Genotype–PhenotypeCorrelations in TuberousSclerosis: Who and Howto Treat

Tuberous sclerosis complex (TSC) is an autosomaldominant hamartomatous syndrome characterized bylesions of the skin, most prominently hypopigmentedmacules and facial angiofibromas (adenoma sebaceum),hamartomatous lesions of the brain (cortical tubers,subependymal nodules, and subependymal giant celltumors), optic gliomas, and benign tumors of the eyes,heart, and lungs.1,2 Tumors of the kidney are alsocommon and may rarely progress to malignancy(�2%) in the form of malignant angiomyolipomas orrenal cell carcinomas.

TSC occurs in about 1 in 6,000 to 10,000 individ-uals.3,4 One third of cases are familial, whereas the ma-jority are sporadic. Early descriptions of the dermato-logical features were recorded by Rayer in 1835(vegetations vasculaires)30 and by Pringle in 1890 (ade-noma sebaceum).5 In 1880, Bourneville described thecerebral pathology and proposed the term sclerose tu-bereuse.6 For most of the twentieth century, Vogt’striad of epilepsy, mental retardation, and adenoma se-baceum has served as a framework. It was not until1979 that the spectrum of clinical and pathologicalfindings was fully described.1,7

Phenotypic variation in TSC is well documented.Among the most severe neurological manifestations ofTSC are intractable seizures, mental retardation, opticglioma, and obstructive hydrocephalus. The frequencyand severity of involvement of other organs is alsohighly variable. Renal angiomyolipoma, cardiac rhab-domyoma, and pulmonary lymphangioleiomyomatosismay be absent or may be severe enough to cause organfailure. The decreased life expectancy associated withTSC is primarily due to neurological manifestations,followed by renal, pulmonary, and cardiovascular dis-ease.8

The neuropathology of TSC is attributed to abnor-malities in proliferation, differentiation, migration, andcytoarchitecture of neurons, starting early in fetal de-velopment and resulting from the disruption of eitherhamartin (TSC1 9q34) or tuberin (TSC2 16p13). Bio-chemical and genetic analyses in Drosophila melano-gaster showed that Tsc1 and Tsc2 proteins were sensorsof nutrient and growth factor signaling, linking theseextracellular cues to the activity of the mammalian tar-get of rapamycin (mTOR), a major regulator of pro-tein translation and cell growth. In mammalian cells,

Tsc similarly couples signaling from growth factor re-ceptors to the activity of mTOR (Fig), whereas nutri-ents (amino acids and glucose) signal to mTOR inde-pendently of the Tsc proteins.9–11

Growth factor receptors act in part to activatephosphatidylinositol-3 (PI3) kinase, generating phospha-tidylinositol-3,4,5-trisphosphate (PIP3). Phosphatidylino-sitol-3,4,5-trisphosphate couples phosphatidylinositol-3kinase to Akt, a serine-threonine kinase, which, in turn,phosphorylates Tsc2, downregulating its GTPase (GAP)activity. Tsc2 is bound to Tsc1 in cells, where Tsc1 sta-bilizes the complex, by blocking ubiquitination and pro-teolysis of Tsc2.12–14 Tsc1/2 drives the small GTPaseRheb into an inactive GDP-bound state, leading to ac-tivation of mTOR. A number of mTOR-independentfunctions have also been ascribed to Tsc, includingmodification of the cytoskeleton, regulation of othersmall G proteins, and interaction with D-type cyc-lins.15–18

Although loss of the retained allele of Tsc1 or Tsc2contributes to the generation of subependymal giantcell tumors in TSC,19 similar loss of heterozygosity ap-pears not to occur in cortical tubers.20 A recent studysuggests that haploinsufficiency may contribute to theneuropathology and clinical manifestations of Tsc, be-cause rodents haploinsufficient for Tsc1 or Tsc2 hadlarge neuronal cell bodies in the hippocampus, enlarge-ment and decreased densities of dendritic spines, andaltered properties at glutamatergic synapses.21 Whethersimilar pathology occurs in human Tsc currently is un-known.

Inhibitors of mTOR are now available clinically, andthey have been shown (in a recent issue of Annals) tocause regression of subependymal giant cell tumors inTsc.22 Although such agents may be of value in treat-ing severely affected patients with Tsc, major questionsremain unanswered. It will be critical to deciphermTOR-dependent and -independent contributions tothe clinical spectrum of this disorder and to address therole of inhibitors of mTOR in the treatment of intrac-table epilepsy observed in many Tsc patients. A morevexing problem is whether chronic pharmacological in-hibition of mTOR will limit the cognitive and behav-ioral abnormalities in TSC. The dosage and schedulingof such therapy is of particular importance, becausecomplete inhibition of mTOR is likely to have pro-found effects on normal growth and development inchildren with TSC.

With these questions in mind, it is of critical impor-tance to identify patients at risk for severe manifesta-tions of Tsc1, to start thinking about therapy for thegroup of children most disabled by this disease. Muta-tional analysis in patients with TSC suggests that awide degree of phenotypic variation can be seen with aparticular genotype.23 Mutations in TSC1 appear to beassociated with a milder phenotype, and these patients

© 2006 American Neurological Association 505Published by Wiley-Liss, Inc., through Wiley Subscription Services

may have lower rates of mental retardation, autistic dis-order, severe facial angiofibroma, seizures (includinginfantile spasms), renal disease, and retinal hamarto-mas.24–28

In this issue of Annals, Jansen and colleagues29 per-form genotype–phenotype correlations in a largeFrench-Canadian kindred with TSC, and in 15 otherfamilies that have mutations at the same codon. Theydescribe functional studies on three different aminoacid substitutions at codon 905 and draw associationsto phenotype. The French-Canadian kindred have aR905Q substitution in Tsc2. The phenotype in thispedigree was mild, with intrafamilial variability rangingfrom isolated hypomelanotic macules to phenotypes in-cluding subependymal giant cell tumors, epilepsy, mildcognitive impairment, and renal angiomyolipoma.Three other families and six sporadic patients with thesame mutation were studied. Their phenotype wasagain found to be mild, with only a minority of pa-tients showing cognitive impairment or severe epilepsy.

For comparison, 10 families with the R905W muta-tion and 1 family with the R905G mutation had cor-tical tubers, subependymal nodules, or subependymalgiant cell tumors. Seizures were prominent and in-cluded infantile spasms and Lennox–Gastaut syn-drome. Skin lesions were more severe. Cognitive im-pairment occurred commonly and was more extreme.Angiomyolipomas, rhabdomyomas, and retinal hamar-tomas also occurred more frequently. In functionalstudies, all three substitutions were found to disruptthe ability of tuberin to antagonize signaling throughmTOR. The R905Q substitution affected tuberin

function to a lesser degree than the R905G andR905W substitutions. This finding is consistent withthe milder phenotype observed in patients with theR905Q substitution.

These results broaden our understanding of the ge-notype–phenotype relationship in TSC. First, patientswith familial mutations in TSC2 mutations tend to bemildly affected.24–28 Second, the lack of strict correla-tion between genotype and phenotype suggests thatother genes modify the severity of this disorder. Third,the effect of a particular mutation on tuberin functioncan be measured biochemically and may help to pre-dict the severity of disease. These findings represent astart in identifying patients at risk for the most severemanifestations of this disorder, so that these patientscan be entered into clinical trials to determine whetherinhibitors of mTOR can be both safe and effective inthe group of patients most impacted by TSC.

This research was supported by the NIH (National Cancer Institute,R01CA 102321, W.A.W.; National Institute of Neurological Dis-orders and Stroke, R21NS5Z161, W.A.W.); The Brain Tumor So-ciety; The Goldhirsh, Pediatric Brain Tumor, Sandler Family, andWaxman Foundations; Clinical Scientist Awards in TranslationalResearch from the Burroughs Wellcome Fund, and Thrasher Funds;and a UC-Genentech Discovery Grant (bio-05-10501).

Suzanne Goh, MD, and William A. Weiss, MD, PhD

Department of NeurologyUniversity of California, San FranciscoSan Francisco, CA

References1. Gomez MR, Sampson JR, Whittemore VH. Tuberous sclerosis

complex. 3rd ed. New York: Oxford University Press, 1999.2. Roach ES, Sparagana SP. Diagnosis of tuberous sclerosis com-

plex. J Child Neurol 2004;19:643–649.3. Sampson JR, Scahill SJ, Stephenson JB, et al. Genetic aspects of

tuberous sclerosis in the west of Scotland. J Med Genet 1989;26:28–31.

4. Osborne JP, Fryer A, Webb D. Epidemiology of tuberous scle-rosis. Ann N Y Acad Sci 1991;615:125–127.

5. Pringle JJ. A case of congenital adenoma sebaceum. J Dermatol1890;2:1–14.

6. Bourneville DM. Sclerose tubereuse des circonvolutions cere-brales. Arch Neurol 1880;1:81–91.

7. Jansen FE, van Nieuwenhuizen O, van Huffelen AC. Tuberoussclerosis complex and its founders. J Neurol Neurosurg Psychi-atry 2004;75:770.

8. Shepherd CW, Gomez MR, Lie JT, Crowson CS. Causes ofdeath in patients with tuberous sclerosis. Mayo Clin Proc 1991;66:792–796.

9. Nobukuni T, Joaquin M, Roccio M, et al. Amino acids medi-ate mTOR/raptor signaling through activation of class 3 phos-phatidylinositol 3OH-kinase. Proc Natl Acad Sci U S A 2005;102:14238–14243.

10. Byfield MP, Murray JT, Backer JM. hVps34 is a nutrient-regulated lipid kinase required for activation of p70 S6 kinase.J Biol Chem 2005;280:33076–33082.

Fig. The tuberous sclerosis complex (TSC) couples signalingfrom growth factor receptors to mammalian target of rapamy-cin (mTOR), thereby regulating cell growth, proliferation, andmetabolism. PI3 � phosphatidylinositol-3; PIP2 � phosphati-dylinositol-4,5-biphosphate; PIP3 � phosphatidylinositol-3,4,5-trisphosphate.

506 Annals of Neurology Vol 60 No 5 November 2006

11. Smith EM, Finn SG, Tee AR, et al. The tuberous sclerosis pro-tein TSC2 is not required for the regulation of the mammaliantarget of rapamycin by amino acids and certain cellular stresses.J Biol Chem 2005;280:18717–18727.

12. Benvenuto G, Li S, Brown SJ, et al. The tuberous sclerosis-1(TSC1) gene product hamartin suppresses cell growth and aug-ments the expression of the TSC2 product tuberin by inhibit-ing its ubiquitination. Oncogene 2000;19:6306–6316.

13. Murthy V, Han S, Beauchamp RL, et al. Pam and its orthologhighwire interact with and may negatively regulate theTSC1.TSC2 complex. J Biol Chem 2004;279:1351–1358.

14. Chong-Kopera H, Inoki K, Li Y, et al. TSC1 stabilizes TSC2by inhibiting the interaction between TSC2 and the HERC1ubiquitin ligase. J Biol Chem 2006;281:8313–8316.

15. Lamb RF, Roy C, Diefenbach TJ, et al. The TSC1 tumoursuppressor hamartin regulates cell adhesion through ERMproteins and the GTPase Rho. Nat Cell Biol 2000;2:281–287.

16. Li Y, Inoki K, Yeung R, Guan KL. Regulation of TSC2 by14-3-3 binding. J Biol Chem 2002;277:44593–44596.

17. Hengstschlager M, Rosner M, Fountoulakis M, Lubec G. Tu-berous sclerosis genes regulate cellular 14-3-3 protein levels.Biochem Biophys Res Commun 2003;312:676–683.

18. Zacharek SJ, Xiong Y, Shumway SD. Negative regulation ofTSC1-TSC2 by mammalian D-type cyclins. Cancer Res 2005;65:11354–11360.

19. Chan JA, Zhang H, Roberts PS, et al. Pathogenesis of tuberoussclerosis subependymal giant cell astrocytomas: biallelic inacti-vation of TSC1 or TSC2 leads to mTOR activation. J Neuro-pathol Exp Neurol 2004;63:1236–1242.

20. Ramesh V. Aspects of tuberous sclerosis complex (TSC) pro-tein function in the brain. Biochem Soc Trans 2003;31:579 –583.

21. Tavazoie SF, Alvarez VA, Ridenour DA, et al. Regulation ofneuronal morphology and function by the tumor suppressorsTsc1 and Tsc2. Nat Neurosci 2005;8:1727–1734.

22. Franz DN, Leonard J, Tudor C, et al. Rapamycin causes re-gression of astrocytomas in tuberous sclerosis complex. AnnNeurol 2006;59:490–498.

23. van Slegtenhorst M, Verhoef S, Tempelaars A, et al. Mutationalspectrum of TSC1 gene in a cohort of 225 tuberous sclerosiscomplex patients: no evidence of genotype-phenotype correla-tion. J Med Genet 1999;36:285–289.

24. Jones AC, Daniells CE, Snell RG, et al. Molecular genetic andphenotypic analysis reveals differences between TSC1 andTSC2 associated familial and sporadic tuberous sclerosis. HumMol Genet 1997;6:2155–2161.

25. Dabora SL, Jozwiak S, Franz DN, et al. Mutational analysis ina cohort of 224 tuberous sclerosis patients indicates increasedseverity of TSC2, compared with TSC1, disease in multiple or-gans. Am J Hum Genet 2001;68:64–80.

26. Sancak O, Nellist M, Goedbloed MA, et al. Mutational analysisof the TSC1 and TSC2 genes in a diagnostic setting: genotype-phenotype correlations and comparison of diagnostic DNAtechniques in tuberous sclerosis complex. Eur J Hum Genet2005;13:731–741.

27. Yamamoto T, Pipo JR, Feng J, et al. Novel TSC1 and TSC2mutations in Japanese patients with tuberous sclerosis complex.Brain Dev 2002;24:227–230.

28. Lewis JC, Thomas HV, Murphy KC, Sampson JR. Genotypeand psychological phenotype in tuberous sclerosis. J Med Genet2004;41:203–207.

29. Jansen AC, Sancak O, D’Agostino D, et al. Unusually mildtuberous sclerosis phenotype is associated with T5C2 R905Qmutation. Ann Neurol 2006;60:528–539.

30. Rayer PFO. Traite theorique et prattique des maladies de lapeau. 2nd Ed, Paris: Bailliere, 1835.

Goh and Weiss: Genetics and Therapy in Tsc 507