Williams syndrome (WS) is a developmental disorder with a characteristic personality and cognitive profile that is associated, in most cases, with a 2 Mb deletion of part of chromosome band 7q11.23. By applying CpG island cloning methods to cosmids from the deletion region, we have identified a new gene, called FZD3. Dosage blotting of DNA from 11 WS probands confirmed that it is located within the commonly deleted region. Sequence comparisons revealed that FZD3, encoding a 591 amino acid protein, is a novel member of a seven transmembrane domain receptor family that are mammalian homologs of the Drosophila tissue polarity gene frizzled. FZD3 is expressed predominantly in brain, testis, eye, skeletal muscle and kidney. Recently, frizzled has been identified as the receptor for the wingless (wg) protein in Drosophila. We show that Drosophila as well as human cells, when transfected with FZD3 expression constructs, bind Wg protein. In mouse, the wg homologous Wnt1 gene is involved in early development of a large domain of the central nervous system encompassing much of the midbrain and rostral metencephalon. The potential function of FZD3 in transmitting a Wnt protein signal in the human brain and other tissues suggests that heterozygous deletion of the FZD3 gene could contribute to the WS phenotype.
Williams-Beuren syndrome (WS, OMIM 194050) is a neuro-developmental disorder with multi-system manifestations, including supravalvar aortic or pulmonic stenosis, connective tissue abnormalities, short stature, variable hypercalcemia in infancy, mild to moderate mental retardation, unique personality and cognitive profile, and characteristic facial features (1,2). Usually sporadic, WS has an estimated incidence of 1 in 20 000 live births (3). A few reports of familial cases confirm autosomal dominant inheritance (4,5). WS is associated with deletion of genes in chromosome band 7q11.23, including those for elastin (ELN) (6–9), replication factor-2C subunit 2 (RFC2) (10) and LIM kinase-1 (L1MK1) (11,12). In addition, a putative RNA binding protein, a gene with homology to restin and several other transcription units have recently been identified (13). It is likely that the wide phenotypic spectrum associated with WS is the consequence of haploinsufficiency for these genes as well as for other genes in the deletion region yet to be identified. Our laboratory has previously defined a 2 Mb common deletion region by typing 65 WS probands with chromosome 7 specific markers (14), and we have begun a systematic search for expressed sequences within the deletion. In this study, we report the isolation of a novel transmembrane receptor gene, FZD3, that is related to the Drosophila tissue polarity gene frizzled (15). Recently, frizzled has been identified as the receptor for the wingless (wg) protein in Drosophila (16). We show that transfected cells expressing FZD3 bind Wg and, therefore, FZD3 is likely to function as a Wnt receptor (16).
Cloning and sequence analysis of FZD3
A chromosome 7 specific cosmid library was screened with inter-ALU PCR products of YAC clones from within the common WS deletion region (Fig. 1) (14). Positive cosmids were confirmed by STS content mapping, including the markers D7S489, D7S613, D7S1870, D7S2472 and ELN. Cosmid ends were sequenced and analyzed by PCR, with respect to presence in the deletion, with somatic cell hybrids containing either the 7q 11.23 deleted or the normal chromosome 7 from a proband with typical WS (10). A modified CpG island cloning method (17) was used to search for coding sequences in the five cosmids that mapped within the deletion region. Cosmid DNA was double digested with EagI (CGGCCG) and PstI, or with SacII (CCGCGG) and HindIII. Digestion products were subcloned into the Bluescript vector and sequenced. Out of the 30 clones sequenced, one clone contained a 700 bp insert that, by database comparison, showed homology to several frizzled family members (15). This clone, called fzd3, was mapped to cosmids 082 and 1124 that are both positive for the marker D7S489B (14) (Fig. 1). D7S489B is deleted in all of our probands who have deletions of ELN and is the most centromeric marker in the common deletion region (14). A fetal brain cDNA library (Stratagene) was screened and a 4.4 kb clone obtained. This clone was found to be chimeric: 2.2 kb of the clone was mapped to chromosome 6 by typing a somatic cell hybrid mapping panel (18) (data not shown). The other 2.2 kb included the original 700 bp sequence of fzd3.
This transcript, called FZD3, encodes a 591 amino-acid open reading frame (ORF) with high homology to members of the frizzled family (Fig. 2c). Although we did not isolate a full-length cDNA clone, we believe we have identified the complete ORF as there was no other methionine codon between the initiator ATG and stop codons in the 5′ untranslated region. A Kyte-Doolittle hydropathic profile suggests that the predicted protein contains seven transmembrane domains (Fig. 2b). In addition, there is a large cysteine-rich region in the N terminal part (Fig. 2a). Both are common structural motifs found in the frizzled family (15,16). When compared to the G protein-coupled seven transmembrane receptor superfamily, FZD3 and other frizzled family members retain some of the features conserved among that superfamily, including N-linked glycosylation sites (Asn-X-Ser/Thr) in the extracellular domain (Asn53 and Asnl58) and two cysteine residues (Cys294 and Cys388) in the second and the third extracellular loops (Fig. 2a) (19). The two highly conserved cysteine residues have been proposed to play a role in maintaining the active conformation of the receptor (20). FZD3 lacks the other conserved features occurring in G protein-coupled receptors, including a palmitylated cysteine residue in the С terminal region that is thought to be involved in activating G proteins (21).
As illustrated by a phylogenetic tree based on amino acid identity among the frizzled family members, FZD3 has the closest similarity to Mfz4 (47% amino acid identity) (Fig. 2c). In Figure 2d, the amino acids of the two proteins have been aligned. We expect that the level of conservation between true human and mouse orthologs should be higher and, therefore, do not propose that FZD3 is the human ortholog of Mfz4. Indeed, a mouse gene with higher homology to FZD3 has recently been identified (data not shown).
Expression and dosage analysis
The expression pattern of FZD3 is distinctive. Northern analysis using the 2.2 kb cDNA probe revealed a single 2.4 kb transcript expressed in skeletal muscle, brain and testis (Fig. 3a). On longer exposures of the autorad a transcript was also seen in pancreas but not in liver and kidney as both these lanes were relatively underloaded. FZD3 is also moderately expressed in most of the endocrine tissues, including pancreas, thyroid, adrenal cortex as well as in small intestine and stomach. Variably-sized weak transcripts were detected in testis (3.5 kb), thyroid (4.4 kb) and pancreas (8 kb). These could be transcripts of related genes. On a Northern blot exposed for only 3 h, the 2.4 kb FZD3 mRNA appears to be equally expressed in all parts of the brain (Fig. 3b).
To study expression patterns in fetal tissues, RT-PCR was carried out using primers amplifying 172 bp of the third transmembrane domain. For eye, brain and testis, the expression patterns appear to be similar to those on Northern blots of adult tissues. In addition, expression was detected in liver and kidney. Interestingly, it has been reported that the incidence of renal abnormalities in WS is 17.7% vs. 1.5% in the normal population (p <0.0003) (22). Whether these two observations are related or not needs further investigation.
Gene dosage analysis was used to confirm the deletion of FZD3 in WS probands. DNA from eight normal controls and eleven WS individuals was digested with EcoRI or XbaI, blotted and hybridized to the 4.4 kb chimeric cDNA clone described above. The chromosome 6-derived component of this clone served as an internal control for DNA loading. A dosage effect was evident for all 11 WS probands, as illustrated for two cases of sporadic WS in Figure 4. The ratio of intensities of the 10 kb fragment (FZD3 specific) and the 5.5 kb fragment (control) is -50% reduced in probands compared to the parental samples. The results provide strong evidence that one copy of FZD3 is deleted in these WS probands. Furthermore, somatic cell hybrids containing either the deleted or the non-deleted copy of chromosome 7 from a WS proband (10) were analyzed by PCR and Southern blotting for presence of the FZD3 gene. The results showed that FZD3 is present as a single copy gene on chromosome 7 and is missing from the del(7)(q11.23q11.23) chromosome (data not shown).
It was recently discovered that the product of the Drosophila frizzled gene Dfz2 functions as a Wg receptor. In the same study, binding of Wg protein was also demonstrated for cells expressing several mammalian frizzled gene family members (16). To determine whether FZD3 binds Wg as well, we used the same Wg binding assay on Drosophila S2 cells and human 293 cells transfected with FZD3 expression constructs (Fig. 5). FZD3, under the control of a metallothionein promoter, was stably transfected into Drosophila S2 cells that do not produce Wg protein. Expression of FZD3 was induced by growing the cells overnight in the presence of copper sulfate. Cells were then incubated with conditioned medium from Wg-producing S2 cells and subsequently with affinity-purified polyclonal antibodies to Wg. The S2 cells expressing FZD3 show surface staining (Fig. 5b), while the S2 control cells not expressing FZD3 show randomly scattered background spots (Fig. 5a).
Human embryonic kidney cells (293T) were cotransfected with a green fluorescence protein (GFP) expressing plasmid and an FZD3 expression construct in pCIS vector (16). After incubation with Wg and anti-Wg antibody, ∼10% of cells with green cytosolic fluorescence (caused by GFP) also had Wg-specific surface staining (Fig. 5e). Control 293 cells that were not transfected with FZD3 showed very little red surface staining (Fig. 5d). Cells transfected with the positive control Drosophila Dfz2 construct revealed positive staining identical to that published previously (data not shown) (16). We conclude from these experiments that stably and transiently expressed human FZD3 receptors bind Wg.
In Drosophila, the frizzled mutant phenotype consists of disruption of the polarity of hairs on the wing, leg, thorax and abdomen. Also, a rough eye phenotype arises secondary to alteration of the orderly arrangement of eye bristles (23,24). Both hemizygotes and homozygotes display an abnormal phenotype, but the severity of alterations varies from one allele to another (23). Interestingly, overexpressed frizzled protein on a wild type background also causes a phenotype similar to frizzled (25). A second Drosophila fizzled gene (Dfz2) is expressed in the developing CNS throughout embryogenesis and functions as a Wg receptor (16). This implies a general scheme in which frizzled proteins are receptors for Wg (or Wnt) signaling molecules. Wg has many different functions in fly development. It is involved in development of the head, of subsets of neurons in each segment of early CNS, in patterning of the segmented ectoderm of the trunk, in patterning of the midgut and growth of the malpighian tubules (26–28).
In the mouse, Wnt-1 (the mouse homolog of Wg) is expressed in the mid-gestation neural tube and in adult testis (29,30). The homozygous Wnt-I knock-out mouse has a different phenotype at each stage of development and dies within 24 h after birth. There is no abnormality outside the CNS at 14.5 days postcoitum (d.p.c); however, a substantial portion of the midbrain fails to develop and the metencephalon is absent (31,32). One homozygous mutant survived to adulthood and displayed ataxia (32). Heterozygous knock-out mice have a well developed cerebellum at 14.5 d.p.c. The phenotype of a spontaneous recessive mutation called swaying (sw) is characterized by ataxia and hypertonia, sw was found to be due to a frameshift mutation in the Wnt-1 gene (33). Numerous other members of the Wnt family have also been identified and characterized (26).
In addition to binding to Drosophila frizzled gene products, Wg binds to the products of some mammalian orthologs (Mfz4, Mfz7, Mfz8 and HfzS), but not to others (Mfz3 and Mfz6) (16). The binding we have observed between FZD3 and Wg appeared to be weaker than the binding between Dfz2 and Wg (16, and control data not shown) or Mfz4 and Wg (16), although it is difficult to quantitate this kind of assay. To understand the function of FZD3 in vivo, the interacting proteins at both N and C termini need to be identified. Human Wnt genes and human homologs of dishevelled (28) would be candidates with which to begin.
Is there a relationship between the WS phenotype and a putative role of FZD3 in Wnt signaling? The cardiovascular pathology and the visuospatial constructive cognition in WS individuals are likely to be due to haploinsufficiency of elastin and LIM kinase-1, respectively (6,11). Although heterozygosity for frizzled mutations in Drosophila does not appear to cause phenotypic abnormalities, the deletion of one copy of FZD3 could well cause some non-lethal and subtle developmental changes in human. Regarding which aspects of the WS phenotype could relate to the haploinsufficiency of FZD3, we speculate that musculoskeletal features could be the result of FZD3 dosage effect in morphogenesis at an early stage of development. Furthermore, haploinsufficiency of the FZD3 receptor in neurons and different parts of the brain could be responsible for neurological manifestations including hyperacusis, personality and behavioral characteristics, hyperreflexia and/or mental retardation. The expression of FZD3 in fetal kidney and endocrine tissues could be involved in the morphological renal abnormalities or the transient infantile hypercalcemia.
In conclusion, we have identified a novel human gene, an ortholog of the Drosophila frizzled homolog, and called it FZD3 following prior nomenclature convention (34). FZD3 is located near the centromeric breakpoint within the WS common deletion region. The expression pattern and possible role for FZD3 in Wnt signaling suggest that FZD3 may function in different tissues during different developmental stages, making FZD3 a strong candidate for involvement in selected aspects of WS pathology.
Materials and Methods
Construction of a library of CpG cluster containing clones from the WS deletion region
Inter-ALU PCR was done as described (35) using YAC HSC7E797 (14) as a template and primers A1:5′-GCGAGACTCCATCTCAAA, ALE3: 5′-CCACTGCACTCCAGCCTGGG, and PDJ33: 5′GCCTCCCAAAGTGCTGGGATTACAGGTGTGAGCCA. PCR products were purified (Promega Wizard kit), random primer labeled (Amersham Multiprime kit) and hybridized to a chromosome 7 specific cosmid library membrane (RLDB, Max-Planck-Institute for Molecular Genetics, Berlin) with Church buffer at 65°C for 18 h. The blot was washed in 1×SSC/1% SDS at room temperature for 30 min, followed by a 0.1×SSC/0.1% SDS wash at 65°C for 30 min. Positive clones were obtained from Hans Lehrach (RLDB, Berlin). Bacteria were grown in LB-kanamycin medium and cosmid DNA was purified following RLDB protocols. Cosmids were typed for region-specific markers as described (14). Cosmids were doubly digested with EagI and PstI, or with SacII and HindIII. Bluescript vectors were linearized and phosphatased using shrimp alkaline phosphatase (USB). Ligation was carried out with 3 U T4 ligase (Promega), 1 μl of 10× buffer and 30 ng of each fragment. One fifth of the reaction was transformed by heat shock into DH5a cells (GibcoBRL) and plated onto LB-ampicillin plates. A library of 120 clones was created and redundant clones were excluded by PCR with M13F/R primers on single colony templates.
PCR products, 300–2000 bp in size, were purified and cycle sequenced on an ABI Prism 377 sequencer with nested primers T7, SP6, KS or SK in pBluescript vector. Sequences were compared with the GenBank database using BLAST software. DNA Star and DNA Strider were used to draw the hydropathy plot, phylogenetic tree and amino acid alignment.
Screening of cDNA library
106 cDNA clones of a fetal brain cDNA library in lambda ZAPII vector (Stratagene) were plated. Phages were replicated onto nylon membranes, hybridized for 16–24 h and washed at a final stringency of 0.1×SSC/0.1% SDS at 65°C. After two to three rounds of screening, pBluescript phagemids were rescued from the positive clones, according to the manufacturer's instructions. Positive clones were PCR amplified and sequenced.
Multiple tissue Northern blot filters were purchased from Clontech and were probed with a 2.2 kb FZD3-specific subclone of the 4.4 kb chimeric cDNA according to the manufacturer's instructions. The filters were washed at a final stringency of 0.1×SSC/0.1% SDS at 65°C. A human beta-actin probe (Clontech) was hybridized as a control for RNA loading. Tissues were obtained from a 20-week-old human male fetus according to an IRB approved human subjects in research protocol and were snap-frozen in liquid nitrogen. Tissues were dispersed using a poly tron homogenizer directly into RNA STAT-60 (Tel-Test'B' Inc.) and extracted as described by the manufacturer. Three μg of total RNA was treated with 10 U RNase-free DNase (Boehringer Mannhein) for 15 min at 37°C, then heat inactivated for 12 min at 70°C. First strand cDNA was synthesized by adding 4 μl 5× buffer, 2 μl 0.1 M DTT, 1 μl 10 mmol hexamer in DEPC (diethylpyrocarbonate) treated water, 1 μl 10 mM dNTP and 200 U SuperScriptII reverse transcriptase (GibcoBRL) to the treated RNA and incubating for 45 min at 42°C, followed by 15 min at 70°C. Twenty μl DEPC treated water was added and 1 μl was used for PCR. The calpain regulatory subunit gene, which is ubiquitously expressed, was used as a positive control (36). Calpain primer sequences are 5′-TCAGCGCCACAGAACTCAT and 5′-TTGAATTCCTCAAAGCCCAG; PCR conditions are: 94°C, 3 min, then 37 cycles at 94°C, 30s; 55°C, 30 s; 72°C, 30 s. The control product is 172 bp. The primers for FZD3 PCR are 5′-GCGCGCTCTACGTGATCCAG and 5′-AGGCAGCCATGTGGAAATAG; PCR conditions are: 94°C, 3 min, then 38 cycles at 94°C, 30 s; 61°C, 30 s; 72°C, 30 s. The 254 bp product covers the third transmembrane domain. Five μl products of each reaction were analyzed by size fractionation on 4% agarose gels.
Genomie DNA samples were digested with EcoRI or XbaI overnight, size fractionated by agarose gel electrophoresis and blotted onto nylon membranes (Amersham Hybond N). The chimeric FZD3 cDNA probe was random primed and hybridized as described (10). The ratio between FZD3 and control bands was determined by computing densitometry (Molecular Dynamics, model 300A) with ImageQuant software.
The 2.2 kb cDNA including the coding region of the human FZD3 gene was stably expressed in S2 cells using the pMtHy expression vector as described (37). S2 cells transfected with a construct containing the heat shock promoter driving wg expression (S2-HSwg cells) were made by Cumberledge and Krasnow (38). Conditioned medium from these S2-HSwg cells contains soluble Wg protein and was collected and used as described (39). Untransfected S2 cells or S2 cells stably transfected with Dfz2 or FZD3 in pMtHy under the control of the metallothionein promoter were induced overnight in the presence of copper sulfate. The cells were washed once in PBS and incubated with 1.5 ml of 10× concentrated Wg conditioned medium for 3 h at 4°C. After 3–10 min washes with cold PBS, the cells were fixed in 2% methanol-free formaldehyde (Polysciences, Inc) for 15 min at room temperature. Following three 10 min washes with cold PBS, affinity purified rabbit anti-Wg antibody diluted 1:25 in 5% donkey serum/PBS was added to the cells and incubated overnight at 4°C. After additional washes in PBS, fluorescent Cy5 secondary antibody (Jackson ImmunoResearch) diluted 1:100 in 5% donkey serum/PBS was added to the cells for 1 h at room temperature. Following additional washes with PBS the cells were mounted in Vectashield mounting medium (Vector).
For transient expression in 293T cells, the Dfz2 and FZD3 coding regions were inserted into the pCIS expression vector under the control of the cytomegalovirus immediate early promoter/enhancer and with an optimized translation-initiation context, and cotransfected with a GFP expression plasmid at a ratio of 5:1 (16). These plasmids were transfected into a 6 cm dish of 293T cells using the calcium phosphate method. Eight hours following transfection, 10 mM chlorate was added to the cells. After 24 h, the medium was changed to 2.5 ml serum-free Schneiders medium and 10 mU of heparatinase (Seikagaku) and 10 mM chlorate were added for 3 h before continuing as above. Confocal and Nomarski images were collected with a Zeiss Axioscope, equipped with a Bio-Rad MRC 1000 confocal laser, and were processed in Adobe Photoshop 3.0.
The authors are grateful to Nadia Pohl and Hans Lehrach (RLDB, Max Planck Institut fur Molekulare Genetik, Berlin) for YACs and chromosome 7 specific cosmids, to Stephen Scherer and Lap-Chee Tsui (Hospital for Sick Children, Toronto) for YACs, to Paige Kaplan, Childrens Hospital of Philadelphia, for clinical information and samples, to Chiping Qian, Erika Valero and Kaye Suyama for excellent technical assistance, to Carolyn Schanen and Marcel Brink for advice and Johannes Schweizer for comments on the manuscript. This work was supported by NIH grant R01 GM00298 (U.F.) and by the Howard Hughes Medical Institute of which R.N. and U.F. are investigators and Y.-K.W. is an associate.