WO2000061636A2

WO2000061636A2 - Aberrantly glycosylated antibodies as marker for cancer

Info

Publication number: WO2000061636A2
Application number: PCT/US2000/009907
Authority: WO
Inventors: Douglas D. Taylor; Cicek Gercel-Taylor
Original assignee: Research Corporation Technologies, Inc.
Priority date: 1999-04-14
Filing date: 2000-04-14
Publication date: 2000-10-19
Also published as: WO2000061636A3

Abstract

The present inventors have found that the level of aberrantly glycosylated IgGs in serum is correlated to the development of cancer. More specifically, the present inventors have found that the ratio of aberrantly glycosylated IgGs versus normal IgGs is significantly increased in cancer patients compared with normal individuals, and that such ratio increases as the cancer progresses. The present inventors have also found that the majority (up to 85%) of the tumor-reactive IgGs in cancer patients are aberrantly glycosylated. In addition, the present inventors have isolated a switching factor that induces the production of aberrantly glycosylated IgGs from antibody-producing cells. Based on these findings, the present invention provides novel methods and compositions useful for diagnosing and treating cancers.

Description

ABERRANTLY GLYCOSYLATED ANTIBODIES: A CANCER DIAGNOSTIC AND PROGNOSTIC INDICATOR

The present invention relates to aberrantly glycosylated antibodies and the correlation thereof with cancers . The present invention further relates to an isolated switching factor that induces the production of aberrantly glycosylated antibodies. Novel methods useful for diagnosing and treating cancers are provided.

Im unosuppression in cancers

Non-specific immunosuppression is one of the manifestations of progressive cancer and is a critical factor in the prognosis of cancer patients, which becomes more pronounced as the disease-_^progresses ⁽Old LJ, Cancer Res . 41: 361-375, 1981; Hellstrom KE, Hellstrom I, Principles of tumor immunity: Tumor antigens, in: DeVita VT, Hellman S, and Rosenberg SA, eds. Biologic therapy of cancer, New York: Lippincott:

35-52, 1991) . Evidence indicates that the tumor itself, exerts these immunosuppressive effects, such as decreased monocyte/macrophage functions (Bast et al., Fed . Proc . 42:1081, 1983) and diminished cellular immunity (manifested as decreased numbers of lymphocytes, reduced responsiveness of lymphocytes to mitogens and decreased delayed cutaneous hypersensitivity) (Braun et al . , Cancer Immunol . immunoth . 15:114, 1983; Brooks et al . , Clin Exp Immunol 74:162, 1988). The im umnosuppression associated with cancer appears to be biphasic: in the presence of localized disease, it is tumor antigen specific (Charest et al., J Biol Chem 258:8769, 1983⁾, while after the disease disseminates, it appears to exhibit a non-specific component (Ebert et al . , J^" Immunol 138:2161, 1987). Thus, as the disease progresses, both the ability to inhibit tumor growth and to overcome infections become compromised. The diminished immune responsiveness associated with early localized cancer may result from antigenic competition for efferent antitumor immune processes and cannot be passively transferred. In contrast, late stage immunosuppression appears to result from inhibition of processes at the afferent level and to be the consequence of immunoregulatory factors released from tumor cells (Taylor et al . , Shed membrane vesicles: a mechanism for tumor induced immunosuppression. In: Immunity to Cancer (eds. AE Reif & MS Mitchell), Academic Press: pp369-373, 1985; Elg et al . , Cancer 71:3938-3941, 1993) and can be passively transferred in the lymphocyte population.

When the tumor growth is limited to the primary site, immune functions can be restored to normal by resection of the primary tumor; however, after the formation of metastases, immune functions generally remain grossly compromised following resection of the primary tumor (Pope BL, Cell Immunol 93:364, 1985). Studies have suggested that this late immunosuppression is mediated by components on or released by metastatic tumor cells arising late in the malignant progression (Elg et al . , Cancer 71:3938-3941, 1993; Somers et al . , Ann Royal Coll Surg Engl 78:103- 109, 1996). Many of these studies have indicated that as the tumor progresses, the cells which arise (whether as selected subpopulations or as cells with newly acquired properties) appear to exhibit a loss of immunostimulatory activities and a gain of immunosuppressive activity (Reif AE, Some key problems for success of classical immunotherapy, in: Immunity to Cancer, AE Reif & MS Mitchell (eds) , Academic Press, pp3-16, 1985) . This differential expression of biologically relevant molecules between primary and late arising metastatic tumor cells may be the determining factor in the altered immune interactions between the host and tumor in early and late stage disease. Since these immunoregulatory molecules are not only expressed on the cell surface, but may also be shed into the extracellular environment, they may be capable of systemic immunosuppression, in addition to localized suppression.

Evaluation of the i-nmunocompetence of patients with ovarian cancer has been utilized to determine whether the tumor has compromised host immunity, to identify specific immune parameters that have prognostic value and a provide a baseline for assessment of immunotherapy. Increasing tumor burden has been associated with decreasing immunocompetence of these patients. T cells and monocyte function are impaired in women with disseminated carcinoma. In untreated ovarian cancer, B cell functions are impaired, as measured by proliferation to PWM and primary response to K H. In early stage ovarian cancer, T cell function is generally normal, as measured by lymphocyte proliferation assays or by delayed cutaneous reactivity to recall antigens. However, reactivity declines with advancing stage and anergy is associated with a poor prognosis.

Humoral responses to tumor associated antigens

While B cell functions are impaired in most ovarian cancer patients, in terms of immunologic eradication of the tumor, studies suggest immune recognition of tumor antigens remains intact. Numerous studies have indicated the presence of tumor- reactive immunoglobulins in cancer patients, including melanoma (Reif AE, 1985; Merimsky et al . , Proc Natl Acad Sci USA, 93:14809-14814, 1996; Conroy et al . ,

Lancet 345: 126, 1995), breast (Barbouche et al . , Europ J Clin Chem Clin Biochem 32: 511-514, 1994), head and neck (Vlock et al . , Cancer Res 49:1361-1365, 1989) and ovarian cancers (Taylor et al . , Am J Reprod Immunol 3:7-11, 1983; Taylor et al . , Am J Reprod Immunol 6:179-

184, 1984) . Cancer patients frequently develop antitumor immune responses, both humoral and cellular, against antigens expressed by their tumors. These immune responses in cancer patients appear to be elicited by antigens which have been altered, overexpressed or inappropriately (temporally or developmentally) expressed (Merimsky et al . , Proc Natl

Acad Sci USA, 93:14809-14814, 1996; Taylor et al . , Am J

Reprod Immunol 3:7-11, 1983; Taylor et al . , Am J Reprod Immunol 6:179-184, 1984). Autologous humoral immune responses represent a patient ' s antitumor immune response against immunogenic epitope(s) of physiologically relevant tumor-associated antigens (Taylor et al . , Am J Reprod Immunol 3:7-11, 1983). Recent studies have indicated that some tumor-reactive immunoglobulins derived from the autologous response may recognize tumor-associated alterations in specific epitopes of commonly expressed antigens. Several antigens have been identified in melanoma by using the "autologous typing" technique; however, the molecular identification of many of these antigens remains elusive. This is, in part, due to poor antibody titer, a lack of sufficient quantities of antibodies and/or formation of immune complexes with circulating antigens. In addition, poor immunologic responses can result from the characteristics of the antigen, such as poor immunogenecity, loss of antigen expression, and antigen masking.

Evidence indicates that several classes of tumor-associated components exist: those shared by all tumors of given histopathologic type, those that are tissue specific and those that represent embryonic antigens. The phenomenon of both shared and distinct tumor-associated antigens has been observed in experimental and human cancers. In an effort to identify the antigens responsible for eliciting these responses, antisera to tumor cells or extracts of tumor cells have been prepared by inoculating other species.

These antisera tend to be developed against strongly immunogenic epitopes on antigens and have tended to not be tumor-specific. Recently, several tumor associated antigens and viral and cellular oncogenes or deregulated tumor suppressor gene products have been shown to elicit autologous antibody responses in a variety of cancers . These immune responses are the consequence of several mechanisms, including aberrant glycosylation (mucins (Kotera et al . , Cancer Res . 54: 2856-2860, 1994), point mutations (in ras (Jung et al . ,

J. Exp. Med. 173: 273-276, 1991) and p53 (Winter et al., Cancer Res . 52: 4168-4174, 1992) or overexpression (of c-erbB-2/her2/neu and p53 protein) . Although these autologous humoral responses appear to be ineffective in the prevention or eradication of the tumor, the minor differences responsible for the immune response may serve as targets for immunotherapy.

The presence of tumor reactive immunoglobulins have been recognized for more than twenty-five years. In tumor models, the presence of tumor-reactive cellular or humoral immune responses is expected to prevent the development of tumors or the metastatic spread of existing tumors. Two seemingly diametrically opposed phenomena exist in patients with ovarian cancer: (1) the development of a humoral response against tumor associated components and (2) the growth and progression of the tumor. In fact, the presence of autoantibodies against certain cellular proteins, including p53 and MUC-1, in advanced ovarian cancer patients is closely linked to poor prognosis and poor survival . While humoral and cellular immunity in these women appear to be functionally suppressed as evidenced by the failure to prevent tumor growth, tumor-reactive humoral responses are clearly demonstrable in the majority of ovarian cancer patients. Since there is evidence of immune recognition of tumor antigens, these phenomena do not involve total suppression of the immune system, but a highly precise process resulting in limited antigen- specific tolerance. Some studies have postulated a

"blocking" role of tumor reactive immunoglobulins, but the mechanism of this negative effect of this patient- derived antibody have not been defined.

Aberrantly glycosylated antibodies

The present invention has identified the presence of aberrantly glycosylated IgG in patients having a cancer at an elevated level. In addition, the presetn inventors have isolated a switching factor which induces the production of aberrantly glycosylated antibodies. Novel methods useful for diagnosing and treating cancers are also provided. Figure 1 depicts the percentage of aberrantly glycosylated IgG (algG) in total IgG in sera of women with ovarian cancer. Women with ovarian cancer were significantly different from normal non-pregnant women at p<0.001.

Figure 2 depicts the percentage of aberrantly glycosylated IgG (algG) in total IgG in sera of women with breast cancer.

Figure 3A depicts a Commassie staining of IgG isolated from normal individuals and patients with prostate cancer. Cancer patients are indicated by "+". Figure 3B depicts percent of aberrantly glycosylated IgGs (algG) in normal individuals and in patients with prostate cancer (+) . Figure 4A depicts a Commassie staining of IgG isolated from normal individuals and patients with colon cancer. Cancer patients are indicated by "+", female patients are indicated by " *" .

Figure 4B depicts percent of aberrantly glycosylated IgGs (algG) in normal individuals and in patients with colon cancer (+) .

Figure 5A depicts levels of aberrantly glycosylated IgG in sera of endometrial cancer patients. Figure 5B depicts levels of aberrantly glycosylated IgG in sera of endometrial cancer patients at various stages.

Figure 6 depicts the amount of aberrant IgG in the sera of samples from normal patients (1) and those with benign (2) ovarian tumors or malignant ovarian tumors (3) . Coded serum samples were supplied by the NCI. Aberrant IgG content of the sera was determined by absorption to concanavalin A and detection with a secondary antibody. Values are the mean absorbance + SEM (n=20 for each mean) . Each of the means is significantly different from the other two ⁽Kruskal-Wallis ANOVA, Turkey's comparison of the means, p<0.05) . Figure 7A depicts the presence of bound IgG on tumor cells, or on membrane fragments isolated from the ascites and sera of 2 ovarian cancer patients, respectively.

Figure 7B depicts the distribution of normal and aberrant IgG in tumor-reactive antibodies eluted from shed membrane fragments obtained from ascitic fluids of ovarian cancer patients.

Figure 8 illustrates a Western Blot of cellular proteins derived from normal ovary (lane 1) and UL-1 ovarian tumor cells (Lane 2) . Serum from an ovarian cancer patient (diluted 1:20) was used as the source of primary antibody and reactive bands were visualized using peroxidase-labeled anti-human Ig and ECL. Figure 9 depicts a Western Blot of serum from an ovarian cancer patient with cellular proteins from UL-1 ovarian tumor cells, before (UnAb) and after (Ab) overnight absorption of serum with proteins from normal ovary. Reactive bands were visualized using peroxidase labeled anti-human Ig and ECL.

Figure 10 depicts a Western Blot analysis of

Cathepsin D isolated from the immune complexes of four ovarian cancer patients. Bands were detected by monoclonal anti-Cathepsin D. Figure 11 depicts the immunoreactivity of monoclonal antiCathepsin D (Panel A) , and two ovarian cancer patients (Panel B & C) with RP-HPLC-separated tryptic fragment of proCathepsin D, isolated from conditioned medium of UL-lovarian tumor cells. "*" denotes positive undigested p53 control, ° denotes peptide recognized by monoclonal antibody, and "₊" denotes fractions sequenced.

Figure 12 depicts the distribution of normally and aberrantly glycosylated IgG heavy chains among the subclasses of IgGs.

Figure 13 depicts the profile of N-linked oligosaccharide chains isolated from the Fc portion of aberrantly and normally glycosylated IgG heavy chains. Figure 14 illustrates the oligosaccharide structures for normal IgG and aberrant IgG.

Figure 15 depicts increases in the level of aberrantly glycosylated IgG in hybridoma culture media after incubation for 7 days with ascitic fluids from ovarian cancer patients (designated #1 through #10) at

10% (v/v) . C indicates the control level of aberrantly glycosylated IgG. Levels of normal symmetric and aberrantly glycosylated IgG were determined by RID.

Figure 16 graphically depicts the procedure for purifying the switching factor.

Figure 17 depicts the reactivity of the supernatants from hybridomas toward aberrantly glycosylated IgG in ELISA.

Figure 18 depicts the effector activities of aberrantly glycosylated IgG and normal IgG in a cytotoxicity assay.

One aspect of the present invention is directed to methods of diagnosing cancers in patients.

The present methods can be used to diagnose cancers including, but not limited to, ovarian cancer, prostate cancer, breast cancer, colon cancer, and endometrial cancer .

One embodiment of the present invention provides a method of diagnosing a cancer in a patient comprising measuring the level of aberrantly glycosylated IgG in the patient serum, and determining an increase in the value relative to a normal individual as indicative of a cancer in the patient. Aberrantly glycosylated IgG molecules can be detected based on their ability to bind Con-A, for example. Aberrantly glycosylated IgG molecules can also be detected by using antibodies that are specific for the aberrantly glycosylated IgG, but not normally glycosylated IgG molecules.

Another embodiment of the present invention provides a method of diagnosing a cancer in a patient by measuring the level of aberrantly glycosylated IgG and total IgG in the patient serum, and determining an increase in the ratio of aberrantly glycosylated IgG over total IgG relative to a normal individual as indicative of a cancer in the patient.

Still another embodiment of the present invention provides a method of diagnosing a cancer in a patient by measuring the level of aberrantly glycosylated IgG and normally glycosylated IgG in the patient serum, and determining an increase in the ratio of aberrantly glycosylated IgG over normally glycosylated IgG relative to a normal individual as indicative of a cancer in the patient.

Another embodiment of the present invention provides a method of diagnosing a cancer in a patient by isolating tumor-reactive IgG from the patient serum, and determining an increased ratio of aberrantly glycosylated IgG versus normal IgG in the tumor- reactive IgG.

Still another embodiment of the present invention provides a method of diagnosing a cancer in a patient by detecting the presence of the switching factor in the patient serum. In another embodiment of the present invention, the above methods of diagnosis can be combined with other cancer detecting methods for diagnosing particular types of cancers. In a further aspect of the present invention, methods of treating cancers are provided.

Still another embodiment of the present invention provides a method of treating a cancer in a patient by eliminating the aberrant IgGs from the patient serum.

Another embodiment of the present invention provides a method of treating a cancer in a patient by removing the aberrant glycosylation on the aberrant IgGs in the patient serum. Still another embodiment of the present invention provides a method of treating a tumor in a patient by administering to the patient, a therapeutically effective amount of aberrantly glycosylated IgG isolated from the patient, preferably, aberrantly glycosylated IgG molecules that are tumor specific, conjugated with an anti-tumor agent.

In another embodiment, the present invention provides a method of treating a cancer in a patient, by administering to the patient, a therapeutically effective amount of an antibody that is specific for aberrantly glycosylated IgGs .

The present invention is further directed to antibodies that are specific for aberrantly glycosylated IgG, i.e., antibodies that bind aberrantly glycosylated IgG and substantially do not bind normally glycosylated IgG. The present invention contemplates both polyclonal and monoclonal antibodies.

Another embodiment of the present invention provides monoclonal antibodies that are specific for aberrantly glycosylated IgG, produced from hybridomas (ATCC deposit # ) .

In a further aspect of the invention, a switching factor has been isolated from cancer patients. The switching factor of the present invention has been characterized as a protein dimer with an estimated Mw of 67,315D. The switching factor migrates as a single band on SDS PAGE under reducing conditions with an apparent Mw of about 26-36 kD. The protein sequence of the factor includes at least one of

SEQ ID NO. 1-10. SEQ ID NO.

1 FPLEAI

2 IPSLADD 3 KTYLELYADX_aaLI (X_aa : unsure)

4 [F or I] EVA [H or E] RFK

5 [E or A] X_aaQRFYVAKG (X_aa : can be Leu or a modif ied amino acid)

6 EQHVKLVNEVTEFA 7 [A or F] QNNGD [F or I] ADIIRKX_aalDX_aa2RL (X_aal : unsure ; X_aa2 can be a modif ied amino acid)

8 VEGA [D , E or V] GT [K or L] [V or Q]

9 VEGA [D, E or V] [G or T] [K or L] [V or Q] [N or T] DEX_aal [Q or A] EVX_aa2GEYTTX_aa3X_aa4 [L or V] (X_aal , X_aa2 and X_aa4 : unsure ; X_aa3 can be a modified amino acid)

10 TFHADIFTLS

The factor of the present invention is further characterized as having a switching activity that induces antibody producing cells to produce aberrantly glycosylated IgGs.

Still another embodiment of the present invention contemplates antibodies raised against the switching factor of the present invention, and in particular, antibodies specific for any one of SEQ ID NOS: 1-10.

The present invention also provides methods of diagnosing a pathological condition such as a cancer in a patient by detecting the presence of the switching factor in the patient serum. Methods of treating cancers based on the discovery of the present switching factor are also contemplated.

The present inventors have found that the level of aberrantly glycosylated IgG in serum is correlated to the development of cancer . More specifically, the present inventors have found that the ratio of aberrantly glycosylated IgG versus normal IgG is significantly increased in cancer patients compared with normal individuals, and that such ratio increases as the cancer progresses. The present inventors have also found that the majority (up to 85%) of the tumor- reactive IgGs in cancer patients are aberrantly glycosylated. In addition, the present inventors have found that a tumor-derived switching factor accounts for the increased level of aberrantly glycosylated IgG in serum. A tumor-derived switching factor has been isolated which is capable of inducing B cells to produce aberrantly glycosylated IgGs . Based on these findings, the present invention provides novel methods and compositions useful for diagnosing and treating cancers .

According to the present invention, the term "aberrant glycosylation" refers to a pattern of glycosylation that deviates from the normal pattern of glycosylation on an Ig molecule, for example, by providing a substantially higher content of mannose in, and/or mannose groups at the terminus of the oligosaccharide chain on an Ig molecule. The aberrant glycosylation has be identified by the present inventors to be associated with the hinge region of the Fc portion of the heavy chain of an Ig molecule, in particular, an IgG molecule.

The aberrantly glycosylated IgG molecules (also referred to herein as aberrant IgGs or algGs, are distinguished from normal IgGs (or nlgGs) by their affinity for Con-A sepharose (algGs bind to Con-A while nlgGs do not) . The aberrant glycosylation may result from a change in the activities and/or levels of glycosylation enzymes or deglycosylation enzymes. The present inventors have proposed that the aberrant glycosylation is due to the failure of α-mannosidase in removing the mannose groups, a step which normally takes place prior to further modification of the oligosaccharide chain, e.g, addition of other carbohydrate groups such as GlcNAc or GalNAc .

One aspect of the present invention is directed to methods of diagnosing cancers .

As used herein, the term "diagnosing" encompasses detecting the onset, as well as monitoring the progression or regression of a cancer.

Cancers which can be diagnosed by the methods of the present invention can be any type of cancer, including, but not limited to, ovarian cancer, breast cancer, endometrial cancer, prostate cancer and colon cancer.

It is believed that tumor-reactive immunoglobulins are present in serum shortly after the induction of a tumor and prior to the appearance of circulating tumor antigens. Circulating tumor antigens are not present in serum until after the tumor has developed into a detectable mass. Thus, the diagnosis methods of the present invention, which are based on the determination of the level of aberrantly glycosylated IgGs in serum, can detect cancers at an earlier stage than existing diagnostic screens which detect the presence of circulating tumor antigens. For example, the present invention provides that the percent of IgGs that are aberrantly glycosylated is about 8% in normal individuals, and is increased to about 25% in Stage I endometrial cancer patients. One embodiment of the present invention provides a method of diagnosing a cancer in a patient by measuring the proportion of aberrantly glycosylated IgG in total IgG in the patient serum, and determining an increase in value relative to a control value as indicative of a cancer in the patient.

In accordance with the present invention, total IgG can be isolated from a patient's blood by a number of well-known procedures. Such procedures include immunoabsorption, Cohn's alcohol fractionation (Cohn et al., -7. Am. Chem . Soc . 68:459-475, 1946 ; Oncley et al., J. Am. Chem. Soc , 71 : 541-550, 1949) , fractionation (Schneider et al . , Vox Sang.31 : 141-151, 1976), ultracentrifugation (Barundern et al . , Vox Sang.

7: 157-174, 1962), the method of Kistler and Nitschmann ( Vox Sang 7: 414-424, 1962), polyelectrolyte affinity adsorption, electrophoresis, ion exchange chromatography, and polyethylene glycol fractionation.

One preferred procedure to isolate total IgG from a patient's blood is immunoabsorption using antibodies specific for human IgGs. Various human IgG- specific antibodies are commercially available, such as mouse anti-human IgG (γ-chain specific) from Sigma

Chemical Co. (St. Louis, MO). The anti-human antibodies can be immobilized on solid support by, e.g., chemical cross-linking. A variety of matrices suitable for chemical cross-linking can be employed, such as activated sepharose or agarose supports, or activated affi-gel supports (Bio-Rad, Hercules, CA) , activated support matrices with spacer arms, as well as matrix materials coupled to magnetic particles. Chemical cross-linking procedures are routine and are described in, e.g., Current Protocols in Molecular Biology (Ausubel et al . , John Wiley & Sons, New York) .

Commercially available matrices immobilized with anti- human IgG antibodies can also be used. To carry out immunoabsorption, a IgG-containing source material, such as a serum or ascitic fluid sample, or a partially purified sample thereof, is brought into contact with the solid support cross-linked with anti- human IgG antibodies. IgG molecules in the sample bind to the solid support material, and the non-specific molecules can be substantially eliminated after extensive washes. The IgG molecules bound to the solid support can then be eluted using solutions that disrupt the binding. Such immunoabsorption procedures are known in the art and can be found in, e.g., Coligan et al. Current Protocols in Immunology, John Wiley & Sons

Inc., New York, New York (1994).

Another procedure which can be employed to isolate IgG from a patient's blood is by chromatography on DEAE-Affi-Gel Blue (Cooper HM, Paterson Y,

Purification of IgG from antiserum, ascites fluid or hybridoma supernatant . In: Short Protocols in Molecular Biology, Wiley & Sons, 11.26, 1995). According to such procedure, the DEAE-Affi-Gel Blue column (BioRad Laboratory) (7ml bed volume/ml serum) is equilibrated with about 3-5 volumes of loading buffer at 4°C. The loading buffer generally contains, e.g., about 20mM Tris-HCl, about 30mM NaCl, pH 8.0, which permits IgG binding to the column. After dialyzed against loading buffer for at least about 24-36 hours, the serum samples are applied to the column. Unbound material is eluted with excessive amount of loading buffer. The bound IgG fraction is removed from the column with elution buffer, such as a solution containing about

20mM Tris-HCl, about 50mM NaCl, pH 8.0, at a flow rate of lOml/hr. Additional purification steps can be employed if desired. For example, these IgG containing fractions collected can be concentrated using an Amicon centrifugal concentrator with a lOOkD cutoff which can remove transferrin, Mw 76kD, which co-elutes with IgG. Total IgG isolated can be quantitated by any of the standard protein assays, e.g., Bradford assay (BioRad) . The purity of isolated IgG can be assessed by SDS-PAGE.

Once total IgG is isolated from a patient, separation of aberrantly glycosylated IgG from normally glycosylated IgG molecules can be achieved by chromatography using Concanavalin A-sepharose. Unlike normally glycosylated IgGs, aberrantly glycosylated

IgGs bind Con-A through their mannose groups . Typically, total IgG (about 1 mg/ml) is mixed with Con A-Sepharose stabilized in binding buffer (about 0.025 M Tris-HCl pH 7.2, about 0.2 M NaCl and trace amounts of CaCl₂, MgCl₂ and MnCl₂) . The mixture is allowed to sit at about 4°C for at least about half an hour, preferably, at least two hours, with periodic shaking. Normally glycosylated IgGs can be washed off using the binding buffer solution. Aberrantly glycosylated IgGs can be eluted with, e.g., binding buffer containing about 0.15 M α-methyl mannoside . Recovered aberrantly glycosylated IgGs can be quantitated by routine protein assays, e.g., the Bradford assay. The percentage of the aberrantly glycosylated IgGs in the total IgG population can then be determined. Alternatively, total IgG isolated from the patient can be separated on a SDS PAGE and visualized by, e.g., Commassie staining or Western Blot Analysis. The aberrantly glycosylated IgG molecules can be distinguished from the normally glycosylated IgGs by a different mobility on SDS PAGE. The percentage of aberrantly glycosylated IgG in total IgG can be quantitated by conventional methods.

In yet another approach, the proportion of the aberrantly glycosylated IgGs in the total IgG population can be determined by employing an antibody that specifically recognize the aberrantly glycosylated IgGs. For example, such an antibody can be coupled to a solid support . Total IgG can be brought into contact with the solid support. The aberrantly glycosylated

IgG molecules bind to the solid support coupled with the antibody molecules, and can thus be separated from the normally glycosylated IgG molecules in total IgG which do not bind to the antibody. The percentage of the aberrantly glycosylated IgGs in total IgGs can then be readily determined.

Once the percentage of the aberrant IgG in total IgG from the patient serum is determined, such value is compared with a control value. The control value can be established from a pool of normal individuals, or from the patient under examination at a time prior to the occurrence of the suspected pathological condition being diagnosed. A rise in such value is indicative of a cancer in the patient. For purpose of monitoring the. progression of a cancer or the efficacy of a therapeutic regimen in a patient, the ratio of aberrantly glycosylated IgG versus total IgG can be compared with the values established from the same patient at earlier times. The present invention provides that the extent of the rise in the percentage of the aberrantly glycosylated IgG in total IgG is indicative of the stage of a cancer. For example, the present invention provides that such percentage increases from about 8-15% for normal individuals, to about 20-25% for Stage I endometrial cancer patients, about 30-35% for Stage II endometrial cancer patients, and about 40-45% for Stage III endometrial cancer patients.

According to the present invention, the level of total IgG in cancer patients is not substantially different from that in normal individuals. Thus, an increase in the absolute level of aberrantly glycosylated IgG in serum is indicative of a cancer. Accordingly, another embodiment of the present invention provides a method of diagnosing a cancer in a patient by measuring the level of aberrantly glycosylated IgG in the patient serum, and determining an increase in the value relative to a control value as indicative of a cancer in the patient. Those skilled in the art can use any of the above-described procedures for measuring the level of aberrantly glycosylated IgG in the patient serum in order to make such diagnostic determination.

Another embodiment of the present invention provides a method of diagnosing a cancer in a patient by isolating tumor-reactive IgGs, and measuring the ratio of aberrant IgG versus normal IgG within such tumor-reactive IgG population.

"Tumor-reactive IgGs" refer to IgGs that recognize and bind antigenic epitopes that are expressed by tumor cells.

In accordance with the present invention, tumor-reactive IgGs can be isolated by a number of methods. One such method involves the use of tumor shed membrane fragments. Typically, the serum sample or the ascitic fluid of a patient is used as the starting material . Shed membrane fragments can be isolated by chromatography, e.g., column chromatography on high exclusion limit agarose based gels followed by ultra-centrifugation at about 100,000xg. Tumor- reactive IgGs can be eluted from the tumor shed membrane vesicles with solutions that disrupt the interactions between IgGs and antigens, such as solutions with low or high pH. The eluted tumor- reactive IgGs can be collected from the supernatant after ultracentrifugation at about 100,000xg, and quantitated using standard assays.

Once the tumor-reactive IgGs are isolated from a patient, aberrantly glycosylated IgGs can be separated from normally glycosylated IgGs using the methods described hereinabove, e.g., by Con-A sepharose-based chromatography. The ratio of the aberrantly glycosylated versus the regularly glycosylated IgGs within the tumor-reactive IgG population can be determined and compared with a control value, which can be established from pools of normal individuals. A rise in the ratio is indicative of the presence or the progression of a tumor in the patient . In another embodiment of the present invention, the above methods of diagnosing a cancer can be combined with other cancer diagnosis methods for determining the particular type of cancer in a patient. For example, the detection of an increased level of aberrantly glycosylated IgGs can serve as the first step of the diagnosis, i.e., as a screening indicator for the presence of a cancer in a patient. In the second step, tests that detect a specific type of cancer are performed, for example, by detecting the presence of an antigen that is specifically associated with a particular type(s) of cancer.

It is known that a number of antigens are associated with one or a small subset of tumors. For example, PSA is specifically associated with prostate cancer. Tyrosinase TAAs are preferentially associated with melanoma. The carcinoembryonic antigen (CEA) and CA125 antigens are preferentially expressed by adenocarcinomas derived from gastrointestinal tract and ovary, respectively (Conry et al . , Cancer Gene Ther. 2:

33-38, 1995; Schlebusch et al . , Hybridoma 14: 167-174,

1995) . Cathepsin D is a useful marker for cancers such as breast cancer (Rochefort, Biological and clinical significance of cathepsin D in breast cancer, in Acta Oncol . 31: 125-130, 1992), ovarian (Scambia et al . ,

Clinical significance of cathepsin D in primary ovarian cancer, in -European Journal of Cancer 30: 935-940,

1994), and endometrial cancer (Scambia et al . , Significance of cathepsin D expression in uterine tumors, in European Journal of Cancer 31: 1449-1454,

1995) . Diagnostic assays which detect the presence of these tumor-specific antigens can be carried out in the second step of the present methods .

Alternatively, a patient may be diagnosed as having a particular type of cancer first. In this case, the detection of the level of aberrantly glycosylated IgGs in serum can be performed as a second step, for monitoring the progression of the cancer or for determining the efficacy of a therapeutic regimen. In an alternative embodiment of the present invention, the above described methods of diagnosis are applied to diagnose a cancer of any type, with the proviso that the cancer at issue is not ovarian cancer. In a further aspect of the present invention, methods of treating a cancer in a patient are provided.

The term "treating a cancer" or "treating a tumor" as used herein means that the growth of cancerous cells is significantly inhibited, which is reflected by, e.g., the tumor volume. Tumor volume may be determined by various known procedures, e.g., obtaining two dimensional measurements with a dial caliper. "Treating a tumor" also encompasses inhibiting the metastasis of the tumor.

Cancers which can be treated by using the methods of the present invention can be any type of cancer, including, but not limited to, ovarian cancer, breast cancer, endometrial cancer, prostate cancer and colon cancer.

According to the present invention, aberrantly glycosylated IgG molecules are able to bind antigens, but are unable to form functional immune complexes which initiate humoral immune responses. Thus, aberrantly glycosylated IgGs compete with normal

IgGs with the same specificity for antigen binding, thereby blocking the function of normal IgG molecules.

Accordingly, one embodiment of the present invention provides a method of treating a cancer in a patient by eliminating the aberrantly glycosylated IgGs from the patient serum.

One approach for eliminating aberrantly glycosylated IgGs is by applying the patient serum to ConA- sepharose, or a solid support material coupled with an antibody specific for aberrantly glycosylated

IgGs . The procedures have been described hereinabove . The serum, once free of aberrantly glycosylated IgGs, can be returned to the patient.

In another embodiment, the present invention contemplates a method of treating a cancer in a patient by removing the aberrant glycosylation in the aberrantly glycosylated IgGs in the patient serum.

Normally, IgG precursors are initially glycosylated with certain number of mannose groups on each of the Fab arms and on the Fc portion of the heavy chains . Certain mannose groups are subsequently cleaved, and the remaining mannose groups are covered by other carbohydrate groups. Due to abnormal- activities of glycosylation or deglycosylation enzymes in cancer patients, the IgG molecules become aberrantly glycosylated by having mannose groups at the terminus of oligosaccharide chain.

Accordingly, a cancer can be treated by removing the aberrant glycosylation in the IgGs, for example, by employing a mannose-cleavage enzyme such as -mannosidase. Such enzyme can be added to the patient's serum in vivo . Alternatively, the patient serum can be treated with such enzyme ex vivo . The serum can be returned to the patient after the treatment and function to provoke immune responses against the cancer in the patient.

Another embodiment of the present invention provides a method of treating a tumor in a patient by administering to the patient, a therapeutically effective amount of an autologous tumor- specific aberrantly glycosylated IgG, conjugated with an anti^¬ tumor agent, for example, a radioisotope or a toxin. A tumor- specific aberrantly glycosylated IgG conjugated with a toxin, also referred to as immunotoxin, can be administered to a patient and targeted to the tumor therein, and executes its toxic effects against the tumor. Examples of such toxins include ricin A chain, and melanocyte stimulating hormone (for treating melanomas) . Such conjugated autologous aberrantly glycosylated IgG can be administered to the patient with a pharmaceutically acceptable carrier. As used herein, a pharmaceutically acceptable carrier includes any and all solvents, including water, dispersion media, culture from cell media, isotonic agents and the like that are non-toxic to the host. Preferably, it is an aqueous isotonic buffered solution with a pH of around 7.0. The administration of a conjugated aberrantly glycosylated IgG to the autologous patient can be carried out in any convenient manner, preferably, by subcutaneous (s.c), intraperitoneal (i.p.), intra- arterial (i.a.), or intravenous (i.v.) injection. In another embodiment, the present invention provides a method of treating a cancer in a patient, by administering to the patient, a therapeutically effective amount of an antibody that is specific for aberrantly glycosylated IgGs . According to the present invention, aberrantly glycosylated IgGs, while able to bind antigens, are unable to initiate the complement pathway and provoke an effective humoral immune response. As a result, tumors are able to evade the surveillance of the immune system. Antibodies specific for aberrant

IgGs, after administered to a cancer patient, can bind such aberrant IgGs which, in turn, can bind to antigens on tumor cells, thereby forming a functional complex capable of initiating the complement pathway and provoking an effective humoral immune response against the tumor. Preferably, antibodies specific for aberrantly glycosylated IgGs for such therapeutic use are partially or fully humanized.

A further aspect of the present invention is directed to antibodies that are specific for aberrantly glycosylated IgG, i.e., antibodies that bind aberrantly glycosylated IgG and substantially do not bind normally glycosylated IgG. The present invention contemplates both polyclonal and monoclonal antibodies. The present invention has identified that the aberrant glycosylation is associated with the Fc portion of IgGs. The Fc portion of aberrantly glycosylated IgGs can be obtained by using conventional procedures, for example, as described in Examples herein below. Purified aberrantly glycosylated Fc fragments can be administered to any appropriate animals for raising antibodies, e.g., rabbit, mouse, sheep and the like. Antibodies that are specific for the aberrantly glycosylated IgG molecules can be produced and isolated.

The present invention provides monoclonal antibodies that are specific for aberrantly glycosylated IgG, produced from hybridomas (ATCC deposit # ) . Functional derivatives of an isolated anti- aberrant IgG antibody are also part of the invention. By "functional derivatives" is meant antibody or fragments derived from an isolated anti-aberrant IgG antibody, where the derivative retains the specificity towards aberrantly glycosylated IgGs. Functional derivatives include, e.g., Fab, Fab', F(ab')₂, humanized antibodies, single chain antibodies and the like. Methods for making such humanized antibodies and single chain antibodies are known in the art. See, e.g., Harris, Humanizing Monocloanl Antibodies for in vivo Use, in Animal Cell Biotech . 6: 259-279, 1994, Academic Press Limited; US Patent 5,585,089 (Humanized Immunoglobulins) ; Mendez et al., Functional Transplant of Megabase Human Immunoglobulin Loci recapitulates human antibody response in mice, in Nature Genetics 15: 146-152, 1997; Bedzyk et al . (1990) J. Biol . Chem. , 265:18615; Chaudhary et al . (1990) Proc . Natl . Acad. Sci . , 87:9491; U.S. Patent No. 4,946,778 to Ladner et al.; and U.S. Patent No. 5,359,046 to Capon et al .

A further aspect of the invention is directed to an isolated switching factor.

The term "switch" as used herein refers to the alteration in glycosylation of IgG molecules produced by B cells, i.e., from normal glycosylation to aberrant glycosylation. The switching factor of the present invention has been shown to induce the switching of antibody-producing cells from producing normal IgGs to producing aberrantly glycosylated IgGs . Antibody-producing cells include B cells isolated from an animal including human, B cell lines or B cell hybridomas .

The switching factor of the present invention has been isolated by applying the ascites of ovarian cancer patients to a series of chromatography.

The switching activity has been detected in the serum and the ascitic fluid from ovarian cancer patients, as well as in the conditioned media of cultured tumor cells established from ovarian cancer patients. Therefore, the switching factor can be produced by tumor cells. The switching activity is essentially not detected in the serum of normal individuals.

The switching factor of the present invention is further characterized as a single peak off a RP-HPLC

C8 300μ 4.6 x 250 mm column. In MALDI mass spectra, such factor appears as a single mass component of 67,315D with an apparent doubly charged ion at 33,534D. The factor appears as a single band on SDS PAGE under reducing conditions with an apparent Mw of about 26-36 kD. Thus, the switching factor of the present invention is a protein dimer held together by disulfide bond(s) .

Protease digestion of the switching factor combined with amino acid sequencing has revealed the following peptide sequences: SEQ ID NO

1 FPLEAI

2 IPSLADD 3 KTYLELYADX_aaLI (X_aa: unsure)

4 [F or I]EVA[H or E] RFK

5 [E or A] X_aaQRFYVAKG (X_aa: can be Leu or a modified amino acid)

8 VEGA [D , E or V] GT [K or L] [V or Q]

9 VEGA [D, E or V] [G or T] [K or L] [V or Q] [N or T] DEX_aal [Q or A] EVX_aa2GEYTTX_aa3X_aa JL or V] (X_aal , X_aa2 and X_aa4 : unsure ; X_aa3 can be a modif ied amino acid)

10 TFHADIFTLS

The switching activity can be detected in an assay as follows. A sample, e.g., serum or ascitic fluid or fractions purified therefrom, is obtained from a cancer patient. Aliquotes of such sample is added to the culture of cells capable of producing IgGs. After culturing the cells for a few days, the supernant of the culture can be removed. The percent of aberrantly glycosylated IgG in total IgG can be determined using any of the methods described hereinabove. The switching activity is measured by determining the increase in such value over a control value. Another embodiment of the present invention contemplates antibodies raised against the switching factor of the present invention.

The antibodies can be generated by using the full-length switching factor or portions thereof as an immunogen, e.g., a peptide of any of SEQ ID NOS : 1-10. Antibodies can be generated by injecting an effective amount of the switching factor or portions thereof into a suitable animal, alone or in combination with a adjuvant. Such animals can include rabbit, chicken, rat, mouse, goat, horse and the like. The present invention contemplates both polyclonal antibodies and monoclonal antibodies. The procedure for making polyclonal and monoclonal antibodies is well known in the art and can be found in, e.g., Harlow, E. and Lane,

D., Antibodies: A Laboratory Manual , Cold Spring Harbor

Press, 1988.

In another embodiment of the present invention, nucleic acid molecules encoding the present switching factor are contemplated. Such nucleic acid molecules can be isolated using any of the molecular cloning techniques available in the art, e.g., screening an appropriate cDNA library by using a primer designed based upon any of SEQ ID NOS: 1-10. The cloned nucleic acid sequences encoding the switching factor can be inserted into recombinant vectors for propagation and/or expression of the proteins or peptides encoded thereby. Such recombinant vectors as well as host cells are also contemplated by the present invention.

According to the present invention, the isolated switching factor of the present invention induces B cells to produce aberrantly glycosylated IgGs and causes an increased level of aberrantly glycosylated IgGs in patients with a cancer, such as ovarian cancer, breast cancer, endometrial cancer, prostate cancer and colon cancer. The aberrantly glycosylated IgGs mask the antigens by binding to antigens without provoking an effective immune responses .

Accordingly, the present invention contemplates methods of treating a cancer by inhibiting the function of the switching factor. The inhibition may be accomplished by, e.g., abolishing the expression of the switching factor, sequesting the circulating switching factor by using antibodies specific for the switching factor, inhibiting the activity of the factor with compounds (e.g., inorganic or organic molecules). Further in accordance with the present invention, the appearance of the switching activity correlates with the onset of a cancer. The present invention provides methods of diagnosing a cancer by detecting the presence of the switching factor in a patient. The detection can be achieved by using assays which measure the switching activity as described hereinabove, or by using antibodies specific for the switching factor.

The present invention is further illustrated by the following examples.

EXAMPLE 1 CELL LINES AND PATIENT MATERIALS

The ovarian tumor cell line, UL-1, was developed from the ascites fluid obtained from a 64 year old patient with stage III ovarian cancer (Owens et al., Proc . Am . Assoc . Cancer Res . 34: 26, 1993).

These cells were grown in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum, 0. ImM nonessential amino acids, ImM sodium pyruvate, 200mM L- glutamine, lOOng/ml streptomycin and 100 IU/ml penicillin in a humidified 5% C0₂ atmosphere. Cell viability was evaluated by trypan blue exclusion. All cultures utilized were >95% viable. Samples of normal ovary were obtained from individuals undergoing hysterectomy and oophorectomy unrelated to ovarian cancer disease by the Gynecology service. The ovarian epithelium was aseptically dissected from the fresh ovary and incubated in trypsin as described by Testa et al . ( Cancer Res . 54: 2778-

2784, 1994) . This procedure selectively removes surface epithelial cells. Sera samples were obtained from ovarian cancer patients undergoing treatment by the Division of Gynecologic Oncology at the University of Louisville and normal control sera were obtained from non-cancer bearing volunteers.

EXAMPLE 2 INCREASED LEVELS OF ABERRANTLY GLYCOSYLATED IgG IN CANCER PATIENTS

An increased level of aberrantly glycosylated

IgG antibodies was detected in female-specific cancer types including ovarian cancer, endometrial cancer and breast cancer.

Ovarian Cancer

The percentage of aberrantly glycosylated IgG in total IgG was analyzed for ovarian cancer patients . Initially, total IgG was quantitated by .a standard commercial radial immunodiffusion kit, normal level. To determine the level of aberrantly glycosylated IgG, total IgG was incubated with ConA-Sepharose in a 1.5ml microcentrifuge tube. After 1 hour, the ConA-Sepharose was pelleted by centrifugation and the supernatant was removed. The ConA-Sepharose pellet was washed twice with 20mM Tris-HCl, 0.5M NaCl, pH7.2 and the asymmetric antibodies eluted by treatment with 0.1M borate buffer, pH6.5. After the borate buffer treatment, the ConA- Sepharose was removed by centrifugation. The supernatant was removed and diluted 1:1 with PBS. This mixture was then assayed by a standard commercial radial immunodiffusion kit, low level. The percentage of aberrantly glycosylated IgG in total IgG was calculated. As shown in Figure 1, the percentage of aberrantly glycosylated IgG in total IgG was approximately 35%, whereas the percentage is about 8% in age and sex matched controls.

Breast cancer

The distribution of normal to aberrant antibodies was analyzed in the sera of women with breast cancer. Initially, total IgG was quantitated by a standard commercial radial immunodiffusion kit, normal level. To determine the level of aberrantly glycosylated IgG, the total IgG was incubated with ConA-Sepharose in a 1.5ml microcentrifuge tube. After

1 hour, the ConA-Sepharose was pelleted by centrifugation and the supernatant was removed. The ConA-Sepharose pellet was washed twice with 20mM Tris- HCl, 0.5M NaCl, pH7.2 and the asymmetric antibodies eluted by treatment with 0.1M borate buffer, pH6.5.

After the borate buffer treatment, the ConA-Sepharose was removed by centrifugation. The supernatant was removed and diluted 1:1 with PBS. This mixture was then assayed by a standard commercial radial immunodiffusion kit, low level. The proportion of aberrantly glycosylated IgG to total IgG was compared between breast cancer patients and age and sex matched controls. The results are shown in Figure 2.

Prostate Cancer

The level of aberrantly glycosylated IgG antibodies was also examined for male-specific cancer types, such as prostate cancer.

Total IgG was isolated from the serum (0.1m) of volunteers (either normal individuals or patients with prostate cancer) by immunoprecipitation with anti- human IgG. After precipitation, the isolated IgGs were separated by SDS-PAGE under reducing conditions. The resulting light chains 22kD and the two heavy chains (55kD [normal] and 65kD [aberrant]) were visualized by

Commassie blue staining as shown in Figure 3A.

Sera (0.25ml) from normal volunteers or patients with prostate cancer were applied to an immunoabsorption column containing anti-human IgG. After the column was washed with 2OX volumes of TBS/Tween 20, it was eluted with 0.1m glycine-HCl, pH2.7. The washing and elution was each monitored by OD₂₈₀. The isolated IgG was dialyzed against TBS overnight. Aliquots of the IgG were added to microtiter plates containing either anti-human IgG or

ConA. The wells were washed and then peroxidase- labeled anti-human IgG was added to all wells. The amount of peroxidase in each well was determined colorimetrically . The amount of peroxidase associated with IgG bound to ConA (aberrant IgG) was divided by the amount of peroxidase associated with total IgG (bound to anti-IgG) . Figure 3B (left panel) presents the ratio for all samples, while the right panel divides the samples between controls and cancer patients.

As can be seen from Figure 3A and 3B together, an increased level of aberrantly glycosylated IgGs was observed in patients with prostate cancer when compared with normal individuals.

Colon Cancer

Experiments were carried out using sera from patients with a non-gender-specific cancer, colon cancer. As shown in Figure 4A and 4B, an increased level of aberrantly glycosylated IgGs was observed in patients with colon cancers when compared with normal individuals.

Progression of Cancer To identify the correlation between the increase in the level of aberrant IgG and the progression of the cancer, patients at various stages of endometrial cancer were examined. The experiments were carried out in essentially the same manner as described above. The results are illustrated in Figure

5A and 5B. Figure 5A presents the ratio of aberrant

IgG over total IgG for all patients, while Figure 5B divided the patients into groups according to stage . As seen from Figures 5A and 5B, an increase in the ratio of aberrant IgG over total IgG was detected in early stage (Stage I) endometrial cancer patients. The level of asymmetric IgG was found to increase with the progression of the cancer. The level of aberrant IgG in 20 patients suffering benign ovarian cancer and 20 patients suffering malignant ovarian cancer was examined. The values were compared to that obtained from normal individuals. The serum samples from benign ovarian cancer patients and malignant ovarian cancer patients were obtained from NCI (National Cancer Institute) . The amount of aberrant IgG in the serum was determined by absorption of ConA and detection with secondary antibody as described above. The results are summarized in Table I and the median values are shown in Figure 6.

Table I . Amount of Aberrant IgG in Serum Samples .

H=46.224 with 2 degrees of freedom. (P=<0.001)

_The differences in the median values among the treatment groups are greater than would be expected by chance; there is a statistically significant difference (P =<0.001). EXAMPLE 3 SPECIFIC ASSOCIATION OF ABERRANTLY GLYCOSYLATED IgGs WITH TUMOR-REACTIVE IMMUNOGLOBULINS Experiments were carried out to compare the amount of tumor reactive IgGs that were able to bind tumor cell-associated antigens, that were bound to membrane fragments (MF) isolated from the ascites or on MFs isolated from the sera of two ovarian cancer patients.

Shed tumor derived membrane fragments were isolated from the ascites or sera of ovarian cancer patients based on a procedure developed by Taylor et al., BBRC 113: 470, 1983. The procedure utilizes column chromatography on high exclusion limit agarose based gels. These tumor-derived membrane fragments were pelleted by centrifugation at 100,000xg. The pellet was resuspended in 0.1M glycine-HCl, pH2.7 and incubated at room temperature for 15 minutes. The membranes and antibody were separated by recentrifugation at 100,000xg. The supernatant, containing eluted IgG, was neutralized with 1. OM Tris and concentrated by ultrafiltration with a lOOkD membrane. The isolated IgG was quantitated by a standard commercial radial immunodiffusion kit, normal level .

The amount of tumor reactive IgGs that were able to bind tumor cell-associated antigens was determined as follows . Immunoabsorbant columns were constructed using lysates from UL-1 ovarian tumor cells conjugated to Sepharose 4B. Sera (1ml) from normal volunteers or patients with ovarian cancer were applied to this immunoabsorption column. After the column was washed with 2OX volumes of TBS/Tween 20, it was eluted with 0.1m glycine-HCl, pH2.7. The washing and elution was each monitored by OD280. The tumor-reactive antibodies were dialyzed against TBS overnight and added to microtiter plates containing anti-human IgG.

As shown in Figure 7A, the MFs isolated from the ascites or from the sera of the cancer patients exhibited higher levels of tumor reactive IgG bound than tumor cells .

The ratio of aberrant versus normal IgG in the tumor reactive IgG population was examined for five ovarian cancer patients. Tumor-reactive IgGs bound to the MFs isolated from the ascites of five ovarian cancer patients were isolated as described above. To determine the level of aberrantly glycosylated IgG, the total tumor-reactive IgGs were incubated with ConA- Sepharose in a 1.5ml microcentrifuge tube. After 1 hour, the ConA-Sepharose was pelleted by centrifugation and the supernatant was removed. The ConA-Sepharose pellet was washed twice with 20mM Tris-HCl, 0.5M NaCl, pH7.2 and the asymmetric antibodies eluted by treatment with 0.1M borate buffer, pH6.5. After the borate buffer treatment, the ConA-Sepharose was removed by centrifugation. The supernatant was removed and diluted 1:1 with PBS. This mixture was then assayed by a standard commercial radial immunodiffusion kit, low level. The proportion of aberrantly glycosylated IgG to total tumor-reactive IgG was then calculated.

As illustrated in Figure 7B, aberrantly glycosylated IgGs account for about 85-95% of the tumor-reactive IgGs.

EXAMPLE 4 IDENTIFICATION OF ANTIGENS RECOGNIZED BY PATIENT'S HUMORAL RESPONSE

Isolation Of Cellular Membranes

Membrane enriched fractions were prepared from UL-1 ovarian tumor cells by sucrose density gradient centrifugation, as follows. Monolayers of UL- 1 cells, grown to confluence, were washed with PBS and harvested from the dishes by scraping. Cells were pelleted by centrifugation at 400 X g for 10 min and resuspended in a hypotonic buffer, consisting of lOmM Tris pH 7.4, lOmM NaCl, ImM EDTA and ImM PMSF at 4°C for 15 min. The cells were homogenized in a dounce homogenizer for 15-20 strokes. The resulting cell homogenate was centrifuged at 1000 X g for 15 min to separate nuclei and unbroken cells (pellet) . The supernatant was centrifuged at 100,000X g for 60 min to isolate crude membranes. Crude membranes were resuspended in 8.5% sucrose and applied to a discontinuous sucrose density gradient of 8.5%, 30%, 35% and 50% sucrose prepared in 10 mM Tris pH 7.4 and ImM PMSF buffer. These crude membranes were centrifuged at 100,000 X g for 2 hours. Four fractions were obtained: at the 8.5% and 30% interface (fraction I), at the 30% and 35% interface (fraction II) , at the 35% and 50% interface (fraction III) and at bottom of the gradient (fraction IV) . Each layer was carefully removed and the four fractions were diluted with lOmM Tris pH 7.4 and ImM PMSF solution and centrifuged at

100,000 X g for one hour as previously described. The pellets were suspended in a buffer, 10 mM Tris pH 7.4, lOOmM NaCl, 10% glycerol, ImM DTT and ImM PMSF and protein concentrations of these fractions were determined by the Bio-Rad protein assay. To evaluate the enrichment of plasma membranes, the level of 5' nucleotidase was determined for each fraction. 5' Nucleotidase was assayed by measuring the released inorganic phosphate from AMP at 37°C for 30 min in a reaction mixture containing 5.6 mM sodium-AMP, 56 mM glycine-sodium hydroxide pH 9.1 and 11.1 mM MgCl₂ (Emmelot et al . , Methods Enzymol . , 31:

75-92, 1974) . The sucrose gradient isolated membrane fraction (8.5/30% interface) used in these studies exhibited a greater than 6 fold enrichment of 5 ' - nucleotidase .

The cell-associated protein components were separated by SDS-PAGE and the presence of components reactive with the humoral immune response of patients with ovarian cancer was assessed by western immunoblot as follows.

Immunoblot analysis of tumor derived immunoreactive proteins Fractions 1 and II (40μg) were separated by

SDS-PAGE under reducing conditions using the method Laemmli et al . (Nature 227: 680-685, 1970), with 3% stacking gel and 12% separating gel. Molecular weight estimates were obtained by simultaneous electrophoresis of prestained molecular weight standards (BioRad,

Hercules, CA) . Following the electrophoretic separation, the proteins were electrophoretically transferred to nitrocellulose membrane. The nitrocellulose membrane was blocked by incubation with 5% nonfat dry milk in 30mM Tris pH 7.4, l50mM ΝaCl,

0.5% Tween 20 (wash buffer) . The blots were incubated with patients' sera diluted to 1/20 for 1 hour. Blots were then washed three times with wash buffer for 15 min each, and incubated with peroxidase conjugated rabbit anti human polyvalent immunoglobulins (Sigma Chemical Co., St. Louis, MO) for 1 hour. Blots were then washed three times with wash buffer for 15 min each. The reactive bands were visualized by enhanced chemiluminescence (Amersham Life Sciences, Arlington Heights, IL) . The resulting i munoreactive bands were compared to the molecular weight standards to determine the molecular weight .

For conformational studies, commercial antibodies against each of these proteins were obtained. The commercial antibodies used were monoclonal anti-Cathepsin D (Calbiochem-Novabiochem International, San Diego, CA) and goat polyclonal anti- GRP78 (Santa Cruz Biotechnology, Santa Cruz, CA) . Western blot analysis was performed with these as described above.

For all ovarian cancer patients, these western immunoblots identified multiple bands ranging in molecular weight from 20 to 140 kD (Figure 8) .

While the sera of all ovarian cancer patients recognized some antigens associated with the UL-l tumor, the number and intensity of the immune interactions varied among patients. The variable intensities of signal on the immunoblot could be due to the stage of disease, immunologic status of the patients, or previous and current treatment, as well as other genetic factors. While the specific proteins recognized varied among patients, some of these bands were recognized by most patients. Two such proteins expressed by the UL-1 ovarian tumor cells had approximate molecular weights of 32 and 71 kD.

Reactivity to these proteins was exhibited by the sera of ovarian cancer patients, but not by normal control sera. As controls for tumor specificity, the reactivity of normal sera were tested with UL-1 proteins. In addition, the reactivity of the cancer patients' sera was tested against with normal ovarian epithelium ⁽Figure 8, lane 1) . Sera from normal (non- tumor bearing⁾ controls failed to recognize proteins isolated from UL-1 tumor cells. When the patients' sera were adsorbed overnight with proteins derive_d from normal human ovarian epithelium and subsequently use_d in western blot analysis against UL-1 derived proteins, many of the reactive bands were removed or diminishe_d ⁽Figure 9) . The reactivity of the 71 kD and 3₂ kD proteins were unchanged by absorption of the patient's serum. Multiple fragments for each protein were selected for sequencing by triple quadruple mass spectrometer. The sequencing technique produced data from the 71 kD and 32 kD proteins (Table II) .

Comparisons of these peptide sequences by the BlastP program identified them as glucose regulated protein 78 (a member of shock protein 70 family) and Cathepsin D, respectively.

Table II

EXAMPLE 5 PRESENCE OF CIRCULATING CATHEPSIN D

Since a humoral response to Cathepsin D could be demonstrated in vivo, the association of the specific Cathepsin D antigens in circulating immune complexes was analyzed by western immunoblot .

Circulating immune complexes were isolated as follows. Serum samples (0.5ml) from cancer patients and normal volunteers were initially incubated with

Affi-Gel Blue 50-150μm (Bio-Rad, Hercules, CA) for 1 hour at room temperature with continuous gentle mixing to remove albumin. These samples were then pre-cleared by incubation with anti-mouse Ig coupled to Sepharose. The serum supernatants were applied to 1ml-Protein G-

Sepharose columns (Bio-Rad, Hercules, CA) , which were previously equilibrated with PBS, pH7.4. After the sera were applied, the columns were washed with 25ml of 20mM phosphate buffer, pH7.4, and the absorbance at 280nm was monitored to insure the complete removal of nonspecifically bound material. Bound immunoglobulins (free and immune complexes) were eluted with lOmM glycine-HCl buffer, pH3.0. The elution was monitored by absorbance at 280 nm and the eluted fractions were immediately neutralized by addition of 1M Tris, pH9.

The eluted fractions were dialyzed and concentrated by speed-vac. These immunoglobulin fractions were separated by SDS-PAGE and the association of specific antigens were assessed by western immunoblotting. The Western immunoblot analysis of Cathepsin

D antigen associated with the immune complexes from patients demonstrated the presence of reactive bands corresponding to the 52kD pro-Cathepsin D and a lower molecular weight Cathepsin D (Figure 10) . In contrast to the mature Cathepsin D observed in UL-1 cells, the molecular weight of the Cathepsin D associated with molecular of the Cathepsin D associated with circulating immune complexes appeared to be approximately 29kD. The presence of Cathepsin D in these immune complexes further confirms the development of Cathepsin D reactive antibodies in these patients.

Example 6 IDENTIFICATION OF in vivo RECOGNIZED ANTIGENIC EPITOPES ON CATHEPSIN D

Since the circulating form of the enzyme originally generated the response, Cathepsin D was isolated from the UL-1 conditioned medium to define the specific antigenic epitopes being recognized by the patients ' humoral responses . The form of the enzyme present in the conditioned medium was primarily the proCathepsin D. Tryptic fragments were obtained from this pro-enzyme and peptide specific ELISA plates were prepared. Each ovarian cancer patient recognized multiple fragments; however, some fragments were recognized by both patients. Four of these common fragments were sequenced. Two sequences corresponded to fragments in the propeptide, one fragment corresponded to the glycosylation site of the mature form, and the fourth corresponded to the carboxy- terminus of the enzyme. To identify the commonly recognized antigenic epitopes, Cathepsin D was isolated by affinity chromatography using pepstatin A. To mimic the circulating form of the enzyme, it was purified from the UL-1 conditioned medium. In contrast to cell- associated Cathepsin D, the primary form released from the cells appeared to be the pro-form (Figure 11) .

The pro-Cathepsin D was digested with trypsin and the resulting fragments separated on RP-HPLC. These separated fragments were analyzed for their immunoreactivity with monoclonal antibody and the patients' sera by a peptide specific ELISA. Two fractions were demonstrated to contain the antigenic site recognized by the monoclonal antibody. While multiple fractions were found to react with the Cathepsin D tryptic fragments, four fractions were recognized by both ovarian cancer patients. These 4 positive fragments were sequenced and their identity as Cathepsin D further confirmed (Table III) . _Peak ₃₄ contained a 5 amino acid fragment corresponding to amino acids 24-28 of the pro-peptide. Peak 4₈ contained an 8 amino acid fragment corresponding to amino acids 404-411 of the C-terminal portion of the heavy chain of mature Cathepsin D. Peak 97 contained a 13 amino acid fragment corresponding to amino acids 134-146 of the light chain of mature enzyme (which contains the glycosylation site) and 16 amino acid fragment corresponding to amino acids 55-68 of propeptide, 69 and 70 of light chain of mature Cathepsin D. The monoclonal antibody utilized in this technique identified the 236-253 amino acid epitope on heavy chain of mature cathepsin D.

TABLE III

The mechanisms underlying the immunogenicity of tumor-derived Cathepsin D appears to be multifaceted. The peptides identified by patients' humoral response did not contain mutations at amino acid level, which is consistent with other studies that Cathepsin D protein has only one conservative point mutation at amino acid 58 of alanine to valine of the propeptide in MCF-7 breast cancer cells. A missense conversion of C→T at position 224 of the Cathepsin D gene is responsible for this mutation, observed in 23- 30% of breast cancer and normal tissues tested and attributed to genetic polymorphism rather than somatic mutation. This phenotypic change of alanine to valine in the propeptide of cathepsin D increases the hydrophobicity of the profragment and this change is implicated in altered routing and/or maturation of the proenzyme (Hatasaka and Varner, Curr. Opin . In Obstet .

Gynecol . , 6: 503-509, 1994). As shown in Table 2, patient antibody identified an epitope containing amino acid 58; however, no change in the amino acid sequence of alanine to valine was observed.

EXAMPLE 7 CHARACTERIZATION OF ABERRANTLY GLYCOSYLATED IgG

⁽I⁾ Characterization of General Glycosylation Patterns I G purification

IgGs were purified by chromatography on DEAE- Affi-Gel Blue from serum (Cooper HM, Paterson Y, Purification of IgG from antiserum, asci tes fluid or hybridoma supernatant . In: Short Protocols in Molecular Biology, Wiley & Sons, 11.26, 1995) . The DEAE-Affi-Gel

Blue column (7ml bed volume/ml serum) was equilibrated with 5 volumes of loading buffer (20mM Tris-HCl, 30mM NaCl, pH 8.0) at 4°C. After dialyzed against loading buffer for 36 hours, the serum samples were applied to the column. Unbound material was eluted with 3 bed volumes of loading buffer. The bound IgG fraction was eluted with 20mM Tris-HCl, 50mM NaCl, pH 8.0 at a flow rate of lOml/hr. Fractions were collected and the initial peak containing IgG was retained, as determined by RID. These IgG containing fractions were pooled and concentrated using an Amicon centrifugal concentrator with a lOOkD cutoff (to remove transferrin, Mw 76kD, which co-elutes with IgG) . IgG was quantitated by protein assay and the purity was assessed by SDS-PAGE.

Glycoform detection by immunoblotting

IgG sugar composition was estimated using a previously described lectin binding assay (Bond et al . , J Autoinnunity, 10: 77, 1997) with reference to standards which were determined by controlled hydrazinolysis and gel-permeation chromatography. Equivalent amounts of IgG (4μg) were dot blotted onto nitrocellulose membranes where terminal galactose, mannose, and GlcNAc residues were detected by the use of biotinylated lectins. The reproducibility and standardization of this assay were demonstrated in comparative trials. Ricinus communis (RCA I) was used to detect galactose, Bandeiraea simplicifolia II (BSII) was used to detect GlcNAc, and ConA was used to bind mannose. Streptavidin-horseradish peroxidase was used as the biotin detector. Samples were measured in triplicate and standardization was carried out for each experiment to allow for variations. Absorbance measurements were made using a Hewlett Packard 4c scanner and analyzed using Collage software.

IgG measurements of galactose, GlcNAc, and mannose were defined as OD units. This was achieved by expressing the OD of each sample as a fraction of an internal standard IgG. The %GO was calculated from the ratio of BSII binding/RCA I binding and BSII binding/ConA binding. The results obtained were descriptive of the levels of specific sugars, the actual measurement was lectin binding to exposed residues on the oligosaccharide chains.

Statistical analysis

Association analysis between %GO (as well as galactose, mannose and GlcNAc units) and disease was achieved using regression analysis and Fisher's exact test. Separate confidence intervals were calculated when appropriate. Multiple comparison procedures were performed; a one way analysis of variance followed by Gabriel's test was used to compare each group mean. %GO data was log transformed since the distribution was positively skewed and the variability increased with the mean. Paired data when measuring galactose, mannose and GlcNAc were compared by simple linear regression and 95% confidence intervals for the difference between the slopes of regression were used, where a zero difference between slopes near the middle of a confidence interval showed no evidence that the two population regression lines had different slopes.

In this manner, specific changes were detected which were associated with the presence of ovarian tumors or which correlate with the progression of disease.

(II) Alteration of glycosylation patterns on portions of IgG molecule Isolation of IgG and separation into subclasses

Total IgG were isolated from serum (20ml) by application to a Protein G-Sepharose column. The sample was recirculated to maximize binding. The column was washed with 20 column volumes of 20mM sodium phosphate, pH 7.4 and then, total IgG were eluted with

0.1M glycine-HCl, pH 2.7. The total serum IgG were dialyzed against 20mM Tris-HCl, 0.5M NaCl, p 7.2, overnight. This total IgG preparation was applied to an anti-human IgG_!-Sepharose column in 20mM Tris-HCl, 0.5M NaCl, pH7.2. The sample was recirculated to maximize binding. The column was washed with 20 column volumes of 20mM sodium phosphate, pH 7.4 and the unbound IgG were concentrated using ultrafiltration with a lOOkD cutoff membrane. The unbound IgG were then applied to an anti-human IgG₂-Sepharose column.

The sample was recirculated to maximize binding. The column was washed with 20 column volumes of 20mM sodium phosphate, pH 7.4 and the unbound IgG were concentrated using ultrafiltration with a lOOkD cutoff membrane. The unbound IgG were applied to an anti-human IgG₃-

Sepharose column. The sample was recirculated to maximize binding. The column was washed with 20 column volumes of 20mM sodium phosphate, pH 7.4 and the unbound IgG were concentrated using ultrafiltration with a lOOkD cutoff membrane. The unbound IgG were then applied to an anti-human IgG₄-Sepharose column. The sample was recirculated to maximize binding. The column was washed with 20 column volumes of 20mM sodium phosphate, pH 7.4. The bound IgG subclass from each column was then eluted with 0.1M glycine-HCl, pH 2.7.

Each IgG subclass was dialyzed against 20mM Tris-HCl, 0.5M NaCl, pH7.2, overnight. The concentration of each IgG subclass was determined using a commercial ELISA technique . IgG was isolated from the serum (0.1ml) from patients with breast (Br) , colon (C) , prostate (P) and ovarian (O) cancers by immunoprecipitation with anti- human IgG. After precipitation, the isolated IgG was separated by SDS-PAGE under reducing conditions. The two heavy chains (55kD [normal] and 65kD [aberrant] ) were analyzed by Western immunoblot using monoclonal antibodies specific for IgGl, IgG2, IgG3 or IgG4 heavy chains (Figure 12) . Although aberrant glycosylation can occur in any subclass of IgG in cancer patients, the aberrantly glycosylated heavy chains appeared to be primarily, if not exclusively, in IgG and IgG₂.

Preparation of F(ab'). fragments

After separation into IgGl, IgG2 , IgG3, and IgG4, 1M citrate buffer, pH3.5 was added and the final pH was adjusted to 3.5 with ImM HC1. The individual IgG groups were digested as described by Parham (On the fragmentation of monoclonal IgGl, IgG2a and IgG2b from BALB/c mice, Journal of Immunology 131: 2895-2902, 1983) by adding 125/xl of a lmg/ml pepsin solution.

After digesting for 18hr at 37°C, the reaction was stopped by addition of 1M Tris to a pH of 7.5. This was dialyzed against 5mM Tris-HCl, pH7.5. The digested material was applied to a DEAE-cellulose column, equilibrated with 5mM Tris-HCl. The initial eluted fraction was the F(ab')₂ fraction and the non-digested

IgG were eluted with 5mM Tris-HCl containing 0. IM NaCl.

Purity of each fraction was assessed by SDS-PAGE as described by Laemmli (Nature, 227: 680-685, 1970).

Preparation of Fab' fragment

Cysteine base (final concentration 0.1M) in lOmM Tris-HCl, pH8.2 was added to lmg of F(ab')₂ in 5mM

Tris-HCl buffer. This solution was incubated at 37°C for 3 hr. The digestion was stopped by addition of

30mM iodoacetamide at room temperature in the dark.

The digest was dialyzed against PBS to remove iodoacetamide .

Fc and Fab fragmentation

Serum (2ml) from a patient with ovarian cancer was applied to an immunoabsorption column containing anti-human IgG. After the column was washed with 2OX volumes of TBS/Tween 20, it was eluted with 0.1m glycine-HCl, pH2.7. The washing and elution was monitored by OD280. The IgG was dialyzed against TBS overnight. The isolated IgG was applied to a ConA- Sepharose column and the non-binding IgG (normal) was removed by washing. The bound IgG (aberrant) was eluted. Both populations were dialyzed and lyophilized. Papain (lOOul of 0.05mg/ml) was added to each population and incubated for 4 hours at 37°C. Iodoacetamide (20ul of 0.3M) was added to stop the reaction. The mixture was applied to a Bio-Gel-P6 column for desalting. The mixture was then applied to

Protein G to bind the Fc portion and the Fab portion did not bind. The non-binding Fab was separated by SDS-PAGE under non-reducing conditions to visualize the presence of distinct aberrant and normal chains. The Fc population was eluted and separated by SDS-PAGE under reducing conditions to divide the two pieces of Fc heavy chain. Only one Fab band was observed_; however, 2 distinct bands were obtained with the Fc portion. Serum (2ml) from a patient with ovarian cancer was applied to an immunoabsorption column containing anti-human IgG. Aliquots (200ug) of Fc fragments obtained as described above were subjected to glycosylation profiling using a N-linked oligosaccharide profiling kit from Bio-Rad

Laboratories. Figure 13 depicts the oligosaccharide obtained from the normal and aberrant Fc.

Affinity chromatography with ConA-Sepharose F(ab)₂, Fab, or Fc fragments were applied to

ConA-Sepharose columns (0.7x5cm) equilibrated with 25mM Tris-HCl, pH7.2 containing 0.2M NaCl, 3mM CaCl₂, 3mM MgCl₂ and 3mM MnCl₂. Samples were incubated overnight at 4°C and unbound proteins were eluted with 25mM Tris- HC1, pH7.2 containing 0.2M NaCl, 3mM CaCl₂, 3mM MgCl₂ and 3mM MnCl₂. Elution of specifically bound IgG fragments was performed using binding buffer containing 0. IM α-methyl mannoside.

Analysis of terminal sugars present in IgG subclasses and their fragments

To qualitatively analyze the terminal sugars present in each IgG subclass and fragments derived from them, a commercially available DIG-Glycan Differentiation kit was used. This method utilized the ability of lectins to identify hydrocarbon structures present on glycoproteins by binding specifically. Lectins was conjugated with digoxigenin and their binding was assessed by means of a colorimetric reaction with alkaline phosphatase-conjugated anti- digoxigenin. Galantus nivalis agglutinin (GNA) was used to recognize terminal mannose, Sambucus nigra agglutinin (SNA) was used to bind terminal sialic acid linked through (2-6) -galactose or N-Ac Galactose, Maackia amurensis agglutinin (MAA) was used to recognize terminal sialic acids linked through α(2-3)- galactose, peanut agglutinin (PNA) was used to bind galactose linked through β (1-3) -N-acetylgalactosamine- Serine, and Datura stramonium agglutinin (DSA) was used to bind galactose- β (1-4) -N-glucosamine. Positive controls included carboxypeptidase Y for GNA, transferrin for SNA, fetuin for SNA, MAA, and DSA, and asialofetuin for PNA and DSA. The IgGs and their fragments were analyzed by slot blotting and absorbance measurements were made using a Hewlett Packard 4c scanner and analyzed using Collage software.

Two possible types of oligosaccharide chain associated with the hinge region are proposed based on the Con-A binding property (Figure 14) . The one on the right most likely represents the aberrant species based on its apparent size.

EXAMPLE 8 INDUCTION OF SYMMETRIC TO ASYMMETRIC IgG SWITCH BY A SWITCHING FACTOR

The ability of the ascitic fluids from ovarian cancer patients to alter the distribution of symmetric and asymmetric antibodies in hybridomas was analyzed. The ascitic fluids were centrifugated to remove cells and other debris and were filtered through both 0.8m and 0.45m filters. Aliquots of these ascitic fluids were added to cultures of hybridoma 9F12 at concentration ranging from 0 to 50% (v/v) . Hybridoma 9F12 is a mouse-human hybrid producing human IgG₂. The hybridoma cultures were incubated for 7 days. After 7 days, aliquots were removed to quantitate the levels of symmetric and aberrantly glycosylated IgG. Total IgG and aberrantly glycosylated IgG were quantitated using a standard commercial radial immunodiffusion kit for mouse IgG, as described above. The percentage of aberrantly glycosylated IgG among total IgG was calculated (Figure 15) . The results indicated a

"switching" activity present in the ascitic fluids of these ovarian cancer patients which led to an increase in the percentage of aberrantly glycosylated IgG among total IgG produced by hybridomas.

The switching activity was also detected in the conditioned media of the tumor cells obtained from ovarian cancer patients and cultured in vitro. Thus, the switching factor was produced by tumor cells, rather than by other cells under the influence of tumor cells.

The switching factor was isolated from ascites or conditioned media through a series of chromatography. The procedure is graphically depicted in Figure 16. Fractions collected after each step of purification were analyzed in bioassays as described above, and the fractions having switching activities were collected for further purification.

A protein component was identified from a single peak fraction off a RP-HPLC microbore C8 column.

The protein SDS-PAGE and the single band appeared to possess a molecular weight of about 26kD-36kD estimated from reduced SDS-PAGE. By MALDI on a sinapini acid matrix (non-reducing mass spectrometry) , a single mass component at 67,315 was found, with an apparent doubly charged ion at 33,534 D. These findings indicated that the protein exists as a dimmer held together by disulfide bonding.

The protein was subject to proteolytic digestion and was found to be trypsin and chymotrypsin resistant. Subsequent digestion with thermolysin at 65°C resulted in the release of two peptide fragments. Peptide sequencing analysis revealed the sequence of such peptides to be FPLEAI (SEQ ID NO: 1) and IPSLADD (SEQ ID NO: 2) . Incubation of the protein with pronase at 65°C results in peptide fragments, the sequences of which were determined to be KTYLELYADX_aaLI (X_aa: unsure) (SEQ ID NO:3), [F or I]EVA[H or E] RFK (SEQ ID NO:4), [E or A] X_aaQRFYVAKG (X_aa: can be Leu or a modified amino acid) (SEQ ID NO: 5), EQHVKLVNEVTEFA (SEQ ID NO: 6), [A or

F]QNNGD[F or I] ADIIRKX_aalDX_aa2RL (X_aal : unsure; X_aa2 can be a modified amino acid) (SEQ ID NO:7), VEGA[D,E or V] GT [K or L] [V or Q] (SEQ ID NO: 8) , VEGA[D,E or V] [G or T] [K or L] [V or Q] [N or T] DEX_aal [Q or A] EVX_aa2GEYTTX_aa3X_aa4 [L or V] (X_aal, X_aa2 and X_aa4 : unsure; X_aa3 can be a modified amino acid) (SEQ ID NO: 9) and TFHADIFTLS (SEQ ID NO: 10).

A Blast Search of the GenBank database revealed that SEQ ID NO: 6 is 92% identical to amino acids encoded by nucleotide #157 to 198 of Macaca mulatta serum albumin mRNA (gb M90463.1) . No similarities were found to any other peptide fragments

EXAMPLE 9 PRODUCTION OF ANTIBODIES SPECIFIC FOR ABERRANTLY GLYCOSYLATED IGG

Antigen (IgG) isolation: IgG from the ascites and sera was isolated using DEAE-Affi-Gel (Bio- Rad Labs) . The DEAE-Affi-Gel column (7ml bed volume/ml serum) was equilibrated with 5 volumes of loading buffer (20mM Tris-HCl, 30mM NaCl, pH 8.0) at 4°C. After the samples were dialyzed against loading buffer for 36 hours, they were applied to the column. Unbound material was eluted with 3 bed volumes of loading buffer. The bound IgG fraction was eluted isocratically with 20mM Tris-HCl, 50ιrιM NaCl, pH 8.0 at a flow rate of lOml/hr. Fractions were collected and the initial peak containing IgG was retained, as determined by radial immunodiffusion (RID) . These IgG containing fractions were pooled and concentrated using an Amicon centrifugal concentrator with a lOOkD cutoff (to remove transferrin, Mr 76kD, which co-elutes with

IgG) . IgG was quantitated by protein assay and purity assessed by SDS-PAGE.

Preparation of Fc and F(ab')₂ fragments:

After separation of IgG, IM citrate buffer, pH3.5 was added and the final pH was adjusted to 3.5 with ImM

HC1. The IgG was digested by adding 125ml of a solution containing lmg/ml pepsin-conjugated to agarose. After digesting for 18hr at 37°C, the reaction was stopped by addition of IM Tris to a pH of 7.5. This material was dialyzed against 5mM Tris-HCl, pH7.5. The digested material was applied to a Protein A column, equilibrated with 5mM Tris-HCl. The unbound fraction contained the F(ab')₂ fraction and the bound material cotained the Fc portion. The Fc fragments were eluted from the Protein A with ImmunoPurea IgG elution buffer (Pierce, Rockford, IL) . The purity of these fractions was assessed by SDS-PAGE.

Separation of aberrantly glycosylated Fc fragments by affinity chromatography with ConA- Sepharose: Eluted Fc fragments were applied to ConA-

Sepharose columns (0.7x5cm) equilibrated with 25mM Tris-HCl, pH7.2 containing 0.2M NaCl, 3mM CaCl₂, 3mM MgCl₂ and 3mM MnCl₂. Samples will be incubated overnight at 4°C and unbound protein was be eluted with 25mM Tris-HCl, pH7.2 containing 0.2M NaCl, 3mM CaCl₂,

3mM MgCl₂ and 3mM MnCl₂. Elution of specifically bound Fc fragments was performed with 0. IM a-methyl mannoside .

Production of monoclonal antibodies: Monoclonal antibodies against the isolated aberrantly glycosylated Fc fragments were produced in mice by intradermal immunization. All animal work was performed by the Maine Biotechnology Services. This production was divided into 4 phases: immunization, fusion, subcloning and antibody production. Initially,

5 mice were serially immunized with aberrantly glycosylated Fc fragments. After each 5-week period, serum was obtained from each mouse to identify the presence and titer of response. After the initial immunization, the mice were boosted 4 additional times.

The 2 mice exhibiting the highest titer were sacrificed and their spleens removed. Single cell suspensions were generated from the spleens and these were fused with F/0 myeloma cells. The presence of antibodies exhibiting positive reactivity with the intact aberrantly glycosylated IgG was assessed by ELISA, described below. Positive fractions were additionally tested for reactivity with symmetric (or nomrally glycosylated) IgG. Seventeen positive antibody- secreting fusion products (positive for aberrantly glycosylated IgG and negative for normal Ig_G) were selected and expanded. Based on these reactivities, products can be subcloned to monoclonality.

Enzyme linked ixnmunosorbent assay (ELIS_A) for reactive antibodies: To quantitate the immunoreactivity of the monoclonal antibodies produce_d against the aberrantly glycosylated Fc fragment, an ELISA technique was established to assess the binding of hybridoma-derived antibodies to the aberrantly glycosylated IgG molecule. The aberrantly glycosylated IgG was equilibrated in coupling buffer, consisting of lOOmM sodium carbonate and 0.5M NaCl, pH8.3 and added to the wells of a 96-well MaxiSorp microtiter plate ⁽Nalge-Nunc International, Naperville, IL) . The plates were subsequently washed and then blocked with SuperBlock (Pierce Chemical Co., Rockford, IL) for 1 hour. Hybridoma culture media (100ml) from the above 17 fusion products were added to quadruplicate wells. The plates were washed and incubated with alkaline phosphate-conjugated anti-mouse IgG. After washing 3 times, the level of bound antibody was determined by incubating the wells with a colorimetric substrate and reading the absorbance at 450nm. The results are formulated in Table IV and graphically depicted in

Figure 17.

Table IV

Example 10 Effector Activity of Antibodies

To assay the effector activity of the aberrant versus normal antibody populations, production of aberrantly glycosylated IgG was induced in the OKT3 hybridoma. IgG was isolated by chromatography on Protein A-Sepharose. Subsequently, the normal and aberrantly glycosylated IgG were separated by additional chromatography on ConA-Sepharose, as previously described. The IgG solutions were assayed for protein and diluted to identical concentrations. Serial dilutions of the IgG solutions were incubated with Jurkat cells (10⁶ cells) for 45 minutes at 4°C. The cells were washed with PBS and then incubated for 30 minutes with 1ml guinea pig serum (as complement source) . Cytotoxicity was measured by CytoTox96 assay kit (Promega) . The results are shown in Figure 18.

Claims

WE CLAIM :

1. A method of diagnosing a cancer in a patient, comprising measuring the proportion of aberrantly glycosylated IgG in total IgG in the serum of said patient, and determining an increase in said patient relative to a normal individual as indicative of a cancer in said patient .

2. A method of diagnosing a cancer in a patient, comprising measuring the level of aberrantly glycosylated IgG in the serum of said patient, and determining an increase in said level relative to a normal individual as indicative of a cancer in said patient .

3. A method of diagnosing a cancer in a patient, comprising isolating tumor-reactive IgG from the patient serum, and determining an increased ratio of aberrantly glycosylated IgG versus normal IgG in the tumor-reactive IgG as indicative of a cancer in said patient .

4. The method according to any of claims 1- 3, further comprising determining the type of the cancer in said patient.

5. The method of claim 4, wherein the type of said cancer is ovarian cancer, endometrial cancer, breast cancer, prostate cancer, or colon cancer.

6. The method according to any of claims 1- ₃, wherein the level of aberrantly glycosylated IgG in the serum is determined by using Con-A sepharose, SDS P_AGE, or an anti-aberrant IgG antibody.

7. A method of monitoring the progression of a cancer, comprising monitoring the change in the ratio of aberrantly glycosylated IgG versus total IgG in t_he serum of said patient.

8. A method of monitoring the progression of a cancer, comprising monitoring the change in the level of aberrantly glycosylated IgG in the serum of said patient.

9. A method of monitoring the progression of a cancer, comprising monitoring the change in the ratio of aberrantly glycosylated IgG versus normal IgG among tumor-reactive IgG.

10. The method according to any of claims 7- 9 , wherein said cancer is selected from the group consisting of ovarian cancer, endometrial cancer, breast cancer, prostate cancer and colon cancer.

11. The method according to any of claims 7- 9, wherein the level of aberrantly glycosylated IgG is determined by using Con-A sepharose, SDS PAGE, or an anti-aberrant IgG antibody.

12. A method of treating a cancer in a patient, comprising eliminating aberrantly glycosylated IgGs from the patient serum.

13. The method of claim 12, wherein said aberrantly glycosylated IgGs are eliminated from the patient serum with Con-A or an antibody specific for said aberrantly glycosylated IgGs.

14. A method of treating a cancer in a patient, comprising removing the aberrant glycosylation on the aberrantly glycosylated IgGs in the patient serum.

15. A method of treating a cancer in a patient, comprising administering to said patient a therapeutically effective amount of autologous aberrantly glycosylated IgGs conjugated with an antitumor agent.

16. A method of treating a cancer in a patient, comprising administering to said patient a therapeutically effective amount of an antibody specific for aberrantly glycosylated IgGs.

17. A method according to any one of claims 12-16, wherein said cancer is selected from the group consisting of ovarian cancer, endometrial cancer, colon cancer, prostate cancer and breast cancer.

18. An isolated antibody specific for aberrantly glycosylated IgGs.

19. The antibody of claim 18, wherein said antibody is a monoclonal antibody (ATCC #) .

20. The antibody of claim 18, wherein said antibody is a polyclonal antibody.

21. An isolated switching factor, wherein said factor is capable of inducing the production of aberrantly glycosylated IgGs from antigen-producing cells, and wherein said factor has a Mw of about 67kD and comprises two subunits each of about Mw 26-36 kD.

22. The switching factor of claim 21, comprising at least one of SEQ ID NOS: 1-10.

23. An isolated antibody specific for the switching factor of claim 21.

24. An isolated antibody specific for a peptide comprising any of SEQ ID NOS: 1-10.

25. A method of diagnosing a cancer in a patient, comprising detecting the presence of said factor of claim 21 in the serum of said patient.

26. A method of treating a cancer in a patient, comprising inhibiting the switching activity of said factor of claim 21 in said patient.