Short Report

Oncogene (2005) 24, 1091–1097. doi:10.1038/sj.onc.1207826 Published online 13 December 2004

A model for GFRalpha4 function and a potential modifying role in multiple endocrine neoplasia 2

Judith B Vanhorne1, Scott D Andrew1, Karen J Harrison1, Sherryl AM Taylor1, Bradley Thomas1, Thomas J McDonald2, Peter J Ainsworth3 and Lois M Mulligan1

  1. 1Department of Pathology, Queen's University, Kingston, ON, Canada K7L 3N6
  2. 2Department of Medicine, University of Western Ontario, London, ON, Canada
  3. 3Department of Biochemistry, University of Western Ontario, London, ON, Canada

Correspondence: LM Mulligan, Department of Pathology, 20 Barrie St, Queen's University, Kingston, ON, Canada K7L 3N6. E-mail:

Received 25 May 2003; Revised 19 April 2004; Accepted 21 April 2004; Published online 13 December 2004.



Mutations of the RET proto-oncogene are found in the majority of patients with the inherited cancer syndrome multiple endocrine neoplasia type 2 (MEN 2). A minority of cases, however, have no detectable RET mutation and there is considerable phenotypic variation within and among MEN 2 families with the same RET mutation, suggesting a role for other loci in this disease. A candidate for such a gene is glial cell line-derived neurotrophic factor receptor alpha 4 (GFRA4), which encodes a cell surface-bound co-receptor (GFRalpha4) required for interaction of RET with its ligand persephin. The GFRA4 gene has multiple alternative splices leading to three distinct protein isoforms that are prominently expressed in thyroid. We postulated that mutations of GFRA4 contribute to MEN 2 in the absence of RET mutations or modify the RET mutation phenotype. We screened patients with MEN 2 or MEN 2-like phenotypes, with and without RET mutations, for variants of GFRA4. We identified 10 variants, one of which was over represented in, and two of which were found exclusively in, our patient populations. One of these was a single-base substitution upstream of the GFRalpha4 coding region, where it may alter gene expression. The second was a 7 bp insertion, which results in a change in reading frame for all three GFRalpha4 isoforms. This would cause a relative shift in membrane bound and soluble forms of GFRalpha4, which would significantly alter the formation of RET signalling complexes. Our data suggest a model of wild-type GFRalpha4 isoform expression that includes both activating and inhibiting co-receptors for RET.


GFRalpha4, GFRA family, thyroid tumours, multiple endocrine neoplasia type 2

The RET receptor tyrosine kinase is required for development of neural crest derived lineages, and of kidney. Activating germline mutations of RET give rise to multiple endocrine neoplasia type 2 (MEN 2), an inherited cancer syndrome characterized by medullary thyroid carcinoma (MTC), a tumour of thyroid C-cells (Hansford and Mulligan, 2000). MEN 2 has three subtypes. Familial medullary thyroid carcinoma (FMTC) is associated only with MTC, and generally has a later disease onset and a less aggressive disease course. MEN 2A is characterized by MTC, pheochromocytoma (PC) and hyperparathyroidism (HPT). Finally, MEN 2B is associated with MTC, PC, and developmental anomalies, and has a generally earlier disease onset and poorer prognosis. More than 95% of MEN 2 families have detectable RET mutations that result in inappropriate activation of the RET receptor (Hansford and Mulligan, 2000). However, there remains a proportion of MEN 2 families or MEN 2-like families in which RET mutations have not been identified, suggesting that other genes contribute to the MEN 2 phenotype in these individuals. Further, even within families with a specific RET mutation, there is considerable variation in age of disease onset and tumour types observed, which suggests that other, modifying loci may contribute to the expression and severity of the disease. Susceptibility alleles or haplotypes at the RET locus have been associated with increased risk for sporadic MTC and may also play a role in age of MEN 2 disease onset (Wiench et al., 2001; Gil et al., 2002), although other MEN-2 modifiers have been harder to find.

Candidate modifying genes have included RET ligand and co-receptor molecules. The RET ligands are the glial cell line-derived neurotrophic factor (GDNF) family ligands (GFL), GDNF, artemin, neurturin, and persephin (Airaksinen et al., 1999). GFLs do not interact directly with RET but first bind the GDNF receptor alpha (GFRalpha) proteins, co-receptors having no intracellular or transmembrane domains, that are linked to the cell surface by glycosylphosphatidylinositol (GPI)-linkage (Airaksinen et al., 1999). Four members of the GFRalpha family have been identified, GFRalpha1-4, each of which forms high affinity interactions with a specific GFL. Each of the GFRalpha family members has a unique tissue-specific and developmentally regulated expression pattern that suggests each has a distinct cell type-specific role in RET-activation (Baloh et al., 2000). Consistent with their common role in binding and activation of the RET receptor, the GFRalpha family members are closely related, sharing 30–45% sequence identity (Airaksinen et al., 1999) as well as similar gene structure (Vanhorne et al., 2001).

GFRalpha4, the RET co-receptor for the GFL persephin (PSPN), is highly expressed in thyroid C-cells and in MTC (Lindahl et al., 2000; Masure et al., 2000). These observations suggested that GFRalpha4 might play a role in development of MTC tumours in the context of MEN 2, and that mutations of this locus might also contribute to MEN 2 phenotypes. In this study, we have characterized the human GFRA4 locus and predicted protein isoforms. We investigated the possibilities that GFRalpha4 could act as a modifier of phenotype in RET-mutation positive MEN 2 patients, or that GFRalpha4 variants might contribute to an MEN 2-like phenotype in the absence of RET mutations.

We first compared existing cDNA and genomic DNA sequences to confirm the exon structure of human GFRA4 and compared this to those of mouse and rat (Figure 1a). GFRA4 has six coding exons that undergo alternative splicing to generate three variants (a, b, and c) in mouse and human but only two (a, b) in rat (Lindahl et al., 2000; 2001; Masure et al., 2000). In humans, introns 2, or 2 and 3, may form part of the GFRA4 coding sequence and exons 3 and 4 can be read in multiple reading frames, while a cryptic splice acceptor site in intron 3 can result in inclusion of an 11 bp sequence (exon 4b) in some transcripts (Figure 1). The three human GFRalpha4 transcripts and isoforms generated are referred to as WTa, WTb, and WTc (Figure 1b). Alternatively spliced forms of exon 1, reported in mouse and rat, and a transmembrane form of the protein reported in mouse, are not encoded by the human gene.

Figure 1.
Figure 1 - Unfortunately we are unable to provide accessible alternative text for this. If you require assistance to access this image, please contact or the author

(a) The GFRA4 locus structure in mouse, rat, and human, showing positions of alternative splicing and coding sequences. Shaded boxes indicate sequences recognized either as coding or intronic sequences in different transcripts. Splicing events restricted to a subset of transcripts are indicated with the appropriate transcript letter. Stop codons for each transcript type are indicated (e.g. *a). The positions of sequence variants detected are indicated relative to the human GFRA4 sequence. (b) The organization of exon and intron sequences included in each alternatively spliced human GFRA4 transcript. Where the same sequence is read in multiple open reading frames, each frame for a given sequence is shaded differently

Full figure and legend (89K)

We next investigated the GFRA4 locus for sequence variants in a panel of 103 patients, and in a control population, by single-strand conformation polymorphism analysis (SSCP) (Vanhorne et al., 2001). To assess whether GFRA4 mutations might act independently of RET or as modifiers of RET-mutation phenotypes, we divided our patient population into patients with MEN 2 who had a known RET mutation (RET +ve) and patients with a MEN 2-like phenotype but no identified RET mutation (RET -ve). This latter group included individuals with a clinical diagnosis of MEN 2, or with multiple MEN 2 related tumours, but without any RET mutation after rigorous screening, and members of families with no RET mutations but with C-cell hyperplasia and fewer than three MTC that did not fit the stringent criteria for FMTC. Finally, a group of MTC tumours were analysed for the occurrence of somatic GFRA4 mutations.

A total of 10 sequence variants were identified, eight of which were found in either our control population alone or in both control and patient populations (Table 1; Figure 1a). Only the c.799G>A variant resulted in an amino-acid substitution (Ala267Thr) in the longest (WTa) GFRalpha4 isoform alone, three others being silent changes and the remaining four variants detected in the 5' and 3' UTR of GFRA4 (Table 1; Figure 1a). The described variants are not predicted to alter splicing, expression, or stability of GFRA4 transcripts. The frequencies of each variant were similar in control and overall patient populations; however, the frequency of the c.867G>A variant differed significantly between the RET mutation-negative and control populations (P=0.025, Fisher's exact test) (Table 1). The variant c.899A>G also had a tendency to differ in frequency between these populations, although the difference was not significant (P=0.098). Our data suggest that the c.867A allele, and possibly the c.899A allele, are under represented in our RET mutation-negative patient population, as compared to either controls or our RET mutation-positive population. To determine whether the association of these alleles with MEN 2 represented the affect of a specific risk haplotype, we reconstructed haplotypes using the relative genotypic data for the four variants in exon 5 and the 3' UTR (Table 2). Although we could not confirm differences in haplotype distribution in our relatively small sample set (P=0.24), we found that haplotypes containing the CAA alleles for variants in the 3' UTR tended to be under-represented in the RET-ve group (P=0.07), suggesting that the TGG haplotype may be associated with an increased risk of an MEN 2-like phenotype in the absence of RET mutations. We saw no significant association of GFRA4 alleles with disease phenotype in our RET +ve population, perhaps reflecting the fact that the stronger phenotypic effect of known RET mutations could mask the milder effects of GFRalpha4 haplotype (ie GFRalpha4 haplotype does not provide a detectable additive risk in these cases).

Two variants were detected exclusively in our patient panel. The first, -52C>T, which lies 52 bp upstream of the GFRalpha4 translation start site (Figure 1a), was identified in a female patient with PC, hyperthyroidism, and HPT, but with no known RET mutation. Comparison with the promoter regions of other GFRA family genes (Hansford et al., 2001) suggests that this variant lies within the GFRA4 proximal promoter and could thus affect gene transcription. Interestingly, Gimm et al. (2001) have shown that a similarly positioned variant in GFRA1 is over represented in sporadic MTC and increases both GFRalpha1 RNA and protein levels in tumours. The GFRA4 -52C>T variant may also alter GFRalpha4 expression, thereby allowing increased RET signalling.

We detected a seven base pair insertion (GCGCCCC) in exon 3 in a patient with a known RET mutation of codon 804 (V804L). This frameshift mutation would alter the coding sequence of all three GFRalpha4 isoforms. Patient SC (II-1 in Figure 2a) in Family C presented with multifocal MTC and C-cell hyperplasia at age 54 but with no other clinical features. Her mother (I-1) who had been diagnosed in her seventh decade with multifocal, bilateral MTC, and subsequently died of unresectable metastatic disease, also had both the GFRA4 insertion and the V804L RET mutation. The GFRA4 insertion was not found in the unaffected son (III-1) who does not have the RET mutation (Figure 2a–c).

Figure 2.
Figure 2 - Unfortunately we are unable to provide accessible alternative text for this. If you require assistance to access this image, please contact or the author

GFRA4 sequence variants detected in patient SC and family. (a) Single-strand conformation polymorphism analysis of exon 3 in Family C and a normal control. Analyses were performed as previously described (Vanhorne et al., 2001). An arrow indicates the insertion allele. (b) Sequence of exon 3 from a wild-type control and patient SC. The position of the 7 bp insertion is indicated. SC is heterozygous for the variant. (c) Comparison of wild type and mutant allele sequence showing the insertion. (d) Predicted GFRA4 transcripts in the presence of the SC insertion (shown in black). Shading of open reading frames corresponds to that shown in Figure 1b. Novel mutant-specific reading frames are indicated in red, green and blue

Full figure and legend (153K)

Mutations at RET codon 804 are primarily found in families with MTC alone, without other MEN 2 disease phenotypes (Lombardo et al., 2002). The V804L mutation has been linked to a late age of disease onset (fourth and fifth decades or later) and less aggressive disease course in some families, but early onset disease and early metastases in others (Shannon et al., 1999; Feldman et al., 2000; Frohnauer and Decker, 2000; Lombardo et al., 2002) have suggested that genetic modifiers that exacerbate the effect of the V804L mutation may exist. In Family C, the co-occurrence of RET V804L and the GFRalpha4 insertion did not appear to affect age of disease onset but may have contributed to a more aggressive disease course.

The nature of the mutation detected in patient SC suggested the possibility of germline inactivation of a GFRA4 allele, which might indicate that GFRalpha4 acts as a tumour suppressor in MTC. To test this, we screened 23 paired normal and MTC tumour samples, including both sporadic and familial tumors, for variants of GFRA4 and for deletion of the surrounding 20p13 region. No additional variants were detected in GFRA4 (Table 1). We next used a panel of four microsatellite markers spanning the 20p13 region close to, and flanking, the GFRA4 locus. We identified partial loss of one allele at all informative loci in tumour DNA from one individual with a known RET mutation (not shown), consistent with previous studies showing that allele loss on chromosome 20 is rare in MTC (Mulligan et al., 1993). Interestingly, GFRA4 lies on chromosome 20p in a region at one time implicated in MEN 2, based on cytogenetically detectable interstitial deletions found in some families (Babu et al., 1984; Van Dyke et al., 1984). It is intriguing to speculate that such deletions could have involved the GFRA4 locus and that these early cytogenetic data may have recognized the location of a MEN 2 modifying locus.

In order to characterize the mechanism by which the 7 bp GFRA4 insertion might contribute to RET-associated phenotypes, we first investigated the predicted protein structures of wild-type GFRalpha4 (WTa, WTb, WTc) and then compared them to the SC mutants (SCa, SCb, SCc). Translation of all alternatively spliced wild-type GFRA4 mRNA forms (Figure 1b) results in three distinct protein isoforms that share the first 131 amino acids (Figure 3). WTa and WTb have a putative GPI-cleavage site, suggesting they are cell surface bound proteins. WTc has a unique C-terminal hydrophobic sequence that also contains a possible GPI-cleavage site (Figure 3), and it is unclear whether this isoform is also GPI-anchored, or whether WTc is soluble. Secondary structure predictions show that GFRalpha1-3 proteins comprise three well-conserved cysteine-rich globular domains (Airaksinen et al., 1999). All GFRalpha4 isoforms lack the first cysteine-rich domain found in these proteins. The WTa isoform is most similar to other GFRalphas, containing two cysteine-rich domains, spanning seven alpha helices, and the conserved GPI-linkage (Figure 3), and has been shown to bind both RET and its soluble ligand PSPN and to stimulate RET phosphorylation and survival of neuronal cell types (Enokido et al., 1998; Masure et al., 2000; Lindahl et al., 2001). Compared to WTa, WTb lacks helices 4 and 5 (Figure 3) and has an odd number of cysteine residues that would result in an unpaired residue, potentially available for novel interactions. WTc contains only the first three helices and has no significant additional structural motifs. Like the WT proteins, all SC isoforms would share three conserved alpha helices in their N-terminal cysteine-rich domain but SCa and SCb would be truncated, resulting in short soluble proteins without hydrophobic C-terminal sequences that would not be membrane bound. The SCc isoform encodes helices 1 to 3 and also C-terminal sequences including helices 6 and 7 and the terminal hydrophobic sequences and GPI-cleavage site found in WTa, but helices 4 and 5 would be absent, and an additional helix not found in other GFRalpha4 isoforms would incorporate novel interstitial sequences (Figure 3).

Figure 3.
Figure 3 - Unfortunately we are unable to provide accessible alternative text for this. If you require assistance to access this image, please contact or the author

Alignment of predicted wild type and variant SC isoform sequences based on multiple alignment algorithms. Predicted signal sequences and GPI-signal peptides are underlined and GPI-cleavage sites are shown in outline. Cysteines are shaded. We identified seven conserved alpha helices, with high reliability (boxed). A dotted box indicates position of the SC insertion mutation. Hydrophobic and positively charged residues critical for GFL binding (Scott and Ibáñez, 2001) are shown in bold

Full figure and legend (190K)

In other members of the GFRalpha family, the regions important for binding GFLs correspond to both helices 1 and 7 of GFRalpha4 (Scott and Ibáñez, 2001). Particularly important are hydrophobic and positively charged residues corresponding in GFRalpha4 to amino acids 83–86 (LLF) and 95–98 (RRR), respectively (Figure 3). Sequences important for RET binding span helices 1–5, but it is likely that the critical region is even more limited. Although the residues in GFRalpha4 critical for interaction with RET and PSPN have not been specifically identified, it seems likely that WTb and WTc will not have the same interactions and binding affinities for these molecules as does WTa. If we can extrapolate from the sequences required for GFL and RET binding in other GFRalpha family members, we might predict that neither WTb nor WTc are able to bind both PSPN and RET efficiently. WTb retains N- and C-terminal helices important for GFRalpha : GFL interactions and may therefore be able to bind PSPN, with or without RET (Figure 3). WTc lacks some of the sequences, including the C-terminal alpha helices, which are critical for GFL binding in other GFRalpha proteins and may thus actually bind RET but not PSPN. Although RET activation generally requires preassociation of GFRalpha and GFL to form a complex that subsequently binds RET, the association of RET and GFRalpha alone also occurs with lower efficiency (Sanicola et al., 1997; Trupp et al., 1998). Thus, WTb and/or WTc may act as negative regulators of RET activation by binding either RET or PSPN and sequestering them, thereby preventing formation of full tripartite complexes.

If, as we predict, GFRA4 encodes a combination of activating and 'blocking' isoforms, the mutation observed in Family C could contribute to the observed disease phenotype by either altering the balance between isoforms, by disabling normal isoform function, or by generation of novel isoform function. The 7 bp insertion identified in patient SC would significantly disrupt the GFRalpha4 protein structure of all three isoforms (Figure 3). SCa and SCb lack the alpha helices required for PSPN binding, would not form GPI-attachments to the cell membrane, and would thus have limited ability to bind RET. SCc retains both N- and C-terminal alpha helices and the GPI-linkage found in WTa and could conceivably form both PSPN and RET interactions. As a result, SCc may act as an activating isoform, while SCa and SCb diffuse from the cell surface and do not sequester RET or ligand to inhibit RET signaling.

Interestingly, Lindahl et al. (2001) have shown that GFRalpha4 isoforms have tissue-specific differences in relative expression levels. In normal thyroid, isoforms WTb and WTc are relatively highly expressed in comparison to WTa. This is consistent with the predicted low level requirements for active RET signaling in maintenance of normal thyroid C-cells (Avantaggiato et al., 1994; Nakamura et al., 1994), which could be achieved by low levels of the high efficiency activating receptor WTa, and higher levels of the low efficiency or inhibitory receptors, WTb and WTc. In MTC tumours, there is almost no isoform WTc but relatively high levels of WTa and WTb (Lindahl et al., 2001) which, in our model, would result in increased relative levels of activating or high efficiency receptor and increased RET signaling. Conversely, the relatively higher levels of the SCc isoform, which we predict is activating, would shift the relative ratio of inhibitory and activating isoforms to better favour RET activation, promoting C-cell proliferation at an early stage of tumorigenesis. Coupled with the V804L mutation in Family C, the result may be more sustained RET signaling, contributing to increased cell proliferation and tumorigenesis.

In this study, we have developed a novel model for GFRalpha4 function, which provides a potential explanation for the expression of different GFRalpha4 isoforms and describes a novel mechanism for regulation of RET signalling. Further, we have shown that risk haplotypes and variants of GFRalpha4 may be associated with an increased risk of MEN 2 phenotypes in the absence of RET mutations or as a modifying effect in the presence of some RET mutations with lower transforming potential. Our data are consistent with a low penetrance modifying effect of GFRalpha4 on MTC development.



  1. Airaksinen MS, Titievsky A & Saarma M. (1999) Mol. Cell. Neurosci. 13: 313−325. | Article | PubMed | ISI | ChemPort |
  2. Avantaggiato V, Dathan NA, Grieco M, Fabien N, Lazzaro D, Fusco A, Simeone A & Santoro M. (1994) Cell Growth Differ. 5: 305−311. | PubMed | ChemPort |
  3. Babu VR, Van Dyke DL & Jackson CE. (1984) Proc. Natl. Acad. Sci. USA 81: 2525−2528. | PubMed | ChemPort |
  4. Baloh RH, Enomoto H, Johnson EM, Jr & Milbrandt J. (2000) Curr. Opin. Neurobiol. 10: 103−110. | Article | PubMed | ISI | ChemPort |
  5. Enokido Y, de Sauvage F, Hongo JA, Ninkina N, Rosenthal A, Buchman VL & Davies AM. (1998) Curr. Biol. 8: 1019−1022. | Article | PubMed | ISI | ChemPort |
  6. Feldman GL, Edmonds MW, Ainsworth PJ, Schuffenecker I, Lenoir GM, Saxe AW, Talpos GB, Roberson J, Petrucelli N & Jackson CE. (2000) Surgery 128: 93−98. | Article | PubMed | ChemPort |
  7. Frohnauer MK & Decker RA. (2000) Surgery 128: 1052−1058. | Article | PubMed | ChemPort |
  8. Gil L, Azanedo M, Pollan M, Cristobal E, Arribas B, Garcia-Albert L, Garcia-Saiz A, Maestro ML, Torres A, Menarguez J & Rojas JM. (2002) Int. J. Cancer 99: 299−304. | Article | PubMed | ChemPort |
  9. Gimm O, Dziema H, Brown J, Hoang-Vu C, Hinze R, Dralle H, Mulligan LM & Eng C. (2001) Oncogene 20: 2161−2170. | Article | PubMed | ChemPort |
  10. Hansford JR & Mulligan LM. (2000) J. Med. Genet. 37: 817−827. | Article | PubMed | ChemPort |
  11. Hansford JR, Samson SL, Naumovski D, Vanhorne JB & Mulligan LM. (2001) Am. J. Hum. Genet. 69S: 366.
  12. Lindahl M, Poteryaev D, Yu L, Arumae U, Timmusk T, Bongarzone I, Aiello A, Pierotti MA, Airaksinen MS & Saarma M. (2001) J. Biol. Chem. 276: 9344−9351. | Article | PubMed | ISI | ChemPort |
  13. Lindahl M, Timmusk T, Rossi J, Saarma M & Airaksinen MS. (2000) Mol. Cell. Neurosci. 15: 522−533. | Article | PubMed | ISI | ChemPort |
  14. Lombardo F, Baudin E, Chiefari E, Arturi F, Bardet S, Caillou B, Conte C, Dallapiccola B, Giuffrida D, Bidart JM, Schlumberger M & Filetti S. (2002) J. Clin. Endocrinol. Metab. 87: 1674−1680. | Article | PubMed | ChemPort |
  15. Masure S, Cik M, Hoefnagel E, Nosrat CA, Van Der Linden I, Scott R, Van Gompel P, Lesage AS, Verhasselt P, Ibanez CF & Gordon RD. (2000) J. Biol. Chem. 275: 39427−39434. | Article | PubMed | ISI | ChemPort |
  16. Mulligan LM, Gardner E, Smith BA, Mathew CGP & Ponder BAJ. (1993) Genes Chromosomes Cancer 6: 166−177. | PubMed | ChemPort |
  17. Nakamura T, Ishizaka Y, Nagao M, Hara M & Ishikawa T. (1994) J. Pathol. 172: 255−260. | PubMed | ISI | ChemPort |
  18. Sanicola M, Hession C, Worley D, Carmillo P, Ehrenfels C, Walus L, Robinson S, Jarworski G, Wei H, Tizard R, Whitty A, Pepinsky RB & Cate RL. (1997) Proc. Natl. Acad. Sci. USA 94: 6238−6243. | Article | PubMed | ChemPort |
  19. Scott RP & Ibáñez CF. (2001) J. Biol. Chem. 276: 1450−1458. | Article | PubMed | ISI | ChemPort |
  20. Shannon KE, Gimm O, Hinze R, Dralle H & Eng C. (1999) J. Endocr. Genet. 1: 39−46.
  21. Trupp M, Raynoschek C, Belluardo N & Ibanez CF. (1998) Mol. Cell. Neurosci. 11: 47−63. | Article | PubMed | ISI | ChemPort |
  22. Van Dyke DL, Babu VR & Jackson CE. (1984) Henry Ford Hosp. Med. J. 32: 266−268. | PubMed | ChemPort |
  23. Vanhorne JB, Gimm O, Myers SM, Kaushik A, von Deimling A, Eng C & Mulligan LM. (2001) Hum. Genet. 108: 409−415. | Article | PubMed | ISI | ChemPort |
  24. Wiench M, Wygoda Z, Gubala E, Wloch J, Lisowska K, Krassowski J, Scieglinska D, Fiszer-Kierzkowska A, Lange D, Kula D, Zeman M, Roskosz J, Kukulska A, Krawczyk Z & Jarzab B. (2001) J. Clin. Oncol. 19: 1374−1380. | PubMed | ChemPort |


We thank the patients and families, clinicians and geneticists who contributed to this study. We thank Andrew Day and Shirley Myers for their assistance. This work was supported the Canadian Institutes of Health Research and the Hospital for Sick Children Foundation (LMM), Queen's University (KJH) and a fellowship from the National Cancer Institute of Canada (SDA).



These links to content published by NPG are automatically generated


One gene???four syndromes

Nature News and Views (27 Jan 1994)