An X Chromosome Gene, WTX, Is Commonly Inactivated in Wilms Tumor

See allHide authors and affiliations

Science  02 Feb 2007:
Vol. 315, Issue 5812, pp. 642-645
DOI: 10.1126/science.1137509


Wilms tumor is a pediatric kidney cancer associated with inactivation of the WT1 tumor-suppressor gene in 5 to 10% of cases. Using a high-resolution screen for DNA copy-number alterations in Wilms tumor, we identified somatic deletions targeting a previously uncharacterized gene on the X chromosome. This gene, which we call WTX, is inactivated in approximately one-third of Wilms tumors (15 of 51 tumors). Tumors with mutations in WTX lack WT1 mutations, and both genes share a restricted temporal and spatial expression pattern in normal renal precursors. In contrast to biallelic inactivation of autosomal tumor-suppressor genes, WTX is inactivated by a monoallelic “single-hit” event targeting the single X chromosome in tumors from males and the active X chromosome in tumors from females.

Wilms tumor (nephroblastoma) is the most common pediatric kidney cancer and is derived from pluripotent renal precursors that produce undifferentiated blastemal cells, primitive epithelial structures, and stromal components [reviewed in (1)]. In 1972, Knudson and Strong proposed that Wilms tumor, like retinoblastoma, may develop as a consequence of two independent rate-limiting genetic events, subsequently defined as biallelic inactivation of a tumor-suppressor gene (1). This prediction was borne out by the identification of WT1, a zinc finger transcription factor gene on chromosome 11p13 that is targeted by germline heterozygous deletions in the WAGR syndrome (Wilms tumor, aniridia, genitourinary abnormalities, and mental retardation), by germline point mutations in the Denys-Drash syndrome (Wilms tumor, pseudohermaphroditism, and nephropathy), and by somatic biallelic inactivation in 5 to 10% of sporadic Wilms tumors (24). Although inactivation of WT1 affects only a small subset of cases, this gene has been shown to encode a master regulator of kidney development, which appears to be required for the survival and subsequent differentiation of renal stem cells (5, 6). Other known abnormalities in Wilms tumor include activating mutations in the β-catenin gene (CTNNB1) on chromosome 3p22, which often coincide with WT1 mutations (7), and epigenetic dysregulation of IGF2 and H19 (8) at the Beckwith-Wiedemann syndrome locus on chromosome 11p15. In the majority of cases, however, no specific genetic abnormalities have been identified.

To search for genetic abnormalities in sporadic Wilms tumor, we performed a detailed genome-wide scan (70-kb median resolution) for DNA copy-number changes in 51 primary tumor specimens using long-oligonucleotide array comparative genomic hybridization (array CGH) (9). In marked contrast to most adult epithelial cancers, the baseline CGH profile in Wilms tumor was quite stable, with an average of only 3.1 large (more than 5 mb) DNA copy-number changes per tumor after accounting for known copy-number polymorphisms (10, 11). As expected, we detected single-copy losses in a subset of cases, including known loci of loss of heterozygosity (LOH) at chromosomes 1p (18%), 16q (14%), 11p (6%), and 7p (4%) (4, 12). LOH at chromosome 11p15 is commonly associated with gene conversion rather than single-copy loss and is therefore underrepresented in our array CGH analysis (13). Notably, we detected small overlapping deletions at chromosome locus Xq11.1 in tumors from 5 out of 26 male patients. The deletions involved only one to three probes in each case, with a minimal area of overlap implicating a single previously uncharacterized gene (FAM123B/FLJ39827) that we named WTX, for “Wilms Tumor gene on the X chromosome” (Fig. 1, A and B). Alldeletions were confirmed by quantitative polymerase chain reaction (qPCR) of genomic DNA, including two cases with deletion breakpoints internal to WTX (Fig. 1C). Genomic deletions were also analyzed by fluorescence in situ hybridization (FISH) (representative example in Fig. 1D).

Fig. 1.

Identification of WTX deletions in Wilms tumor. (A) Array CGH profile for a representative WTX deletion in Xq11.1 in a male patient. All probes on the X chromosome are shown. (B) Minimal region of overlap for deletions targeting WTX. Genes flanking WTX (arrows depict the direction of transcription), array CGH probes (squares), and a confirmatory qPCR marker (circle) are shown. The boundaries of deletions detected in five male Wilms tumor cases define a minimal region of overlap centered on WTX (box). (C) Quantitative PCR of genomic DNA from five male Wilms tumors and one normal male kidney control, with the use of primers for the 5′ and 3′ ends of WTX. Tumors T13, T25, and T41 show complete absence of WTX, whereas tumor T15 lacks only the 5′ end and tumor T32 lacks only the 3′ end of the gene. Error bars indicate standard deviation. (D) FISH analysis demonstrating a somatic deletion in a male Wilms tumor (case 13) with the use of a probe for WTX (red) and a probe for the X chromosome centromere (green) as a control. The tumor lacks the red WTX signal, whereas both signals are present in matched normal tissue. Scale bars, 5 μm.

The endogenous WTX transcript differed from that predicted by database annotation in the 3′ end and required assembly with the use of rapid amplification of cDNA ends and reverse transcriptase (RT)–PCR from human and mouse cDNA (fig. S1). The full-length transcript (7.5 kb) encodes a protein of 1135 amino acids, containing a nuclear localization signal, two coiled-coil domains, an acidic domain that overlaps the first coiled coil, and a proline-rich domain (Fig. 2A). WTX orthologs are present in vertebrates, including zebrafish, but do not share substantial homology with other genes of known function (fig. S2).

Fig. 2.

Analysis of WTX mutations in Wilms tumor. (A) Schematic representation of WTX and potential functional domains, including the nuclear localization signal (NLS), the two coiled-coil domains (CC), and the proline-rich domain (PR). The relative positions of all point mutations are shown. aa, amino acids. (B) Representative nucleotide sequence tracings of a frame shift (left) and a nonsense mutation (right) of WTX in male Wilms tumors and matching germline tissue. Arrows indicate the position of the two hemizygous mutations, 439 insertion T leading to 157Ter and a substitution of T for C at position 1072, leading to Arg358→Ter. (C) Representative FISH analysis of a Wilms tumor from a female patient showing a deletion of the WTX allele on the active X chromosome. Probes for the X centromere (blue), WTX (red), and Xist (green) were used. Only one copy of WTX is present, associated with the inactive (Xist positive) X chromosome (Xi). The active X chromosome (Xa) is marked by the X centromere probe but lacks both WTX and Xist. The same pattern was observed in all tumor cells, consistent with tumor clonality. Scale bar, 5 μm. (D) Sequence tracings of genomic DNA and cDNA from a female Wilms tumor case with a heterozygous WTX mutation. Both alleles are detectable in genomic DNA but only the mutant allele is present in cDNA, indicating that only the mutant WTX allele is transcribed. (E) Schematic representation of Wilms tumor cases with mutations in WTX (red), WT1 (yellow), or β-catenin (blue). Although WT1 and β-catenin mutations can be present in the same tumor, there is no overlap between mutations in WTX and mutations in these two genes. WTX mutations are listed in table S1. WT1 and β-catenin mutations are listed in table S2. Tumors 1 through 51 (underlined) were tested for both deletions and point mutations in WTX. The remaining tumors were tested for point mutations only.

To search for intragenic point mutations, we sequenced the entire coding region of WTX in 82 Wilms tumor specimens, including the 51 cases previously analyzed by array CGH. Intragenic truncating mutations were identified in six tumors, including one nonsense mutation (Arg358→Ter, where Ter is the termination of the chain) observed in two independent cases (table S1). All predicted protein truncations were N terminal to the second coiled-coil domain encoded by WTX (Fig. 2A). A seventh case harbored a missense mutation (Lys292→Asn292) affecting an evolutionarily conserved residue. Consistent with somatic events, WTX mutations were absent in matched normal tissue in all four cases where such tissue was available (Fig. 2B) (table S1). Mutations identified in cases without matched normal tissue were not detected in control DNA from 269 healthy individuals.

In contrast to the classical biallelic “two-hit” Knudson model, the identification of somatic mutations affecting an X chromosome gene in sporadic Wilms tumor raises the possibility of “one-hit” inactivation of a tumor-suppressor gene. This applies to the hemizygous deletions and point mutations detected in males and, given that WTX is located in a chromosomal region subject to X inactivation (14), it would also apply to the heterozygous mutations detected in females if the active copy of the X chromosome is selectively targeted. To test this hypothesis, we used FISH analysis to search for heterozygous WTX deletions in female cases and to determine whether they affected the active or inactive X chromosome. Heterozygous WTX deletions in female cases were more reliably identified by FISH, compared with the initial array CGH analysis, presumably because of high signal-to-noise ratios for single-copy changes involving a small number of probes. Indeed, among 25 female Wilms tumor cases analyzed, 6 showed deletion of one WTX allele, as assessed by FISH with a probe for WTX and control probes for the X chromosome centromere and a telomeric Xq locus (table S1) (fig. S3). To determine whether these deletions targeted the active X chromosome allele, we performed simultaneous FISH analysis for WTX, the X chromosome centromere, and the inactive X chromosome coating transcript Xist (15). In all four cases tested, the intact WTX gene was present in the inactive X chromosome coated by Xist, whereas no WTX signal was detected in the active X chromosome that lacked Xist hybridization (Fig. 2C). Heterozygous WTX deletions in female Wilms tumors therefore target the active X chromosome, leading to gene inactivation by a single event.

We extended this analysis to an intragenic mutation of WTX (Glu334→Ter) in a female Wilms tumor in which it was possible to compare the nucleotide sequence of PCR products generated from genomic DNA or cDNA. Indeed, although the mutation was heterozygous in genomic DNA, only the mutant sequence was detected in the tumor-derived transcripts, consistent with monoallelic expression of the mutant copy from the active X chromosome (Fig. 2D). Taken together with the FISH analysis, these data indicate that intragenic mutations and gross chromosomal deletions of WTX occur at comparable frequencies in male and female Wilms tumor cases and that in females they exclusively target the active X chromosome.

Overall, of 51 tumors tested for both gene-copy alterations and intragenic mutations, 11 (21.6%) had WTX deletions, and 4 (7.8%) had point mutations (total 15 out of 51, 29.4%) (table S1). In these tumors, WT1 mutations were detected in three cases (5.9%) and β-catenin mutations in four cases (7.8%). As expected (7), there was overlap between WT1 and β-catenin mutations (two out of three cases with WT1 mutations). In contrast, no tumor with a deletion or point mutation in WTX contained mutations in WT1 or β-catenin (Fig. 2E). Whether inactivation of WTX defines a distinct subset of nephroblastomas remains to be investigated.

In contrast to other tumor-suppressor genes, WT1 has a developmentally regulated pattern of expression in the organ in which the tumor arises. The restricted expression of WT1 in renal blastemal stem cells and glomerular podocyte precursors is consistent with the presumed cell type of origin of Wilms tumor and highlights the key physiological role of WT1 in normal kidney development (16). In the mouse, WTX expression is relatively high in the neonatal brain and kidney and then declines substantially in the mature organs (Fig. 3A). Lung and spleen also express WTX, but with a less notable developmental profile. The temporal patterns of WTX and WT1 expression within the kidney are virtually identical, consistent with a wave of differentiation that is ongoing at the time of birth and is completed by postnatal week 3 (Fig. 3B). As assessed by RNA in situ hybridization, WT1 and WTX display a high, but not complete, degree of overlap in expression in the kidney. Similar to WT1, WTX is expressed in the condensing metanephric mesenchyme and in early epithelial structures that are precursors to glomeruli (Fig. 3, C and D). Thus, both genes are present in the pluripotent cells that are the presumed precursors of Wilms tumor.

Fig. 3.

Regulated expression of WTX. (A) qPCR quantitation of WTX mRNA in mouse neonatal and adult tissues. (B) Comparison of the developmental time course of WT1 and WTX expression in mouse postnatal kidneys. In (A) and (B), expression levels were normalized to glyceraldehyde-3-phosphate dehydrogenase (GAPDH) and displayed relative to neonatal kidneys. Error bars indicate standard deviation. (C and D) RNA in situ hybridization of human embryonic kidneys (week 13) showing similar but not identical expression patterns of WTX and WT1. Condensing metanephric mesenchyme (M) and glomerular precursors (G) are indicated. Scale bars, 250 μm. Insets are higher-magnification images of the mesenchyme (M). Scale bars, 100 μm. (E) Suppression of colony formation after ectopic expression of WTX in HEK-293 cells and U2OS cells. Cells were cotransfected with a WTX expression plasmid (or empty vector) and a plasmid encoding a drug-resistance marker (puromycin). Experiments were performed in triplicate and drug-resistant colonies were stained after 2 weeks. Representative plates and mean colony numbers are shown (± standard error of the mean).

Functional studies of WT1 and other genes implicated in Wilms tumorigenesis have been hampered by the absence of either mouse tumor models or experimentally manipulable Wilms tumor cell lines. Thus, we ectopically expressed WTX in cancer cell lines that have been used to model WT1 function. In human embryonic kidney (HEK) 293 cells and in U2OS human osteosarcoma cells, transfection of WTX ledtoa marked suppression of colony formation (Fig. 3E). Apoptosis was evident in HEK-293 cells 48 hours after ectopic expression of WTX (fig. S4). Studies in such heterologous cell types point to functional properties consistent with a tumor suppressor, but a more complete understanding of protein function will require in vivo studies of this developmentally regulated gene in the appropriate cellular context (5, 6).

Our data suggest that WTX is a Wilms tumor–suppressor gene with a potentially important role in normal kidney development. In addition, the localization of WTX on the X chromosome allows for complete inactivation by one mutational event targeting the single X allele in males or the active X allele in females. X-linked familial syndromes with increased cancer risk (1719) preferentially affect males. In contrast, WTX is frequently altered in sporadic tumors by a single somatic event that affects both sexes equally. One-hit inactivation of a tumor-suppressor gene on the X chromosome is a departure from the traditional biallelic Knudson model and has been postulated but never documented (20, 21). Although single-hit gene inactivation in Wilms tumor could lead to greatly increased tumorigenesis, its restriction to a limited pool of pluripotent target cells within a specific developmental window would mitigate this effect. As for WTX, high-resolution copy-number analysis and direct sequencing may be required for the identification of other X chromosome tumor suppressors subject to monoallelic inactivation, given that they are not marked by the characteristic secondary allelic loss (LOH) traditionally used for mapping. Together with recently described X chromosome abnormalities in a subset of breast cancer (22), the frequency of single-hit gene inactivation exemplified by WTX suggests that X chromosome genes may play unappreciated roles in human cancer.

Supporting Online Material

Materials and Methods

Figs. S1 to S4

Tables S1 to S4


References and Notes

View Abstract

Stay Connected to Science

Navigate This Article