Purification and characterization of a novel restricted antigen expressed by normal and transformed human colonic epithelium

Nanobody-Based EGFR-Targeting Immunotoxins for Colorectal Cancer Treatment

Immunotoxins (ITXs) are chimeric molecules that combine the specificity of a targeting domain, usually derived from an antibody, and the cytotoxic potency of a toxin, leading to the selective death of tumor cells. However, several issues must be addressed and optimized in order to use ITXs as therapeutic tools, such as the selection of a suitable tumor-associated antigen (TAA), high tumor penetration and retention, low kidney elimination, or low immunogenicity of foreign proteins. To this end, we produced and characterized several ITX designs, using a nanobody against EGFR (V_HH 7D12) as the targeting domain. First, we generated a nanoITX, combining V_HH 7D12 and the fungal ribotoxin α-sarcin (αS) as the toxic moiety (V_HHEGFRαS). Then, we incorporated a trimerization domain (TIE^XVIII) into the construct, obtaining a trimeric nanoITX (TriV_HHEGFRαS). Finally, we designed and characterized a bispecific ITX, combining the V_HH 7D12 and the scFv against GPA33 as targeting domains, and a deimmunized (DI) variant of α-sarcin (BsITXαSDI). The results confirm the therapeutic potential of α-sarcin-based nanoITXs. The incorporation of nanobodies as target domains improves their therapeutic use due to their lower molecular size and binding features. The enhanced avidity and toxic load in the trimeric nanoITX and the combination of two different target domains in the bispecific nanoITX allow for increased antitumor effectiveness.

A New Optimized Version of a Colorectal Cancer-Targeted Immunotoxin Based on a Non-Immunogenic Variant of the Ribotoxin α-Sarcin

Due to its incidence and mortality, cancer remains one of the main risks to human health and lifespans. In order to overcome this worldwide disease, immunotherapy and the therapeutic use of immunotoxins have arisen as promising approaches. However, the immunogenicity of foreign proteins limits the dose of immunotoxins administered, thereby leading to a decrease in its therapeutic benefit. In this study, we designed two different variants of non-immunogenic immunotoxins (IMTXA33αSDI and IMTXA33furαSDI) based on a deimmunized variant of the ribotoxin α-sarcin. The inclusion of a furin cleavage site in IMTXA33furαSDI would allow a more efficient release of the toxic domain to the cytosol. Both immunotoxins were produced and purified in the yeast Pichia pastoris and later functionally characterized (both in vitro and in vivo), and immunogenicity assays were carried out. The results showed that both immunotoxins were functionally active and less immunogenic than the wild-type immunotoxin. In addition, IMTXA33furαSDI showed a more efficient antitumor effect (both in vitro and in vivo) due to the inclusion of the furin linker. These results constituted a step forward in the optimization of immunotoxins with low immunogenicity and enhanced antitumor activity, which can lead to potential better outcomes in cancer treatment.

Multistep Compositional Remodeling of Supported Lipid Membranes by Interfacially Active Phosphatidylinositol Kinases

The multienzyme catalytic phosphorylation of phosphatidylinositol (PI) in a supported lipid membrane platform is demonstrated for the first time. One-step treatment with PI 4-kinase IIIβ (PI4Kβ) yielded PI 4-phosphate (PI4P), while a multistep enzymatic cascade of PI4Kβ followed by PIP 5-kinase produced PI-4,5-bisphosphate (PI(4,5)P2 or PIP2). By employing quartz crystal microbalance with dissipation monitoring, we were able to track membrane association of kinase enzymes for the first time as well as detect PI4P and PI(4,5)P2 generation based on subsequent antibody binding to the supported lipid bilayers. Pharmacologic inhibition of PI4Kβ by a small molecule inhibitor was also quantitatively assessed, yielding an EC50 value that agrees well with conventional biochemical readout. Taken together, the development of a PI-containing supported membrane platform coupled with surface-sensitive measurement techniques for kinase studies opens the door to exploring the rich biochemistry and pharmacological targeting of membrane-associated phosphoinositides.

A hypermorphic epithelial β-catenin mutation facilitates intestinal tumorigenesis in mice in response to compounding WNT-pathway mutations

Activation of the Wnt/β-catenin pathway occurs in the vast majority of colorectal cancers. However, the outcome of the disease varies markedly from individual to individual, even within the same tumor stage. This heterogeneity is governed to a great extent by the genetic make-up of individual tumors and the combination of oncogenic mutations. In order to express throughout the intestinal epithelium a degradation-resistant β-catenin (Ctnnb1), which lacks the first 131 amino acids, we inserted an epitope-tagged ΔN(1-131)-β-catenin-encoding cDNA as a knock-in transgene into the endogenous gpA33 gene locus in mice. The resulting gpA33(ΔN-Bcat) mice showed an increase in the constitutive Wnt/β-catenin pathway activation that shifts the cell fate towards the Paneth cell lineage in pre-malignant intestinal epithelium. Furthermore, 19% of all heterozygous and 37% of all homozygous gpA33(ΔN-Bcat) mice spontaneously developed aberrant crypt foci and adenomatous polyps, at frequencies and latencies akin to those observed in sporadic colon cancer in humans. Consistent with this, the Wnt target genes, MMP7 and Tenascin-C, which are most highly expressed in benign human adenomas and early tumor stages, were upregulated in pre-malignant tissue of gpA33(ΔN-Bcat) mice, but those Wnt target genes associated with excessive proliferation (i.e. Cdnn1, myc) were not. We also detected diminished expression of membrane-associated α-catenin and increased intestinal permeability in gpA33(ΔN-Bcat) mice in challenge conditions, providing a potential explanation for the observed mild chronic intestinal inflammation and increased susceptibility to azoxymethane and mutant Apc-dependent tumorigenesis. Collectively, our data indicate that epithelial expression of ΔN(1-131)-β-catenin in the intestine creates an inflammatory microenvironment and co-operates with other mutations in the Wnt/β-catenin pathway to facilitate and promote tumorigenesis.

Preparation of an engineered safer immunotoxin against colon carcinoma based on the ribotoxin hirsutellin A

Immunotoxins are chimeric proteins composed of an antibody domain that specifically directs the action of the toxic domain, resulting in the death of the targeted cells. Over recent years, immunotoxins have been widely studied and the number of different constructions has increased exponentially. Protein engineering has allowed the design of optimized versions of immunotoxins with an improved tumor binding affinity, stability or cytotoxic efficacy, although sometimes this has compromised the safety of the patient in terms of undesirable adverse secondary reactions. A triple mutant at three Trp residues (HtA3ΔW) of the ribotoxin hirsutellin A retains its specific ribonucleolytic activity, although cell internalization capacity is lacking. This toxin variant has been fused to the single chain variable fragment A33 (scFvA33). This immunoconjugate (IMTXA33HtA3ΔW) was produced in the methylotrophic yeast Pichia pastoris and purified using nickel-nitrilotriacetic acid affinity chromatography. Both target and toxic domains were characterized. The immunotoxin showed an exquisite specific binding against GPA33-positive culture cells, which results in the death of the targeted cells because of specific ribonucleolytic activity against ribosomes of the engineered hirsutellin A variant. IMTXA33HtA3ΔW represents a promising structure in the search for an improved immunotoxin without compromising the safety of patients.

Quantitative intratumoural microdistribution and kinetics of (131)I-huA33 antibody in patients with colorectal carcinoma

BACKGROUND

The ability of recombinant antibodies to adequately penetrate into tumours is a key factor in achieving therapeutic effect; however, the behaviour of antibodies at a cellular level in tumours is poorly understood. The purpose of this study was to investigate those factors that influence the macroscopic and microscopic intratumoural distribution of an IgG1-humanized antibody, huA33, in colorectal tumours.

METHODS

Twelve patients were infused with radiolabelled huA33 at 7 days prior to elective surgery for colorectal carcinoma. Macroscopic huA33 uptake was determined by both gamma well counter and autoradiography measurements of the resected tumour specimens. Microscopic uptake was then quantitated at a cellular level and compared to vascular penetrance. The impact of variation in tumour antigen (GPA33) expression, tumour size, specimen type (primary vs metastatic), presence of macroscopic necrosis, and tumour vasculature on huA33 uptake were examined.

RESULTS

The I-huA33 uptake in whole tumour sections was (mean ± SD) 5.13 ± 2.71 × 10(-3)% injected dose per gram (ID/g). GPA33 was expressed in all viable tumour cells, and huA33 uptake was excellent regardless of tumour size and specimen type. In tumours with macroscopically evident central necrosis (n = 5), huA33 uptake in tumour necrotic centres was lower than in viable peripheries (0.606 ± 0.493 vs 2.98 ± 2.17 × 10(-3)%ID, p = 0.06). However, when corrected for low cell viability in necrotic centres, uptake of huA33 at the cellular level was highly comparable to that in the more viable tumour periphery (7.10 ± 5.10 × 10(-9) vs 3.82 ± 3.67 × 10(-9)%ID/cell, p = 0.4). In the five patients who exhibited macroscopic necrosis in their tumours, huA33 showed excellent tissue penetration, with a maximum penetration distance of 26 μm in peripheral tumour regions and 118 μm in central regions. No correlation was observed between (131)I-huA33 uptake in tumour on a cellular basis and tumour vascularity.

CONCLUSIONS

In patients with colorectal carcinoma, monoclonal antibody huA33 effectively targets viable tumour cells in all cellular milieus examined, including effective penetration into necrotic tumour centres, a novel and therapeutically important finding.

Two distinct populations of exosomes are released from LIM1863 colon carcinoma cell-derived organoids

Exosomes are naturally occurring biological nanomembranous vesicles (∼40 to 100 nm) of endocytic origin that are released from diverse cell types into the extracellular space. They have pleiotropic functions such as antigen presentation and intercellular transfer of protein cargo, mRNA, microRNA, lipids, and oncogenic potential. Here we describe the isolation, via sequential immunocapture using anti-A33- and anti-EpCAM-coupled magnetic beads, of two distinct populations of exosomes released from organoids derived from human colon carcinoma cell line LIM1863. The exosome populations (A33-Exos and EpCAM-Exos) could not be distinguished via electron microscopy and contained stereotypical exosome markers such as TSG101, Alix, and HSP70. The salient finding of this study, revealed via gel-based LC-MS/MS, was the exclusive identification in EpCAM-Exos of the classical apical trafficking molecules CD63 (LAMP3), mucin 13 and the apical intestinal enzyme sucrase isomaltase and increased expression of dipeptidyl peptidase IV and the apically restricted pentaspan membrane glycoprotein prominin 1. In contrast, the A33-Exos preparation was enriched with basolateral trafficking molecules such as early endosome antigen 1, the Golgi membrane protein ADP-ribosylation factor, and clathrin. Our observations are consistent with EpCAM- and A33-Exos being released from the apical and basolateral surfaces, respectively, and the EpCAM-Exos proteome profile with widely published stereotypical exosomes. A proteome analysis of LIM1863-derived shed microvesicles (sMVs) was also performed in order to clearly distinguish A33- and EpCAM-Exos from sMVs. Intriguingly, several members of the MHC class I family of antigen presentation molecules were exclusively observed in A33-Exos, whereas neither MHC class I nor MHC class II molecules were observed via MS in EpCAM-Exos. Additionally, we report for the first time in any extracellular vesicle study the colocalization of EpCAM, claudin-7, and CD44 in EpCAM-Exos. Given that these molecules are known to complex together to promote tumor progression, further characterization of exosome subpopulations will enable a deeper understanding of their possible role in regulation of the tumor microenvironment.

Production and characterization of a colon cancer-specific immunotoxin based on the fungal ribotoxin α-sarcin

A single-chain fusion protein that directed the cytolytic activity of α-sarcin to A33 tumor antigen expressing cells was constructed and shown to effectively kill targeted cells. Glycoprotein A33 (GPA33) is a well-known colon cancer marker and a humanized antibody against it was used to target the α-sarcin. The fungal ribotoxin α-sarcin is one of the most potent and specific toxins known. It is small, protease resistant, thermostable and highly efficient towards the inactivation of ribosomes. This work describes the production and characterization of an immunotoxin resulting from fusing the single-chain variable fragment (scFv) of the monoclonal antibody that targets GPA33 to fungal α-sarcin. This chimeric protein (scFvA33αsarcin), produced in Pichia pastoris and purified in high yield was proven to be properly folded, active, specific and stable. It showed high specific toxicity against GPA33-positive tumoral cell lines providing scientific evidence to sustain that scFvA33αsarcin is a good immunotherapeutic candidate against GPA33-positive colon carcinomas.

The discovery and validation of colorectal cancer biomarkers

Colorectal cancer is currently the third most common malignancy in the world. Patients have excellent prognosis following surgical resection if their tumour is still localized at diagnosis. By contrast, once the tumour has started to metastasize, prognosis is much poorer. Accurate early detection can therefore significantly reduce the mortality from this disease. However, current tests either lack the required sensitivity and selectivity or are costly and invasive. Improved biomarkers, or panels of biomarkers, are therefore urgently required. We have addressed current screening strategies and potential protein biomarkers that have been proposed. The role of both discovery and hypothesis-driven proteomics approaches for biomarker discovery and validation is discussed. Using such approaches we show how multiple reaction monitoring (MRM) can be successfully developed and used for quantitative multiplexed analysis of potential faecal biomarkers.

The therapeutic value of monoclonal antibodies directed against immunogenic tumor glycoproteins

Monoclonal antibodies developed against immunogenic proteins (Tumor Specific Antigens/TSA's) that are expressed in human cancers, display a unique behavioral pattern. They appear to serve in a dual role. This includes the early recognition of these immunogenic membrane proteins that can serve as diagnostic markers, and the targeting of such markers for the destruction of the tumor, primarily thru ADCC.

The monoclonals (mAbs) that we have developed against specific immunogenic tumor membrane proteins have been studied in detail. These tumor proteins, when first defined, were referred to as tumor associated antigens. With the ability of the mAbs to demonstrate therapeutic antitumor activity in those patients with relatively advanced malignancies, the term tumor specific was introduced. Monoclonals that we were able to develop from tumor specific proteins derived from colon and pancreas cancer were found capable of targeting those tumors to induce apoptosis. We were also able to define immunogenic membrane proteins from lung (squamous and adenoCa) as well as prostate neoplasms. Monoclonals developed from these tumor antigens are in the initial phases of investigation with regard to their specificity and antitumor activity.

Mabs capable of targeting the malignancies noted above were produced following immunization of BALBc mice with the Tumor Specific Antigens. The hybridomas that were screened and found to express the antibodies of interest appeared for the most part as IgG2a's. It became apparent after a short period of time that stability of the Fab CDR loops as well as the therapeutic efficacy of the hybridoma mAbs could be lost. Stability was achieved by chimerization and or humanization. The resulting mAbs were found to switch their isotypes to an IgG1 subsequent to chimerization and or humanization, when expressed in CHO cells. The monoclonals, so produced, were not only more efficient in controlling tumor growth but minimized the development of a HAMA response.

Because of 1) the specificity of this group of monoclonal antibodies in targeting well defined immunogenic proteins that were expressed on the tumor cell membrane,2) their lack of cross reactivity to normal tissue, 3) relatively low toxicity when delivered intravenously, 4) rapid targeting of tumor cell populations (4-6 hrs in vitro) and their 5) ability to destroy xenograft transplants (in vivo) within days of delivery, these mAbs were felt to be ideal for possible use in the treatment of patients with recurrent and or metastatic tumors.

Initial clinical studies have been planned for following the filing of an IND. It is required by FDA that the potential effects of tumor control and toxicity be defined using the naked antibodies produced under GMP conditions, In those situations where patients with recurrent malignancies are to be studied we have come to realize that a number of factors can influence the response to monoclonal therapy. This includes the amount of shed antigen in the serum at the time of treatment that could initiate immune complex formation as well as the shedding of inhibitory material into the serum possibly effecting an immune response. As such we plan to eventually employ the therapeutic mAbs in combination with chemotherapy as a means of enhancing the immunogenicity of the tumor system being treated and to possibly weaken the malignant growth for easier destruction by the mAb. We will also look at the combination of mAbs with immunostimulants such as GMCSF and IL-2 (fusion proteins) and eventual conjugation of the mAbs with alpha and possibly B-emitters to help in targeting bystander cells. The present paper reviews the potential therapeutic value of such mAbs in the treatment of recurrent malignancies, especially those having failed chemotherapy in established clinical trials.

KLF4-dependent, PPARgamma-induced expression of GPA33 in colon cancer cell lines

The glycoprotein A33 (GPA33) is a colon cancer antigen. Phase I trials with 131I and 125I monoclonal antibody A33 in colon carcinoma patients showed excellent localization to colorectal cancer and some evidence of tumor response. Using DNA microarrays, we have identified the GPA33 gene as a target of PPARgamma in HT29-Cl.16E colon cancer cells. Treatment of HT29-Cl.16E, Caco2, SW1116 and LS174T colon cancer cells with the PPARgamma agonist GW7845 induced a 2- to 6-fold increase in GPA33 mRNA as determined by real-time PCR. This induction was also found in HT29-Cl.16E cells treated with rosiglitazone and ciglitazone and was prevented by cotreatment with the PPARgamma antagonist GW9662, indicating that this regulation was PPARgamma dependent. No canonical PPAR responsive element was found in the GPA33 promoter. We therefore analyzed the expression of transcription factors involved in GPA33 expression. CDXl, CDX2 and KLF5 expression was not modified by PPARgamma activation. By contrast, a significant increase in KLF4 was seen, both at mRNA and protein levels. Furthermore, chromatin immunoprecipitation studies demonstrated that an increased amount of KLF4 protein was bound to the GPA33 promoter in cells treated with rosiglitazone. Finally, downregulation of KLF4 expression by siRNA reduced rosiglitazone-induced GPA33 expression. This indicates that PPARgamma activation induces KLF4 expression, which in turn increases GPA33 expression. We also demonstrate that PPARgamma activation leads to increased (p21WAF1/Cip1 and keratin 19) or decreased (cyclin D1) expression of known KLF4 targets, suggesting that KLF4 is a nodal player in a network of PPARgamma-regulated genes.

Murine fecal proteomics: a model system for the detection of potential biomarkers for colorectal cancer

Tumor related products shed into the feces offer a potential source of biomarkers for the detection of colorectal cancer (CRC). Using SDS-PAGE followed by nanoflow reversed-phased LC-MS/MS to analyse fecal samples from Apc(Min/+) mice (that develop spontaneous multiple intestinal neoplasia with age) we have identified 336 proteins (115 proteins of murine origin, 201 from fecal bacteria, 18 associated with food intake and 2 of apparent parasitic origin). 75% of the murine proteins identified in this study are predicted to be extracellular or associated with the cell plasma membrane. Of these proteins, a number of the murine homologues of colorectal cancer associated proteins (CCAP) such as hemoglobin, haptoglobin, hemopexin, alpha-2-macroglobulin and cadherin-17 have been identified, demonstrating the potential of fecal proteomics for detecting potential biomarkers and paving the way for subsequent MS/MS based biomarker studies on similar human samples.

Clearance kinetics and external dosimetry of 131I-labeled murine and humanized monoclonal antibody A33 in patients with colon cancer: radiation safety implications

The monoclonal antibody (mAb) A33 detects a membrane antigen that is expressed on greater than 95% of metastatic human colorectal cancers. Previous studies have shown excellent tumor-targeting of (131)I-labeled murine and humanized forms of the mAb. A retrospective analysis of whole-body clearance in the murine form was performed for comparison to the humanized form. Serial whole-body dose rate measurements were obtained for 55 treatments on 30 patients participating in phase I/II dose escalation studies of therapeutic (131)I-murine A33 mAb. Whole-body retention fractions over time were derived. Each treatment was fit with exponential curves to determine the effective half-lives and corresponding clearance fractions. There was a large variability in the calculated mono-exponential clearance effective half-life time, with a mean value of 36.5 h +/- 8.5 h. A bi-exponential fit of all combined data shows that 60% of the administered dose rapidly clears with a biological half-time of 23.9 h and 40% clears with a slower biological half-time of 101.2 h. The whole-body clearance proved to be more rapid in the murine form when compared with recent studies on the humanized form of radiolabeled A33 mAb. The variability in whole-body clearance reinforces the need for patient-specific tracer dosimetry for clinical care and radiation safety precautions. In addition, the slower clearance of the humanized form of the A33 mAb requires longer term radiation safety precautions than the earlier murine form. As other monoclonal antibodies progress from murine to humanized forms, radiopharmacokinetics should be evaluated for clinical and radiation safety implications.

Monoclonal antibody CC188 binds a carbohydrate epitope expressed on the surface of both colorectal cancer stem cells and their differentiated progeny

PURPOSE

Recently, cancer stem cells (CSC), undifferentiated cancer progenitor cells, have been successfully isolated from colorectal tumors. Targeting both CSCs and differentiated, rapidly proliferating tumor cells with therapeutic drugs provides a focused strategy to treat cancer. In this study, we isolated the monoclonal antibody (mAb) CC188 and characterized the epitope recognized by mAb CC188, which is useful for developing biological reagents that target CSCs.

EXPERIMENTAL DESIGN

We used a hybridoma technique to generate mAbs and an immunomagnetic method to isolate colon CSCs. We characterized mAb CC188 binding epitope and examined the epitope distribution in normal and tumor tissues, particularly in CSCs using tissue arrays and immunofluorescence staining method. We also evaluated the effect of mAb CC188 on invasiveness of NSY tumor cells.

RESULTS

mAb CC188 was generated and 98.9% (187 of 189 cases) of colon cancer were positively stained by mAb CC188. "+", "++," and "+++" staining were 25.9%, 28.6%, and 43.4%, respectively. The mAb CC188 binding epitope was identified as a carbohydrate, which was expressed on the surface of colon CSCs (CD133+), differentiated colon cancer cells (CD133-), and cells from various types of epithelial tumors. In contrast, the expression of the carbohydrate epitope was low in normal prostate muscle and pancreatic acinar cells, as well as in some normal epithelial cells of the breast duct, cervix, and skin. A functional study indicated that mAb CC188 suppressed the invasiveness of NSY tumor cells.

CONCLUSION

mAb CC188 selectively targets a carbohydrate epitope expressed on cancer cells, providing a viable method for specific tumor imaging and targeted therapy.

Surface expression of gpA33 is dependent on culture density and cell-cycle phase and is regulated by intracellular traffic rather than gene transcription

The cell-surface marker, gpA33, a new member of the immunoglobulin superfamily, is expressed by gastrointestinal cells and by 95% of colon cancers. It has become a promising target of immunologic therapy strategies, but its biologic function and potential role in tumorigenesis are unknown. In this study, we have investigated the expression of gpA33 on the mRNA and cell-surface protein levels by quantitative reverse transcriptase polymerase chain reaction and flow cytometry, respectively, in response to cell density in the culture and to cell-cycle arrest in the G1, S, or G2/M phases. As internalization of the surface protein had previously been reported, we also investigated the binding and intracellular migration of an anti-gpA33 fluobody with green fluorescent protein (A33scFv::GFP) by laser confocal microscopy. Contrary to intuition, we found that gpA33 surface expression and mRNA levels do only partly correlate under the conditions tested. Dependent on cell density in culture, gpA33 surface expression peaked at the point of confluence. Dependent on cell-cycle phase, it peaked in the G2/M phase but was lowest in the S phase, whereas mRNA levels were highest in S, but almost absent in G1. Laser confocal microscopy clearly demonstrated the intracellular uptake of A33scFv::GFP and showed the migration of microvesicles over time. These findings are, in part, concordant with the putative role of gpA33 as an adhesion molecule. However, intracellular traffic and recycling to the cell surface appear to play a major role in its function and to have an influence on its surface density superseding translational regulation.

Use of multidimensional separation protocols for the purification of trace components in complex biological samples for proteomics analysis

The routine detection of low abundance components in complex samples for detailed proteomics analysis continues to be a challenge. Whilst the potential of multidimensional chromatographic fractionation for this purpose has been proposed for some years, and was used effectively for the purification to homogeneity of trace components in bulk biological samples for N-terminal sequence analysis, its practical application in the proteomics arena is still limited. This article reviews some of the recent data using these approaches, including the use of microaffinity purification as part of multidimensional protocols for downstream proteomics analysis.

A phase I radioimmunolocalization trial of humanized monoclonal antibody huA33 in patients with gastric carcinoma

In order to determine the in vivo characteristics of huA33, an open label dose escalation biopsy-based phase I clinical trial and radioimmunolocalization study were conducted with a complement determinant region-grafted humanized monoclonal antibody against the A33 antigen in patients with gastric carcinoma. Thirteen patients were entered onto one of four dose levels (1.0, 2.0, 5.0 or 10.0 mg/m(2)). Patients with locally advanced (UICC-TNM [International Union Against Cancer-tumor, node, metastasis] stage over 2 but resectable at clinical diagnosis) gastric carcinoma received a single infusion of (131)I-huA33 1 week prior to surgery. Adverse events were monitored, and imaging studies with gamma camera plus ex vivo scintigraphy of the resected specimen, biodistribution study by dosimetry analysis of the biopsied and resected tissues, and immunohistochemical analysis were carried out and evaluated. No dose-limiting toxicity was observed during the trial. Therefore, the maximum tolerated dose was not reached. Although cancer tissues with + intensity and <25% extent by immunostaining in biopsied frozen sections did not show positive imaging or postoperative dosimetry findings, cancers with ++ or +++ intensity or wide (>25%) extent by frozen and paraffin sections in the biopsied specimen showed positive ex vivo tumor images and positive antigen expression in resected gastric cancer specimens, and the biodistribution analysis showed tumor uptake of (131)I-huA33. In conclusion, humanized monoclonal antibody huA33 demonstrated selective localization to gastric cancer that expressed A33 antigen strongly. These excellent targeting characteristics of huA33 indicate potential for targeted therapy of advanced gastric cancer that is refractory to cytotoxic therapy, and could also be exploitable for curatively resected early gastric cancer in an adjuvant setting.

Purification and characterization of a high specificity polyclonal antibody to the adenomatous polyposis coli tumour suppressor protein

Recombinant proteins, commonly expressed in fusion with an affinity tag to facilitate purification, are often used as immunogens for polyclonal antibody production. Careful immunopurification of the antibody product is often the key to obtaining a high-specificity polyclonal antibody against the protein domain of interest. This study describes the purification and characterization of such an antibody directed against the adenomatous polyposis coli (APC) tumour suppressor. We used a combination of affinity chromatography and biosensor analysis to optimize and monitor antibody purification. This antibody was then characterized by immunoprecipitation, proteomic analyses and immunofluorescence staining and shown to be a valuable reagent for the study of APC biology. Using this antibody we successfully isolated and identified APC, using MS/MS, from transfected cell lines. A novel phosphorylation site on APC was identified at ser 1436. Similar strategies involving multiple immuno-affinity steps coupled with surface plasmon resonance (SPR), immunoprecipitation proteomic and immunofluorescence analyses should be generally applicable for the purification and characterization of other polyclonal antibodies.

A phase I trial of humanized monoclonal antibody A33 in patients with colorectal carcinoma: biodistribution, pharmacokinetics, and quantitative tumor uptake

PURPOSE

To determine the in vivo characteristics of huA33, a CDR-grafted humanized antibody against the A33 antigen, we have conducted an open-label, dose escalation, biopsy-based phase I trial of huA33 in patients with colorectal carcinoma.

EXPERIMENTAL DESIGN

Patients with colorectal carcinoma were infused with [131I]huA33 (400 MBq: 10 mCi) and [125I]huA33 (40 MBq: 1 mCi) 1 week before surgery. There were four huA33 dose levels (0.25, 1.0, 5.0, and 10 mg/m2). Adverse events, pharmacokinetics, biodistribution, tumor biopsies, and immune responses to huA33 were evaluated.

RESULTS

There were 12 patients entered into the trial (6 males and 6 females; age range, 39-66 years). No dose-limiting toxicity was observed. The biodistribution of huA33 showed excellent uptake of [131I]huA33 in metastatic colorectal carcinoma. Pharmacokinetic analysis showed no significant difference in terminal half-life (T1/2beta) between dose levels (mean +/- SD, 86.92 +/- 22.12 hours). Modeling of colon uptake of huA33 showed a T1/2 of elimination of 32.4 +/- 8.1 hours. Quantitative tumor uptake ranged from 2.1 x 10(-3) to 11.1 x 10(-3) %ID/g, and tumor/normal tissue and tumor/serum ratios reached as high as 16.3:1 and 4.5:1, respectively. Biosensor analysis detected low-level human anti-human antibody responses in four patients following huA33 infusion.

CONCLUSIONS

huA33 shows selective and rapid localization to colorectal carcinoma in vivo and penetrates to the center of large necrotic tumors, and colon elimination half-life of huA33 is equivalent to basal colonocyte turnover. The excellent targeting characteristics of this humanized antibody indicate potential for the targeted therapy of metastatic colorectal cancer in future trials.

Biosensor-Based Micro-Affinity Purification for the Proteomic Analysis of Protein Complexes

A biosensor-based micro-affinity purification method to recover protein binding partners and their complexes for down stream proteomics analysis has been developed using the BIAcore 3000 fitted with a prototype Surface Prep Unit (SPU). The recombinant GST-intracellular domain of E-cadherin or the recombinant GST-beta-catenin binding domain of Adenomatous Polyposis Coli (APC) were immobilized onto the SPU and used to affinity purify binding partners from chromatographically enriched SW480 colon cancer cell lysates. A GST- immobilized surface was used as a control. Samples recovered from the SPU were subjected to SDS-PAGE with sensitive Coomassie staining followed by automated in-gel digestion and LC-MS/MS. The results obtained using the SPU were compared with similar experiments performed using Sepharose beads.

Transcriptional regulation of A33 antigen expression by gut-enriched Krüppel-like factor

A33 antigen is a membrane-bound protein that is expressed only in intestinal epithelium and in most human colon cancers. Thus, A33 antigen has been explored as a potential therapeutic target for the treatment of colon cancers. However, little is known about the mechanism responsible for the tissue-specific pattern of its expression. In this report, we demonstrate that gut-enriched Krüppel-like factor (GKLF) binds to the promoter region of A33 antigen gene in colonic carcinoma cells and that mutations in the GKLF binding sequence in this region lead to diminished expression of A33 antigen. In addition, the expression of GKLF is linked to the expression of A33 antigen and blocking the expression of GKLF leads to the abolishment of A33 antigen expression. These results suggest that GKLF is a critical regulator in inducing the expression of A33 antigen in intestinal epithelium. While it has been suggested that GKLF is a regulator in inducing cell growth arrest and differentiation of the intestine, our observation that A33 antigen gene is a downstream target for GKLF suggests a more complex and diverse role for GKLF in the gut.

Comprehensive proteomic analysis of breast cancer cell membranes reveals unique proteins with potential roles in clinical cancer

Proteins associated with cancer cell plasma membranes are rich in known drug and antibody targets as well as other proteins known to play key roles in the abnormal signal transduction processes required for carcinogenesis. We describe here a proteomics process that comprehensively annotates the protein content of breast tumor cell membranes and defines the clinical relevance of such proteins. Tumor-derived cell lines were used to ensure an enrichment for cancer cell-specific plasma membrane proteins because it is difficult to purify cancer cells and then obtain good membrane preparations from clinical material. Multiple cell lines with different molecular pathologies were used to represent the clinical heterogeneity of breast cancer. Peptide tandem mass spectra were searched against a comprehensive data base containing known and conceptual proteins derived from many public data bases including the draft human genome sequences. This plasma membrane-enriched proteome analysis created a data base of more than 500 breast cancer cell line proteins, 27% of which were of unknown function. The value of our approach is demonstrated by further detailed analyses of three previously uncharacterized proteins whose clinical relevance has been defined by their unique cancer expression profiles and the identification of protein-binding partners that elucidate potential functionality in cancer.

The use of coiled-coil interactions for the analysis and micropreparative isolation of adenomatous polyposis coli protein complexes

The coiled coil is a common structural motif found both as the dominant structure in fibrous proteins and as an oligomerization domain in a variety of cytoskeletal and extracellular matrix proteins. Coiled-coils typically consist of two to four helices that are supercoiled around one another in either parallel or antiparallel orientations. In the past few years our knowledge of the structure and specificity of coiled coil interactions has increased, allowing the de novo design and preparation of coiled-coils with well-defined structure and specificity. Indeed, the design and synthesis of a peptide that binds specifically to a single coiled-coil-containing protein, adenomatous polyposis coli (APC) has been reported. We have optimized solid-phase synthesis techniques to produce a modified form of the anti-APC peptide that contains a biotin moiety specifically placed so as to allow selective orientation onto the surface of a biosensor or affinity support. These peptide surfaces have been used to both monitor and purify APC and APC complexes from cellular extracts.

Biophysical characterization of interactions involving importin-alpha during nuclear import

Proteins containing the classical nuclear localization sequences (NLSs) are imported into the nucleus by the importin-alpha/beta heterodimer. Importin-alpha contains the NLS binding site, whereas importin-beta mediates the translocation through the nuclear pore. We characterized the interactions involving importin-alpha during nuclear import using a combination of biophysical techniques (biosensor, crystallography, sedimentation equilibrium, electrophoresis, and circular dichroism). Importin-alpha is shown to exist in a monomeric autoinhibited state (association with NLSs undetectable by biosensor). Association with importin-beta (stoichiometry, 1:1; K(D) = 1.1 x 10(-8) m) increases the affinity for NLSs; the importin-alpha/beta complex binds representative monopartite NLS (simian virus 40 large T-antigen) and bipartite NLS (nucleoplasmin) with affinities (K(D) = 3.5 x 10(-8) m and 4.8 x 10(-8) m, respectively) comparable with those of a truncated importin-alpha lacking the autoinhibitory domain (T-antigen NLS, K(D) = 1.7 x 10(-8) m; nucleoplasmin NLS, K(D) = 1.4 x 10(-8) m). The autoinhibitory domain (as a separate peptide) binds the truncated importin-alpha, and the crystal structure of the complex resembles the structure of full-length importin-alpha. Our results support the model of regulation of nuclear import mediated by the intrasteric autoregulatory sequence of importin-alpha and provide a quantitative description of the binding and regulatory steps during nuclear import.

Proteomic analysis of the human colon carcinoma cell line (LIM 1215): development of a membrane protein database

The proteomic definition of plasma membrane proteins is an important initial step in searching for novel tumor marker proteins expressed during the different stages of cancer progression. However, due to the charge heterogeneity and poor solubility of membrane-associated proteins this subsection of the cell's proteome is often refractory to two-dimensional electrophoresis (2-DE), the current paradigm technology for studying protein expression profiles. Here, we describe a non-2-DE method for identifying membrane proteins. Proteins from an enriched membrane preparation of the human colorectal carcinoma cell line LIM1215 were initially fractionated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE, 4-20%). The unstained gel was cut into 16 x 3 mm slices, and peptide mixtures resulting from in-gel tryptic digestion of each slice were individually subjected to capillary-column reversed phase-high performance liquid chromatography (RP-HPLC) coupled with electrospray ionization-ion trap-mass spectrometry (ESI-IT-MS). Interrogation of genomic databases with the resulting collision-induced dissociation (CID) generated peptide ion fragment data was used to identify the proteins in each gel slice. Over 284 proteins (including 92 membrane proteins) were identified, including many integral membrane proteins not previously identified by 2-DE, many proteins seen at the genomic level only, as well as several proteins identified by expressed sequence tags (ESTs) only. Additionally, a number of peptides, identified by de novo MS sequence analysis, have not been described in the databases. Further, a "targeted" ion approach was used to unambiguously identify known low-abundance plasma membrane proteins, using the membrane-associated A33 antigen, a gastrointestinal-specific epithelial cell protein, as an example. Following localization of the A33 antigen in the gel by immunoblotting, ions corresponding to the theoretical A33 antigen tryptic peptide masses were selected using an "inclusion" mass list for automated sequence analysis. Six peptides corresponding to the A33 antigen, present at levels well below those accessible using the standard automated "nontargeted" approach, were identified. The membrane protein database may be accessed via the World Wide Web (WWW) at http://www.ludwig. edu.au/jpsl/jpslhome.html.

The murine A33 antigen is expressed at two distinct sites during development, the ICM of the blastocyst and the intestinal epithelium

The expression pattern of the murine A33 antigen has been defined during development using wholemount immunohistochemistry. Two temporally and spatially distinct sites of expression were identified: the inner cell mass of the blastocyst and the endoderm cell layer of the intestinal tract where expression is initiated at E14.5 in the hindgut and subsequently extends throughout the length of the intestine. The onset of mA33 antigen expression in the gut occurs at the beginning of an extensive phase of cell movement involved in the conversion of the endoderm cell layer to a single cell layer of polarized epithelium. Expression of mA33 antigen is then maintained into adulthood, where it is a definitive marker of intestinal epithelium.

Characterization of mouse A33 antigen, a definitive marker for basolateral surfaces of intestinal epithelial cells

The murine A33 antigen is emerging as a definitive marker of intestinal epithelial cells. Cloning and sequence determination of cDNAs encoding mA33 antigen predict a novel type 1 transmembrane protein of 298 amino acids, comprising an extracellular domain with two immunoglobulin-like domains, a single-span transmembrane domain, and a highly acidic cytoplasmic domain. On the basis of conservation of amino acid sequence and genomic structure, the mA33 antigen is a member of a growing subfamily within the immunoglobulin superfamily, which includes transmembrane proteins CTX/ChT1, CTM/CTH, and CAR. During embryonic development, mA33 antigen expression is first observed in the inner cell mass of blastocysts before implantation. Intestinal expression of mA33 antigen is initiated in the hindgut at E14.5 and increases steadily throughout late embryonic and postnatal life into adulthood. The protein is specifically expressed on the basolateral surfaces of intestinal epithelial cells of all lineages, independent of their position along the rostrocaudal and crypt-villus axes. Thus the mA33 antigen appears to be a novel marker for both proliferating and differentiating intestinal epithelial cells.

The rabbit antibody repertoire as a novel source for the generation of therapeutic human antibodies

The rabbit antibody repertoire, which in the form of polyclonal antibodies has been used in diagnostic applications for decades, would be an attractive source for the generation of therapeutic human antibodies. The humanization of rabbit antibodies, however, has not been reported. Here we use phage display technology to select and humanize antibodies from rabbits that were immunized with human A33 antigen which is a target antigen for the immunotherapy of colon cancer. We first selected rabbit antibodies that bind to a cell surface epitope of human A33 antigen with an affinity in the 1 nm range. For rabbit antibody humanization, we then used a selection strategy that combines grafting of the complementarity determining regions with framework fine tuning. The resulting humanized antibodies were found to retain both high specificity and affinity for human A33 antigen.

Proteomic analysis of the human colon carcinoma cell line (LIM 1215): development of a membrane protein database

The proteomic definition of plasma membrane proteins is an important initial step in searching for novel tumor marker proteins expressed during the different stages of cancer progression. However, due to the charge heterogeneity and poor solubility of membrane-associated proteins this subsection of the cell's proteome is often refractory to two-dimensional electrophoresis (2-DE), the current paradigm technology for studying protein expression profiles. Here, we describe a non-2-DE method for identifying membrane proteins. Proteins from an enriched membrane preparation of the human colorectal carcinoma cell line LIM1215 were initially fractionated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE, 4-20%). The unstained gel was cut into 16 x 3 mm slices, and peptide mixtures resulting from in-gel tryptic digestion of each slice were individually subjected to capillary-column reversed phase-high performance liquid chromatography (RP-HPLC) coupled with electrospray ionization-ion trap-mass spectrometry (ESI-IT-MS). Interrogation of genomic databases with the resulting collision-induced dissociation (CID) generated peptide ion fragment data was used to identify the proteins in each gel slice. Over 284 proteins (including 92 membrane proteins) were identified, including many integral membrane proteins not previously identified by 2-DE, many proteins seen at the genomic level only, as well as several proteins identified by expressed sequence tags (ESTs) only. Additionally, a number of peptides, identified by de novo MS sequence analysis, have not been described in the databases. Further, a "targeted" ion approach was used to unambiguously identify known low-abundance plasma membrane proteins, using the membrane-associated A33 antigen, a gastrointestinal-specific epithelial cell protein, as an example. Following localization of the A33 antigen in the gel by immunoblotting, ions corresponding to the theoretical A33 antigen tryptic peptide masses were selected using an "inclusion" mass list for automated sequence analysis. Six peptides corresponding to the A33 antigen, present at levels well below those accessible using the standard automated "nontargeted" approach, were identified. The membrane protein database may be accessed via the World Wide Web (WWW) at http://www.ludwig. edu.au/jpsl/jpslhome.html.

Micropreparative ligand fishing with a cuvette-based optical mirror resonance biosensor

We have previously demonstrated the role of an optical biosensor (BIAcore 2000) as a specific detector to monitor chromatographic fractions during the purification and characterisation of ligands for orphan biomolecules. We have now extended this application to perform micropreparative ligand fishing directly on the sensor surface using an automated cuvette-based optical biosensor (Iasys Auto+) equipped with a high-capacity carboxymethyldextran surface (surface area 16 mm2). Using a F(ab)2' fragment of the A33 monoclonal antibody as bait, we have recovered microgram quantities of essentially homogeneous A33 ligand from the sensor surface in a form suitable for subsequent sensitive and specific down stream analysis (micropreparative HPLC, sodium dodecyl sulphate-polyacrylamide gel electrophoresis and Western blotting). The design of the cuvette-based system facilitates recovery of desorbed material from the constrained workspace in small volumes at high concentration. The use of on-surface detection allows the surface viability to be continuously monitored and permits direct quantitation of both bound and recovered material.

Instrumental biosensors: new perspectives for the analysis of biomolecular interactions

The use of instrumental biosensors in basic research to measure biomolecular interactions in real time is increasing exponentially. Applications include protein-protein, protein-peptide, DNA-protein, DNA-DNA, and lipid-protein interactions. Such techniques have been applied to, for example, antibody-antigen, receptor-ligand, signal transduction, and nuclear receptor studies. This review outlines the principles of two of the most commonly used instruments and highlights specific operating parameters that will assist in optimising experimental design, data generation, and analysis.

CTX, a Xenopus thymocyte receptor, defines a molecular family conserved throughout vertebrates

CTX, a cortical thymocyte marker in Xenopus, is an immunoglobulin superfamily (Igsf) member comprising one variable and one constant C2-type Igsf domain, a transmembrane segment and a cytoplasmic tail. Although resembling that of the TCR and immunoglobulins, the variable domain is not encoded by somatic rearrangement of the gene but by splicing of two half-domain exons. The C2 domain, also encoded by two exons, has an extra pair of cysteines. The transmembrane segment is free of charged residues, and the cytoplasmic tail (70 amino acids) contains one tyrosine and many glutamic acid residues. ChT1, a chicken homologue of CTX, has the same structural and genetic features, and both molecules are expressed on the thymocyte surface. We cloned new mouse (CTM) and human (CTH) cDNA and genes which are highly homologous to CTX/ChT1 but not lymphocyte specific. Similarity with recently described human cell surface molecules, A33 antigen and CAR (coxsackie and adenovirus 5 receptor), and a number of expressed sequence tags leads us to propose that CTX defines a novel subset of the Igsf, conserved throughout vertebrates and extending beyond the immune system. Strong homologies within vertebrate sequences suggest that the V and C2 CTX domains are scions of a very ancient lineage.

Micro-sequencing strategies for the human A33 antigen, a novel surface glycoprotein of human gastrointestinal epithelium

Monoclonal antibody (mAb) A33, which recognizes a M(r) approximately 43,000 differentiation antigen (A33) expressed in normal human colonic and small bowel epithelium as well as in 95% of colon cancers, shows specific targeting of colon cancer in humans and is currently being evaluated for clinical use. Here, we describe strategies for the purification and structural analysis of the A33 antigen from the human colorectal carcinoma cell lines LIM1215 and SW1222. Edman degradation of the intact protein and nine peptides, derived by proteolytic digestion of the A33 antigen with Asp-N endoproteinase, thermolysin, trypsin and pepsin followed by micropreparative reversed-phase high-performance liquid chromatography, allowed the unambiguous sequence assignment of 153 amino acid residues; these data reveal one N-glycosylation sequeon in Asp-N endoproteinase peptide D4, and a disulfide linkage between peptides D1 and D4. This amino acid sequence information has facilitated the cloning and subsequent sequencing of a cDNA for the A33 antigen which demonstrates that it is a novel human cell surface molecule of the immunoglobulin superfamily.

Characterization of posttranslational modifications of human A33 antigen, a novel palmitoylated surface glycoprotein of human gastrointestinal epithelium

Monoclonal antibody (mAb) A33 recognizes a differentiation antigen (A33) expressed in normal human gastrointestinal epithelium and in 95% of human colon cancers. Murine mAb A33 shows specific targeting of colon cancer in humans and a humanized A33 antibody is currently being evaluated in the clinic. The cDNA for the human A33 antigen has recently been cloned, and sequence comparison indicated that the A33 antigen is a novel human cell surface molecule of the immunoglobulin superfamily. Because mAb A33 recognizes a conformational epitope, only a partial characterization of the A33 antigen has been carried out to date. In this report we show that the A33 antigen is (I) N-glycosylated, containing approximately 8 K of N-linked carbohydrate and there is no evidence for O-glycosylation, sialylation or glycophosphatidylinositol, and (ii) S-acylated in vitro, incorporating [3H] palmitic acid linked through a hydroxylamine-sensitive thioester bond. The S-palmitoylation may be involved in regulating the internalization process initiated by binding of mAb A33 to cell surface A33 antigen.

Electrophoretic analysis of the novel antigen for the gastrointestinal-specific monoclonal antibody, A33

The murine monoclonal antibody A33 (mAbA33) recognises a human cell membrane-associated antigen selectively expressed in epithelial cells of the lower gastrointestinal tract and > 90% of colonic cancers, but is not detected in a wide range of other normal tissues by immunohistochemical analysis. In phase I/II clinical triasl, mAbA33 has been shown to target advanced colon cancers and the humanized version is currently being evaluated in therapy studies. Although the mAbA33 has been well characterised by immunohistochemical and clinical studies, until recently, the target antigen has remained poorly defined. This was largely attributable to the antigenic determinant recognised by mAbA33 being dependent on the native spatial conformation of the A33 antigen which impeded its identification by conventional two-dimensional electrophoresis (2-DE) and immunoblot analysis. We have developed an immunoblot method, based on nonreducing/non-urea precast 2-DE gels, that has facilitated the purification of the detergent (0.3% Triton X-100) solubilised A33 antigen from the human colon cancer cell lines LIM1215 and SW1222. Under these 2-DE conditions, the A33 antigen electrophoreses with an apparent M(r) approximately 41000 and pI 5.0-6.0. Attempts to isolate the A33 antigen from 2-DE gels for direct structural analysis were unsuccessful, due to its co-electrophoresis with actin and cytokeratin proteins. However, using Western blot and biosensor detection the A33 antigen has been purified chromatographically and N-terminal sequence analysis was possible. Using polyclonal antibodies raised against a synthetic peptide corresponding to the N-terminal region of the A33 antigen we have used Western blot analysis to localise the molecule in our master 2-DE protein database for normal human colon crypts and several colon carcinoma cell lines (URL address: http:(/)/www.ludwig.edu.au). Under reducing 2-DE conditions, the A33 antigen electrophoresis as 6 differentially charged isoforms (pI 4.6-4.8) with a single molecular weight species at M(r) approximately 55000.

A two-dimensional gel database of human colon carcinoma proteins

The master two-dimensional gel database of human colon carcinoma cells currently lists cellular proteins from normal crypts and the colorectal cancer cell lines LIM 1863, LIM 1215 and LIM 1899 (Ward et al., Electrophoresis 1990, 11, 883-891; Ji et al., Electrophoresis 1994, 15, 391-405). Updated two-dimensional electrophoretic (2-DE) maps of cellular proteins from LIM 1215 cells, acquired under both nonreducing and reducing conditions, are presented. Fifteen cellular proteins are identified in the reducing 2-DE gel map, and seven in the nonreducing gel map, along with a tabular listing of their M(r)/pI loci and mode of identification. We also include our mass spectrometric based procedures for identifying 2-DE resolved proteins. This procedure relies on a combination of capillary column (0.10-0.32 mm internal diameter) reversed-phase HPLC peptide mapping of in-gel digested proteins, peptide mass fingerprinting, sequence analysis by either collision-induced dissociation or post-source-decay fragmentation, and protein identification using available database search algorithms. These data, and descriptions of the micro-techniques employed in this laboratory for identifying 2-DE resolved proteins can be accessed via the internet URL: http:(/)/www.ludwig.edu.au.

The human A33 antigen is a transmembrane glycoprotein and a novel member of the immunoglobulin superfamily

The mAb A33 detects a membrane antigen that is expressed in normal human colonic and small bowel epithelium and > 95% of human colon cancers. It is absent from most other human tissues and tumor types. The murine A33 mAb has been shown to target colon cancer in clinical trials, and the therapeutic potential of a humanized antibody is currently being evaluated. Using detergent extracts of the human colon carcinoma cell lines LIM1215 and SW1222, in which the antigen is highly expressed, the molecule was purified, yielding a 43-kDa protein. The N-terminal sequence was determined and further internal peptide sequence obtained following enzymatic cleavage. Degenerate primers were used in PCRs to produce a probe to screen a LIM1215 cDNA library, yielding clones that enabled us to deduce the complete amino acid sequence of the A33 antigen and express the protein. The available data bases have been searched and reveal no overall sequence similarities with known proteins. Based on a hydrophilicity plot, the A33 protein has three distinct structural domains: an extracellular region of 213 amino acids (which, by sequence alignment of conserved residues, contains two putative immunoglobulin-like domains), a single hydrophobic transmembrane domain, and a highly polar intracellular tail containing four consecutive cysteine residues. These data indicate that the A33 antigen is a novel cell surface receptor or cell adhesion molecule in the immunoglobulin superfamily.