, akinesia, hyporeflexia, and mild cognitive impairment. The magnetic resonance imaging showed atrophy of the cerebellum without involvement of the brainstem7 and basal ganglia (data not shown). Twenty-three family members were examined by at least one of us (D.D. and A.D.), and 15 met the disease criteria defined as the presence of gait or limb ataxia. After informed consent (approved by the institutional ethics committee of CHRU) was obtained, blood samples were drawn from 29 subjects (Fig). Genotype Analysis From the 1Unité Fonctionnelle de Neurobiologie, Laboratoire de Biochimie et Biologie moléculaire; 2Institut National de la Santé et de la Recherche Médicale U422; 3Fédération de Neurologie, Hôpital R. Salengro, Faculté de Médecine and Centre Hospitalier Régional et Universitaire; 4CNRS UPRESA 8090, Institut de Biologie de Lille, Lille; and 5Centre National de Génotypage, Evry, France. Received Mar 18, 2002, and in revised form Jun 18. Accepted for publication Jun 21, 2002. Published online Aug 25, 2002, in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/ana.10344 Address correspondence to Dr Sablonnière, Unité Fonctionnelle de Neurobiologie, Laboratoire de Biochimie et Biologie moléculaire, Hôpital R. Salengro, Centre Hospitalier Régional et Universitaire, Bd du Professeur Leclerc, 59037 Lille Cedex, France. E-mail: email@example.com 666 © 2002 Wiley-Liss, Inc. A genomewide screening was performed with the ABI PRISM Linkage Map Set (version 2.0 PE; Applied Biosystems, Foster City, CA) consisting of 400 highly polymorphic fluorescent-labeled microsatellite markers chosen from the Généthon linkage map8 with an average spacing of 10 to 20cM. After a multiplex polymerase chain reaction (PCR) protocol, the markers were coamplified with an average of four markers per PCR. The amplifications were done with the GeneAmp 9700 PCR system, with a total volume of 10l containing 10% PCR buffer, 3.0mM MgCl2, 0.25mM each dNTP, between 0.65 and 4.5pmol each primer, 20ng DNA, and 0.4 units AmpliTaq Gold. Samples were subjected to an initial denaturation step of 95°C for 12 minutes, followed by 30 cycles of 94°C for 15 seconds, 55°C for 15 seconds, and 72°c for 30 seconds with a final extension of 72°C for 10 minutes. PCR products were pooled, purified, Fig. Pedigree of the family in this study. (squares) Men; (circles) women. (filled symbols) Affected subjects; (open symbols) unaffected subjects; (symbol with a dot) possible carrier. (diagonal line) Deceased. Haplotypes are presented below symbols. Alleles corresponding to D7S481, D7S2514, D7S2464, D7S2557, D7S503, D7S673, D7S516, D7S2515, D7S2848, and D7S2252 are shown from top to bottom. Black bars represent the disease haplotype. Informative recombination events place the SCA21 disease locus between markers D7S2514 and D7S2515. and separated with a MegaBace 1000 automated sequencer (Amersham Biosciences, Arlington Heights, IL). Alleles were sized using the GENETIC PROFILER Program (version 1.1). Linkage Analysis Linkage analysis was performed using the LINKAGE Program Package (version 5.2),9 assuming an autosomal dominant inheritance and a disease frequency of 1 to 100,000. Pairwise logarithm of odds (LOD) scores and multipoint LOD scores were calculated using programs from FASTLINK 3.0P. Penetrance in each subject was assigned to the four liability classes determined from the age at onset: 0.95 at age 40 years or more; 0.85 at age 31 to 40 years; 0.65 at age 21 to 30 years; and 0.50 at age 20 years or less. The allele frequencies of each marker for LOD score analysis were those defined by Généthon in the white population and reported in the Genome Database. Results Clinical Features A summary of the clinical profile of the 15 patients is given in Table 1. Four patients (III-3, III- 5, III-7, and III-12) classified as unaffected on initial examination 3 years ago further developed ataxia as well as hyporeflexia and thus were classified as affected. At the time of examination, gait or limb ataxia was present in all affected subjects. The most striking features was a variable onset and a slow progression of the disease. Average age at onset was 17.4 ⫾ 7.4 years (n ⫽ 14) and ranged from 6 to 30 years. Age at onset tends to be earlier through successive generations. In 14 informative parent-child pairs, the mean anticipation was 10.6 ⫾ 4.6 years, ranging from 4 to 18 years. Cerebellar testing disclosed gait and limb ataxia associated with dysgraphia and dysarthria. Slight to mild akinesia of the limbs was often present with inconstant rigidity, postural and rest tremor of the limbs without response to dopa. There was no motor deficit, no cranial nerve palsy, no pyramidal sign, and no sensory loss, but there was hyporeflexia in all but two Vuillaume et al: Spinocerebellar Ataxia (SCA21) 667 Table 1. Summary of Clinical Findings Patient Characteristics Age at examination (yr) Age at onset (yr) Gait ataxia Limb ataxia Dysarthria Dysgraphia Akinesia Rigidity Tremor Hyporeflexia Hyperreflexia Mental status (MMS or IQ*) I-2 II-3 II-5 II-8 II-10 II-12 II-13 II-15 III-2 III-3 III-5 III-7 III-10 III-12 III-13 65 30 ⫹ ⫹ ⫹ ⫹ ⫹ ⫹ ⫹ ⫹ ⫺ 24/30 45 19 ⫹ ⫹ ⫹ ⫹ ⫹ ⫹ ⫹ ⫹ ⫺ 28/30 42 25 ⫹ ⫹ ⫹ ⫹ ⫹ ⫹ ⫹ ⫹ ⫺ 25/30 41 25 ⫹ ⫹ ⫹ ⫹ ⫹ ⫺ ⫺ ⫹ ⫺ 27/30 37 20 ⫹ ⫹ ⫹ ⫺ ⫹ ⫺ ⫹ ⫹ ⫺ 28/30 36 12 ⫹ ⫹ ⫹ ⫹ ⫹ ⫺ ⫹ ⫹ ⫺ 24/30 34 25 ⫹ ⫺ ⫹ ⫺ ⫹ ⫺ ⫺ ⫹ ⫺ 29/30 32 20 ⫹ ⫹ ⫺ ⫹ ⫹ ⫺ ⫹ ⫺ ⫹ 28/30 24 7 ⫹ ⫹ ⫺ ⫹ ⫹ ⫹ ⫹ ⫹ ⫺ ND 21 20 ⫹ ⫺ ⫺ ⫺ ⫺ ⫺ ⫺ ⫹ ⫺ ND 22 21 ⫹ ⫹ ⫹ ⫹ ⫹ ⫺ ⫺ ⫹ ⫺ ND 11 10 ⫹ ⫺ ⫹ ⫺ ⫺ ⫺ ⫺ ⫹ ⫺ ND 9 6 ⫹ ⫹ ⫹ ⫹ ⫹ ⫺ ⫹ ⫹ ⫺ 64* 13 13 ⫹ ⫹ ⫺ ⫺ ⫺ ⫺ ⫺ ⫹ ⫺ ND 14 9 ⫹ ⫹ ⫹ ⫹ ⫹ ⫹ ⫹ ⫺ ⫹ 70* MMSE ⫽ Mini-Mental State Examination; ND ⫽ not determined. patients. The ophthalmological examination showed intermittent microsaccadic pursuit and square wave jerks without nystagmus nor retinopathy. Neuropsychological examination shows a cognitive impairment, moderate in two patients (I-2 and II-12) and more severe in two young children with IQs of 64 and 70, respectively.7 Reevaluation of clinical data in affected members of generation II showed no obvious worsening of neurological features 3 years later. Contrary to the other family members, II-3 had a positive medical history of neuroleptic treatment, carbon monoxide intoxication, and early beginning alcoholism leading to several hospitalizations and the loss of the custody of her children. The cerebellar syndrome was unchanged, but there were orofacial and head dyskinesia associated to a severe extrapyramidal syndrome with axial and limbs rigidity associated to rest and action tremor. Genotypic and Linkage Data The first convincing linkage evidence for the disease locus was obtained from D7S516, for which we detected a positive LOD score of 1.93 at recombination fraction of 0.08. We explored this result further by using 13 additional markers from this region, selected from the Généthon linkage map8 and Cooperative Human Linkage Center database10 to confirm and refine the genetic localization. LOD scores greater than 3.20 were obtained for six consecutive markers around D7S503 (Zmax ⫽ 3.26 at ⌰ ⫽ 0.05), suggesting that the disease locus is linked to these markers (Table 2). Changing the allele frequencies did not have a significant effect on LOD scores. As expected by the results of two-point LOD score analysis which suggest a candidate interval with an identical Zmax value of 3.26 at ⌰ ⫽ 0.05 for six consecutive markers, multipoint analysis did not add information on the localization of SCA21 locus inside this interval (data not shown). Recombination events, visualized by haplotype reconstruction, confirmed the results obtained from linkage analysis in this family. Flanking markers were defined by recombinations in affected Subjects II-15 (D7S2514, telomeric) and III-2 (D7S2515, centromeric), delimiting an approximately 24cM region on chromosome 7p21.3-p15.1 (see Fig). Subject III-6, aged 19 years, had the same haplotypes as the affected members but had not presented with cerebellar ataxia nor dysmetria yet but with an isolated hyporeflexia of the four limbs and could be an asymptomatic carrier. Subject II-3 was classified as affected on the basis of clinical criteria. She reported a significant 668 Annals of Neurology Vol 52 No 5 November 2002 history of toxic exposure since the onset of symptoms at age 19 years. Her cerebellar syndrome was generally more severe than that of other members of similar age, with atypical associated features such as dyskinesia. All eight of her children were examined, and none of them had ataxia nor hyporeflexia. Subject II-3 essentially lacks the haplotype shared by the affected, and is considered a phenocopy. However, on the basis of her phenotype, assigned before the genetic analysis, she is classified as affected in the LOD score calculations. Discussion Clinical signs, age at onset, and duration of the disease vary substantially between and within ADCA families.3,4,11 The currently favored classification for late-onset ataxia is based on genetics. As these diseases are mapped to specific genetic loci, they are assigned specific numbers SCA1 to 17. Many of these SCAs fall within Harding’s ADCA type I. Because there is considerable clinical overlap among SCAs, most of them do not truly fit into the Harding classification. It thus appears wiser to identify them by their SCA designation.12 Patients in this family present a distinct and slowly progressive ataxia syndrome with extrapyramidal signs such a tremor, akinesia, and rigidity that prompt us to classify this kindred as ADCA type I. The signs that are specifically associated with other pure cerebellar ataxia such as seizures in SCA10,13 mental retardation in SCA13,14 myoclonus in SCA14,15 and essential head tremor in SCA1616 did not represent specific features of our disease. Many of the known SCA mutations are unstable expansions of a CAG repeat, and intergenerational expansion of the repeat has provided the basis for anticipation in SCA1, 2, 3, and 7. Our family shows earlier onset of the disease in successive generations, although this apparent anticipation may be attributable to ascertainment biases. The presence of a dynamic mutation such as a trinucleotide repeat expansion appeared unlikely because RED assays were inconclusive.7 Obviously, the possibility of a small trinucleotide repeat expansion or another type of expansion causing the disease cannot be excluded. In conclusion, we have mapped a locus on chromosome 7 responsible for a distinct slowly progressive ataxia (SCA21) associated with a prominent extrapyramidal syndrome, a mild cognitive impairment, and near-normal eye movements. Several genes have been identified in the 7p21.3-p15.1 region, and if none of them appeared to be Table 2. Pairwise LOD Scores between the Disease Locus and Chromosome 7p Markers Recombination Rate () Markers 0.00 0.01 0.05 0.1 0.2 0.3 0.4 Zmax max D7S481 D7S2514 D7S2464 D7S513 D7S2557 D7S503 D7S493 D7S673 D7S2525 D7S516 D7S2515 D7S2848 D7S2252 D7S2250 2.56 ⫺2.43 0.12 0.97 0.96 0.97 0.95 0.98 0.96 ⫺0.50 ⫺0.70 0.03 ⫺1.25 ⫺1.07 2.53 0.57 0.50 2.86 2.86 2.86 2.86 2.86 2.86 1.39 0.86 0.65 0.52 0.72 2.37 1.70 0.91 3.26 3.26 3.26 3.26 3.26 3.26 1.88 1.95 1.69 1.38 1.58 2.16 1.96 1.00 3.17 3.17 3.17 3.17 3.17 3.17 1.91 2.18 1.95 1.63 1.81 1.72 1.81 0.89 2.62 2.62 2.62 2.62 2.62 2.62 1.63 1.99 1.81 1.56 1.70 1.22 1.34 0.62 1.84 1.84 1.84 1.84 1.84 1.84 1.15 1.45 1.33 1.16 1.26 0.66 0.68 0.25 0.90 0.90 0.90 0.90 0.90 0.90 0.53 0.70 0.65 0.57 0.62 2.56 1.96 1.00 3.26 3.26 3.26 3.26 3.26 3.26 1.93 2.18 1.95 1.63 1.81 0.00 0.10 0.09 0.05 0.05 0.05 0.05 0.05 0.05 0.08 0.10 0.10 0.10 0.10 LOD ⫽ logarithm of odds. an evident candidate for SCA21, three genes could represent potential candidates owing to their function or pattern of expression. SCIN (SCINDERIN) encodes a Ca⫹⫹ regulated member of the gelsolin superfamily of actin filaments severing proteins that appears to function in the mechanism of release of neurotransmitters.17 SP4 transcription factor (MIM 600540) is a nuclear transcriptional activator mainly expressed in brain which belongs to the SP1 family of zinc-finger proteins.18 As found in TATAbinding protein, its transactivation domain contains glutamine-rich regions suggesting the possibility of a trinucleotide repeat expansion mutation within these sequences. A third gene, KOC1, encoding the IGF-II mRNA binding protein 3 (MIM 146732), could represent a potential candidate, because its mouse homolog is known to regulate neuronal differentiation.19 Identification of the specific mutation in the family in this study will contribute to the diagnosis and genetic counseling of at-risk subjects, as well as to our understanding of the molecular mechanisms of SCA21. The defined candidate region (approximately 24cM) is still large and our current challenge is to narrow this interval using additional families with linkage to this locus, so that the search for disease-causing mutation is more effective. This work was supported in part by the University of Lille 2, and by grants from the Centre Hospitalier Régional et Universitaire (Délégationà la Recherche), the Centre National de Génotypage, and the Association pour l’Etude et la Prévention des Maladies du Système Nerveux. We are grateful to the family members for their participation and Drs M. Delpech and M. Lathrop for constructive comments on the manuscript. We are grateful to Dr P. Vermersch for referring the first ataxia patient identified from this family to us. References 1. Harding AE. Clinical features and classification of the inherited ataxias. Adv Neurol 1993;61:1–14. 2. Subramony SH, Filla A. Autosomal dominant spinocerebellar ataxias ad infinitum? Neurology 2001;56:287–289. 3. Klockgether T. Recent advances in degenerative ataxias. Curr Opin Neurol 2000;13:451– 455. 4. Stevanin G, Dürr A, Brice A. Clinical and molecular advances in autosomal dominant cerebellar ataxias: from genotype to phenotype and physiopathology. Eur J Hum Genet 2000;8: 4 –18. 5. Nakamura K, Jeong SY, Uchihara T, et al. SCA17, a novel autosomal dominant cerebellar ataxia caused by an expanded polyglutamine in TATA-binding protein. Hum Mol Genet 2001;10:1441–1448. 6. Matsuura T, Yagamata T, Burgess DL, et al. Large expansion of the ATTCT pentanucleotide repeat in spinocerebellar ataxia type 10. Nat Genet 2000;26:191–194. 7. Devos D, Schraen-Maschke S, Vuillaume I, et al. Clinical features and genetic analysis of a new form of spinocerebellar ataxia. Neurology 2001;56:234 –238. 8. Dib C, Faure S, Fizames C, et al. A comprehensive genetic map of the human genome based on 5,264 microsatellites. Nature 1996;380:152–154. 9. Lathrop GM, Lalouel JM. Easy calculations of lod scores and genetic risks on small computers. Am J Hum Genet 1984;36: 460 – 465. 10. Sheffield VC, Weber JL, Buetow KH, et al. A collection of triand tetranucleotide repeat markers used to generate high quality, high resolution human genome-wide linkage maps. Hum Mol Genet 1995;4:1837–1844. 11. Schöls L, Amoiridis G, Büttner T, et al. Autosomal dominant cerebellar ataxia: phenotypic differences in genetically defined subtypes? Ann Neurol 1997;42:924 –932. 12. Paulson H, Ammache Z. Ataxia and hereditary disorders. Mov Dis 2001;19:759 –782. 13. Rasmussen A, Matsuura T, Ruano L, et al. Clinical and genetic analysis of four Mexican families with spinocerebellar ataxia type 10. Ann Neurol 2001;50:234 –239. 14. Herman-Bert A, Stevanin G, Netter JC, et al. Mapping of spinocerebellar ataxia 13 to chromosome 19q13.3-13.4 in a family with autosomal dominant cerebellar ataxia and mental retardation. Am J Hum Genet 2000;67:229 –235. Vuillaume et al: Spinocerebellar Ataxia (SCA21) 669 15. Yamashita I, Sasaki H, Yabe I, et al. A novel locus for dominant cerebellar ataxia (SCA14) maps to a 10.2-cM interval flanked by D19S206 and D19S605 on chromosome 19q13.4-qter. Ann Neurol 2000;48:156 –163. 16. Miyoshi Y, Yamada T, Tanimura M, et al. A novel autosomal dominant spinocerebellar ataxia (SCA16) linked to chromosome 8q22.1-24.1. Neurology 2001;57:96 –100. 17. Trifaro JM. Scinderin and cortical F-actin are components of the secretory machinery. Can J Physiol Pharmacol 1999;77: 660 – 671. 18. Hagen G, Denning J, Preiss M, et al. Functional analyses of the transcription factor Sp4 reveal properties distinct from Sp1 and Sp3. J Biol Chem 1995;270:24989 –24994. 19. Mori H, Sakakibara S, Imai T, et al. Expression of mouse Igf2 mRNA-binding protein 3 and its implications for the developing central nervous system. J Neurosci Res 2001;64: 132–143. Oligodendrocytic Polyglutamine Pathology in DentatorubralPallidoluysian Atrophy 1 Subjects and Methods Patients and Transgenic Mice 2 Mitsunori Yamada, MD, PhD , Toshiya Sato, MD, Shoji Tsuji, MD, PhD,2 and Hitoshi Takahashi, MD, PhD1 White matter degeneration is one of the pathological conditions of dentatorubral-pallidoluysian atrophy. Autopsy brains exhibited a reduced number of glial cells in the lesions and an involvement of oligodendrocytes in nuclear inclusion formation, which previously has been recognized only as a pathological hallmark in neurons. Dentatorubral-pallidoluysian atrophy transgenic mice showed an increased number of affected glias with increasing age and with larger expansions of CAG repeats. These findings suggest that glial cells in dentatorubralpallidoluysian atrophy also are involved in the polyglutamine pathogenesis. Ann Neurol 2002;52:670 – 674 From the Department of 1Pathology and 2Neurology, Brain Research Institute, Niigata University, Niigata, Japan. Received Mar 29, 2002, and in revised form Jun 24. Accepted for publication Jun 30, 2002. Published online Sep 25, 2002, in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/ana.10352 Address correspondence to Dr Yamada, Department of Pathology, Brain Research Institute, Niigata University, 1 Asahimachi, Niigata 951-8585, Japan. E-mail: firstname.lastname@example.org 670 © 2002 Wiley-Liss, Inc. Dentatorubral-pallidoluysian atrophy (DRPLA)1 is a hereditary neurodegenerative disease caused by the expansion of a CAG repeat.2,3 The major neuropathology of this disorder is a combined degeneration of the dentatorubral and pallidoluysian systems1,4,5; however, it has shown that some DRPLA patients also present with white matter lesions.6,7 The changes are not correlated with the onset age, disease duration, or disease severity6 but appear to be a function of patient age.7 Neuropathologically, the lesions showed diffuse myelin pallor but lacked obvious ischemic changes,5,6 and the pathogenesis remains unclear. In this study, we conducted immunohistochemical and morphometric analyses on human DRPLA brains to determine whether glial cells in the white matter are involved in the accumulation of expanded polyQ stretches, which is a pathogenic abnormality common to neurons of all the CAG repeat diseases. To explore the chronological changes of glial cells and the influence of the CAG repeat size, we further extended the study to DRPLA transgenic mice with 76 or 129 glutamines. Twelve patients with DRPLA and six controls served as the subjects. We divided our 12 DRPLA patients into three subgroups according to classifications by Naito.8 Clinical findings of the patients are summarized in Table 1. Ten of the 12 patients also were examined genetically under informed consent. We also examined the transgenic mice harboring a single copy of a full-length human mutant DRPLA gene with 76 CAG repeats9 or with 129 CAG repeats.10 The Q76 mice were examined at 14 (n ⫽ 3) and more than 100 weeks of age (n ⫽ 5), but because the Q129 mice died by 16 weeks of age,10 examinations were restricted to 14 weeks of age (n ⫽ 3). Instead of aged Q129 mouse brains, we examined a brain of the mosaic mouse at 129 weeks of age carrying 76 and 129 CAG repeats. Immunohistochemistry Formalin-fixed, paraffin-embedded brain sections were stained with hematoxylin and eosin and Klüver-Barrera stains. For immunohistochemistry, sections were immunostained using mouse monoclonal antibodies against neurofilament (1:50; Sanbio, Uden, Netherlands) or expanded polyQ stretches (1C2, 1:16,000; Chemicon, Tenecula, CA as described previously.11 We used diaminobenzidine as the chromogen. To characterize the 1C2-immunopositive cells, we also subjected the sections to double immunostaining combined with rabbit anti 䡠 sera against glial fibrillary acidic protein (GFAP; DAKO, Glostrup, Denmark; 1:2,000) or against transferrin (TF, 1:2,000; DAKO) as the marker of oligodendroglia.12 GFAP or TF was detected using a VECTASTAIN ABC-AP kit (Vector Laboratories, Burlingame, CA) and Vector Blue as the chromogen. For negative controls, the primary antibodies were replaced with normal sera.