WO1996003527A2

WO1996003527A2 - Diagnostic method and probe

Info

Publication number: WO1996003527A2
Application number: PCT/GB1995/001721
Authority: WO
Inventors: David Tarin; Yasuhiro Matsumura
Original assignee: Isis Innovation Limited
Priority date: 1994-07-21
Filing date: 1995-07-21
Publication date: 1996-02-08
Also published as: CA2195404A1; WO1996003527A3; JPH10504188A; EP0771361A2

Abstract

Introns and a new exon, Exon 9a, are formed as transcripts during uncontrolled expression of the CD44 gene in tumour cells. Novel probes comprise labelled oligonucleotides which hybridize to complementary strands synthesized on introns. A diagnostic method comprises hybridizing these probes with nucleic acids recovered from mammalian cells. Intron In-9 having a 474 bp sequence and Exon 9a having a 436 bp sequence.

Description

DIAGNOSTIC METHOD & PROBE

Eukaryotic gene expression begins with assembly of a homologous RNA strand on the DNA template. The RNA transcript is then edited in the cell nucleus so that non-coding regions (introns) are

10 excised and the remaining coding sequences are spliced together to form the mature messenger RNA molecule. This moves through the nuclear membrane and carries the encoded information which directs protein assembly to the synthetic apparatus in the cell cytoplasm.

'-' In previous work we and other groups have provided evidence that the function of the CD44 gene is severely deranged from very early pre-invasive stages of tumour development and that this can be used as a means of tumour detection and diagnosis on solid tissue

2^υ specimens (1-5) and on exfoliated cells in clinically obtained excreta and body fluids (2) . The gene is some 50-60kb in size, resides on chromosome 11 and is known to be composed of at least 20 exons, 10 or more of which can be alternatively spliced to produce various

P³ isoforms (9) . These are the variant exons. The abnormal products of the gene can be detected with the technique of reverse-transcription/PCR followed by

Southern hybridization as described before (1-5) or by monoclonal or polyclo al antibodies to peptide

3 sequences encoded by the variant exons of the gene (6- 8) . The abnormal and over-abundant activity of this genetic locus in tumour tissue relative to normal counterparts, can be detected by a distinctive pattern when its amplified and electrophoretically separated

35 products are blotted and hybridised with homologous probes. In tracks containing products from tumour tissues very long intense smears are observed, whereas in tracks containing normal samples there are only occasional and discrete bands or no signal at all. Urine samples from patients with bladder cancer, which contain exfoliated cancer cells, also show similar deranged CD44 activity. (It should be noted that, depending on the positions of the primers chosen, distinctive large molecular weight bands can also be produced from tumour cell-containing samples with this technique, as shown previously (2) but the long dense smears are more obvious indicators of severely deranged gene transcription.)

We reasoned that one explanation for the smears could be that the over-abundant production of transcripts from the gene could overwhelm editing and splicing mechanisms in the cell and lead to the accumulation of detectable quantities of immature or defective transcripts containing intronic sequences. To test this hypothesis we obtained a short intron sequence by PCR across the gap between two exons in genomic DNA with appropriate CD44 primers which we designed. The genomic DNA which we used as a template for this procedure was obtained from a CD44 clone picked from a genomic library. Using this amplified sequence as a labelled probe, we demonstrated that the hypothesis that abundant intron-containing RNA sequences are present in tumour cells but not in their normal counterparts, is correct.

From this discovery, it is deduced that that editing and splicing failures may be characteristic consequences of chaotic over-expression of genes in general in tumour cells in general . It is deduced that intron-containing RNA sequences may be characteristic of tumour cells in general. It follows from this that probes for intronic sequences can be used for the specific diagnosis of tumours. In one aspect this invention provides a probe comprising an oligonucleotide which is substantially homologous or hybridises to a complementary strand synthesised on an intron of a mammalian gene and which is labelled with a signal moiety. Preferably the probe comprises a mixture of two or more such oligonucleotides. The term oligonucleotide is here used to describe single stranded chains of at least 8 e.g. tens, hundreds or even thousands of nucleotides. Such a probe is substantially homologous or hybridises to the RNA which is a complementary strand synthesised on an intron of a mammalian gene. Such RNA would be removed and broken down during the editing and splicing process which takes place in a normal cell; its continuing and accumulating presence is therefore characteristic of a tumour cell.

At the present time, our demonstration has been confined to several different introns of the CD44 gene. However as noted, it is plausible that corresponding effects will occur in genes in general in tumour cells in general. For diagnostic purposes, it may be preferable to use probes derived from introns of tissue-specific genes. For instance one could plausibly deduce the origin of a metastasis when the site of the primary tumour is not clinically evident. Examples of genes likely to be of interest are the fibronectin gene and the alpha-fetoprotein gene both in connection with liver cancer; and the surfactant gene in connection with lung cancer. Where a gene is well characterised and its exons sequenced, it is a straightforward matter to sequence the introns linking those exons, and to devise probes based on those introns.

The probe is labelled with a signal moiety, whose nature is not material to the invention.

Suitable probes include radioactive isotopes, haptens of antigens for binding to antibody, biotin for binding to avidin or streptavidin, fluorescent moieties, or components of light- or colour- generating enzyme systems. The technology for labelling oligonucleotides with signal moieties is well established.

In another aspect, the invention provides a diagnostic method which comprises recovering nucleic acids from a sample of mammalian cells, contacting the nucleic acids under hybridising conditions with a probe which comprises at least one oligonucleotide which is substantially homologous or hybridises to a complementary strand synthesised on an intron of a mammalian gene and detecting whether hybridisation has taken place. The method is generally performed in order to diagnose neoplasia or metastasis. The mammalian cells are usually human cells. The mammalian gene is preferably a human gene, e.g. the CD44 gene as discussed above. The nucleic acids may be of extranuclear origin, e.g. messenger RNA. The nucleic acids may be amplified, e.g. by use of the reverse transcriptase polymerase chain reaction, prior to being contacted with the probe. The nucleic acids may advantageously be immobilised, e.g. in a gel or by blotting on a membrane, prior to being contacted with the probe. For ease of determination, the probe may be labelled with a signal moiety as discussed above, and this labelling may be effected before or after hybridisation with the target nucleic acids. he presence in messenger RNA of intron-based sequences carries further implications for diagnosis and therapy. Such intron-based sequences are expected to be translated into novel polypeptide chains. Further, intron-based sequences may alter the reading frame of the mRNA so that a wide variety of novel tumour-specific proteins and polypeptides may be generated by translation. These proteins and polypeptides will themselves be antigenic. It will be possible to develop antibodies, polyclonal or monoclonal, and to use those antibodies for assays or destruction of tumour cells. In all such assays, both hybridisation assays using oligonucleotide probes, and also immunoassays using antibody probes, the use of a mixture or cocktail of probes rather than a single probe is expected to yield more specific and sensitive results. Similarly for destroying tumours, the use of a mixture or cocktail of antibodies rather than a single antibody is expected to yield superior results.

Reference is directed to the accompanying drawings, in which:- Figure 1 is a map of CD44 gene products together with positions of primers used in the Examples.

Figure 2 is the sequence of the intron In-9.

Figure 3 is the sequence of Exon 9a. Figure 4 is a partial sequence at the 5 ' -end of the intron In-15.

Figure 5 is a partial sequence at the 3 ' -end of the intron In-15.

Figures 6 and 7 are ethidium bromide electrophoresis gels.

The CD44 gene is a complex one, with at least 20 exons already known and more still being discovered. About half of the 20 exons are constitutive with the remainder (Nos. 6-15) being variable and alternatively spliced. The structure and numbering are shown in Figure 1.

The intron following exon 10 for example is herein called intron In-10.

We have sequenced part or all of several introns of the CD44 gene. The intron sequence following exon 9 (In-9) is 474 bp long. Figure 2 shows this 474 bp sequence of intron In-9.

Figure 4 shows a partial sequence at the 5'- end of the intron In-15.

Figure 5 shows a partial sequence at the 3 ' - end of the intron In-15.

Figure 3 shows a new Exon 9a described below. This was first thought to be an intron In-8. But it has emerged that the sequence shown in Figure 3 is not an intron at all, but rather the 5 '-end of exon 9 which is sometimes edited out following transcription and which is herein called Exon 9a.

In a further aspect, the invention provides nucleic acids which consist of or comprise sequences substantially the same as, or substantially complementary to, all or a characteristic part of any of these sequences. Each has been used successfully to discriminate between cancer cells and normal cells.

The sample on which the assay is performed may be a small piece of tissue, a fine needle aspirate of cells from a tumour or a sample of urine, stool, sputum or other body fluid.

Example 1

Fresh urine samples were obtained from 14 patients with bladder cancer and from 14 volunteers with no known urological condition or symptoms. The cells were sedimented by centrifugation and the messenger RNA from the cell pellet extracted using a MicroFast Track Kit. Complementary DNA was synthesised from the messenger RNA template using the complementary DNA Cycle Kit (Invitrogen) and amplification was performed with appropriate primers and parameters using 2.5 units of Taq polyτnerase in 50/μl reaction mixture. The primers used in the study, at 1 pmol/μl reaction, were P3 (5 ' -TGGATCACCGACAGCACAGAC) , P4 (5 ^■ -GATGCCAAGATGATCAGCCATTCTGGAAT) .

The intron sequence used as a probe labelled with the ECL detection kit (Amersha ) was that following exon 9 (In-9) of the CD44 gene. This is 474 bp long and we have identified its base sequence to be as shown in Figure 2.

On probing amplified CD44 cDNA PCR products from the human bladder cancer cell line RT112 with this probe a distinct band corresponding to a size of 1.8 kbp was identified. There was also a smear in this track. The main band corresponds exactly in size to the sum of the sizes of all the varian exons of the CD44 genes (exon 6 to 14 inclusive) together with the flanking sequences on either side of the splice junction site defined by Primers P3 and P4 plus 473bp (the s ze of intron In-9) . The smear results from the presence of several other CD44 mRNA transcripts of different sizes containing In-9. Similar results were obtained with 11 of the 14 urine samples from patients with bladder cancer. None of the samples from normal subjects showed any homology with the probe.

When the filter was stripped and probed with a short nucleotide corresponding to the 3 ' section of In-15 some of the cancer urine tracks which were negative with the intron In-9 probe now showed a positive signal, but none of the normal tracks did so.

The 100% specificity of this method of detection for bladder cancer, in the samples examined, opens the pathway to high specificity, high sensitivity, cancer diagnosis using cocktails of intron probes for various relevant genes, in samples from different tissues and body sites. Example 2

We conducted PCR across the gaps between various exons in cDNA reverse-transcribed from mRNA extracted from the RT112 bladder cancer cell line (Figure 1) . This cell line was chosen because its histogenetic origin is appropriate for this study and because preliminary work showed that it heavily overexpresses large CD44 transcripts. It was found, as shown in Figures 2 and 3 , that two extra stretches of nucleotides were present in some of the CD44 poly A RNA transcripts from this cell line; one being a 436bp section inserted between exon 8 and exon 9, the other being a 474bp segment between exon 9 and exon 10. We performed comparative sequencing studies on the abnormal cDNA transcripts which we found in RT112 and on the corresponding stretch of a CD44 genomic DNA clone (c2311) . Amplification of both of these with identical sets of primers, revealed that the 436bp section between exon 8 and exon 9 is a new variant exon of the CD44 gene, with a consensus splice donor sequence at its 5' terminus. Since the 3' terminus of this 436bp sequence was found to be directly spliced to the former exon 9, we gave the new exon the designation exon 9a and renamed the former exon 9 as exon 9b.

The 474bp insert between exons 9 and 10 was found to correspond exactly to the whole of intron 9 (i.e. the intron following exon 9 - see Figure 2) of the CD44 genomic clone c2311, demonstrating that this intron was not edited out during intra-nuclear processing of the poly A RNA transcribed from the CD44 gene of the RT112 cell line. A specific probe for intron 9 was obtained by PCR of genomic clone c2311 DNA with primers designed to anneal to exon 9 and exon 10. This hybridised to the larger (700bp) band seen on ethidium bromide gels (see legend for Fig. 2) , confirming the retention of this region in the abnormal transcripts from this cancer cell line.

Subsequent hybridisation studies with probes for intron 9 and exon 9a on filters of RT-PCR products from fresh bladder cancer biopsies (Figure 6B,C) revealed typical large molecular weight smears that gave clear discrimination between cancer samples and normal samples. Probes for various other variant exons also produced similar smears with cancer samples but not with normal tissues (Figure 6A) as described previously (1,2) .

Further studies with the RT-PCR/Southern hybridisation technique on exfoliated cells sedi ented from urine samples from patients with bladder cancer (see Table 1 for patient details) and from tumour-free individuals, using probes for variant CD44 exons (9a, 12) or for intron 9, also discriminated clearly between urinary sediment from the patients with bladder cancer and those from the tumour-free individuals. As shown in Figure 7 and Table 1, amplified CD44 gene products (mRNA) from urine cells of 23 (77%) of 30 patients with bladder cancer and only 2 (5%) of 41 controls showed long intense smears with a probe for exon 12, confirming our previous findings (2) . in studies involving hybridisation with the intron 9 probe, only 1(2.5%) of 41 normal samples gave a positive signal . On the other hand, samples from 18(60%) of 30 patients with bladder cancer showed positive bands and smear patterns (Table 1) . It should e noted that two tumour urine samples which showed a smear pattern with intron 9 (e.g. lane 15, Figure 7) were from patients with very early stage (pTaGl) bladder cancer. Preliminary studies in this laboratory further show that introns 14 and 15 (following exons 14 and 15 respectively) of the variant region and introns between the 5 constitutively expressed exons at the 5 ' end of the gene, are also incorporated in transcripts from tumour samples in various patients and in cell lines (data not shown) . The presence and intensity of the smear obtained with any of the probes was not simply related to the numbers of cells present (Table ID .

From the new data presented here we can deduce that the abnormal (smear) pattern seen in biopsy samples of cancer tissue, in exfoliated cancer cells and in cancer cell lines, results from at least four distinct mechanisms acting in combination. These include increased expression of each already known variant exon as well as an increased number of combinations of many variant exons, one of which is the newly identified 436bp exon 9a. This results in numerous products of graded sizes which, because they are overexpressed, form thick bands which merge to form smears. Additionally, we found that the entire body of intron 9 was not spliced out of mRNA transcripts in many bladder tumour samples, as well as in exfoliated urine cells from the urine of patients with bladder cancer. This contributes to further products of different sizes crowding the gels. Fourthly, we found by probing the smears with sense and anti-sense oligonucleotides to intron 15 (data not shown) , that the retention of introns in the poly A RNA from tumour samples results in linear (asymmetric) PCR reactions because the templates become too long to be spanned during the extension phase of the PCR reaction. This, again, results in a range of products of graded sizes. Amalgamation of all of these factors leads to the formation of intense smears detectable by hybridisation with several variant exon probes.

We have previously emphasised that over- expression of any individual variant exon is not itself tumour specific, but that the smear pattern seen on Southern blots after hybridisation with any exon is highly characteristic of clinical samples containing tumour cells. Now we have presented evidence suggesting that defective editing and splicing of transcripts, resulting in retention of introns, could be a hallmark of neoplasia, at least of the bladder. As the occurrence of this phenomenon has been confirmed in other human tumour tissues and tumour cell types (see below), it provides novel, non-invasive, opportunities for early cancer diagnosis and screening.

Using the RT112 cell line, we separately extracted the poly A RNA from the nucleus and from the cytosol, with the method described previously (10) . Each fraction was subjected to the RT-PCR technique. Similar Southern hybridisation patterns were obtained from nuclear and cytosolic samples, when the respective filter membranes were hybridised with the intron 9 probe. Our data therefore confirm that the defective transcripts in tumour cells traverse the nuclear membrane and enter the cycoplasm. Further experiments involving digestion of nuclear and cytosolic extracts with RNase and DNase confirmed that the intron-9 sequence was observed in mRNA and that this result was not artefactually produced by contamination of the extracts with genomic DNA. If the transcripts are translated, it can be expected that a family of uniquely unusual proteins may be synthesized. The presence of such intronic inserts could result in the incorporation of specific new peptide sequences in the corresponding proteins, in truncated variants (if new stop codons are introduced) , or in shifts in the reading frame, downstream from the inserted elements. The consequences of such profound disturbances in protein assembly are hard to predict, but could be very useful diagnostically and therapeutically.

Interestingly, the retention of introns in transcripts was found even in pre-invasive cancer (lane 15, Figure 7 and Table 1) . This indicates that clinical application of this information can enable one to diagnose early stage bladder cancer and detect recurrences reliably, without invasive instrumentation (i.e. cystoscopy) of the patient. However, some advanced cases with haematuria did not show the intense smear pattern (Figure 7, Lane 24 and Table 1) and also did not display intron retention. We speculate that there could be two reasons for this, one being that the RNA extraction kit which we used cannot cope with more than 5 x 10⁶ cells simultaneously, because of the amount of ribonuclease released during the extraction procedure; the other being that the PCR primers annealed mainly to the mRNA from the numerous lymphocytes and other blood leukocytes in the sample, which only express the standard part of the CD44 molecule. Competition, by abundant "standard form" transcripts, for the available supplies of this primer set (competitive PCR) , could interfere with amplification of the relatively small numbers of variant exons or introns contributed by the tumour cells (11) . It may be possible to overcome these problems by perfecting better methods for mRNA extraction and by designing new primers which can anneal to other CD44 regions, including introns. However, this is not a significant clinical necessity, because persistent haematuria should, in any case, be investigated by conventional methods. The major priority is to detect asymptomatic bladder tumours and the observations described above demonstrate that, if appropriate precautions are taken to obtain high quality, intact RNA from fresh samples, RT-PCR is a sensitive and potentially powerful technique for this purpose. It enabled the simultaneous amplification of normal and abnormal RNA isoforms of the target molecule, which resulted in the identification of specific new signs of deranged gene expression, indicative of disease.

Example 3

The expression of intron 9 has also been detected in colon cancer cell lines (SW480 and HT-29) and in fresh tissue samples from colon carcinomas by RT-PCR. In the initial studies 5 microgram of total RNA from each sample was converted to cDNA and subsequently amplified with the primers PI (at the beginning of exon 3) and P4 (at the end of exon 18) . PI is GACACATATTGCTTCAATGCTTCAGC. The presence of the intron 9 sequence in amplified transcripts in HT-29 and SW 480 colon carcinoma cell lines was detected on blots of the electrophoresed PCR products using a probe complementary to intron 9. The sizes of the bands were above 1.3 kb. Although fresh tissue specimens from 24 colonic tumours and 24 corresponding normal colonic mucosas were examined in the same manner, no expression was observed after hybridisation with the intron 9 probe. Therefore, to increase the sensitivity of the assay, 200 ng of poly A selected RNA was used for the RT-PCR reactions in subsequent experiments on colon carcinomas and normal mucosas. In this work the primer on the 5' side (VPl) was located at the beginning of exon 8 and the other primer was at the end of intron 9 (VP4) . In this scheme, if intron 9 is not present in the transcripts no amplification occurs because the 3 ' primers cannot anneal to them. However, if amplification does occur, the amplicon consists of exon 8, exons 9a and 9b and intron 9. Accordingly, a probe for exon 9b was used for detection of specific amplicons after electrophoresis of PCR products. The results showed that much higher expression of intron 9 was observed in all 5 tumours than in normal mucosas. The presence of weak signals from some of the normal colon tissue samples is explained by the presence of some nuclear pre mRNA. These results corroborate the work described above on intron retention in mRNA transcripts from bladder cancer specimens. This shows that the abnormal CD44 splicing process which results in this phenomenon also occurs in cancers of other tissues.

In the colon cancer cell lines the presence of intron 9 could be detected using total cellular RNA as the template for amplification. This sufficed because such lines are pure cultures of cancer cells. However, more sensitive methods were needed for colon cancer tissue specimens, because of the presence of many normal cells diluting the abnormal signal.

We have also obtained evidence of unusual retention of intron 9 CD44 transcripts in several oesophageal, gastric and breast cancer cell lines and tissue biopsies from patients with these types of tumours.

Figure legends

Figure 2

Intron retention and sequence of CD 4 Intron 9 mRNA was purified from 100 μg total RNA of RT112 cells using Oligotex dT (Qiagen) and cDNA was synthesised using the cDNA Cycle Kit (Invitrogen) . cDNA was amplified by PCR with primer Vp2 and Vp3.

4 μl of the completed cDNA solution was used for PCR with primers

Vp2: TCAACCACACCACGGGCTTTTGAC and

Vp3 : GCTTGTAGAATGTGGGGTCTCTTC Thirty five cycles PCR were then conducted.

The cycle conditions were 94° 30 seconds, 55°C 1 minute, 72°C 2 minutes. A Hot Start procedure was adopted. A 700bp band was obtained in addition to the expected 225bp band . The latter corresponds in size to exon 9 plus exon 10. A 700bp band was also obtained by PCR with the same primer set when the genomic CD44 clone c2311 was used as a template. The genomic clone c2311 was screened from a PI genomic library using PCR (Genome Systems Inc) with the following primer set, which anneals to CD44 exon 5.

5' sense primer: AGTGAAAGGAGCAGCACTTCACGA and 5' antisense primer: AGCAGGGATTCTGTCTGTGCTGTC

Both of these 700bp bands were extracted from the gel, reamplified with the same primer set and subcloned into the TA vector. Determination of the nucleotide sequence for each 700bp band revealed that the amplified DNA in each case contained the whole 474bp intron 9.

Figure 3

New CD44 exon 9a and sequence cDNA was amplified by PCR with primers Vpl and Vp4 to study the sequence of the junction between exon 8 and intron 9 as described in (13) . A 109Obp band (a) was obtained in addition to the expected 650bp band. The latter corresponded to the combined sizes of exons 8 & 9 (b) .

Each band was extracted and re-amplified by PCR with primers Vpl and Vp5. This resulted in a 600bp band as well as the expected 177bp band confirming that they are truncated versions of (a) and (b) . The 600bp band was subcloned into the TA vector and sequenced. The new 436 base nucleotide was obtained between exon 8 and exon 9. Since this new 436 base nucleotide directly connected to the exon 9, we designated the new nucleotide sequence as exon 9a and the old one as exon 9b To study the sequence of the splicing donor part of exon 9a, the PCR was conducted with Vpl and Vp6 using the genomic clone c2311 as a template and the PCR product was sequenced. Primers used were

Vpl GACAGACACCTCAGTTTTTCTGGA Vp4 CTGCTGTGATGATGGTTAAATAC Vp5 AGTCATCCTTGTGGTTGTCTGAAG Vp6 CCAGGGACTTAGTCACTGTGT

Figure 6

Southern hybridisation with probes for exon 12 (A) , intron 9 (B) , exon 9a (C) and the standard form (D) labelled with ECL system. 5 μg of total RNA from a) the RT112 cell line (track 1) , b) normal bladder tissue (track 2) , c) bladder cancer tissue (track 3) , d) normal urinary cell sediment (track 4) and e) urine cell sediment from a patient with advanced bladder cancer (track 5) were reverse transcribed and amplified by PCR using primers Spl and Sp2, using the BRL superscript kit and thirty cycles PCR. The cycle conditions were the same as above. Primers used were:

Spl: TGGATCACCGACAGCACAGACAGA and Sp2 : GATGCCAAGATGATCAGCCATTCTGGAAT The membrane was hybridised with peroxidase labelled probes (ECL direct nucleic acid labelling and detection systems, Amersham) . Each probe was made by amplifying the TA plasmid clones containing variant exons, standard exons or introns of CD44 using related primers. Each PCR product was then extracted and directly labelled with peroxidase to produce the chemiluminescence probe, used in the ECL System. The lengths of the probes used were:

Standard section probe 483bp Intron 9 probe 474bp Exon 9a probe 436bp Exon 12 probe 72bp

One litre of normal urine from a healthy volunteer and 250ml of urine from a cancer patient were collected to get enough RNA for the study. Between each hybridisation, the blotted filter was stripped by boiling with 0.5% SDS solution.

Figure 7 Southern hybridisation of urinary samples amplified with primer Spl and Sp2. Tracks 1-14 show results with urine samples from tumour-free people. Tracks 15-28 show results with urine samples from bladder cancer patients. Naturally voided urine samples (of about 50ml) were collected and processed both for PCR and for counting of viable cells after fluorescein diacetate/ethidium bromide staining as described previously (2) . In PCR studies thirty five cycles were performed. The cycle conditions were the same as above. Panel A, B, C, D, E and F show the results of hybridization of the same filter with intron 9, exon 9a, exon 7, exon 12, exon 15 and standard section probe respectively. The filter was stripped of the previous probe between each fresh hybridization. Patient details are given in Table 1.

References

1. Y Matsumura and D Tarin. Lancet 340, 1053 (1992) .

2. Y Matsumura, D Hanbury, J C Smith, D Tarin. Brit Med J 308, 619 (1994) .

3. V J M Wielenga et al . Cancer Res. 53, 4754 (1993) . 4. K H Heider et al . Cancer Res. 53, 4197 (1993) .

5. K K Tanabe, L M Ellis, H Saya. Lancet 341, 725 (1993) .

6. G Koopman et al . J Exp Med 177, 897 (1993) 7. B Mayer et al. Lancet 342, 1019 (1993) .

8. Y Guo et al. Cancer Res. 54, 422 (1994) . 9. C Tolg, M Hofman, P Herrlich, H Ponta. Nucleic Acids Res. 21, 1225 (1993) . and G R Screaton, M V Bell, J I Bell, D G Jackson. J. Biol. Chem. 268, 12235 (1993) .

10. C W Anderson, J B Lewis, J F Atkins, R F Gesteland. Proc. Natl. Acad. Sci. USA, 71, 2756 (1974) .

J M Gudas, G B Knight, A B Pardee. Proc. Natl. Acad. Sci. USA, 85, 4705 (1988) .

11. A Innis, D H Gelfand, J J Sninsky, T J White. PCR protocols: A guide to methods and applications (Academic Press, 1990) .

TABLE 1

Details of 30 patients with bladder cancer

RT-PCR Southeπi hybridisation results

Patient No. Sex and Histological sage No of live celk per Intense smear Positive signal Lane in age (years) and grade of 10^*4ml concentrated with exon 12 with Intron 9 R| 3 tumours urine

1 F79 pTaGl <1 ♦ ♦ IS

2 F74 pTIGl 2-3 - . 16

3 F78 pTlG2 30 ♦ + 17

4 F70 pTlG2 1 ♦ . 18

5 M78 pTlG2 <1 ♦ ♦ 19

6 M77 pTlG2 <1 ♦ . 20

7 M73 pTlG3 <1 + + 21

8 M81 pTlG3 2-3 ♦ ♦ 22

9 M70 pTlG3 2 + + 23

10 M86 pT2G3 Haematuria • . 24

11 M57 pT2G3 3-4 ♦ ♦ 25

12 M77 pT2G2 10 ♦ ♦ 26

13 M73 pT3G3Lx 1 ♦ + 27

14 M82 pT2G3 5 ♦ . 28

15 M90 NS 4-5 ♦ ♦

16 M81 pT3G3Ll 1-2 ♦ .

17 F74 pTlG2 5 ♦ +

18 F84 pTlG2 >10 ♦ ♦

19 M58 pTlG2 <1 - .

20 F70 PT1G2L0 1 ♦ ♦

21 M60 pT2G3Ll <1 - .

22 M69 PT1G3L0 Haemauiria ♦ +

23 M65 pT0G2LO 10 ♦ +

24 M70 _PT1G2L0 Haematuria • .

25 M8S PT1G2L0 Haematuria .

26 M72 pTaGl <1 • .

27 M30 pT2G3 10 ♦ +

28 M64 pTaGlLO <1 ♦ .

29 M49 pTaGl 2-3 ♦ ♦

30 M83 NS Haematuria ♦ ♦

Cancer patient total 30 23 18

Normal patient total 41 2 1

Tumour stage and grade were evaluated according to classification of International Union against Cancer (IUCC): pTis = preinvasive carcinoma (carcinoma in situ); pTa = papillary non-invasive cancer, pTl = tumour not extending beyond the lamina propria; pT2 = tumour invading superficial smooth muscle; pT3 = tumour invading deep muscle; G 1 = high degree of differentiation; G2 = medium degree of differentiation; G3 = low degree of differentiation; L0 = no lymphatic invasion; LI = invasion of superficial lymphatics; L2 = invasion of deep lymphatics; NS = not pathologically staged.

Urine samples were concentrated 10- fold by dilution. The viability and quantity of cells in the sample were assessed by fluorescence microscopy with fluorescein diacetate and eihidium bromide as described previously (4). In samples with haematuria the numbers of viable ceUs are increased by leukocytes from the contaminating blood and the numbers of live urothelial cells are difficult to measure. TABLE II Results of molecular analysis of urine samples from 30 patients with bladder cancer and 41 controls related to cell number in urine. Values are numbers of subjects unless stated otherwise. Urine samples were concentrated 10-fold by centrifugation.

Results of molecular analysis

Cancer patients Controls

No of live cells per Intense smear Positive Negative Intense smear Positive Negative 10^*4ml concentrated with exon 12 with with with exon 12 with with urine intron 9 intron 9 intron 9 intron 9

>10 4 4 3 0 1 1

2-10 9 8 1 1 0 7

1-2 5 3 0 0 0 16

<1 5 3 3 1 0 14

Total number 23 18 7 2 1 38

Claims

1. A probe comprising an oligonucleotide which is substantially homologous or hybridises to a complementary strand synthesised on an intron of a mammalian gene and which is labelled with a signal ^u moiety.

2. A probe comprising a mixture of two or more oligonucleotides each of which is substantially homologous or hybridises to a complementary strand synthesised on an intron of a mammalian gene and each 5 of which is labelled with a signal moiety.

3. A probe as claimed in claim 1 or claim 2, wherein the mammalian gene is the human CD44 gene.

4. A diagnostic method which comprises recovering nucleic acids from a sample of mammalian 0 cells, contacting the nucleic acids under hybridising conditions with a probe which comprises at least one oligonucleotide which is substantially homologous or hybridises to a complementary strand synthesised on an intron of a mammalian gene and detecting whether 5 hybridisation has taken place.

5. A method as claimed in claim 4, wherein neoplasia or metastasis is diagnosed.

6. A method as claimed in claim 5, wherein bladder cancer or colon cancer or oesophageal cancer or 0 gastric cancer or breast cancer is diagnosed.

7. A method as claimed in any one of claims 4 to

6, wherein the mammalian gene is the human CD44 gene.

8. A method as claimed in any one of claims 4 to

7, wherein the nucleic acids are amplified by the 5 reverse transcriptase polymerase chain reaction prior to being contacted with the probe.

9. A method as claimed in any one of claims 4 to

8, wherein the nucleic acids are immobilised prior to being contacted with the probe.

10. A method as claimed in any one of claims 4 to

9, wherein the or each single stranded nucleic acid of the probe is labelled with a signal moiety, and after contact between the nucleic acids and the probe the signal moiety is detected in order to determine whether hybridisation has taken place.

11. Nucleic acids which consist of or comprise sequences substantially the same as, or substantially complementary to, all or a characteristic part of the sequence In-9 listed herein.

12. Nucleic acids which consist of or comprise sequences substantially the same as, or substantially complementary to, all or a characteristic part of the Exon 9a listed herein.