Germline mutations of the gene encoding bone morphogenetic protein receptor 1A in juvenile polyposis

Article metrics


Juvenile polyposis (JP; OMIM 174900) is an autosomal dominant gastrointestinal hamartomatous polyposis syndrome in which patients are at risk for developing gastrointestinal cancers1,2. Previous studies have demonstrated a locus for JP mapping to 18q21.1 (ref. 3) and germline mutations in the homolog of the gene for mothers against decapentaplegic, Drosophila, (MADH4, also known as SMAD4) in several JP families4. However, mutations in MADH4 are only present in a subset of JP cases5, and although mutations in the gene for phosphatase and tensin homolog (PTEN) have been described in a few families6,7, undefined genetic heterogeneity remains. Using a genome-wide screen in four JP kindreds without germline mutations in MADH4 or PTEN, we identified linkage with markers from chromosome 10q22–23 (maximum lod score of 4.74, θ=0.00). We found no recombinants using markers developed from the vicinity of the gene for bone morphogenetic protein receptor 1A (BMPR1A), a serine–threonine kinase type I receptor involved in bone morphogenetic protein (BMP) signaling8. Genomic sequencing of BMPR1A in each of these JP kindreds disclosed germline nonsense mutations in all affected kindred members but not in normal control individuals. These findings indicate involvement of an additional gene in the transforming growth factor-β (TGF-β) superfamily in the genesis of JP, and document an unanticipated function for BMP in colonic epithelial growth control.


We chose four families with multiple affected members but without mutations in MADH4 or PTEN for linkage analyses, and found lod scores >1 with the markers D10S2327, GATA115E01 and D10S677 from chromosome 10q. We genotyped additional markers from this region and calculated a maximum lod score of 2.33 at θ=0.10 with D10S573 (Table 1). This region corresponded to chromosome 10q22–23 and was near both PTEN (refs. 9,10) and a putative JP locus called JP1 (ref. 11). Because we had already excluded mutations in PTEN in these kindreds by direct sequencing, we searched the Unigene and LocusLink databases from the National Center for Biotechnology Information server for other genes mapping to 10q22–23. One of these genes was BMPR1A (also known as ALK3), which mediates BMP intracellular signaling through MADH4 (refs. 12,13). Given the association between MADH4 and JP in other kindreds, we analyzed this gene in detail. We first developed two new microsatellite markers from a bacterial artificial chromosome (BAC) clone containing BMPR1A. One of these was a GGAA repeat (ALK3GGAA) located 76.3 kb upstream of BMPR1A exon 1 with eight alleles and a heterozygosity of 72%, and the other was a CA repeat (ALK3CA) 49.4 kb upstream of BMPR1A exon 1 with 12 alleles and a heterozygosity of 68%. We found maximum lod scores of 4.74 and 4.17 at θ=0.00 with ALK3CA and ALK3GGAA, respectively, by linkage analysis in these JP families. We placed BMPR1A 3.1 cR telomeric to WI-5226 and 7.9 cR centromeric to AFM225YD12 (with D10S1242 lying just centromeric to AFM225YD12) by radiation-hybrid mapping, and placed D10S573 3.5 cR telomeric to D10S1427 and 2.9 cR centromeric to WI-5226. Using haplotype analysis for the markers listed in Table 1, we identified seven affected individuals with essential recombination events, further refining the JP locus to between D10S573 and D10S1242. We found no recombinants with ALK3CA and ALK3GGAA in these four families.

Table 1 Chromosome 10q22–23 markers in four juvenile polyposis kindreds (lod scores)

BMPR1A consists of 11 exons distributed over 52,157 bp. We determined the complete sequence of each exon of BMPR1A in selected members of each kindred. In the D kindred, we detected a 4-bp deletion in exon 1 (44–47delTGTT), resulting in a stop codon at nucleotides 104–106 (Table 2). In the E kindred, there was a transition at nucleotide 715, changing codon 239 from a glutamine to a stop codon (Q239X). In the B kindred, there was a transition at nucleotide 812, changing a tryptophan to a stop codon (W271X). In the S kindred, there was a 1-bp deletion in exon 8 (961delC), creating a stop at the next codon. (Fig.1).

Table 2 Germline mutations in BMPR1A found in four unrelated juvenile polyposis kindreds
Figure 1: BMPR1A sequence variants in four JP kindreds.

Top strand, wild-type sequence; bottom strand, mutant sequence; box surrounds area of change. a, Four-bp deletion in the D kindred. b, Substitution in the E kindred. c, Substitution in the B kindred. d, One-bp deletion in the S kindred.

All eight affected members of the D kindred had the four-bp deletion, as did one individual at risk (43 years of age). We did not identify this mutation in four other family members at risk without a diagnosis of JP (5–51 years of age; Fig. 2a). All eight affected members of the E kindred had the substitution in exon 7, whereas none of five family members (41–71 years of age) without JP had the mutation (Fig. 2b). In the B kindred, all five affected family members had the substitution in exon 7, as did one (35 years of age) of four family members without a diagnosis of JP (8–70 years of age; Fig. 2c). In the S kindred, all five affected kindred members had the deletion in exon 8, and we did not find this mutation in any of the six family members without JP (19–77 years of age; Fig. 2d). We did not find any of the four mutations described here in 250 normal control individuals (Table 2). We identified a polymorphism at nucleotide 4 (4A>C), and in 109 unrelated individuals, allele frequencies were 12% for the A/A genotype, 35% for A/C and 53% for C/C. Accordingly, this encoded for threonine at amino acid 2 (ACT, as in the published cDNA sequence)8 in 29% of chromosomes and for proline (CCT) in 71%.

Figure 2: Mutation testing in four JP kindreds.

Black, known affected individuals;?, at-risk individuals of all ages who have not been diagnosed with JP; arrows, bands corresponding to mutations. a, Denaturing gel of exon 1 in the D kindred. b, Singe-strand conformation polymorphism (SSCP) gel of exon 7 in the E kindred. c, SSCP gel of exon 7 in the B kindred. d, Denaturing gel of exon 8 in the S kindred.

The existence of a locus for JP on 10q22–23 was first indicated by loss of heterozygosity (LOH) studies of sporadic and familial juvenile polyps. These studies documented deletions mapping to a 3-cM interval between D10S219 and D10S1696 (ref. 11), 2.68 cM centromeric to D10S573 on the Center for Medical Genetics sex-averaged map. The gene PTEN, which predisposes to Cowden syndrome, was subsequently mapped to 10q22–23 (refs. 9,10). Cowden syndrome is a hamartomatous polyposis syndrome in which the colonic polyps are histologically indistinguishable from those seen in JP patients, but patients with Cowden syndrome also generally develop trichilemmomas, breast and thyroid neoplasms14. There has been confusion regarding whether JP may also be caused by mutations in PTEN (ref. 15). However, mutations in PTEN have not been identified in standard JP families, and PTEN was excluded as the JP gene in our families. Moreover, the discovery of germline nonsense mutations in BMPR1A provides compelling evidence that this is the gene on chromosome 10q22–23 responsible for JP in the families we studied. Thus, there are two closely linked genes on chromosome 10q22–23 that can cause hamartomatous polyps, but these genes have no apparent functional relationship and the extra-intestinal phenotypes associated with mutations in these genes are distinct.

It has been assumed that polyps develop in JP patients through a tumor-suppressor mechanism. After an individual with JP was found to have an interstitial deletion of 10q22–24, examination of markers from 10q22–23 for LOH in juvenile polyps demonstrated some degree of LOH in 39 of 47 juvenile polyps from 16 patients11. Fluorescent in situ hybridization analysis indicated that these deletions were specific to lymphocytes and macrophages in the lamina propria rather than in epithelial cells11. Other studies have determined that deletions occur in the epithelium and fibroblasts of juvenile polyps, and have suggested that these cells may have a common clonal origin16. We assessed LOH of 10q markers in both epithelial and stromal fractions from six juvenile polyps from four patients after laser-capture microdissection. Although an occasional epithelial or stromal fraction showed partial LOH, no distinct or consistent result emerged. Either the cells in the polyps we examined did not undergo LOH or there was an admixture of cell types with varying clonal origin in our microdissected fractions, which prevented us from finding LOH. This can be more definitively addressed in the future by in situ analysis using antibodies against the carboxyl terminus of BMPR1A, once they become available.

Historically, TGF-β was thought to be the principal member of the TGF-β superfamily controlling the growth of colonic epithelial cells. However, some data have supported the idea that other, unspecified members of this family may be more important than TGF-β (refs. 1719). Although the involvement of BMP has not been indicated before in colorectal tumorigenesis, BMP receptors are widely distributed and are not confined to bone20. Moreover, there is evidence that BMPs can negatively regulate neoplastic growth21,22. To determine whether BMPR1A alterations occur in sporadic cancers, we evaluated 139 sporadic colorectal cancers for LOH at 10q22–23 using three simple tandem repeat polymorphisms spanning the BMPR1A region (CHLC.GATA81F06, ALK3GGAA and D10S1242). We detected LOH in 34 of 139 (24%) tumors analyzed. We then performed genomic sequencing of all 11 BMPR1A exons and intron–exon boundaries in 22 tumors showing LOH. We identified no somatic mutations in the remaining alleles of these tumors, arguing against the idea of the involvement of this gene in colorectal neoplasia unassociated with JP.

BMPR1A is a type I receptor of the TGF-β superfamily, with a cysteine-rich extracellular region, an intracellular glycine–serine-rich (GS) domain near the plasma membrane and an intracellular kinase domain23. There is a broad range of ligands in the BMP family, including BMPs 2–11, and these bind to specific type II BMP receptors. The type II receptors bind these ligands and activate the type I receptors through phosphorylation of their GS domains24. When BMPR1A is activated through phosphorylation by the type II receptor, it then phosphorylates MADH1 (SMAD1; refs. 12,25,26), MADH5 (SMAD5) and possibly MADH6 (SMAD8) (ref. 23), which then associate with cytoplasmic MADH4 (refs. 12,27). These MADH4–MADH1, −5 or −8 complexes then migrate to the nucleus, associate with DNA-binding proteins and regulate the transcription of DNA sequences12. The nonsense mutations reported in these four JP kindreds encode BMP receptors that lack the intracellular serine–threonine kinase domain8, and are predicted to result in loss of BMP-mediated intracellular signaling. The finding that germline mutations in both MADH4 and BMPR1A result in the JP phenotype raises the question of whether the effects of MADH4 are mediated through alterations in BMP signaling rather than through other TGF-β family members. In addition to their important implications for the diagnosis and management of families with JP, our results provide the first genetic evidence, to our knowledge, that BMPs may play a key role in controlling epithelial neoplasia.


Patients and families.

We obtained blood samples from kindred members after obtaining informed consent, reviewed medical records (including pathology, endoscopy and surgical reports) to confirm the diagnosis of JP, and reviewed pathology slides, where available. We classified individuals as 'affected' if they had histologic evidence of upper gastrointestinal or colorectal juvenile polyps, and as 'unknown' if there was no definitive histologic diagnosis of juvenile polyps. We selected control patient samples at random from blood samples obtained from anonymous donors at the University of Iowa outpatient laboratory.


We extracted DNA from whole-blood samples using a salting-out procedure28 and did the genome screen using the Weber screening set 8a simple tandem repeat polymorphisms markers (Research Genetics), as described previously3.

Linkage analysis and mapping.

We did two-point linkage calculations assuming autosomal dominant inheritance, a gene frequency of 1 in 100,000, and 95% penetrance3, using the MLINK subroutine of the FASTLINK 2.3 version of the LINKAGE program package29. We constructed haplotypes manually for each family assuming the least possible number of recombination events. We ordered markers according to the Whitehead Human Physical Mapping project ( and the Center for Medical Genetics maps ( We used the GeneBridge 4 radiation hybrid DNA panel (Research Genetics) and the WICGR mapping service ( to map BMPR1A exon 1 and D10S573.

Generation of simple tandem repeat polymorphisms from BMPR1A region.

We did a basic local alignment search tool search ( using the BMPR1A cDNA sequence8, which showed homology to the 173,675-bp BAC clone RP11-420K10. We searched the BAC sequence for simple tandem repeat sequences by alignment of 20-bp sequences of di-, tri- and tetranucleotide repeat elements. We then selected the following primer pairs flanking these repeat elements using the Primer3 program ( ALK3CA-1a (5′–GATCCAGAAACCAAGGGAAA–3′) and ALK3CA-1b (5′–TGGTAGATGGAGGTGGGGTA–3′) (186-bp product); and ALK3GGAA-1a (5′–CACACTGCAGGTGCTCTACAA–3′) and ALK3GGAA-1b (5′–CTTGGGCAACAGAGCAAGAT–3′) (210-bp product). We used these primers to amplify DNA from 100 control patients, then separated the products by electrophoresis through 6% denaturing polyacrylamide gels and silver-stained the gels. We determined the allele frequencies and heterozygosity for each marker, then used each for genotyping in JP families.

Definition of BMPR1A intron–exon boundaries.

We defined BMPR1A intron–exon boundaries by alignment of short segments of the BMPR1A cDNA sequence to the RP11-420K10 BAC sequence using the Sequencher program (Gene Codes Corporation, v.3.0.1). After defining the intron–exon boundaries, we selected the following primers from the introns flanking each exon using the Primer3 program: exon1, ALK3-1a (5′–TGTCAAGTGCTTGCGATCTT–3′) and ALK3-1b (5′–GGCTGGGCCTAACTATTCAA–3′) (289-bp product); exon 2, ALK3-2a (5′–TTGTCACGAAACAATGAGCTTT–3′) and ALK3-2b (5′–AACTCTTAAGAAGGGCTGCAT–3′) (257-bp product); exon 3, ALK3-3a (5′–AGGCCATCTGTACCTGTTCAC–3′) and ALK3-3b (5′–ATATGGCCCCTCCCTTCTTT–3′) (246-bp product); exons 4 and 5, ALK3-4/5a (5′–TAAAATTTGCAGGCCCCTTT–3′) and ALK3-4/5b (5′–GCTTTACAAACAGCGGTTGA–3′) (519-bp product); exon 6, ALK3-6a (5′–GGATTCTTTCTGAGGGAAGGA–3′) and ALK3-6b (5′–TCCACCATCATGAGGACACA–3′) (317-bp product); exon 7, ALK3-7a: 5′–CCCTTTGCCAGTCTTAATGG–3′, ALK3-7b: 5′–AGGCTTCCACCTGTACCTCA–3′(323-bp product); exon 8, ALK3-8a (5′–TGAGCATTACTTCTCCCTAGCC–3′) and ALK3-8b (5′–TTCAAAACAGTGGGGCAAAG–3′) (394-bp product); exon 9, ALK3-9a (5′–CAACTTGGACCTTGGCTTTC–3′) and ALK3-9b (5′–CATGGCATGCCTGTATCAAA–3′) (361-bp product); exons 10 and 11, ALK3-10/11a (5′–AACCATTTTTGTGCCCATGT–3′) and ALK3-10/11b (5′–CACTCTAATTCCACCCATGC–3′) (456-bp product). We optimized PCR conditions for each primer pair using control DNA samples.

DNA sequencing and mutation analysis.

We amplified DNA from kindred members using BMPR1A primers in a 30-μl reaction volume, then separated products by electrophoresis through 2% agarose gels. We sequenced gel-purified PCR products (QIAquick, Qiagen) in both directions with dye terminators (Applied Biosystems Prism Cycle Sequencing), using the PCR primers as sequencing primers, and determined their sequences using a Model 373 automated sequencer and ABI analysis software (Applied Biosystems). Additional family members were tested for mutations by sequencing, SSCP and/or denaturing gel electrophoresis of specific exons. We resolved the four-bp deletion in exon 1 in the D kindred by electrophoresis through 6% denaturing polyacrylamide gels and silver-stained the gels. We evaluated the mutation in exon 7 in the E kindred by SSCP using the primers 5′–CAGCGAACTATTGCCAAACA–3′and 5′–ATGGCGCATTAGCACAGTTT–3′(177-bp product); for the B kindred, the SSCP primers were 5′–AAGTATGGATGGGCAAATGG–3′and 5′–ATGGCGCATTAGCACAGTTT–3′(116-bp product). To examine the one-bp deletion in exon 8 in the S kindred, we amplified family members using the primers 5′–CAGGTTCCTGGACTCAGCTC–3′and 5′–CTTTCCTTGGGTGCCATAAA–3′(170-bp product), then separated the products by electrophoresis through 10% denaturing polyacrylamide gels and identified them by silver-staining. We examined the exon 1 polymorphism by amplifying DNA from each individual using the ALK3-1a and ALK3-1b primers, then digested the products with Hinf1. We separated these products by electrophoresis through 6% polyacrylamide gels and identified them by silver staining. For mutational analysis of sporadic cancers, we gel-purified the PCR products (Qiagen) and analyzed the sequencing reactions on an SCE-9610 96-well capillary electrophoresis system (SpecrtruMedix Corporation).

Laser capture microdissection.

We cut paraffin-embedded tissue blocks containing juvenile polyp tissue from JP family members into sections 5 μm in thickness and stained slides with hematoxylin and eosin. We used a Pixcell II image archiving workstation (Arcturus Engineering) to obtain separate laser captures of lamina propria and epithelial cells using an amplitude of 50 mW, a duration of 800 μs and a 7.5-μm beam. We extracted DNA from the Capsure lids (Arcturus Engineering) containing microdissected tissue in 50 μl lysis buffer (10 mM Tris, pH 8.0, 1 mM EDTA, 1% Tween 20 and 0.1 mg/ml proteinase K) incubated overnight at 37 °C.

LOH studies.

For sporadic colorectal cancers, we amplified the markers ALK3GGAA, CHLC.GATA81F06 and D10S1242 in a 10-μl volume using 4 ng genomic DNA and DNA extracted from cancer xenografts or cell lines, as described30. We diluted 5 μl of a 1:20 dilution of each PCR product with 45 μl Hi-Di formamide (Applied Biosystems), then analyzed each sample using an SCE-9610 capillary electrophoresis system (SpectruMedix Corporation). For juvenile polyps, we used 5–10 ng genomic DNA and DNA extracted by laser capture microdissection from lamina propria and epithelium to amplify D10S573, ALK3CA, ALK3GGAA, the mutated exon of BMPR1A and D10S1242. We separated the PCR products by electrophoresis through 6% polyacrylamide gels and identified them by silver staining.

GenBank accession numbers.

BMPR1A cDNA, NM004329; BAC clone RP11-420K10, AC021036; ALK3CA, AF364128; ALK3GGAA, AF364128.

Accession codes




  1. 1

    Jarvinen, H.J. & Franssila, K.O. Familial juvenile polyposis coli: Increased risk of colorectal cancer. Gut 25, 792–800 (1984).

  2. 2

    Howe, J.R., Mitros, F.A. & Summers, R.W. The risk of gastrointestinal carcinoma in familial juvenile polyposis. Ann. Surg. Onc. 5, 751–756 (1998).

  3. 3

    Howe, J.R. et al. A gene for familial juvenile polyposis maps to chromosome 18q21.1. Am. J. Hum. Genet. 62, 1129–1136 (1998).

  4. 4

    Howe, J.R. et al. Mutations in the SMAD4/DPC4 gene in juvenile polyposis. Science 280, 1086–1088 (1998).

  5. 5

    Houlston, R. et al. Mutations in DPC4 (SMAD4) cause juvenile polyposis syndrome, but only account for a minority of cases. Hum. Mol. Genet. 7, 1907–1912 (1998).

  6. 6

    Olschwang, S., Serova-Sinilnikova, O.M., Lenoir, G.M. & Thomas, G. PTEN germ-line mutations in juvenile polyposis coli. Nature Genet. 18, 12–14 (1998).

  7. 7

    Lynch, E.D. et al. Inherited mutations in PTEN that are associated with breast cancer, Cowden disease, and juvenile polyposis. Am. J. Hum. Genet. 61, 1254–1260 (1997).

  8. 8

    ten Dijke, P. et al. Activin receptor-like kinases: a novel subclass of cell-surface receptors with predicted serine/threonine kinase activity. Oncogene 8, 2879–2887 (1993).

  9. 9

    Nelen, M.R. et al. Localization of the gene for Cowden disease to chromosome 10q22-23. Nature Genet. 13, 114–116 (1996).

  10. 10

    Liaw, D. et al. Germline mutations of the PTEN gene in Cowden disease, an inherited breast and thyroid cancer syndrome. Nature Genet. 16, 64–67 (1997).

  11. 11

    Jacoby, R.F. et al. A juvenile polyposis tumor suppressor locus at 10q22 is deleted from nonepithelial cells in the lamina propria. Gastroenterology 112, 1398–1403 (1997).

  12. 12

    Kretzschmar, M., Liu, F., Hata, A., Doody, J. & Massague, J. The TGF-beta family mediator Smad1 is phosphorylated directly and activated functionally by the BMP receptor kinase. Genes Dev. 11, 984–995 (1997).

  13. 13

    Henningfeld, K.A., Rastegar, S., Adler, G. & Knochel, W. Smad1 and Smad4 are components of the bone morphogenetic protein-4 (BMP-4)-induced transcription complex of the Xvent-2B promoter. J. Biol. Chem. 275, 21827–21835 (2000).

  14. 14

    Gentry, W.C., Jr., Eskritt, N.R. & Gorlin, R.J. Multiple hamartoma syndrome (Cowden disease). Arch. Dermatol. 109, 521–525 (1974).

  15. 15

    Eng, C. & Ji, H. Molecular classification of the inherited hamartoma polyposis syndromes: Clearing the muddied waters. Am. J. Hum. Genet. 62, 1020–1022 (1998).

  16. 16

    Woodford-Richens, K. et al. Allelic loss at SMAD4 in polyps from juvenile polyposis patients and use of fluorescence in situ hybridization to demonstrate clonal origin of the epithelium. Cancer Res. 60, 2477–2482 (2000).

  17. 17

    Dai, J.L., Bansal, R.K. & Kern, S.E. G1 cell cycle arrest and apoptosis induction by nuclear Smad4/Dpc4: phenotypes reversed by a tumorigenic mutation. Proc. Natl. Acad. Sci. USA 96, 1427–1432 (1999).

  18. 18

    Sirard, C. et al. Targeted disruption in murine cells reveals variable requirement for Smad4 in transforming growth factor beta-related signaling. J. Biol. Chem. 275, 2063–2070 (2000).

  19. 19

    Fink, S.P. et al. Transforming growth factor-beta-induced growth inhibition in a smad4 mutant colon adenoma cell line. Cancer Res. 61, 256–260 (2001).

  20. 20

    Iwasaki, S. et al. Distribution and characterization of specific cellular binding proteins for bone morphogenetic protein-2. J. Biol. Chem. 270, 5476–5482 (1995).

  21. 21

    Raida, M. et al. Expression, regulation and clinical significance of bone morphogenetic protein 6 in esophageal squamous-cell carcinoma. Int. J. Cancer 83, 38–44 (1999).

  22. 22

    Kleeff, J. et al. Bone morphogenetic protein 2 exerts diverse effects on cell growth in vitro and is expressed in human pancreatic cancer in vivo. Gastroenterology 116, 1202–1216 (1999).

  23. 23

    Heldin, C.-H., Miyazono, K. & Ten Dijke, P. TGF-β signaling from cell membrane to nucleus through SMAD proteins. Nature 390, 465–471 (1997).

  24. 24

    Massague, J. TGFβ signaling: Receptors, transducers, and Mad proteins. Cell 85, 947–950 (1996).

  25. 25

    Hoodless, P.A. et al. MADR1, a MAD-related protein that functions in BMP2 signaling pathways. Cell 85, 489–500 (1996).

  26. 26

    Liu, F. et al. A human Mad protein acting as a BMP-regulated transcriptional activator. Nature 381, 620–623 (1996).

  27. 27

    Lagna, G., Hata, A., Hemmati-Brivanlou, A. & Massague, J. Partnership between DPC4 and SMAD proteins in TGF-β signalling pathways. Nature 383, 832–836 (1996).

  28. 28

    Miller, S.A., Dykes, D.D. & Polesky, H.F. A simple salting out procedure for extracting DNA from nucleated cells. Nucleic Acids Res. 16, 1215 (1988).

  29. 29

    Cottingham, R.W., Idury, R.M. & Schaffer, A.A. Faster sequential linkage computations. Am. J. Hum. Genet. 53, 252–263 (1993).

  30. 30

    Thiagalingam, S. et al. Mechanisms underlying losses of heterozygosity in human colorectal cancers. Proc. Natl. Acad. Sci. USA 98, 2698–2702 (2001).

Download references


We thank R. Smith and E. Stone for their support in this project; and E. Lemyre, C. Gilpin, J. Peters and C. Prows for referring members of these JP families for genetic studies. This work was supported by a grant from the Roy J. Carver Charitable Trust, the Clayton Fund, the American College of Surgeons Owen H. Wangensteen Faculty Research Fellowship, and National Institutes of Health grants CA43460 and CA62924.

Author information

Correspondence to James R. Howe.

Rights and permissions

Reprints and Permissions

About this article

Further reading