Epigenetic inactivation of LKB1 in primary tumors associated ... - Nature

ã

Oncogene (2000) 19, 164 ± 168 2000 Macmillan Publishers Ltd All rights reserved 0950 ± 9232/00 $15.00 www.nature.com/onc

SHORT REPORT

Epigenetic inactivation of LKB1 in primary tumors associated with the Peutz-Jeghers syndrome Manel Esteller1, Egle Avizienyte2, Paul G Corn1, Ragnhild A Lothe3, Stephen B Baylin1, Lauri A Aaltonen2 and James G Herman*,1 1

Tumor Biology, The Johns Hopkins Oncology Center, 424 North Bond Street, Baltimore, Maryland, MD 21231, USA; Department of Medical Genetics, Haartman Institute, PO Box 21, (Haartmaninkatu 3), 00014 University of Helsinki, Finland; 3 Department of Genetics, Institute of Cancer Research, The Norwegian Radium Hospital, 0310 Oslo, Norway 2

Germ-line mutations of the LKB1 gene cause PeutzJeghers syndrome (PJS) characterized by mucocutaneous pigmentation, predisposition to benign hamartomas of the gastrointestinal tract and also to several types of tumors. However, somatic mutations of this gene are very rare. To examine inactivation of LKB1 by epigenetic mechanisms, we investigated a series of primary tumors and cancer cell lines, for hypermethylation aecting the CpG island located in the 5' region of the LKB1 gene using Methylation-speci®c PCR (MSP). First, we screened 51 cancer cell lines. Only three colorectal and one cervical carcinoma cell lines were methylated at LKB1, and loss of the LKB1 transcript was demonstrated. Treatment with the demethylating agent 5-aza-2'-deoxycytidine restored LKB1 expression. To address the incidence of LKB1 epigenetic inactivation in primary tumors, we analysed colorectal, breast, gastric, pancreatic, thyroid, bladder and testicular carcinomas (n=195). Normal tissues from the mentioned organs were unmethylated in this region. Among the described tumors, only one colorectal carcinoma and three testicular tumors displayed LKB1 promoter hypermethylation. Further study of those histological types more commonly associated with PJS, demonstrated that LKB1 promoter hypermethylation was present in ®ve of 11 (45%) papillary breast carcinomas. Finally, in three patients with a strong family story suggestive of PJS disease, abnormal LKB1 methylation was found in four of 22 (18%) hamartomatous polyps lesions. Our ®ndings provide an alternative pathway for inactivation of the LKB1 tumor suppressor gene involving promoter hypermethylation. Oncogene (2000) 19, 164 ± 168. Keywords: LKB1; hypermethylation; Peutz-Jeghers; human cancer

Germ-line mutations of the LKB1 gene, also known as STK11, located in the chromosomal region 19p13.3 cause Peutz-Jeghers syndrome (PJS) (Hemminki et al., 1998; Jenne et al., 1998). PJS is a rare, autosomal

*Correspondence: JG Herman Received 20 July 1999; revised 8 September 1999; accepted 14 September 1999

dominantly inherited condition characterized by mucocutaneous pigmentation, as well as predisposition to gastrointestinal hamartomatous polyposis (Tomlinson and Houlston, 1997). Although hamartomas are characteristic of the syndrome, hyperplastic and adenomatous polyps are occasionally detected. Carriers of the genetic alteration are also predisposed to tumors of the gastrointestinal tract, breast, testis, ovary and cervix (Tomlinson and Houlston, 1997; Giardiello et al., 1987; Boardman et al., 1988) and an 18-fold increase risk of cancer has been reported in PJS patients (Giardiello et al., 1987). Interestingly, some rare tumor types can be found relatively frequently in PJS patients, including cervical adenoma malignum, ovarian sex cord tumors with annular tubes and testicular Sertoli cell tumors (Tomlinson and Houlston, 1997; Giardiello et al., 1987; Boardman et al., 1988). LKB1 encodes a widely expressed serine/ threonine kinase highly conserved in mouse and Xenopus (Hemminki et al., 1998). The large majority of mutations in LKB1 detected in PJS patients are predicted to truncate the protein and hence inactivate it (Hemminki et al., 1998; Jenne et al., 1998). Supporting the LKB1 tumor suppressor role, loss of LKB1 kinase activity has been recently demonstrated in the mutant proteins derived from PJS families (Mehenni et al., 1998; Ylikorkala et al., 1999). Because many genes that predispose to cancer are also mutated somatically in non-familial cases, the presence of somatic mutations has been examined in sporadic colorectal, pancreatic, gastric, ovarian, cervical, lung, soft tissue, renal, breast and testicular tumors but somatic mutations are very rare (Avizienyte et al., 1998, 1999; Bignell et al., 1998; Dong et al., 1998; Resta et al., 1998; Wang et al., 1998). Thus, the role of LKB1, if any, in the development of sporadic neoplasms remains unclear. An alternative mechanism for the inactivation of a tumor suppressor gene is promoter hypermethylation (Baylin et al., 1998; Jones et al., 1999). Methylation is the main epigenetic modi®cation in humans (Baylin et al., 1998) and changes in patterns of this process play an important role in tumorigenesis. In particular, hypermethylation of normally unmethylated CpG islands located in the promoter regions of many tumor suppressor and DNA repair genes such as p16, p15, Rb, VHL, E-cadherin, GSTP1, MGMT and hMLH1 correlates with its loss of expression in cancer cell lines and primary tumors (Baylin et al.,

LKB1 promoter hypermethylation in human cancer M Esteller et al

1998; Herman et al., 1996b; Jones et al., 1999; Kane et al., 1997; Herman et al., 1998; Esteller et al., 1998a,b, 1999a). In the cases of p16 and hMLH1, germ line point mutations are responsible for genetic diseases with increased cancer risk involving melanoma and colorectal/endometrial/gastric malignancies respectively, but somatic mutations of these genes are infrequent in sporadic tumors. In these nonfamilial malignancies, silencing of p16 and hMLH1 by promoter methylation is the more frequent alteration (Baylin et al., 1998; Kane et al., 1997; Herman et al., 1998; Esteller et al., 1998b). To determine the possible inactivation of LKB1 by epigenetic mechanisms, we investigated primary tumors and cancer cell lines for hypermethylation aecting the CpG island located in the 5' region of the LKB1 gene. Our results demonstrate the presence of hypermethylation-associated inactivation of LKB1 in cancer cell lines and in a subset of primary tumors observed in PJS patients. DNA methylation patterns in the CpG island located in the 5' region of LKB1 gene were ®rst screened in 51 cancer cell lines, including 15 colon, ®ve breast, seven ovary, one cervix, 11 lung, three thyroid, three leukemia, three brain and three prostate tumor types using Methylation-speci®c PCR (MSP). Only three colorectal and one cervical carcinoma cell lines displayed abnormal LKB1 methylation: DLD-1, HCT15 and HeLaS3 were nearly completely methylated at LKB1, while HCT116 was partially methylated at this locus showing the presence of unmethylated and methylated alleles. The methylation status was also con®rmed by restriction cut analysis of the MSP products with BstUI, as previously described (Herman et al., 1996a). A representative example of the MSP results is shown in Figure 1a. Interestingly, the DLD-1 cell line was established from a colorectal carcinoma with coexisting papillary adenoma (Tompkins et al., 1974; Vermeulen et al., 1998). The colorectal carcinoma cell line HCT15, isogenic to DLD-1 (Vermeulen et al., 1998), had the same pattern of methylation. Minimal expression of LKB1 was found in the methylated cell line DLD-1, while normal lymphocytes and the unmethylated colon cancer cell lines LoVo and HT-29 expressed abundant LKB1 transcript (Figure 1b). Treatment of DLD-1 with the demethylating agent 5-aza-2'-deoxycytidine, restored LKB1 expression (Figure 1b) coincident with the appearance of unmethylated LKB1 alleles (Figure 1c). Thus, the methylation event was the cause underlying the loss of LKB1 in this cell line. The cell line HeLaS3, derived from a cervical carcinoma with features common to those observed in PJS patients (Jones et al., 1971), demonstrated complete methylation of the LKB1 promoter, and the LKB1 transcript is absent in this cell line (Nezu J, personal communication). The functional importance of the partial methylation observed in HCT-116 is unknown, since the cell line expressed LKB1 mRNA. However, other cell lines with mismatch repair defects, such as RKO, SW48, and LoVo are completely unmethylated, suggesting that methylation of LKB1 is not dependent on mismatch repair capacity. Together, these data demonstrate the silencing by hypermethylation in the 5' end of LKB1 gene is one of mechanism associated with loss of expression, and thus inactivation, in cancer cell lines.

To address the incidence of LKB1 epigenetic inactivation in primary tumors, we examined unselected tumors including 12 colorectal, 32 breast, 30 gastric, 18 pancreatic, 12 thyroid and 32 bladder carcinomas. These tumors did not have LKB1 methylation, nor did normal tissues from these organs (Figure 2a,b). Furthermore, 18 colorectal and 28 testicular carcinomas with no LKB1 mutations (Avizienyte et al., 1998) and an additional 13 colorectal carcinomas with loss of heterozygosity (LOH) at LKB1 (Avizienyte et al., 1998) were also examined. One colon carcinoma and three testicular tumors (a seminoma, a non-seminoma and a carcinoma in situ, precursor of both types) displayed LKB1 promoter

165

Figure 1 (a) Analysis by Methylation-speci®c PCR (MSP) of the promoter region of LKB1. Primers used for the unmethylated reaction were 5'-GGA TGA AGT TGA TTT TGA TTG GGTT3' (sense) and 5'-ACC CAA TAC AAA ATC TAC AAA CCA ACA-3' (antisense) and for the methylated reaction 5'-ACG AAG TTG ATT TTG ATC GGG TC-3' (sense) and 5'-CGA TAC AAA ATC TAC GAA CCG ACG-3' (antisense). The sense primer of the unmethylated reaction begins at bp 15 and the sense primer of the methylated reaction begins at bp 17 from GenBank sequence AF 035625. The unmethylated product is 122 bp long and the methylated product is 117 bp. The presence of a visible PCR product in those lanes marked U indicates the presence of unmethylated genes of LKB1; the presence of product in those lanes marked M indicates the presence of methylated genes. In vitro methylated DNA (IVD) was used as a positive control for methylated LKB1 alleles. DNA from normal lymphocytes (Lymph) was used as negative control for methylated LKB1 alleles. (a) LKB1 is methylated DLD-1, but unmethylated in the other cell lines on the left. Hct-15 and Hct-116 also demonstrate methylation of LKB1. (b) LKB1 expression determined by reverse transcription PCR in normal lymphocytes, colon cancer cell lines LoVo and HT-29, and DLD-1 before and after 5-aza-2'deoxycytidine treatment, showing LKB1 reactivation in DLD-1. GAPDH expression demonstrates relatively equal amounts of initial mRNA. RT ± PCR reactions (+) and RT ± PCR minus (7) as negative control. One hundred ng of this cDNA were ampli®ed by PCR. Primers for LKB1 were 5'-GGC ATG CAG GAA ATG CTG GAC AGC-3' (sense, exon 3) and 5'-GTG TCC AGG CCG TTG GCA ATC TCG-3' (antisense, exon 5) and for GAPDH were 5'-CGG AGT CAA CGG ATT TGG TCG TAT (sense) and 5'-AGC CTT CTC CAT GGT GGT GAA GAC-3' (antisense). (c) MSP of LKB1 in the colorectal cancer cell line DLD-1 before and after treatment with the demethylating agent 5-aza-2'deoxycytidine Oncogene


166

hypermethylation (Table 1). This colorectal tumor also had LOH, suggesting complete loss of LKB1 function. We then studied a third set of sporadic tumors composed exclusively of the speci®c histologic subtypes typical of PJS (Tomlinson and Houlston, 1997; Giardiello et al., 1987). LKB1 promoter hypermethyla-

Figure 2 Analysis by Methylation-speci®c PCR (MSP) of the promoter region of LKB1 in primary tumors, as described in Figure 1. The presence of a visible PCR product in those lanes marked U indicates the presence of unmethylated genes of LKB1; the presence of product in those lanes marked M indicates the presence of methylated genes. In vitro methylated DNA (IVD) was used as a positive control for methylated LKB1 alleles. DNA from normal lymphocytes (Lymph) was used as negative control for methylated LKB1 alleles. (a) MSP of LKB1 in normal tissues; (b) MSP of LKB1 in unselected colorectal, pancreatic and breast primary tumors; (c) MSP of LKB1 in hamartomatous polyps (HP) and papillary breast carcinomas (PB)

Table 1 Methylation of LKB1 in cell lines and primary tumors Tumor Type

Methylated

Unmethylated

Cell lines Colon Breast Cervical Lung Ovary Other

3 0 1 0 0 0

12 5 0 11 7 12

Primary Tumors Colon Breast Gastric Pancreatic Thyroid Bladder Colona Testiculara Colonb Adenoma Malignumc Ovarian Sex Cordc Papillary Breastc Colon Hamartomac

0 0 0 0 0 0 0 3 1 0 0 5 4

12 32 30 18 12 32 18 25 12 3 1 6 18

a

Tumors examined and found to lack LKB1 mutations. bTumors with LOH at LKB locus. cTumors characteristic of Peutz-Jeghers syndrome

Oncogene

tion was absent in three adenoma malignum of the uterine cervix and one sex cord ovarian tumor with annular tubes. However, LKB1 methylation was present in ®ve of 11 (45%) papillary breast carcinomas (Figure 2c, Table 1). The methylation status was also con®rmed by restriction cut analysis of the MSP products. Interestingly, bilateral breast cancer, papillary breast carcinoma and breast papillomas are not uncommon ®ndings in PJS patients (Tomlinson and Houlston, 1997; Giardiello et al., 1987; Boardman et al., 1988; Riley and Swift, 1980). However, papillary breast carcinoma is very rare in the general population accounting for less than 5% of the diagnosed breast carcinomas. Finally, since hamartomas and adenocarcinomas from patients with PJS carrying LKB1 germ line mutations have LOH of 19p markers near LKB1 in 70% of tumors (Gruber et al., 1998), epigenetic silencing of LKB1 may account for the second hit for LKB1 inactivation in the remaining 30% of cases. We studied 22 hamartomas collected from three unrelated patients with a strong family history suggestive of PJS disease, one with a well characterized LKB1 frameshift germ line mutation (Hemminki et al., 1998). Abnormal LKB1 methylation was found in four of 22 (18%) of these hamartomatous polyps (Figure 2c and Table 1). Biallelic inactivation of LKB1 by combination of the genetic and the epigenetic events was suggested in two of these hamartomas, each having one of the mentioned germ line frameshift mutation combined with promoter hypermethylation in the absence of LOH (Table 2). Knudson's tumor suppressor/recessive oncogene theory predicts that genes which confer a risk of cancer as a result of germ-line mutations are likely to be somatically mutated in sporadic cancers of the same type. This has proved to be the case for Rb and p53. However, other hereditary cancer genes such as BRCA1, BRCA2, and ATM do not appear to completely conform this model. Until recently the same exception was applied to the mismatch repair genes due to the fact that hMLH1 and hMSH2 mutations are present only in 10% of sporadic carcinomas with microsatellite instabilty (MSI), the characteristic feature of the tumors developed in patients with Hereditary Non-Polyposis Colorectal Carcinoma. However, it has been shown that an alternative mechanism of inactivation involving silenTable 2 Summary of tumors commonly associated to PJS showing LKB1 epigenetic inactivation Sample

Histologic type

LKB1 Mutation

LOH at LKB1 locus

6269 6280 PJ3 PJ4 C378 TE56 TE136 TE54 PJ14 PJ15 PJ16 PJ17 PJ18

Hamartomatous Polyp Hamartomatous Polyp Hamartomatous Polyp Hamartomatous Polyp Colorectal Carcinoma Testicular germ cell tumor Testicular germ cell tumor Testicular germ cell tumor Papillary breast carcinoma Papillary breast carcinoma Papillary breast carcinoma Papillary breast carcinoma Papillary breast carcinoma

germline germline NA NA Absent Absent Absent Absent NA Absent NA NA NA

Absent Absent NA NA Present NA NA NA NA NA NA NA NA

NA, not analysed


cing of the hMLH1 by promoter hypermethylation is a frequent event in sporadic colon and endometrial carcinomas with MSI (Kane et al., 1997; Herman et al., 1998; Esteller et al., 1998b). In a similar way, somatic LKB1 mutations in tumors are extremely rare, but our ®ndings of the existence of silencing of LKB1 by promoter hypermethylation provides an alternative pathway to LKB1 inactivation in sporadic tumors. The two hits for the inactivation of tumor suppressor genes are generally thought to be intragenic mutations and loss of chromosomal material (LOH or homozygous deletion). Promoter hypermethylation should now be considered one of the `hits' suered by tumor suppressor genes (Baylin et al., 1998; Jones et al., 1999). Thus, silencing by abnormal promoter methylation of Rb, VHL, MLH1, p15, and p16 associated with inactivation of the other allele by a `classical hit' such as intragenic mutation or LOH are a relatively common ®nding in human cancer (Baylin et al., 1998; Jones et al., 1999). Our ®ndings ®t this model, demonstrating in one sporadic tumor LKB1 promoter hypermethylation associated with LOH in the absence of point mutation, and in a patient with LKB1 germ line mutation, inactivation of LKB1 by abnormal CpG island methylation in the absence of LOH. LKB1 epigenetic inactivation displays a highly specialized distribution. The speci®c pattern of promoter hypermethylation aecting genes involved in cancer covers a wide spectrum from p16, hypermethylated in many tumor types (Baylin et al., 1998), through MGMT or GSTP1, which epigenetic inactivation is found in carcinogen or hormone-related carcinomas respectively (Esteller et al., 1998a, 1999a). At the other extreme, Rb hypermethylation occurs only in retinoblastoma and VHL hypermethylation is limited to clear cell renal carcinoma and hemangio-

blastoma (Baylin et al., 1998). Hypermethylationassociated inactivation of LKB1, like Rb and VHL, appears only in the rare and particular subtypes of neoplasm associated with PJS, which can be seen in Table 1. For instance, none of the common type of breast carcinomas analysed had abnormal LKB1 methylation, but it was present in half of the more unusual papillary type. Future examination of the LKB1 promoter methylation status in other sporadic neoplasms typically linked to PJS, including larger series of ovarian sex cord tumors with annular tubes, sertoli cell testicular tumors and adenomas malignum of the uterine cervix may reveal additional evidence of LKB1 inactivation. Interestingly, LOH at 19p13.3 where LKB1 is located has been described in the majority of adenomas malignum of the uterine cervix (Lee et al., 1998), and abnormal LKB1 methylation could inactivate the other allele. Our ®ndings provide an alternative pathway for LKB1 inactivation involving hypermethylation of its 5' CpG island with a very speci®c pattern, further supporting epigenetic inactivation as a `hit' for LKB1 tumor suppressor gene abrogation.

167

Acknowledgments The authors thank Enrique Lerma, Jaime Prat, Xavier Matias-Guiu,, Francesc Treserra, Angel Garcia and Sophie D Fossa for providing tumor samples. M Esteller is a recipient of a Spanish Ministerio de Educacion y Cultura Award. JG Herman is a Valvano Scholar. SB Baylin and JG Herman receive research funding and are entitled to sales royalties from ONCOR, which is developing products related to research described in this paper. The terms of this arrangement have been reviewed and approved by The Johns Hopkins University in accordance with its con¯ict of interest policies.

References Avizienyte E, Loukola A, Roth S, Hemminki A, Tarkkanen M, Salovaara R, Arola J, Butzow R, Husgafvel-Pursiainen K, Kokkola A, Jarvinen H and Aaltonen LA. (1999). Am. J. Pathol., 154, 677 ± 681. Avizienyte E, Roth S, Loukola A, Hemminki A, Lothe RA, Stenwig AE, Fossam SD, Salovaara R and Aaltonen, LA. (1998). Cancer Res., 58, 2087 ± 2090. Baylin SB, Herman JG, Gra JR, Vertino PM and Issa JP. (1998). Adv. Cancer Res., 72, 141 ± 196. Bignell GR, Barfoot R, Seal S, Collins N, Warren W and Stratton MR. (1998). Cancer Res., 5, 1384 ± 1386. Boardman LA, Thibodeau SN, Schaid DJ, Lindor NM, McDonnell SK, Burgart LJ, Ahlquist DA, Podratz KC, Pittelkow M and Hartmann LC. (1988). Ann. Inter. Med., 128, 896 ± 899. Dong SM, Kim KM, Kim SY, Shin MS, Na EY, Lee SH, Park WS, Yoo NJ, Jang JJ, Yoon CY, Kim JW, Kim SY, Yang YM, Kim SH, Kim CS and Lee JY. (1998). Cancer Res., 58, 3787 ± 3790. Esteller M, Corn PG, Urena JM, Gabrielson E, Baylin SB and Herman JG. (1998a). Cancer Res., 58, 4515 ± 4518. Esteller M, Garcia A, Martinez-Palones JM, Xercavins J and Reventos J. (1999b). Am. J. Clin. Pathol., 111(3), 336 ± 342. Esteller M, Hamilton SR, Burger PC, Baylin SB and Herman JG. (1999a). Cancer Res., 59, 793 ± 797. Esteller M, Levine R, Baylin SB, Ellenson LH and Herman JG. (1998b). Oncogene, 17, 2413 ± 2417.

Giardiello FM, Welsh SB, Hamilton SR, Oerhaus GJ, Gittelsohn AM, Booker SV, Krush AJ, Yardley JH and Luk GD. (1987). N. Engl. J. Med., 316, 1511 ± 1514. Gruber SB, Entius MM, Petersen GM, Laken SJ, Long PA, Boyer R, Levin AM, Mujumdar UJ, Trent JM, Kinzler KW, Vogelstein B, Hamilton SR, Polymeropoulos MH, Oerhaus GJ and Giardiello FM. (1998). Cancer Res., 58, 5267 ± 5270. Hemminki A, Markie D, Tomlinson I, Avizienyte E, Roth S, Loukola A, Bignell G, Warren W, Amino M, Hoglund P, Jarvinen H, Kristo P, Pelin K, Ridanpaa M, Salovaara R, Toro T, Bodmer W, Olschwang S, Olsen AS, Stratton MR, de la Chapelle A and Aaltonen LA. (1998). Nature, 391, 184 ± 187. Herman JG, Gra JR, Myohanen S, Nelkin BD and Baylin SB. (1996a). Proc. Natl. Acad. Sci. USA, 93, 9821 ± 9826. Herman JG, Jen J, Merlo A and Baylin SB. (1996b). Cancer Res., 56, 722 ± 727. Herman JG, Umar A, Polyak K, Gra JR, Ahuja N, Issa JP, Markowitz S, Willson JK, Hamilton SR, Kinzler KW, Kane MF, Kolodne, RD, Vogelstein B, Kunkel TA and Bayli SB. (1998). Proc. Natl. Acad. Sci. USA, 95, 6870 ± 6875. Jenne DE, Reimann H, Nezu, J, Friedel W, Lo S, Jeschke R, Muller O, Back W and Zimmer M. (1998). Nat. Genet., 18, 38 ± 43. Jones Jr HW, McKusick VA, Harper PS and Wuu KD. (1971). Obstet. Gynecol., 38, 945 ± 949. Oncogene


168

Oncogene

Jones PA and Laird PW. (1999). Nat. Genet., 21, 163 ± 167. Kane MF, Loda M, Gaida GM, Lipman J, Mishra R, Goldman H, Jessup JM and Kolodner R. (1997). Cancer Res., 57, 808 ± 811. Lee JY, Dong SM, Kim HS, Kim SY, Na EY, Shin MS, Lee SH, Park WS, Kim KM, Lee YS, Jang JJ and Yoo NJ. (1998). Cancer Res., 58, 1140 ± 1143. Mehenni H, Gehrig C, Nez, J, Oku A, Shimane M, Rossier C, Guex N, Blouin JL, Scott HS and Antonarakis SE. (1998). Am. J. Hum. Genet., 63, 1641 ± 1650. Resta N, Simone C, Mareni C, Montera M, Gentile M, Susca F, Gristina R, Pozzi S, Bertario L, Bufo P, Carlomagno N, Ingrosso M, Rossini FP, Tenconi R and Guanti G. (1998). Cancer Res., 58, 4799 ± 4801. Riley E and Swift M. (1980). Cancer, 46, 815 ± 817.

Tomlinson IP and Houlston RS. (1997). J. Med. Genet., 34, 1007 ± 1011. Tompkins WA, Watrach AM, Schmale JD, Schultz RM and Harris JA. (1974). J. Natl. Cancer Inst., 52, 1101 ± 1110. Vermeulen SJ, Chen TR, Speleman F, Nollet F, Van Roy FM and Mareel MM. (1998). Cancer Genet. Cytogenet., 107, 76 ± 79. Wang ZJ, Taylor F, Churchman M, Norbury G and Tomlinson I. (1998). Am. J. Pathol., 153, 363 ± 366. Ylikorkala A, Avizienyte E, Tomlinson IPM, Tiainen M, Roth S, Loukola A, Hemminki A, Johansson M, Sistonen P, Markie D, Neale K, Phillips R, Zauber P, Twama T, Sampson J, Jarvinen H, Makela TP and Aaltonen LA. (1999). Hum. Mol. Genet., 8, 45 ± 51.