Patent Application
20050260604
The present invention relates to DNA probes for detecting a tandemly-repeated nucleotide sequence in the gene encoding mucin glycoprotein expressed by human mammary epithelial cells, to the use of the probe in diagnosis and in “fingerprinting” individuals, to the polypeptides expressed by the corresponding mucin gene, to antibodies against the polypeptides and to the use of the polypeptides and antibodies in the diagnosis and therapeutic treatment of cancer.
Normal and malignant human mammary epithelial cells express high molecular weight glycoproteins (gps) which are extensively glycosylated and very antigenic. As a result, many of the monoclonal antibodies (MAbs) selected for reactivity with human breast cancer and other carcinomas are found to react with molecules which are produced in abundance by the fully differentiated human mammary tissue and are found in the milk fat globule (MFG) and in milk. However, the level of expression of a particular antigenic determinant may be different in the gps produced by the normal differentiated cell and in the similar molecules produced by breast cancers. This means that some antibodies can show a certain specificity for reacting with tumour gps.
The molecules bearing the epitopes recognised by these antibodies are complex and have been difficult to analyse, both because they are large and heavily glycosylated (>250,000 daltons) and because of the complex pattern of expression. Two of the MAbs, HMFG-1 and -2, react with a component in human milk which appears to be greater than 400,000 daltons, whereas the molecules found in sera and tumours are smaller, although the dominant components are still greater than 200,000 daltons on immunoblots. The large glycoprotein produced by the differentiated mammary epithelial cells found in human milk or in the milk fat globule has been purified and shown to have some of the characteristics of the mucins. This component contains a large amount of carbohydrate joined in O-linkage to serine and threonine residues via the linkage sugar N-acetylgalactosamine. Moreover, the core protein contains high levels of serine, threonine and proline and low levels of aromatic and sulphur containing amino acids.
These mucin-like glycoproteins are also secreted by a number of other normal epithelial cells. The monoclonal antibody HMFG-1 is highly reactive with the milk mucin and evidence suggests that the epitope recognised by this antibody is more abundant on the fully processed mucin, characteristic of normal differentiation.
In tumours, the molecular weight of the molecules carrying these antigenic determinants differs among individual tumours and, in the case of the components recognised by the HMFG-2 antibody, can range from 80-400K daltons. Although it appears that the differences observed in the mobility of the high molecular weight bands are due to genetic polymorphism this probably does not explain variations in the size of the lower bands. It has been proposed that these may be the result of aberrant processing occurring in the tumour cell possibly within the glycosylation pathways.
For the majority of the monoclonal antibodies reacting with this group of molecules the exact nature of the antigenic epitopes remains unclear but circumstantial evidence has suggested that carbohydrate may at least be partly involved in many of the epitopes. Moreover, from previously available data it was not known whether the mucin found in the normal differentiated cells, and that observed in the tumours, contain the same core protein, or just carry common carbohydrate determinants.
Mucin has now been isolated from human milk by affinity chromatography enabling identification of the core protein and the gene encoding the protein. This has been found to be a highly polymorphic gene defined by the peanut urinary mucin (PUM) locus [see Swallow et al., Disease Markers, 4, 247, (1986) and Nature, 327, 82-84 (1987)]. The gene product, which is hereafter referred to as human polymorphic epithelial mucin or HPEM, has been detected in breast tumours and other carcinomas as well as in some normal epithelial tissues.
It has now been found that the HPEM core protein has epitopes which also appear in the aberrantly processed gps produced by adenocarcinoma cells. Certain of these epitopes are not exposed in the fully processed mucin glycoprotein produced by the lactating mammary gland.
In one aspect the present invention therefore provides an antibody against a human mucin core protein which antibody substantially does not react with a fully processed human mucin glycoprotein.
As used herein the term “antibody” is intended to include fragments of antibodies bearing antigen binding sites such as the F(ab′)2 fragments.
Antibodies according to the present invention react with HPEM core protein, especially as expressed by colon, lung, ovary and particularly breast carcinomas, but have reduced or no reaction with the corresponding fully processed HPEM. In a particular aspect the antibodies react with HPEM core protein but not with fully processed HPEM glycoprotein as produced by the normal lactating human mammary gland.
Antibodies according to the present invention have no significant reaction with the mucin glycoproteins produced by pregnant or lactating mammary epithelial tissues but react with the mucin proteins expressed by mammary epithelial adenocarcinoma cells. These antibodies show a much reduced reaction with benign breast tumours and are therefore useful in the diagnosis and localisation of breast cancer as well as in therapeutic methods.
The antibodies may be used for other purposes including screening cell cultures for the polypeptide expression product of the human mammary epithelial mucin gene, or fragments thereof, particularly the nascent expression product. In this case the antibodies may conveniently be polyclonal or monoclonal antibodies.
Antibodies according to the present invention may be produced by innoculation of suitable animals with HPEM core protein or a fragment thereof such as the peptides described below. Monoclonal antibodies are produced by the method of Kohler & Milstein (Nature 256, 495-497/1975) by immortalising spleen cells from an animal innoculated with the mucin core protein or a fragment thereof, usually by fusion with an immortal cell line (preferably a myeloma cell line), of the same or a different species as the innoculated animal, followed by the appropriate cloning and screening steps.
In a particular aspect the present invention provides the monoclonal antibodies designated SM3 against the HPEM core protein. In another aspect the invention provides the hybridoma cell line which secretes the antibodies SM3 and has been designated HSM3. Samples of HSM3 have been deposited with ECACC on 7th January 1987 under accession number 87010701.
Using antibodies according to the invention it has been possible to screen a phage library constructed from mRNA isolated from a human breast cancer cell line to identify sequences coding for portions of the mucin core protein. Complementary DNA sequences have been constructed and from these it has surprisingly been found that the gene encoding the core protein contains multiple tandem repeats of a 60 base sequence leading to considerable polymorphism sufficiently extensive that cDNA fragments corresponding to the repeat sequence would be useful for fingerprinting DNA. The fingerprinting thus made possible has applications in for instance ascertaining whether bone marrow growth after transplants is from the host or the donor and in forensic medicine for identifying individuals using body tissues or fluids.
Accordingly the present invention also provides a nucleic acid fragment comprising at least 17 nucleotide bases the fragment being hybridisable with at least one of
a) the DNA sequence
b) DNA complementary to the DNA of a) , i.e. of sequence
The sequences in (a) and (b) each include a double tandem repeat sequence of 120 bases. Fragments according to the invention may correspond to any portion of this sequence including portions bridging the start point of the repeat.
Fragments according to the invention will hybridise under conditions of low stringency with the DNA and RNA sequences (a) to (d) above. Preferred fragments are those which also hybridise under conditions of high stringency. The most preferred fragments of the invention are those which have sequences exactly identical to, or exactly complementary to the sequences (a) to (d) above.
Normally a given DNA or RNA fragment according to the invention will be capable of hybridising with both DNA according to a) and RNA according to c) or with both DNA according to b) and RNA according to d) above.
Preferably the nucleic acid fragment according to the present invention will comprise a portion of at least 30 nucleotide bases capable of hybridising with at least one of
When the existence of a tandem repeat sequence was first identified it was believed that the sequence consisted of 59 base pairs corresponding with the sequences indicated in (a) and (b) above except for the lack of the base indicated with “*”.
Many fragments according to the invention as originally defined in British Patent Application No. 8700269 also conform with the new definition of fragments as set-out herein and those fragments of sequences defined under (a), (b), (c) or (d) above which do not include the bases marked “*” form a particular aspect of the present invention. Such fragments have sequences corresponding to at least a portion of the sequences
In the human genome the DNA tandem repeat sequence comprises antiparallel double stranded DNA, one strand having sequence (a) and being paired with a strand having sequence (b).
As mentioned above the nucleic acid fragments of the invention may be used as a probe for detecting one or other strand of the DNA tandem repeat sequence in the human genome, or RNA transcribed from either strand and hence for identifying the gene or genes for human mucin core proteins, mRNA transcribed therefrom and complementary DNA and RNA. For such purposes it may be convenient to use the complete normal gene comprising at least one tandem repeat sequence, or mRNA transcribed therefrom or to attach non-complementary fragments to either or both the 5′ and 3′ ends of a fragment according to the invention and/or to attach detectable labels (such as radioisotopes, fluorescent or enzyme labels) to the probe or to bind the probe to a solid support. All of these may be achieved by conventional methods and the nucleic acid fragments of the invention may be produced de novo by conventional nucleic acid synthesis techniques.
The nucleic acid fragments of the present invention may also be used in active immunisation techniques. In such methods the fragment codes for a polypeptide chain substantially identical to a portion of the mucin core protein and may be extended at either or both the 5′ and 3′ ends with further coding or non-coding nucleic acid sequences including regulatory and promoter sequences, marker sequences and splicing or ligating sites. Coding sequences may code for corresponding portions of the mucin core protein chain or for other polypeptide chains. The fragment according to the invention, together with any necessary or desirable flanking sequences is inserted, in an appropriate open reading frame register, into a suitable vector such as a plasmid or a viral genome (for instance vaccinia virus genome) and is then expressed as a polypeptide product by conventional techniques. In one aspect the polypeptide product may be produced by culturing appropriate cells transformed with a vector, harvested and used as an immunogen to induce active immunity against the mucin core protein. In another aspect the vector, particularly in the form of a virus, may be directly innoculated into a human or animal to be immunised. The vector then directs expression of the polypeptide in vivo and this in turn serves as an immunogen to induce active immunity against the mucin core protein.
The invention therefore provides nucleic acid fragments as hereinbefore defined for use in methods of treatment of the human or animal body by surgery or therapy and in diagnostic methods practised on the human or animal body. The invention also provides such methods for treatment of the human or animal body by surgery or therapy and diagnostic methods practised in vivo as well as ex vivo and in vitro.
The invention further provides a polypeptide comprising a series of residues encoded by the DNA tandem repeat sequence, the sequence shown at (b) above being the coding sequence. Polypeptides according to the invention are selected from any of those having 5 or more amino acid residues represented by the hollowing amino acid sequence
Polypeptides according to the invention may have a sequence corresponding with any portion of the 40 residue sequence above and may include the start point of the repeat sequence.
Other polypeptides according to the invention include three or more repeats of the 20 amino acid repeat sequence. Such polypeptides may include minor variations by way of substitution of individual amino acid residues.
The invention further provides polypeptides as defined above modified by addition of N-acetyl galactosamine (a linkage sugar) on serine and/or threonine residues and by addition of oligosaccharide moieties to that or via other linkage sugars and/or fragments linked to carrier proteins such as keyhole limpet haemocyanin, albumen or thyroglobulin.
Preferably the polypeptide comprises at least 10 amino acid residues of the sequence above, more preferably 20 residues. The polypeptide may comprise the full sequence above. Such polypeptides may further comprise additional amino acid residues, preferably conforming to the amino acid sequence of HPEM core protein.
In a further aspect the present invention provides the HPEM core protein. This is encoded by the PUM gene and may be produced by recombinant DNA techniques and expressed without glycosylation in human or non-human cells. Alternatively it may be obtained by stripping carbohydrate from native human mucin glycoprotein which itself may be produced by isolation from samples of human tissue or body fluids or by expression and full processing in a human cell line. The HPEM core protein may be used for raising antibodies in animals for use in passive immunisation, diagnostic tests and tumour localisation and in active immunisation of humans.
The invention further provides antibodies (monoclonal or polyclonal), and fragments thereof, against any of the polypeptides described above. Such antibodies may be obtained by conventional methods and are useful in diagnostic and therapeutic applications.
The invention further provides antibodies (monoclonal or polyclonal), or fragments thereof, linked to therapeutically or diagnostically effective ligands. For therapeutic use of the antibodies the ligands are lethal agents to be delivered to cancerous breast or other tissue in order to incapacitate or kill transformed cells. Lethal agents include toxins, radioisotopes and ‘direct killing agents’ such as components of complement as well as cytotoxic or other drugs. Further therapeutic uses of the antibodies inclusive passive immunisation.
The invention further provides therapeutic methods comprising the administration of effective non-toxic amounts of such antibodies or fragments thereof and antibodies or fragments thereof for use in therapeutic treatment of the human or animal body. Especially in therapeutic applications it may be appropriate to modify the antibody by coupling the Fab region thereof to the Fc region of antibodies derived from the species to be treated (e.g. such that the Fab region of mouse monoclonal antibodies may be administered with a human Fc region to avoid immune response by a human patient) or in order to vary the isotype of the antibody.
In the diagnostic field the antibodies may be linked to ligands such as solid supports and detectable labels such as enzyme labels, chromophores and fluorophores as well as radioisotopes and other directly or indirectly detectable labels. Preferably monoclonal antibodies or fragments thereof are used in diagnosis.
The invention further provides a diagnostic assay method comprising contacting a sample suspected to contain abnormal human mucin glycoproteins with an antibody as defined above. Such methods include tumour localisation involving administration to the patient of the antibody or fragment thereof bearing a detectable label or of an antibody or fragment thereof and, separately simultaneously or sequentially in either order a labelling entity capable of selectively binding the antibody or fragment thereof. The invention also provides antibodies or fragments thereof for use in diagnostic methods practised on the human or animal body.
Particular uses of the antibodies include diagnostic assays for detecting and/or assessing the severity of breast, ovary and lung cancers.
Diagnostic test kits are provided for use in diagnostic assays and comprise antibody or a fragment thereof, optionally suitable labels and other reagents and, especially for use in competitive assays, standard sera.
The invention will now be illustrated by the following Examples and with reference to the figures of the accompanying drawings in which
Purification of the Milk Mucin
The milk mucin was purified from human skimmed milk by passage through an HMFG-1 affinity column followed by size exclusion chromatography. The HMFG-1 monoclonal antibody was purified from tissue culture supernatant using a protein A column (1). The purified antibody was coupled to cyanogen bromide activated sepharose (Pharmacia) as described in the manafacturer's instructions. Human skimmed milk was passed in batches of 100 ml. through the antibody column followed by extensive washing with PBS. Bound antigen was eluted from the column using 0.1 M glycine pH 2.5 and the fractions registering an optical density at 280 mn were pooled, dialysed against 0.25 M acetic acid and lyophilized. Batches of about 20 mgs were dissolved in 0.25 M acetic acid and passed through a G75 Sephadex column (1×100 cm) which had been previously equilibrated with acetic acid. The column was washed with 0.25 M acetic acid 1 ml fractions collected. The peak fractions which were eluted in the void volume were pooled, lyophilized A the dry powder stored at 4° C. Amino acid analysis was performed using a Beckman 6300 amino acid analyser.
Deglycosylation of the Milk Mucin
To remove the O-linked carbohydrate from the milk mucine the molecule was treated with anydrous hydrogen fluoride as described by Mort and Lamport (21), for either 1 hour at 4° C. which produced a partially stripped preparation, or 3 hours at room temperature which produced the extensively stripped mucin.
Iodination of the Milk Mucin
Iodinations of the purified mucin, the partially or extensively stripped mucin were carried out using the Bolton and Hunter method (51). Briefly, the mucin, 2.5 μg in 20 μl 0.1M borate buffer pH PH 8.5, was added to the dried Bolton and Hunter reagent (1 mCi, Amershan International plc) and incubated at room temperature for 15 minutes. The reaction was stopped by the addition of 0.5 ml of 0.2M glycine in borate buffer and after a further minutes incubation, free Bolton and Hunter reagent was removed by passage through a G25 Sephadex column (PD10 columns Pharmacia) previously equilibriated in PBS.
Iodination of Lectins
Wheat gear agglutinin (WGA), peanut agglutinin (PNA) (Vector Labs) and Helix pomatia agglutinin (HPA) (Boehringer) were iodinated as described by Karlsson et al. (52) using the chloramine T method.
Polyacrylamide Gels and Western Blots
Polyacrylamide gel electrophoresis and immunoblotting was performed as described previously (1). Briefly, samples were run on 5-15% polyacrylmaide gels and then electrophorutically transferred to nitrocellulose paper (Schleicher and Schuell) at 50 volts overnight at 4° C. (36). In the immunoblotting experiments the paper was reacted with monoclonal antibodies and binding detected with an ELISA method sing 4-chloro-1-naphthol as the substrate. For lectin binding studies the Western blots were reacted with the iodinated lectins as described by Swallow et al. (48).
Production of Monoclonal Antibodies
A female BALB/c mouse was immunized with 5 μg of the partially stripped milk mucin in Freund's complete adjuvant and 5 months later boosted with a further 5 μg of the same preparation in Freund's incomplete adjuvant. After a or 20 days, 5 μg of the mucin extensively stripped of its carbohydrate was given intravenously in saline solution. The spleen was removed 4 days later, and fused with the NS2 mouse myeloma cell line (53).
Screening of Hybridoma Supernatant end Immunopreciptitations
Screening aesay was a modification of that described by Melero add Conzalez-Rodriguez (54). Multiwell plates were coated with 50 μl of 0.1 =g/ml protein A (Pharmacia Pine Chemicals) in PBS and allowed to dry overnight at 37° C. The plates were blocked with 5% BSA for 1 hour at 37° C. followed by the addition of 50μl of rabbit anti-mouse immunoglobulin (DAKO, diluted 1:10 in PBS/BSA=PBS/BSA). After incubating for 2 hours at 37° C. the plates were washed twice with PBS containing 1% BSA and 50 μL of hybridoma supernatant added. The plates were incubated overnight at 4° C., washed twice with PBS/BSA and 50 μL of iodinated partially stripped mucin containing 100,000 cpm added to each well. The plates were then incubated at room temperature for 4 hours, washed 4 times with PBS/BSA and the individual wells counted in a gamma counter. ?or immunoprecipitation experiments 50 μl of SDS sample buffer containing dithiothreitol was added to each of the wells which were then boiled for 3 minutes and the buffer loaded onto 5-15% polyacrylamide gradient gals.
Staining of Tissue Sections
Tissues from primary mammary carcinomas, benign breast biopsies, normal breast, and pregnant lactating breast tissue were fixed in methacarn (methanol chloroform and acetic acid 60:30:10) and embedded into paraffin wax. Sections ware stained with the antibodies using the indirect peroxidase anti peroxidase method as previously described (47).
Purification of the Milk Mucin
The milk mucin was purified from human skimmed milk on an HMFG-1 antibody affinity column. Iodination of the eluted material revealed the presence of a large molecular weight component and a 68KD band. Precipitation of the affinity purified material with antibodies HMFG-1 and HMFG-2 (tracks 2 and 5) followed by gel electrophoresis showed that both the high molecular weight components and the 68KD component were precipitated by both antibodies (less effectively by HMFG-2). Since the 68KD component was also precipitated by two unrelated antibodies (
A high molecular weight glycoprotein (PAS-O) containing more than 50% carbohydrate in O-linkage has been purified from the human milk fat globule by Shimiru and Iamauchi (8). To see whether this component was similar to the main isolated from milk by affinity chromatography on an HMFG-1 affinity column, the amino acid composition of the purified HMFG-1 reactive mucin was determined and compared to the amino acid composition of the purified PAS-O component. Table 1 shows that there is good correspondence between the two sets of data, indicating that the core proteins of PAS-O and the mucin purified bare are the same.
Isolation of the Core Protein of the Milk Mucin
As there are no enzymes easily available that are efficient at removing O-linked sugars, and β elimination often results in damage to the protein core, the oligosaccharides were removed by treatment of the mucin with anhydrous hydrogen fluoride. This treatment has been shown by Mort and Lamport (21) to be effective in removing sugars from pig submaxillary mucin vithout daaging the protein core. Amino acid analysis of the material produced after HF treatment of the milk mucin suggested that the protein core was also in this case undamaged, since the composition was the same as that seen in the intact mucin (Table 1) .
Initially the milk mucin was exposed to HP for only 1 hour at 4°, but analysis of the product showed that there was only partial removal of the sugars with such treatment, and it was necessary to treat the mucin at room temperature for 8 hours to obtain a molecule which showed no lectin binding ability.
To test for the presence of sugars on the intact mucin and on the products produced after the two different HF treatments each preparation was subjected to acrylamide gel electropheresis, transferred to nitrocellulose paper and reacted with 125I-labelled lectins. The lectins used were peanut lectin (PNA) which reacts with galactose linked to N-acetyl galactosamine, wheat germ (WGA) reactive with N-acetyl glucosamine and Helix pomatia agglutinin (HPA) which reacts with the linkage sugar in O-linked glycosylstion, N-acetylgalactosamine.
Generation of Monoclonal Antibodies to the Milk Mucin Core Protein
A fusion was carried out using the spleen of a mouse that had been immunized with two injections of the partially stripped milk mucin followed by a boost with the extensively stripped mucin. The clones were initially screened against the 125I partially stripped material using protein A plates (see Methods). Four hybridomas were selected and cloned, and table 2 shows their spectrum of reactivity with the intact, partially and extensively stripped mucin. As can be seen from this table three of the hybridomas which were isolated showed a strong reaction with the partially and extensively stripped mucin and did not react with the intact mucin. These appeared to be good candidates for monoclonal antibodies to the protein core and two, SM-3 and SM-4, were selected to be characterised further.
It can also be seen from table 2 that the HMFG-1 and HMPG-2 antibodies reacted very strongly with the mucin stripped of its arbobydrate. Theme two antibodies were, in fact, developed using the intact mucin (from the milk fat globule) as immunogen and, in the case of HMFG-2, growing mammary epithelial cells (14). Their reaction with the stripped mucin was unexpected, as circumstantial evidence had previously led to the belief that carbohydrate might form at least part of their antigenic epitopes.
Molecular Weight of Molecules Carrying Antigenic Determinants
The antibody SM-3 was shown to be of the IgGI1 subclass, while the SM-4 antibody was found to be IgM. We therefore chose to use the SM-3 antibody in subsequent experiments since antibodies of the IgM class can present problems in some appliction. Immunoprecipitation of the extensively stripped material with SM-3 showed a reaction with the lectin-unreactive 68K component (FIG. 5A, track 3). The monoclonal antibody HMPG-2 can also be seen to immune precipitate the lectin-unreactive 68K component (track 2). The antibodies were reactive with antigen on immunoblots and
We have previously shown that the molecular weight of the components in breast cancer cells carrying determinants found on the milk mucin is lower than 400K and can vary from one tumour to another (1). Reaction of antibody SM-3 with Western blots of gal separated extracts of breast tumour cells shows that this antibody reacts with components of similar molecular weight to those reactive with antibody HMFG-2 (data not shown). Because the antibody SM-3 differs from the antibodies HMFG-1 and 2 in that it does not react with the intact mucin processed by the lactating glad and yet reacts with molecules processed by breast cancer cells, it was appropriate to examine the reaction of SM-3 with a range of breast cancers.
Reactivity of SM-3 with Breast Tissues and Tumours
The antibody SM-3 reacted with paraffin embedded tissues provided these were fixed in methacarn (not formal saline). Using this method for preparation of tissue sections, the reaction of the antibody was compared to that of HMFG-2 on breast tissues and tumours with an indirect immunoperoxidase staining method. This analysis shoved a dramatic difference in the staining pattern of SM-3 compared to that seen with HMFG-2. Thus, although a strong positive reaction was seen in 20/22 breast cancers stained with SM-3 (as compared to 22/22 stained with HMFG-2), normal resting breast, pregnant or lactating tissues and most benign lesions were largely unstained with SM-3 but were stained with HMFG-2. Some examples of staining patterns of breast tissues and tumours are illustrated in
Twenty-two primary carcinomas and fourteen benign lesions were examined and the reaction of SU-3 compared to the staining with WG-2 in each case. In the primary carcinomas, staining with SM-3 was heterogeneous but generally quite strong and always confined to tumour cells; connective tissue and stroma showed no reaction (see
In contrast to HMPG-1 and HMPG-2 which strongly stain lactating and pregnant breast, SM-5 was totally negative with three out of six cases of pregnant or lactating breast (see.
SM-3 was also shown to be negative on sections of normal liver, lung, thymus, sweat gland, epididymus, prostate, bladder, small intestine, large intestine, appendix, thyroid and skin. The antibody showed weak positive staining only with the distal tubules of the kidney, the occasional chief cell of the stomach, the occasional duct cell of the salivary gland and the sebaceous gland.
Large molecular weight mucin molecules are expressed by many carcinomas and carry many of the tumour associated antigenic determinants recognized by monoclonal antibodies. These epitopes may also be expressed by some normal epithelium, and some monoclonal antibodies like HMFG-1 react particularly well with a mucin found in normal human milk (1,17). As long as the study of the mucins is restricted to their detection with antibodies reactive with undefined epitopes, the knowledge of their structure, expression and proccessing will also be restricted. We have begun to investigate the structure and expression of the mammary mucin by isolating the core protein and developing antibodies which have allowed us to select partial cDNA clones for the gene coding for the core protein. This Example describes the production and characterization of these antibodies.
Treatment of the HMFG-3G-1 affinity purified milk mucin with hydrogen fluoride resulted in the appearance of a dominant band of about 68E daltons and a minor species of about 72KD on SDS acrylamide gels. These bands showed no reactivity with lectins, including Helix pomatia agglutinin which is specific for N-acetyl galactosamine, the first sugar in 0-linked glycosylation (55). It therefore seems probable that is 68K dalton polypeptide represents the core protein of the mucin. Supportive evidence for this comes from the observation that the antibodies described here, which are reactive with the stripped 68K component, can precipitate a molecule of this size from the in vitro translation products of mRNA isolated from breast cancer cells expressing the mucin
As the milk mucin contains at least 50% carbohydrate (16) ,a protein core of only 68ED appears too small if the intact molecule has an observed molecular weight greater than 400KD. However, mucins can b composed of small subunits which aggregate and are held together by some form of non-covalent interactions, as yet not understood. For example, although the molecular weight of the opine submaLill=7 mucin has been reported to be greater than 1×106 daltons (45), it has a protein core of only 650 amino acids with a molecular weight of 58,300 daltons (46).
An unexpected finding was that the antibodies HMFG-1 and HMFG-2 which react with the milk mucin, also show a positive reaction with the extensively stripped material which showed no lectin binding capability. Previous indirect evidence, including the resistance to fixation, boiling and reduction, the repetitive nature of their epitopes and the appearance of several bands on immunoblots, had led to the belief that carbohydrate present on the milk mucin was involved in these epitopes This idea was reinforced by the observation that lectins could block the binding of HMFG-1 and 2 (1). While it is not possible to exclude the possibility that some sugars, undetected by the lectin binding experiments, remain on the extensively stripped mucin described here, this is unlikely to be the explanation for the reactivity of the antibodies HMFG-1 and 2. This can be said since both antibodies have recently been shown to react positively with β-galactosidase fusion proteins expressed by phage carrying DNA coding for the core protein of the mammary mucin. It appears therefore that at least of each of the epitopes recognised by HMFG-1 and HMFG-2 contain amino acids but it rust be assumed that some of these epitopes on the core protein are exposed, i.e. not masked in the fully glycosylated molecule. The HMFG-2 epitope is however less abundant on the milk mucin than the HMFG-1 epitope, while it is readily detectable on the mucin molecules expressed by tumours (1); These molecules have a smaller molecular weight and may be less heavily glycosylated or polymerized.
Here we have reported the development of new antibodies which are reactive with the protein core of the mucin and with the partially deglycosylated molecule, but which are unreactive with the fully processed mucin produced by the lactating mammary gland. One of these antibodies SM-3, which is an IgG1, has been studied in more detail. It has been shown to react with the mucin molecules which are produced by breast cancer cells and are recognised by many antibodies developed against the intact milk mucin. It should be emphasized however that the epitope recognised by SM-3 which is on the core protein and is exposed in the mucin as processed by tumour cells, is not exposed on the normally processed milk mucin. This feature offers the possibility of enhanced tumour specificity, and a pilot immunohistochemical study of breast tumours and tissues has shown that indeed the SM-3 antibody reacts strongly with the majority of primary breast cancers (91%) but shows little or no reaction with benign breast tumours, resting or lactating breast, and most normal tissues.
There are several implications of the work described here which may be important for both basic and clinical studies in breast cancer. The observation that parts of the core protein (detectable by antibodies) are exposed on the mucins as processed by breast cancer but masked on the mucin as processed by cells in normal breast and benign tumours implies that there is an alteration in the processing of the mucin in malignancy. A more detailed study of the processing of the mucin in normal and malignant cells may then give basic information for defining the malignant cell. Moreover, since the specificity of the reaction of the antibody SM-3 for tumours is better than that of antibodies developed against the intact mucin, this antibody may prove to be a more effective diagnostic tool for the detection of breast cancer cells in tissue sections, tissue fluids and cells. The reactive components are membrane associated as wall as intracellular and in vivo localisation of tumours may also be possible.
The abbreviations used are: HMPG, human milk fat globule; PBS, phosphate-buffered saline (153 mM NaCl, 3 mM XCL, 10 mM Na2HPO4, 2 mM KM2PO4 pH 7.4);WGA, wheat gem agglutinin; PNA, peanut agglutinin; HPA, Helix pomatia agglutinin; BSA, bovine serum, albumin; SDS, sodium dodecyl sulfate.
Purification and deglycosylation of human milk mucin was conducted as in Example 1 mucin was purified on an HMFG-1 antibody.
The stripped mucin preparations were separated by electrophoresis through NaDodSO4/polyacrylamide gels (10%) and silver stained by two methods, one of which can be used to stain highly glycosylated proteins (22,23).
Preparation of Polyclonal Rabbit Antiserum to Stripped Core Protein
One Now Zealand White rabbit was immunized with 100 μg of the partially stripped core protein in complete Freund's adjuvant (Gibco). Booster injections of 500 μg of the totally stripped core protein were administered in incomplete Freund's adjuvant (Gibco) 3 and 4 weeks after the initial injection and the rabbit was bled one week later. Ten microliters of immune serum (75 μg/ml protein) precipitated 200 ng of fully stripped core protein in a Protein A assay (24) and detected it on immunoblots. The immunoglobulin fractions of rabbit preimmune and rabbit anti-mucin core protein were prepared by adding ammonium sulfate to 50% saturation. The resulting pellet was resuspended in one-half the original serum volume of PBS and dialyzed against the same buffer. After dialysis, only residual precipitate was removed by centrifugation. Immunoglobulin fractions were stored in aliquots at −20° C.
Description of MAbs Used
In addition to the polyclonal antiserum used for initial screening, a cocktail of two MAbs, SM-3 and SM-4 (see Example 1) which recognise the mucin core protein (20) and HMPG-1 and HMPG-2(1,14) were used to screen the purified plaques, the β-galactosidase fusion proteins and for immunoprecipitations from in vitro translated proteinsa. Other MAbs used were a monoclonal anti-β-galalctosidase antibody (25) which was a gift from H. Durbin (ICRF, London) an anti-interferon antibody, ST254 (24), LE61, a keratin antibody (26) and M18 which recognizes a carbohydrate structure on the milk mucin (27).
aThe MAbs SM-3 and SM-4 (SM refers to stripped mucin) show strong reactivity with the partially and fully stripped core protein but no reactivity with the fully glycosylated mucin (20).
In Vitro Translation of Proteins
RNA was isolated from the human breast cancer cell line MCF-7 using the guanidium isothiocyanate method of Chirgwin et al. (28) and poly(A)30 RNA was purified by chromatography using oligo (dT)-cellulose (New England Bio Labs). The poly(A)+ RNA was translated in a reticulocyte lysate system (Amersham) in the presence of [35S] methionine (1000 Ci/mole; 1 Ci=37 GBq, Amersham). Samples containing 5×104 acid insoluble cpm were precipitated in a protein A assay (24) using MAbs SM-3, SM-4, HMFG-1, HMFG-2 and a control antibody to human interferon, The antibody-selected proteins were then separated on a 10% NaDodSO4/polyacrylamide gel, impregnated with Amplify (Amersham) and exposed to XAR-5 film (Kodak) at −76° C.
Antibody Screening of γgt11 Library and Protein Blotting
The γgt11 library used in this study was constructed from mRNA isolated from the human breast cancer cell line MCF-7 and was generously provided by Philippe Walter and Pierre Chambon (Strasbourg, France). The poly (A)+ RNA used for the preparation of the randomly primed library was prepared from mRNA that sedimented faster than 28S rRNA and was enriched in estrogen receptor (29). The library was made essentially as described by Huynh et al. and Young and Davis (30-32) and contained approximately 1×106 recombinants per μg of RNA. Between 85% and 95% of the plaques contained inserts.
The phage library was plated onto bacterial strain Y11090 and grown for 3 hr at 42° C. After isopropyl β-D-thiogalactoside (IPTG) induction and 3 hr of growth at 37° C., filters were prepared from each plate and screened with anti-mucin core protein antibody by the method of Young and Davis (32). The first antibody used in screening was the rabbit antiserum raised against the stripped core protein prepared as described above. Prior to use in screening, the antiserum was diluted 1:200 in PBS containing 1% bovine serum albumin (PBS/BSA). Preabsorption with Y1090 bacterial lysate was not found to be necessary. The nitrocellulose filters (Schleicher and Schuell) were blocked by incubation in PBS containing 5% BSA for 1 hr at room temperature with gentle agitation. The filters were incubated at room temperature overnight with a 1:200 dilution of antiserum in heat sealed plastic bags. The filters were washed 5×5 min in PBS/BSA, and bound antibody was detected by using horseradish peroxidase-conjugated sheep anti-rabbit antiserum (Dako) diluted 1:500 with PBS/BSA and incubated for 2 hr with the filters. The filters were washed 5×5 min in PBS/BSA and 1×10 min in FBS before color detection using 4-chloro-1-naphthol (1). Immunoreactive bacteriophage were picked and purified through two additional rounds of screening. Subsequently, bacteriophage inserts were subcloned into the EcoRI sites of pUCB (33) producing the plasmid used most extensively, pMUC 10. The plasmids were maintained in DH1 cells.
To examine the β-galactosidase-cDNA fusion proteins for immunoreactivity, cell lysates were derived. Lysogens were prepared as described in Young and Davis (34). Cells were pelleted, suspended in Laemmli sample buffer (35) and separated by electrophoresis through NaDodSO4/polyacrylamide gels (10%) and transferred onto nitrocellulose filters as described (1,36). The filters were treated as above for antibody screening.
Northern Analysis
RNA was isolated from tissue culture calls and frozen tissues by the guanidinium isothiocyanate method of Chirgwin et al. (28). Total RNA (10 μg per lane) was denatured by heating at 55° C. for 1 hr in deionized glyoral and fractionated by electrophoresis through a 1.3% glyoral gel (38). The RNA was transferred to nitrocellulose (Schleicher and Schuell), prehybridized and hybridized as described by Thomas (34). Filters were wished down to 0.1×SSC with 0.1% EDS at 65° C .and exposed to XAR-5 film (Kodak) at −70° C. with intensifying screens,
Southern Analysis
High molecular weight genomic DNA was prepared from white blood cells and cell lines (39,40). These genomic DNAs (10μg) were cleaved with restriction enzymes following the manufacturer's a recommended conditions and fractioned through 0.6% and 0.7% agarose gels. Cloned plasmid DNA was cleaved and fractionated on 1.3% agarose. The gels were denatured, neutralized and transferred to nylon membranes (Biodyne) according to the manufacturer's instructions. The EcoR1 insert from pMUC10 was separated on a 1% low melting point agarose (Biodyne) gel and labelled with [α-32P]dCTP by the method of random priming (41) and hybridized to filters at 42° C. Pilters were washed down to 0.1×SSC with 0.1% SDS at 55° C. and exposed to XAR-5 film (Kodak) at −70° C. with intensifying screens.
Purification and Deglycosylation of Mucin Glycoprotein
Mucin glycoprotein reactive with the monoclonal antibody HMFG-1 was prepared from pooled human breast milk by using an HLPG-1 antibody affinity column, followed by molecular sieve chromatography on Sephadex G-75 in order to remove lower molecular weight components (
The purified material was subjected to treatment with hydrogen fluoride to remove the O-linked sugars that are characteristic of mucin glycoproteins. Two different reaction conditions were used which resulted in a partially deglycosylated core protein (treated at 0° C. for 1 hr) and a fully deglycosylated core protein (treated at room temperature for 3 hr) as determined by iodinated lectin binding following separation by gel electrophoresis and transfer to nitrocellulose paper (20). The partially deglycosylated core protein was reactive with wheat germ agglutinin, peanut agglutinin and helix pommatia lectin (which recognizes the linkage sugar N-acetylgalactosamine) whereas the fully stripped protein showed no reactivity with any of these three lectins.
The hydrogen fluoride treated core protein was separated by electrophoresis through NaDodSO4/polyacrylamide gels (10%) and silver stained. Silver staining revealed that the predominent component of the partially stripped mucin was a high molecular weight band of about 400 kd although faint bands of lower molecular weight could also be observed (
Antibody Reactive Proteins Produced by MCF-7 Cells
The MCF-7 breast cancer cell line expresses large amounts of HMFG-1 and -2 reactive material on its cell surface (14) and was thus judged to be a suitable source of mRNA for a cDNA library. Before proceeding to screen the MCF-7 library with the monoclonal antibodies, they were tested for their ability to precipitate a component from in vitro translation products produced from MCF-7 mRNA. Poly (A)+ RNA from MCF-7 was prepared and translated in vitro. Proteins from the translation reaction were immunoprecipitated using the monoclonal antibodies HMFG-1, HMFG-2, SM-3 and SM-4 and displayed by polyacrylamide gel electrophoresis and fluorography (
The abundance of the core protein mRNA in total cellular poly (A)+RNA was 4% as estimated by comparing the amount of (35S)methionine present as immunoprecipitated protein to the amount of methionine incorporated into total protein during in vitro translation.
Screening the cDNA Library
The γgt11 cDNA library made from size selected MCF-7 mRNA (see Methods) was screened initially with the polyclonal antiserum made to the mucin core protein which had been stripped of its carbohydrate. Screening of 2×106 plaques resulted in 11 positive clones, 7 of which were taken successfully through two further rounds of plaque purification.
To demonstrate that the reactivity of the phage clones with the antibody probes was due to antigenic determinants on the cDNA translation product, β-galactosidase fusion proteins were made from all 7 clones. The proteins were separated by electrophoresis, transferred to nitrocellulose paper and probed with a variety of antibodies to the stripped mucin, including the polyclonal antiserum which was used initially to select the clones and a cocktail of SM-3 and SM-4. In addition. HMFG-1 and HMFG-2, the two monoclonal antibodies which originally detected this differentiation and tumour-associated epithelial mucin (1,14) were tested. All 7 clones yielded fusion proteins which were specifically recognized by the polyclonal antiserum, the monoclonal cocktail, and HMFG-2. HMFG-1 antibody reacted with 6 of the 7 fusion proteins and failed to recognize the protein from clone 9 which contains the smallest insert. In every case the strongest signal was given by the HMFG-2 antibody and this reaction is shown in
Characterization of cDNAs and RNA Blot Analysis
The inserts from the λ clones were designated pMUC3-10 (omitting pMUC5) and were subcloned into the vector pUC 8 for easier manipulation. The 7 clones were compared to each other for sequence homology. Each of the plasmids was digested with EcoRI and the insert separated on a 1.4% agarose gel. The largest cDNA insert from pMUC10 was used to probe the inserts and found to hybridize to all 6 inserts (
As shown by agarose gel electrophoresis (
Because the λMUC clones were identified only by antibody binding, we needed additional assurance that they were indeed coding for the breast epithelial mucin. To determine the authenticity of pMUC10, we correlated the presence of mRNA hybridizing to the clone with mucin expression in various cell lines. As shown in
Genomic DNA Blot Hybridization and Detection of a Restriction Fragment Length Polymorphism (RFLP)
Genomic DNA was prepared from a panel of ten individuals consisting of six unrelated individuals and a family of four, and from three cell lines. The DNAs which were digested with HinfI or EcoRI and blotted and hybridized to the radiolabelled pMUC10 insert, exhibit restriction fragment length polymorphisms. The restriction fragments from the ten individuals and three cell lines are shown in
The cDNA clones described here which were obtained from the MCF-7 λgt11 library were selected using polyclonal and monoclonal antibodies prepared against a normal cellular product, the milk mucin in its deglycosylated form. This was done because it was easier to obtain large quantities of the mucin for stripping than to prepare similar quantities of immunologically related glycoproteins expressed by breast cancer calls (44). The fact that the antibodies did select for cDNA coding for nonglycosylated core protein molecules in MCF-7 cells, strongly suggests that the glycoproteins in these cells, which were originally detected by their reaction with antibodies to the milk mucin, contain the same core protein as this mucin. This in confirmed by the detection of mRNAs of approximately the same sizes in the normal and malignant cells, using one of the probes isolated from the MCF-7 library. We will therefore refer to the antibody reactive glycoproteins on breast cancer cells as mucins, bearing in mind that their processing may be different resulting in molecules of different molecular weights but with the same core protein as that of the milk mucin.
Seven clones were obtained from the MCF-7 library of which the largest was 1800 kb. This clone cross hybridized with the other 6 smaller clones. The β-galactosidase fusion proteins expressed by six of the cross-hybridizing lambda clones were reactive with the polyclonal antiserum directed against the mucin core protein as well as with four well-characterized monoclonal antibodies directed to various epitopes on the stripped core protein, SM-3, SM-4, HMFG-1 and HMFG-2 (14,20). The smallest lambda clone, λMUC9, produced a β-galactosidase fusion protein which reacted with three of the four monoclonal antibodies and with the polyclonal antiserum.
The surprising result that the extensively characterized HMFG-1 and HMFG-2 monoclonal antibodies reacted strongly with the lambda plaques and the fusion proteins and could immunoprecipitate proteins from in vitro translated mRNA provides strong evidence that these clones do indeed code for a portion of the mucin core protein. Although previous evidence such as resistance to fixation, boiling, treatment with dithiothreitol and NaDodSO4 and the presence of multiple epitopes on the molecule suggested that these were carbohydrate (1), it has now been established that the epitopes of the HMFG-1 and HMFG-2 monoclonal antibodies are definitely protein in nature. Carbohydrate may be required to obtain the strongest binding, either as part of the epitope or by conferring some conformational change on the protein portion, but part of the antigenic determinant must consist of an amino acid sequence. Since these two MAbs are reactive with the fully glycosylated milk mucin as well as the stripped core protein, this data means that the intact molecule contains areas of naked peptide which contribute to the antigenic sites for these two antibodies.
Confirmatory evidence that pMUC10 codes for the mammary mucin core protein is provided by RNA blots. The relative abundance of mRNA in the breast cancer cell lines MCF-7, T47D, ZR-75-1 and in normal mammary epithelial cells corresponds to the antigen expression by these cells as measured by the binding of the HMFG-1 and HMFG-2 monoclonal antibodies. Cell types which are negative for antigen expression such as human fibroblasts, Daudi cells and HS578T, a carcinosarcoma line derived from breast (14), are negative in RNA blot hybridizations. A fortuituous observation made with the ZR-75-1 cells yielded indirect strong evidence that pMUC10 does indeed code for the mucin glycoprotein core protein. This cell line, which routinely expresses large amounts both of mRNA and antigen, yielded one preparation of RNA which was unexpectedly negative by blot hybridization. It was subsequently found that those particular ZR-75-1 cells from which the RNA had been made had lost the expression of the antigen as well at this time (as determined by reaction with HMFG-1 and 2). Different passage numbers of the ZR-75-1 cells were recovered and shown once again to express both antigen and message. The sizes of the messages, 4.7 kb and 6.4.kb, are quite large, since a 68 kd or 92 kd protein would need only about 3 kb to code for the protein portions. This suggests that a large portion of the mRNA maybe untranslated. Efforts are underway to obtain a full-length clone.
Thus, the cDNA clones presented here represent a portion of the gene coding for the human mammary mucin which is expressed by differentiated breast tissue as well as by most breast cancers. The major proteins precipitated from in vitro translation products of RNA from MCF-7 cells by antibodies to the milk mucin core protein (68 Kd) have an apparent molecular weight of 68 Kd and 92 Kd. These proteins, produced by the breast cancer cell therefore share epitopes with the 68 Kd core protein of the milk mucin (20). Whether a similar 92 Kd protein is also produced by normal mammary epithelial cells, and is truncated or destroyed by HF treatment is not yet clear. MCF-7 cells biosynthetically labelled with 140 amino acids yield upon immunoprecipitation with HMFG-1 and HMFG-2 antibodies, two glycosylated proteins of 320 kd and 430 kd, and it is possible that each of these glycoproteins utilizes only one core protein of either 68 kd or 92 kd. Alternatively, each of the glycoproteins could contain both the 92 kd and 68 kd proteins either in different proportions or variably glycosylated. Further screening of the library may yield full length cDNAs coding for both sizes of the immunologically related core proteins. Since there appears to be only a single gene (based on Southern blot data obtained by using a partial cDNA probe), it is probable that the multiple messages arise by alternative RNA splicing and this would explain the fact that they contain colon sequences. Although a core protein of 68 kd appears to be small to yield a fully glycosylated molecule of greater than 300 kd which contains 50% carbohydrate, there is evidence that such a structure for mucins is possible. Ovine submaxillary mucin has a reported molecular weight of 1×106 daltons (45), yet its protein core consists of 650 amino acids resulting in a molecule of 58 kd (46).
The mucins which are detected with HMFG-1 and HMFG-2 MAbs on immunoblots of tumours and breast cancer cell lines show variations in size from 80 kd to 400 kd in the molecular weights of the tumour mucin molecules (1,47). Using these same antibodies which detect high molecular weight mucins present in normal urine, a polymorphism has indeed been shown to be genetically determined (48). Although the very low molecular weight components are likely to represent precursor forms of the mucin which appears to be incompletely processed in many tumour cells (20), the variations in the higher molecular weight components are likely to be due to this genetic polymorphism. It was unclear, however, whether the structural basis of the polymorphism was due to the genetically determined protein or to the carbohydrate portion of the mucin. The detection of restriction fragment length polymorphisms in the Southern blotting experiments using the mucin probe suggest that the mucin polymorphism occurs at the level of the DNA which codes for the protein. Preliminary sequence data suggest that the basis for this polymorphism is a region of variable tandem repeats present in the protein coding sequences. This structural feature may be responsible for the generation of the many allelic restriction fragments at the mucin locus. We are presently investigating the basis of the mucin polymorphism by a Southern blot survey of DNA from white blood cells of normal, related individuals whose inheritance pattern of urinary mucins has been determined. In addition, we are examining DNA preparations made from the white blood cells and tumours of individual breast cancer patients. to determine if there is any discordance between genotype in the paired samples, since tandemly repeated DNA may provide an unstable site where recombination or amplification could occur.
The presence of mucins in the majority of carcinomas and their association with the differentiation of mammary epithelial cells makes it particularly important to identify regions involved in the tissue specific and developmental regulation of the gene. Moreover, the introduction of a functional mucin gene into cells should provide insights into the role of this molecule in breast epithelial differentiation and possibly enable us to identify any alterations in the function or expression of the mucin which are related to malignant transformation in the human breast.
The abbreviations are as follows: PBS, phosphate-buffered saline; MAb, monoclonal antibody; IPTG, isopropyl β-D-thiogalactoside; bp, base pairs(s); Kb, kilobase(s).
125I cpm bound
The binding of the antibodies to iodinated intact, partially and totally deglycosylated milk mucin was assayed using the protein A plate method as described in Materials and Methods.
Cell. Physiol. 102, 317-321.
55. Clamp, J. R., Allan, A., Gibbons, R. A. and Roberts, G. Chemical aspects of mucins. Br. Med. Bull. 34:25-41, 1978.
1* =
2* =
3* =
4* =
*meant to be Huynh - See 47 - but looks like wrong book or wrong ed'n.
*meant to be Huynh - See 47 - but looks like wrong book or wrong ed'n.
| Number | Date | Country | Kind |
|---|---|---|---|
| 8700269 | Jan 1987 | GB | national |
| 8700279 | Jan 1987 | GB | national |
| 8726172 | Nov 1987 | GB | national |
| Number | Date | Country | |
|---|---|---|---|
| Parent | 08456919 | Jun 1995 | US |
| Child | 09729226 | Dec 2000 | US |
| Parent | 08134992 | Oct 1993 | US |
| Child | 08456919 | Jun 1995 | US |
| Number | Date | Country | |
|---|---|---|---|
| Parent | 09729226 | Dec 2000 | US |
| Child | 10766020 | Jan 2004 | US |
| Parent | 07381663 | Sep 1989 | US |
| Child | 08134992 | Oct 1993 | US |
