SASH1: a candidate tumor suppressor gene on chromosome 6q24.3 is downregulated in breast cancer

Abstract

Loss of heterozygosity (LOH) and in silico expression analysis were applied to identify genes significantly downregulated in breast cancer within the genomic interval 6q23–25. Systematic comparison of candidate EST sequences with genomic sequences from this interval revealed the genomic structure of a potential target gene on 6q24.3, which we called SAM and SH3 domain containing 1 (SASH1). Loss of the gene-internal marker D6S311, found in 30% of primary breast cancer, was significantly correlated with poor survival and increase in tumor size. Two SASH1 transcripts of approximately 4.4 and 7.5 kb exist and are predominantly transcribed in the human breast, lung, thyroid, spleen, placenta and thymus. In breast cancer cell lines, SASH1 is only expressed at low levels. SASH1 is downregulated in the majority (74%) of breast tumors in comparison with corresponding normal breast epithelial tissues. In addition, SASH1 is also downregulated in tumors of the lung and thyroid. Analysis of the protein domain structure revealed that SASH1 is a member of a recently described family of SH3/SAM adapter molecules and thus suggests a role in signaling pathways. We assume that SASH1 is a new tumor suppressor gene possibly involved in tumorigenesis of breast and other solid cancers. We were unable to find mutations in the coding region of the gene in primary breast cancers showing LOH within the critical region. We therefore hypothesize that other mechanisms as for instance methylation of the promoter region of SASH1 are responsible for the loss of expression of SASH1 in primary and metastatic breast cancer.

Introduction

Breast cancer (BC) represents the most common malignant neoplasm in Western countries affecting one in eight to one in ten woman (Casey, 1997). The knowledge of molecular genetic mechanisms underlying breast tumorigenesis has increased since the discovery of the breast cancer susceptibility genes BRCA1 and BRCA2 (Miki et al., 1994; Wooster et al., 1995). Germline mutations in one of those genes account for about 30% of hereditary BC in the German population (German Consortium for Hereditary Breast and Ovarian Cancer, 2002). However, not all patients with a strong family history of this malignancy display alterations in these high penetrance genes. Moreover, there is no evidence of a correlation between mutations in BRCA1/BRCA2 and cases of sporadic BC, which make up to 90–95% of all breast cancers. The search for yet uncharacterized genes commonly involved in the development and/or progression of BC is complicated because of the great inter- and intratumor heterogeneity of the disease (Devilee and Cornelisse, 1994). To date, a number of variations in different loci were found to contribute to an increase in the lifetime breast cancer risk. These genes are associated with rare genetic syndromes (TP53, ATM, PTEN, LKB1), metabolic or immunomodulatory pathways (such as GSTM1, CYP19) or proto-oncogenic loci (for instance HRAS1) (De Jong et al., 2002). Genomewide loss of heterozygosity (LOH) and comparative genome hybridization (CGH) studies showed that large fractions of the human genome are aberrant in primary breast tumors and may harbor additional breast cancer-associated genes.

In our studies, we focused on the chromosomal region 6q23–25, which is a common site of allelic loss in BC (Fujii et al., 1996; Rodriguez et al., 2000) and also in other cancers like cervical (Acevedo et al., 2002), ovarian (Shridhar et al., 1999; Wan et al., 1999), pancreatic (Barghorn et al., 2001) and prostate tumors (Srikantan et al., 1999). As LOH is an important hallmark in the process of tumor suppressor gene (TSG) inactivation (Knudson, 1971), the 6q23–25 region appears to be a hot spot for several solid tumors and may contain one or more candidate genes involved in tumor suppression. This hypothesis was strongly supported by microcell-mediated chromosome transfer studies performed by our group, which showed that the introduction of a normal human chromosome 6 fragment encompassing the microsatellite marker D6S310 can suppress the neoplastic phenotype of the BC cell line CAL51 (Theile et al., 1996).

In this report, we narrowed down the location of the putative TSG candidates by allelotyping primary breast tumors. Simultaneously, an in silico approach was applied to obtain information about expressed genes within the defined interval and about their in silico expression profile. Combination of these methods revealed two target sequences showing a significantly reduced expression in breast tumor versus normal tissue. The first candidate gene was located next to the marker D6S310, and matched the putative TSG LOT-1/ZAC (Abdollahi et al., 1997; Spengler et al., 1997). The second candidate gene encompassing the expressed sequence tag (EST) Z28844 was located adjacent to D6S311. This marker is part of the gene KIAA0790 (Nagase et al., 1998). We describe here the complete genomic structure of the gene which we called SAM and SH3 domain containing 1 (SASH1). We show that SASH1 expression is downregulated in both breast tumors and BC cell lines. An attempt to detect mutations in tumor specimens failed. However, various sequence variations in the coding region of the SASH1 gene were identified. SASH1 encodes a protein containing sterile α module (SAM) and Src homology domain 3 (SH3) domains that are predominantly seen in signaling molecules, adapters and scaffold proteins. Since the functional domains of the protein suggested a role in signaling pathways, we further characterized the expression of SASH1 in primary BC and other solid tumors.

Results

Refined deletion mapping within 6q23–25 and association with clinicopathological parameters

Nine polymorphic microsatellite markers covering the genomic region 6q23–25 were selected for LOH analysis (Figure 1) in a panel of 61 sporadic breast tumors (Table 1). The deletion pattern of tumors with partial allelic loss (Figure 2) allowed the identification of three independent smallest regions of overlapping deletions (SRO). Samples S186, S183A and S193 carried an interstitial deletion of the marker D6S310 which defined the SRO I of 560 kb, delimited by D6S1648 and AFM269zg5. A second SRO (SRO II) of 1.98 Mb was defined by the samples S237, S189 and S210 showing loss of the marker D6S311 and retention of heterozygosity at the adjacent markers D6S978 and D6S1637. The third subregion SRO III encompassing 4.92 Mb was marked by interstitial loss of the D6S1654 locus in samples S200 and S209 while both alleles of markers D6S1637 and D6S441 were retained. Moreover, the defined SROs were sites of common allelic loss in tumors showing more complex deletion patterns. We observed 31% LOH around marker D6S310, 30% at the marker D6S311 and 33% at the D6S1654 locus.

Figure 1
figure1

Schematic representation of LOH frequencies in sporadic breast tumors at the chromosomal region 6q23–25. On the left, the surveyed microsatellite markers are listed in their relative order from centromer (cen) to telomer (tel). Horizontal bars indicate LOH frequencies (% LOH). On the right, LOH frequencies (% LOH), the number of tumors tested (T), the number of tumors being informative (I) and the absolute number of tumors showing LOH (L) are documented

Table 1 Clinicopathological features
Figure 2
figure2

Schematic representation of independently deleted smallest regions of overlap (SRO I-III) in 22 breast tumors showing partial deletions at 6q23–25 and candidate TSGs identified by expression analysis in silico. On the left, microsatellite markers and their relative map positions (http://www.ncbi.nlm.nih.gov/mapview/maps.cgi?orq=hum&chr=6) are listed. Numbers on top identify the tumor samples. On the right, candidate genes are shown which were significantly downregulated in breast tumors versus normal breast tissue. LOT-1/ZAC is located in SRO I while candidate SASH1 was found to be situated in SRO II. The region spanning SASH1 is outlined by a gray box. Black circles denote LOH, white circles retention of heterozygosity, gray circles homozygous loci, speckled circles represent untested tumors and hatched circles indicate microsatellite instability

Furthermore, we looked for associations between LOH and selected clinicopathological parameters, including survival time, histological type, estrogen receptor (ER) status and tumor size. A statistically significant correlation was observed for LOH at the D6S311 locus with survival time of patients (P-value=0.034). Four of 14 patients with a heterozygous deletion of D6S311 did not survive the clinical follow-up (of up to 53 months). In the group without LOH of D6S311, however, one of 33 patients died within this interval of time. Also, LOH at the D6S311 locus was significantly associated with growing tumor sizes (P-value=0.028). The rate of LOH was lowest among patients with small tumors (T<2 cm). LOH increased linearly (P-value=0.041) with tumor size in patients and occurred most frequently in patients with tumors >5 cm. Except for D6S311, there was no further significant correlation between LOH at the remaining eight loci and any of the investigated clinicopathological features.

D6S311 resides within SASH1, a gene with potential signaling properties

The association between LOH at D6S311 and reduced survival time prompted us initially to search for TSG candidates around this marker. Searching public databases revealed that the marker D6S311 was located in the first intron of a novel gene, known as KIAA0790 derived from its cDNA entry. No other genes in the immediate vicinity to D6S311 were identified by either gene prediction methods from genomic sequence nor by comparison of EST/cDNA sequences with genomic sequence entries.

The SASH1 cDNA was assembled from a combination of ESTs including EST Z28844 and KIAA0790 mRNA (REFSEQ Accession number. XM_044015). The complete SASH1 cDNA of 7709 bp, which we submitted to EMBL (Accession number BN000088), displays two polyadenylation signals at positions 4407 and 7685 bp, respectively (Figure 3). The human gene comprises 20 exons (Figure 3) and contains an open reading frame of 1247 amino-acid residues with a molecular weight of approximately 140 kDa. Determination of functional properties of the putative SASH1 gene product using protein databases suggested a role in signal transduction. The deduced protein harbors two sterile α module (SAM) domains and a Src homology domain 3 (SH3) indicative of adapter or scaffolding functions. Furthermore, a proline-rich sequence spanning 25 residues is located in the immediate amino-terminal portion of the human protein. These findings supported its putative function as a TSG. Thus, we subjected the gene to further characterization. Based on the functional domains, we named the gene SAM and SH3 domain containing 1 (SASH1). Alignment of SASH1 mouse (ENSMUSP00000015449) and human amino-acid sequences revealed high conservation throughout the conceptual protein (85% homology).

Figure 3
figure3

Schematic view of the genomic organization of SASH1. Genomic structure of SASH1 showing the 20 exons (boxes) to scale; sizes of introns (horizontal lines) are represented by vertical numbers. Gene regions encoding the SH3 domain (striped) and the two SAM domains (gray shaded) are shown. Hatched areas mark the 5′UTR and 3′UTR, respectively. Exon 1 contains the translation start site and the stop codon is situated in exon 20 followed by the polyadenylation signal (black circle). Alternatively, transcripts with an expanded 3′UTR can be transcribed via use of a second polyadenylation signal located in the region of the EST cluster around Z28844. The BAC containing the 5′ orientated region of the gene and the PAC clone spanning main parts of SASH1 are denoted by the lines on top. The accession numbers of the corresponding clone sequences are given above the line

Distribution of SASH1 in normal tissues reveals two transcripts of 4.4 and 7.5 kb

To elucidate the relationship between KIAA0790 and the consensus sequence around Z28844, we looked for corresponding transcription products. Determination of KIAA0790 expression in a variety of normal human tissues by Northern blotting showed two transcripts of approximately 4.4 and 7.5 kb (Figure 4a). Tissue distribution suggested ubiquitous expression of the 4.4 kb transcript, which was most abundant in lung, placenta, spleen and thymus. The larger transcript was expressed to a lesser extent in all tissues surveyed except for brain. The finding of two transcripts by Northern analysis is in agreement with the use of both polyadenylation sites predicted from the SASH1 mRNA. Although the observed size of approximately 7.5 kb for the longer transcript did not directly correspond to a full-length transcript of 7.7 kb, we supposed that the EST cluster around Z28844 was part of an extended 3′UTR of SASH1. Therefore, we performed Northern blot analysis using Z28844 as a probe (Figure 4b). A single transcript of about 7.5 kb could be detected in all tissues but not in brain. We also looked for the existence of a long SASH1 transcript using reversed transcription–polymerase chain reaction (RT–PCR) with primers located in exon 4 and at the end of the assembly in cell lines derived from normal and tumor breast tissues. RT–PCR analysis of MCF10A, HBL100 and CAL51 confirmed the presence of a 6.8 kb transcript (data not shown). Sequencing of the cloned 6.8 kb products revealed exactly the assembled SASH1 cDNA. Thus, we verified that the cluster around Z28844 is part of an alternatively processed SASH1 mRNA with a long 3′UTR. In addition, comparison between the 3′UTR of mouse (EMBL Accession number. AJ507736) and human SASH1 showed remarkable similarity (data not shown). Moreover, we found no evidence of alternative splicing of SASH1 transcripts by program-based prediction, EST sequences and RT–PCR experiments.

Figure 4
figure4

Expression of SASH1 and Z28844 in human tissues. Northern blots containing 1 μg of poly(A)+RNA per lane from adult tissues (Clontech) were hybridized with (a) a SASH1 cDNA fragment encompassing nucleotides 834–4726 and designated SASH12-18. One major transcript of approximately 4.4 kb was observed in all tissues, while an additional transcript of about 7.5 kb was present in all tissues to a lower extent, (b) a cDNA probe (J12417) for the EST cluster around Z28844 comprising the 1.7 kb EcoRI/PacI insert within the IMAGE clone 212843 (GenBank Accession number H70389). The same filters were probed with a cDNA fragment of β-actin (Clontech) as a loading control

In silico profiling and expression analysis shows significant downregulation of the SASH1 transcript in primary breast cancer and other solid tumors

We then used an in silico approach to analyse the expression pattern of the SASH1 transcript in comparison with other transcripts mapped within the genomic region between D6S453 (141.0 Mb) and D6S442 (155.3 Mb). This approach consists of two steps. First, 150 different ESTs that were identified to map between markers D6S453 and D6S442 (GeneMap’98) were automatically extended using the software program ‘AUTEX’ (Schmitt et al., 1999). In a second step, the resulting extended EST consensus sequences were then subjected to an electronic profiling method which we call electronic Northern analysis or eNORTH. eNORTH allows to determine the in silico expression pattern of a cDNA fragment between normal and tumor tissues by counting the distribution of ESTs between these tissues, and normalizing counts for each tissue in respect to the total number of ESTs available for each tissue. This analysis can be performed easily in silico as the tissue origin for each EST is available in the header information. We used several millions of ESTs obtained by sequencing from 30 human tissues and provided either by the CGAP project at NCI or by Incyte Genomics. In total, we identified 62 differentially expressed cDNA sequences in 16 different tissues across the 14 MB. We then focused on those transcripts that were differentially expressed only in breast cancer tissue versus breast normal epithelial cells. With this approach, we were able to narrow down the number of candidate cDNAs down to 10 sequences. Eight sequences were upregulated in breast tumor versus normal tissue, while the remaining two cDNAs showed a significantly reduced expression in breast tumor versus normal tissue. One was the known TSG candidate LOT-1/ZAC (12 ESTs from breast normal, one EST from breast tumor tissue, P-value=0.0406, EST pool sizes: 120738 and 67596, respectively). The other extended target cDNA sequence was assembled around the EST Z28844 containing the marker D6S1118E and represented parts of the SASH1 transcript. The electronic northern profile shows that SASH1 is downregulated in breast tumors as 15 ESTs were identified from normal breast tissue but only two ESTs from breast tumor tissue (EST pool sizes: 120 738 and 67 596, respectively). The P-value of 0.0425 for this observation is significant. The EST cluster around Z28844 showing downregulation in breast tumors was localized 1 kb downstream of KIAA0790. As this EST assembly did not show a significant open reading frame, we assumed that it was part of a longer KIAA0790 transcript and represented its 5′ or 3′UTR. This was later confirmed by RT–PCR experiments.

In silico expression analysis indicated that SASH1 transcripts are significantly reduced in breast tumor tissues. In order to confirm this finding, we employed an RT–PCR approach using nine different BC cell lines and primers specific for exons 14 and 15. As controls we used two immortalized mammary epithelial cell lines MCF10A and 184A1 (Figure 5). SASH1 expression is completely lost in the two of the nine BC cell lines MDA-MB-231 and CAMA-1. In MCF7, BRC230, MT-1 and CAL51 the mRNA level is significantly reduced while no change of the transcript level was observed in MT-3, R193 and BT-20. These observations partly correspond to FISH analysis data on deletions of 6q in BC cell lines. All BC cell lines with deletion of the region 6q24.3 (Zhang et al., 1998) had reduced or missing levels of SASH1 transcripts (BRC230, MT-1, MDA-MB-231 and CAMA-1). However, decreased expression was also observed in cell lines without loss of 6q24.3 (MCF7, CAL51).

Figure 5
figure5

Expression of SASH1 in a panel of human cell lines (two normal breast (BN), nine breast cancer (BC) cell lines) detected by RT–PCR. A SASH1 cDNA fragment of 321 bp generated from exons 14 and 15 and the positive control for RNA integrity, a 126 bp portion of PBGD coding region, were amplified in a multiplex RT–PCR and visualized on a polyacrylamid gel stained with ethidium bromide (size marker (M))

We then asked the question if downregulation of SASH1 transcripts was also a relevant event in vivo during tumorigenesis. To address these questions, we quantified the level of SASH1 mRNA in a panel of 15 primary breast tumors matched to normal breast tissues from the same patient using a highly sensitive fluorescence-based RT–PCR assay. Primary breast tumors that we had previously used for the LOH study at the SASH1 locus could not be re-analysed because of shortage of tissue. Comparison of the relative mRNA levels in the matched samples showed a reduced expression in all tested tumors. In nine of the 15 samples we observed at least a twofold reduction of SASH1 transcripts as compared to the corresponding normal breast tissues (Figure 6).

Figure 6
figure6

Expression ratio of SASH1 in 15 primary breast tumors as compared to matched normal breast epithelia. Horizontal bars show the expression ratio of the matched samples which were normalized to PBGD. The codes of the tumors analysed are given on the left

To confirm these results, we investigated a larger panel of tumors using a Cancer Expression Profiling Array (Clontech), which contained cDNAs from 50 different breast tumor tissues matched with the corresponding normal control tissue. SASH1 expression was examined by hybridization with a probe covering 697 bp from exons 4 to 10. SASH1 transcripts were significantly reduced in 74% (37/50) of all breast tumors as compared to the corresponding normal tissue (Figure 7). The gene is also strongly downregulated in the three breast tumor metastasis samples spotted onto the Clontech Array. The array also allowed us to study SASH1 expression profiles across matched pairs from a number of other solid tumors. SASH1 transcripts were strongly reduced in 19 of the 21 spotted lung tumor samples (90%) and in five of the six spotted thyroid tumors. SASH1 expression was also downregulated in a few colon carcinoma samples (Figure 7).

Figure 7
figure7

SASH1 expression in normalized cDNAs from normal and tumor tissues. The Cancer Profiling Expression Array (Clontech) was hybridized with a SASH1 cDNA probe (SASH1-Ex4-10) covering nucleotides 834 – 1531. To verify equally spotted amounts of cDNA, the array was also hybridized with a cDNA probe for ubiquitin (Clontech) (data not shown). Cell lines: HELA (cervix cancer), Daudi (Burkitt's lymphoma), K562 (chronic myelogenous leukemia), HL-60 (promyelocytic leukemia), G361 (melanoma), A549 (lung carcinoma), MOLT-4 (lymphobic leukemia), SW480 (colorectal adenocarcinoma), Raji (Burkitt's lymphoma). Controls in vertical order from top to bottom: ubiquitin cDNA, yeast total RNA; yeast tRNA; Escherichia coli DNA; poly(A); human C0t-1 DNA; human genomic DNA. N, normal tissue; T, tumor tissue. The outlined groups of dots represent normal, metastatic (spotted side by side) and tumor tissue from the same patient. The arrow indicates one sample of matched normal and tumor colon tissue in the column of stomach samples

Mutation analysis in sporadic and familial BC

To investigate whether sequence alterations of SASH1 could be responsible for its downregulation in primary breast tumors and BC cell lines, we analysed 66 primary breast tumors for somatic mutations in the entire coding region and splice junctions of the gene. The panel included tumors subjected to LOH analysis at the region 6q23–25, as described above. We also investigated the SASH1 gene for mutations in blood samples from 44 patients with familial breast cancer as the region 6q23–25 may also be prone to harbor a gene involved in inherited BC (Zuppan et at., 1991).

As shown in Table 2, we observed a total of 18 sequence alterations in the SASH1 gene. These alterations include four missense, nine silent and four intronic changes as well as one insertion of two amino acids. The substitutions in intronic sequences of SASH1 did not affect splice sites. With the exception of the 6 bp insertion in exon 1, three silent mutations in exon 18 (3640AG, 3748GA, 3802GA) and two missense mutations in exon 19 (3896GA, 3932CT), the observed variations represent single-nucleotide polymorphisms (SNPs) with similar frequencies in patients and controls. Interestingly, an insertion of six nucleotides (exon 1) causing an in-frame insertion of two amino acids was found in various patients and controls. This insertion predicts a new variant of the protein with an additional proline (P) and glutamic acid (E) residue in the proline-stretch region (PE-region).

Table 2 Frequencies of sequence variations in SASH1 in sporadic BC, familial BC and controls

Upon mutation screening we did not detect tumor-specific mutations in sporadic BC. In detail, substitutions in exon 18 (i.e. 3559CA, 3640AG, 3748GA, 3802GA) and exon 19 (3896GA) were not only found in the tumor but also in the corresponding normal breast tissue of the patients. No mutations were revealed within the functional domains of the gene. As the homozygous GA substitution at position 2287 (SH3 domain) observed in two BC samples and missing in controls does not cause an amino-acid change, it was not regarded to be critical.

Across the set of hereditary BC we observed a heterozygous mutation (3932CT) which did not occur in sporadic BC and controls. The substitution 3932CT leads to a heterozygous amino-acid change from leucin to phenylalanine. Although this position is conserved in mouse and human, both amino acids display a hydrophobic character and therefore the influence of this variant seems to be moderate.

In summary, we could not find inactivating events such as premature stop codons in sporadic and familial BC. Strikingly, we could neither demonstrate mutations in patients with LOH at D6S311/D6S1637 nor in BC samples showing reduced expression of SASH1. These data suggest mechanisms other than mutations in coding sequences to be involved in downregulation of SASH1.

Discussion

Allelic losses in the chromosomal region 6q23–25 have frequently been associated with BC and many other solid tumors, which supports the hypothesis that this region harbors one or more TSG possibly involved in tumor formation and progression (Tirkkonen et al., 1998; Huang et al., 2000; Stallmach et al., 2002). In addition, we previously showed that microcell-mediated transfer of a small chromosomal fragment around the marker D6S310 on 6q23 causes loss of the tumorigenic phenotype in CAL51 breast cancer cells (Theile et al., 1996). However, we failed to detect any candidate TSG after complete sequence analysis of the PAC dJ181M7, which contains marker D6S310 and covers a large genomic region around this marker (unpublished results). To date, the gene LOT-1/ZAC located near the marker D6S310 has been suggested as a potential TSG for ovarian and breast cancer. LOT-1/ZAC encodes a zinc-finger protein which was shown to exhibit antiproliterative activities by regulating apoptosis and cell cycle arrest. Expression of the gene is reduced or lost in ovarian and breast cancer cell lines and primary tumors (Abdollahi et al., 1997; Bilanges et al., 1999). In the absence of loss-of-function mutations, inactivation of the putative TSG was suggested to be mediated via promoter methylation and imprinting status (Bilanges et al., 1999; Fagotto et al., 2000).

Here we report on the identification and initial characterization of a novel TSG candidate for BC, which we refer to as SASH1. This gene was identified by a combination of LOH mapping in primary BC and expression profiling analysis in silico within the critical region of 6q23–25 flanked by the markers D6S1648 and D6S442, proximal and distal of D6S310, respectively. Fine mapping revealed three regions of smallest overlapping deletions. These were located next to the markers D6S310 (SRO I spanning 560 kb, 31% LOH), D6S311 (SRO II encompassing 1.98 Mb, 30% LOH) and D6S1654 (SRO III encompassing 4.92 Mb, 33% LOH). We performed correlation studies in order to show association between LOH around microsatellite markers and specific clinicopathological parameters. Loss of D6S311 was significantly associated with shorter survival time and greater tumor size of BC patients in comparison with patients with retention of both alleles. This finding suggested that loss of a novel gene located in the vicinity of D6S311 may be responsible for a growth advantage to breast tumor cells and may contribute to the progression of BC. In this respect, D6S311 might serve as a prognostic marker as LOH of D6S311 would indicate a poor outcome of the disease.

On the basis of the above findings we initiated an in silico expression analysis of all ESTs anchored around the marker D6S311 and identified a novel TSG candidate in the vicinity of this marker. This TSG candidate contains EST Z28844 and is downregulated in primary breast cancers. Detailed sequence analysis showed that this TSG candidate is identical with the long 7.5 kb transcript of a novel gene which we named SASH1 and which was derived and annotated from the KIAA0790 cDNA sequence. The SASH1 gene harbors the microsatellite marker D6S311 and is located in 6q24.3.

The SASH1 gene shows a high degree of conservation among mouse and human on the nucleotide and protein level. Based on the presence of an SH3 and two SAM domains, the putative protein may have a function in signal transduction cascades. SH3 domains mediate protein–protein interactions by binding to polyproline motifs and are involved in a broad spectrum of processes such as tyrosine kinase signaling and cytosceletal organization (Pawson, 1995; Buday, 1999). A similar function is attributed to SAM domains that not only show the ability to self-associate (Carroll et al., 1996) but can also heterotypically interact with other proteins (Serra-Pages et al., 1995). Moreover, Kim et al. (2002) recently suggested that polymerization of SAM domains in transcriptional repressers yields repressive chromatin structure and is involved in transcriptional control. Based on the functional domains SASH1 may belong to the previously described novel family of putative adapter and scaffold proteins. This family includes the SAM-and SH3-containing genes HACS1 (Claudio et al., 2001), NASH1 (Uchida et al., 2001) and SLY (Beer et al., 2001).

The functional features of SASH1 support the idea that the gene is possibly involved in tumor suppression. Therefore, we analysed the expression of the gene in vitro and in vivo. SASH1 expression was verified in a variety of tissues, preferentially in breast, lung, thymus, thyroid, placenta and spleen. Northern analysis showed that two mRNA transcripts of approximately 4.4 and 7.5 kb that differ only in the length of the 3′UTR exist. The length of the shorter 4.4 kb transcript was in agreement with the annotated KIAA0790 cDNA. The 7.5 kb isoform is expressed at lower levels in the examined tissues except for brain.

In six of nine breast cancer cell lines, SASH1 is significantly reduced or completely lost at the mRNA level. Most of these cell lines displayed chromosomal loss of the region 6q24.3, which suggested that downregulation may partially be caused by loss of the gene. Using a sensitive RT–PCR assay and expression analysis with a panel of spotted primary breast cancers and corresponding normal controls, we proved a strong decrease of SASH1 mRNA in about 70% of total 65 primary BC samples. In addition, SASH1 was also significantly reduced in primary tumors of lung and thyroid. Loss of the long arm of chromosome 6 is a frequent event in lung tumors and cell lines (Wurster-Hill et al., 1984; Virmani et al., 1998). Interestingly, Goeze et al., (2002) suggested that deletion of 6q21–qter is associated with the metastatic phenotype in primary lung adenocarcinomas. This implies that a putative TSG from this area could be involved in more aggressive stages of lung tumors when lost. We did not observe a decreased expression of SASH1 in most of the primary tumors although high frequencies of LOH have been reported for SASH1-internal markers D6S311 and D6S1637 in ovarian cancer (Shridhar et al., 1999).

To further investigate whether SASH1 may function as a TSG in BC, we searched for the ‘second hit’ according to Knudson's ‘two hit hypothesis’. Characterization of genetic alterations in 66 sporadic breast tumors and lymphocytes from 44 BC patients with a strong family history of breast cancer did not reveal a loss-of-function mutation. We identified 14 new sequence alterations, six of which were silent substitutions, and four were detected in intronic regions. We found three new missense substitutions and one in-frame insertion of two amino acids. The observed silent (3748GA, 3802GA) and missense mutations (3896GA) were not specific to the sporadic tumors and did not occur in tumors with LOH of the wild-type allele. In hereditary BC, the role of the missense mutation 3932CT needs further determination in additional affected family members. The frequency of the observed sequence variations did not significantly differ in BC and controls. Although functional studies have not been performed, we assume that most of these variations do not alter the function of SASH1 and, in consequence, do not play an important role in BC. However, the observed polymorphism in the first exon results in a new variant of the protein carrying an additional proline and glutamic acid residue in the PE-region. Since the PE-region of SASH1 could be of functional importance regarding interactions with other proteins or with the SASH1-internal SH3 domain, we speculate that the binding properties of the mutant protein variant may be changed because of altered conformation.

Our results show that reduced expression of SASH1 may not be attributed to somatic mutations in the coding sequences of SASH1. Other mechanisms are likely responsible for the loss of expression of SASH1 at this locus. Transcriptional mechanisms including epigenetic changes, genetic variations in the promoter as well as negative or positive expression of binding factors may be responsible for the inactivation of SASH1. Furthermore, alterations in the 3′UTRs of the different transcripts might change post-transcriptional regulation of gene expression. On the other hand, inactivation of SASH1 may provide cancer cells with an advantage in microevolution, as proposed for the TSG E-cad (Cheng et al., 2001). SASH1-specific polyclonal and monoclonal antibodies may be versatile tools to provide more insight into the gene's function. As SASH1 belongs to a novel family of putative adapter and scaffold molecules and has been shown by us to be downregulated in primary breast cancer and other solid tumors, it will be interesting to investigate if other members of this family also possess a similar potential to act as tumor suppressors.

Materials and methods

Materials

A total of 78 tumor specimens and corresponding nontumor epithelial tissues from breast cancer patients who had undergone surgery were collected at the Robert Roessle Hospital (Berlin, Germany). In total, 61 samples were utilized for LOH studies, 66 for mutation analysis and 12 were selected for semiquantitative RT–PCR. Additionally, blood samples from 44 breast cancer patients with a family history of BC, collected at the Max Delbrueck Center (Berlin, Germany), were investigated for SASH1 mutations. Blood samples (50) from German patients, who had attended Berlin hospitals, served as controls. The study was performed with the approval of the local ethics committee.

The cell lines used in this study were obtained from the American Type Culture Collection (ATCC) (Rockville, MD) (MCF10A, 184A1, MDA-MB-231) or were kindly provided by J Gioanni (Centre Antoine-Lacassagne, Nice, France) (CAL51), D Amadori (Morgagni-Pierantoni Hospital, Forli, Italy) (BRC230), I. Fichtner (MT-1, MT-3) and M. Theile (MDC, Berlin, Germany) (CAMA-1, MCF7, BT-20) and Dr Arnold (atugen, Berlin, Germany) (R193).

To confirm their identity, the cell lines were screened with microsatellite markers D16S539, D5S818 and D7S820 and the resulting fingerprints were compared with the corresponding entities in the ATCC database of DNA fingerprints of STR (http://www.atcc.org/).

DNA/RNA extraction

DNA was extracted using standard phenol/chloroform methods. Total RNA was prepared using the Perfect RNA, Eukaryotic, Mini Kit (Eppendorf) or the TRIZOL reagent (Life Technologies) following the manufacturer's instructions.

Frozen tissue samples were sectioned with a cryostat microtome. The tumor cell content, which was assessed in an adjacent hematoxylin–eosin stained section, was estimated to be more than 50%.

LOH analysis

Primers for the nine polymorphic markers were synthesized according to published primer sequences (Genome Data Base, 1998). The order of the markers was derived from existing consensus maps (Cedar Genetics, 1999; http://cedar.genetics.soton.ac.uk/pub/chrom6).

PCR were carried out with 40 ng DNA in PCR buffer (Perkin-Elmer), 0.25 μ M of each primer (one primer of each pair was fluorescence labeled), 375 μ M of dNTPs each and 0.25 U Taq-Polymerase (Perkin-Elmer). Denaturation at 94°C for 5 min was followed by 30–35 cycles of 30 s at 94°C, 30 s at annealing temperature (between 52 and 65°C) and 30 s or 1 min at 72°C, with a final extension of 10 min, 72°C.

The PCR products were mixed with an internal standard size marker and fractionated on a denaturating 6% polyacrylamide gel using an ABI 377 DNA sequencer. LOH data were automatically analysed by comparing nontumor and tumor tissue allele peak sizes, heights and area ratios. Intensity or signal ratio differences of 35% or more were considered sufficient for LOH assignment.

Statistical analysis

All statistical analyses were done using the statistical program SPSS. To evaluate associations between LOH and clinicopathological variables, we applied Fisher's exact test (Crosstabs statistics) in the case of categorical variables and the Mann–Whitney test or the H-test of Kruskal and Wallis in the case of noncategorical variables. Survival curves calculated by Kaplan–Meier estimations were compared according to the log-rank test (Sachs, 1992). For all statistical tests, we used a comparison related significance level of P<0.05.

Expression analysis

Expression analysis in silico

Data were obtained following the procedure decribed by Schmitt et al. (1999) using EST information from public (NCI) and proprietary sources (Incyte Genomics). The set of data comprised a total of 2 017 017 ESTs, which included 120 738 ESTs from breast normal and 67 596 ESTs from breast tumor tissue.

Northern blots

Multiple tissue northern (MTN) filters were purchased from Clontech, which were loaded with 1 μg of poly(A)+RNA per lane. Probes corresponding to cDNA fragments SASH12-18, J12417 and β-actin (Clontech) were labeled with 32P-dCTP using Prime-a-Gene system (Promega). Each blot was stripped and reprobed with β-actin after successful hybridization with SASH1-specific fragments, respectively. Prehybridization, hybridization and washing were performed as recommended by the manufacturer.

RT–PCR

Gene expression was measured determining the exponential range of amplification of the PCR for SASH1 and the normalization control gene porphobilinogen deaminase/hydroxymethylbilane synthase (PBGD) (Yoo et al., 1993). To confirm product specificity, PCR products were subjected to cDNA sequencing. Primers used for SASH1 were U1824 (5′-IndexTermCGGGAAAGCGTCAAGTCGGA-3′) and L2145 (5′-IndexTermCCAGGAGATCCTCCACAGAC-3′). RT–PCR was carried out in a one-step reaction using QIAGEN One-Step RT–PCR Kit or in a two step-reaction using QIAGEN Omniscript RT Kit according to the manufacturer's instruction. In one-tube multiplex RT–PCR, 100 ng RNA were used in a 25 μl reaction volume. In the two-tube RT–PCR, 400 ng RNA were subjected to randomly primed RT in a 25 μl PCR reaction. Initial denaturation was performed at 95°C for 5 min, followed by 40 cycles of 30 s at 95°C, 30 s at 62°C and 1 min at 72°C, with a final extension at 72°C for 10 min.

Semiquantitative fluorescent RT–PCR analysis

Expression analysis of primary breast tumors and matched tissues was performed with fluorescent dye-labeled primers as described above (U1824-TET, PBGD-HEX). For quantification, PCR products were applied to an ABI 377 DNA sequencer. Data were collected automatically and analysed by the Genescan 3.1 software (ABI), which provides quantification based on peak size and peak area.

The reproducibility of the method was confirmed by two independent PCRs for two matched samples. The PCR products were subjected to quantification six times. Means and s.d.'s were evaluated and found within a reproducible range.

Cancer profiling array

Expression analysis of SASH1 was performed using a Cancer Profiling Array (Clontech), which contained matched cDNA pairs of individual patients. In addition, cDNA from cancer cell lines and several positive and negative controls were immobilized on the array. All cDNAs were normalized based on the expression of four housekeeping genes. SASH1 (SASH1-Ex4-10) and ubiquitin (Clontech) cDNA fragments were 32P-random-primer labeled (Random Labeling Kit, Roche) and used to probe the same filter. Prehybridization, hybridization and washing were performed following the manufacturer's protocol.

Mutation analysis

Exons 1–20 of SASH1 were amplified by PCR using intronic primer sequences (available upon request) and examined by single-strand conformation polymorphism analysis (SSCP). PCR was carried out with 100 ng DNA, 15 μl Taq PCR Master Mix (QIAGEN), 15 pmol primer each in a final volume of 30 μl. Cycling parameters included an initial denaturation step at 95°C for 5 min, 35 cycles of denaturation at 95°C for 30 s, annealing at optimal temperature (ranging from 51 to 68°C) for 40 s and extension at 72°C for 1 min 30 s, followed by a final extension at 72°C for 10 min. PCR products showing mobility shifts in the SSCP-Gel were purified with a QIAquick PCR purification Kit (QIAGEN) and then sequenced on ABI 377 sequencers.

References

  1. Abdollahi A, Roberts D, Godwin AK, Schultz DC, Sonoda G, Testa JR and Hamilton TC . (1997). Oncogene, 14, 1973–1979.

  2. Acevedo CM, Henriquez M, Emmert-Buck MR and Chuaqui RF . (2002). Cancer, 94, 793–802.

  3. Barghorn A, Speel EJ, Farspour B, Saremaslani P, Schmid S, Perren A, Roth J, Heitz PU and Komminoth P . (2001). Am. J. Path., 158, 1903–1911.

  4. Beer S, Simins AB, Schuster A and Holzmann B . (2001). Biochim. Biophys. Acta, 1520, 89–93.

  5. Bilanges B, Varrault A, Basyuk E, Rodriguez C, Mazumdar A, Pantaloni C, Bockaert J, Theillet C, Spengler D and Journot L . (1999). Oncogene, 18, 3979–3988.

  6. Buday L . (1999). Biochim. Biophys. Acta, 1422, 187–204.

  7. Carroll M, Tomasson MH, Barker GF, Golub TR and Gilliland DG . (1996). Proc. Natl. Acad Sci. USA, 93, 14845–14850.

  8. Casey G . (1997). Curr. Opin. Oncol., 9, 88–93.

  9. Cheng CW, Wu PE, Yu JC, Huang CS, Yue CT, Wu CW and Shen CY . (2001). Oncogene, 20, 3814–3823.

  10. Claudio JO, Zhu YX, Benn SJ, Shukla AH, McGlade CJ, Falcioni N and Stewart AK . (2001). Oncogene, 20, 5373–5377.

  11. De Jong MM, Nolte IM, Te Meerman GJ, Van der Graaf WTA, Oosterwijk JC, Kleibeuker JH, Schaapveld M and De Vries EGE . (2002). J. Med. Genet., 39, 225–242.

  12. Devilee P and Cornelisse CJ . (1994). Biochim. Biophys. Acta., 1198, 113–130.

  13. Fujii H, Zhou W and Gabrielson E . (1996). Genes Chromosomes Cancer, 16, 35–39.

  14. German Consortium for Hereditary Breast and Ovarian Cancer (2002). Int. J. Cancer, 97, 472–480.

  15. Goeze A, Schluns K, Wolf G, Thasler Z, Petersen S and Petersen I . (2002). J. Pathol., 196, 8–16.

  16. Huang LW, Garrett AP, Muto MG, Colitti CV, Bell DA, Welch WR, Berkowitz RS and Mok SC . (2000). Hum. Pathol., 31, 367–373.

  17. Kim CA, Gingery M, Pilpa RM and Bowie JU . (2002). Nature Struct. Biol., 9, 453–457.

  18. Knudson AG . (1971). Proc. Natl. Acad. Sci. USA, 68, 820–823.

  19. Miki Y, Swensen J, Shattuck-Eidens D, Futreal PA, Harshman K, Tavtigian S, Liu Q, Cochran C, Bennett LM, Ding W, Bell R, Rosenthal J, Hussey C, Tran T, McClure M, Frye C, Hattier T, Phelps R, Haugen-Strano A, Katcher H, Yakumo K, Gholami Z, Shaffer D, Stone S, Bayer S, Wray C, Bodgen R, Dayananth P, Ward J, Tonin P, Narod S, Bristow P, Norris FH, Helvering L, Morrison P, Rosteck P, Lai M, Barret JC, Lewis C, Neuhausen S, Cannon-Albright L, Goldgar D, Wiseman R, Kamb A and Skolnick MH . (1994). Science, 266, 66–71.

  20. Nagase T, Ishikawa K, Suyama M, Kikuno R, Miyajima N, Tanaka A, Kotani H, Nomura N and Ohara O . (1998). DNA Res, 5, 277–286.

  21. Pagotto U, Arzberger T, Theodoropoulou M, Grübler Y, Pantaloni C, Saeger W, Losa M, Journot L, Stalla GK and Spengler D . (2000). Cancer Res., 60, 6794–6799.

  22. Pawson T . (1995). Nature, 373, 573–580.

  23. Rodriguez C, Causse A, Ursule E and Theillet C . (2000). Genes Chromosomes Cancer, 27, 76–84.

  24. Sachs L . (1992). Angewandte Statistik. Springer Verlag: Berlin, Heidelberg, New York, Tokio.

  25. Schmitt AO, Specht T, Beckmann G, Dahl E, Pilarsky CP, Hinzmann B and Rosenthal A . (1999). Nucl. Acids Res., 27, 4251–4260.

  26. Serra-Pages C, Kedersha NL, Fazikas L, Medley Q, Debant A and Streuli M . (1995). EMBO J., 14, 2827–2838.

  27. Shridhar V, Staub J, Huntley B, Cliby W, Jenkins R, Pass HI, Hartmann L and Smith DI . (1999). Oncogene, 18, 3913–3918.

  28. Spengler D, Villalba M, Hoffmann A, Pantaloni C, Houssami S, Bockaert J and Journot L . (1997). EMBO J., 16, 2814–2825.

  29. Srikantan V, Sesterhenn IA, Davis L, Hankins GR, Avallone FA, Livezey JR, Connelly R, Mostofi FK, McLeod DG, Moul JW, Chandrasekharappa SC and Srivastava S . (1999). Int. J. Cancer, 84, 331–335.

  30. Stallmach I, Zenklusen P, Komminoth P, Schmid S, Perren A, Roos M, Jianming Z, Heitz PU and Pfaltz M . (2002). Virchows Arch., 440, 77–84.

  31. Theile M, Seitz S, Arnold W, Jandrig B, Frege R, Schlag PM, Haensch W, Guski H, Winzer KJ, Barrett JC and Scherneck S . (1996). Oncogene, 13, 677–685.

  32. Tirkkonen M, Tanner M, Karhu R, Kallioniemi A, Isola J and Kallioniemi OP . (1998). Genes Chromosomes Cancer, 21, 177–184.

  33. Uchida T, Nakao A, Nakano N, Kurasamu A, Saito H, Okumura K and Ogawa H . (2001). Biochem. Biophys. Res. Commun., 288, 137–141.

  34. Virmani AK, Fong KM, Kodagoda D, Mclntire D, Hung J, Tonk V, Minna JD and Gazdar AF . (1998). Genes Chromosomes. Cancer, 21, 308–319.

  35. Wan M, Sun T, Vyas R, Zheng J, Granada E and Dubeau L . (1999). Oncogene, 18, 1545–1551.

  36. Wooster R, Bignell G, Lancaster J, Swift S, Seal S, Mangion J, Collins N, Gregory S, Gumbs C, Micklem G, Barfoot R, Hamoudi R, Patel S, Rice C, Biggs P, Hashim Y, Smith A, Connor F, Arason A, Gudmundsson J, Ficence D, Kelsell D, Ford D, Tonin P, Bishop DT, Spurr NK, Ponder BAJ, Eeles R, Peto J, Devilee P, Cornelisse C, Lynch H, Narod S, Lenoir G, Egilsson V, Barkadottir RB, Easton BF, Bentley DR, Futreal PA, Ashworth A and Stratton M . (1995). Nature, 378, 789–792.

  37. Wurster-Hill DH, Cannizzaro LA, Pettengill OS, Sorenson GD, Gate CC and Maurer LH . (1984). Cancer Genet. Cytogenet., 13, 303–330.

  38. Yoo HW, Warner CA, Chen CH, and Desnick RJ . (1993). Genomics, 15, 21–29.

  39. Zhang Y, Matthiesen P, Siebert R, Harder S, Theile M, Scherneck S and Schlegelberger B . (1998). Human Genet., 103, 727–729.

  40. Zuppan P, Hall JM, Lee MK, Ponglikitmongkol M and King MC . (1991). Am. J. Hum. Genet., 48, 1065–1068.

Download references

Acknowledgements

We thank K Poppe and S Werner for the excellent technical assistance. This work was supported by: Deutsche Krebshilfe (Grant number: 10-1249).

Author information

Correspondence to Constanze Zeller.

Rights and permissions

Reprints and Permissions

About this article

Keywords

  • breast cancer
  • LOH analysis
  • 6q24.3
  • tumor suppressor genes
  • SAM
  • SH3

Further reading