Proc. Natl. Acad. Sci. USAVol. 93, pp. 7252-7257, July 1996Medical Sciences

Surface-epitope masking and expression cloning identifies thehuman prostate carcinoma tumor antigen gene PCTA-1 a memberof the galectin gene family

(DNA cotransfection-nude mouse tumors/reverse transcription-PCR/monoclonal antibodies/immunohistochemistry/human tumorgalectin-8)

ZAO-ZHONG SU*t, JIAN LINt4, RUOQIAN SHEN*t, PETER E. FISHERt, NEIL I. GOLDSTEIN§, AND PAUL B. FISHER*t¶Departments of tPathology and *Urology, Columbia-Presbyterian Cancer Center, Columbia University, College of Physicians and Surgeons, New York,NY 10032; and §Department of Immunology/Monoclonal Antibodies, ImClone Systems Inc., New York, NY 10014

Communicated by Allan H. Conney, Rutgers, The State University of New Jersey, Piscataway, NJ, March 6, 1996 (received for reviewNovember 29, 1995)

ABSTRACT The selective production of monoclonal an-tibodies (mAbs) reacting with defined cell surface-expressedmolecules is now readily accomplished with an immunologicalsubtraction approach, surface-epitope masking (SEM). UsingSEM, prostate carcinoma (Pro 1.5) mAbs have been developedthat react with tumor-associated antigens expressed on hu-man prostate cancer cell lines and patient-derived carcino-mas. Screening a human LNCaP prostate cancer cDNA ex-pression library with the Pro 1.5 mAb identifies a gene,prostate carcinoma tumor antigen-1 (PCTA-1). PCTA-1 en-codes a secreted protein of -35 kDa that shares "40%sequence homology with the N-amino terminal region ofmembers of the S-type galactose-binding lectin (galectin) genefamily. Specific galectins are found on the surface of humanand murine neoplastic cells and have been implicated intumorigenesis and metastasis. Primer pairs within the 3'untranslated region of PCTA-1 and reverse transcription-PCR demonstrate selective expression of PCTA-1 by prostatecarcinomas versus normal prostate and benign prostatichypertrophy. These findings document the use of the SEMprocedure for generating mAbs reacting with tumor-associated antigens expressed on human prostate cancers.The SEM-derived mAbs have been used for expression cloningthe gene encoding this human tumor antigen. The approachesdescribed in this paper, SEM combined with expressioncloning, should prove of wide utility for developing immuno-logical reagents specific for and identifying genes relevant tohuman cancer.

The early detection of prostate cancer and the accurateprediction of its clinical course is not possible by using currentmethodologies. Contemporary approaches, including physicalexamination, tissue biopsy, monitoring serum prostate-specificantigen (PSA) levels, and ultrasound and bone scans do notensure early prostate cancer detection and are of only limitedvalue in predicting disease progression. Of immense value forthe accurate diagnosis and potentially for the therapy ofhuman prostate cancer is the identification of immunologicaland genetic reagents displaying the appropriate specificity thatwill permit a clear distinction between prostate carcinomaversus normal prostate and benign prostatic hypertrophy(BPH). Immunological reagents that can detect and bind totumor-associated antigens on the surface of prostate carci-noma cells, and that display limited expression on normalprostate and BPH tissue, may prove of particular benefit forthe detection and ultimately for the therapy of prostate cancer.

Production of monoclonal antibodies (mAbs) reacting withantigens present on the surface of tumor cells, but displayingrestricted expression on normal cells, is often a difficult andinefficient process (1-3). A procedure based on immunologicalsubtraction, surface-epitope masking (SEM), has been devel-oped that in principle can obviate many of the limitationspreventing efficient mAb development toward molecules ex-pressed on the cell surface (3, 4). SEM is based on the selectiveblocking of antigens on a genetically modified target cell (i.e.,tester) with polyclonal antibodies produced against the sameunmodified cell line (i.e., driver). Antigen-blocked cells areinjected into BALB/c mice, and sensitized spleen cells areremoved and fused to NS1 myeloma cells, and hybridomassecreting reactive mAbs are isolated (3). The SEM approachhas been successfully used for a number of applications,resulting in the production of mAbs specific for surfaceexpressed molecules with known and unknown functions (3, 4).An improved procedure has been developed for identifying

and cloning dominant acting oncogenes, termed rapid expres-sion cloning (5, 6). This approach involves transfecting highmolecular weight human tumor DNA into a new acceptor cellline, CREF-Trans 6, selecting cells expressing genes inducingtumors in nude mice, and using molecular biological ap-proaches, such as differential RNA display, to clone theputative oncogene (5, 6). Tumor-derived CREF-Trans 6 cellshave also proven useful in combination with the SEM proce-dure for identifying tumor-associated antigens expressed onthe cell surface of cancer cells serving as the initial source fortransforming DNA (5, 6). Expression cloning of cDNAs usingantibodies reacting with their encoded proteins represents adirect means of identifying and cloning functional genes (2).This approach has now been used to identify and clone a gene,prostate carcinoma tumor antigen 1 (PC1TA-1), encodingtumor-associated antigens (TAAs) recognized by the prostatecarcinoma mAb Pro 1.5. Sequence analysis indicates thatPCTA-1 consists of a cDNA of 3.85 kb with 83% nucleotide(nt) and 81% amino acid (aa) sequence homology to an-1.25-kb cDNA, rat galectin-8, cloned from a rat liver cDNAexpression library (7). Some overlapping homologies are also

Abbreviations: PSA, prostate-specific antigen; BPH, benign prostatichypertrophy; SEM, surface-epitope masking; PCTA-1, prostate car-cinoma tumor antigen 1; TAA, tumor-associated antigen; RT-PCR,reverse transcription-PCR; GAPDH, glyceraldehyde-3-phosphate de-hydrogenase, FACS, fluorescence-activated cell sorter; PIN, prostaticintraepithelial neoplasia.Data deposition: The sequence reported in this paper has beendeposited in the GenBank data base (accession no. L78132).tZ.-z.S., J.L., and R.S. contributed equally to this study.STo whom reprint requests should be addressed at: Departments ofPathology and Urology, Columbia University, College of Physiciansand Surgeons, PH STEM-10, 630 West 168th Street, New York, NewYork 10032.

7252

The publication costs of this article were defrayed in part by page chargepayment. This article must therefore be hereby marked "advertisement" inaccordance with 18 U.S.C. §1734 solely to indicate this fact.

Dow

nloa

ded

by g

uest

on

Nov

embe

r 17

, 202

0

Proc. Natl. Acad. Sci. USA 93 (1996) 7253

detected with several small noncontiguous partial cDNA se-quences (of less than 500 nt) previously identified as expressedhuman sequence tags (8). Analysis of protein structure indi-cates that PCTA-1 has -40% homology to specific structuraldomains of the S-type lectin family of galactose-binding pro-teins, the galectins (9-11). These highly homologous proteinsinclude a carbohydrate-binding 35-kDa protein (CBP35) ex-pressed on NIH 3T3 fibroblasts, a 34-kDa surface antigenpresent on metastatic murine tumors, a 31-kDa surface proteinon metastatic human tumors, a 30-kDa carbohydrate-bindingprotein (CBP30) found on baby hamster kidney cells, rat andhuman lung 29-kDa galactose-binding lectins, an IgE-bindingprotein of rat basophilic cells, and a 32-kDa surface antigenMac-2 found on thioglycollate-elicited murine macrophages(9-11). The roles of the galectin proteins are diverse andimpinge upon important biological processes, including cellsignaling, proliferative control, cell adhesion, and cell migra-tion (9-11). In this context, the demonstration that PCTA-1encodes a protein with specific homologies to the galectinssuggests that this molecule, a human tumor homologue of ratgalectin-8 (i.e., galectin-8HT), may contribute to human pros-tate cancer development and evolution.

MATERIALS AND METHODSCell Lines. The LNCaP cell line was derived from metastatic

deposits from a patient with advanced prostate cancer (12).CREF-Trans 6 and LNCaP DNA-transfected nude mousetumor-derived CREF-Trans 6 cells, CREF-Trans 6:4 NMT,were isolated as described (5). The hormone independentprostatic carcinoma cell lines DU-145 and PC-3 were obtainedfrom the American Type Culture Collection. Conditions forgrowing the various cell types were as described (5, 6, 12).cDNA Library Construction, Expression Cloning, and Se-

quencing. An LNCaP cDNA library was constructed in theUni-ZAP XR vector (Stratagene) (6, 13, 14). The cDNAlibrary was screened by using the Pro 1.5 mAb following theprotocol in the picoBlue Immunoscreening Kit (Stratagene).Host bacterial cells (SURE) were plated on fifteen 150 x15-mm NZY plates (Strategene) to yield -20,000 plaques/plate. After 3.5 hr incubation at 42°C, nitrocellulose filterssoaked in 10 mM isopropyl 13-D-thiogalactoside solution wereadded to the top of colonies for plaque lifts. The filters werewashed three to four times with TBST buffer (20mM Tris HCl,pH 7.5/150 mM NaCl/0.05% Tween-20) and soaked in block-ing solution [1% bovine serum albumin in TBS buffer (20 mMTris*HCl, pH 7.5/150 mM NaCl)] and incubated for 1 hr atroom temperature. The filters were then transferred into freshblocking solution containing Pro 1.5 ascites (1:500 dilution)and incubated for 3 hr at room temperature with gentleagitation. After washing four times with TBST buffer, thefilters were transferred into fresh blocking solution containingAb-AP conjugate (1:2000 dilution) and incubated for 1 hr atroom temperature. Positive colonies were identified by devel-oping the filters in a solution containing 0.3 mg of nitrobluetetrazolium per ml, 0.15 mg of 5-bromo-4-chloro-3-indolylphosphate per ml, 100 mM Tris-HCl (pH 9.5), 100 mM NaCl,and 5 mM MgCl2. The reaction was terminated by adding stopsolution (20 mM Tris-HCl, pH 2.9/1 mM EDTA). PCTA-1positive plaques were isolated and the complete nt sequence ofPCTA-1 was obtained by using the Sanger method (15). Aseries of oligonucleotides synthesized from both sides of thePCTA-1 insert in the pBluescript vector were used as primers.The final nucleotide sequence was verified by using an AppliedBiosystems (model 373A, version 1.2.1) sequencer.

In Vitro Translation of PCTA-1. The plasmid DNA contain-ing PCTA-1 was linearized by digestion with XmaIII and usedas a template to synthesize mRNA using the mCAP mRNAcapping kit (Stratagene). In vitro translation of PCTA-1 was

performed by using a rabbit reticulocyte lysate translation kitwith conditions as described by GIBCO/BRL (6).

Preparation of Mouse Polyclonal Antibodies and SEM.CREF-Trans 6 polyclonal antibodies were prepared as de-scribed (3). The CREF-Trans 6 polyclonal antibodies wereused to coat CREF-Trans 6:4NMT cells by the SEM approach,resulting in the production of hybridomas secreting Pro 1.5mAbs (3).

Fluorescence Cell Staining and Immunostaining of TissueSections with Pro 1.5 and PSA. Fluorescence staining with Pro1.5 mAb was as described (16). Tissue sections were preparedfrom fresh tissues frozen in liquid nitrogen. Serial sections forseveral tissues were used to prepare RNA for reverse tran-scription (RT)-PCR (6). Tissue sections were stained by usingstandard protocols [Super Sensitive Detection System (Bio-Genex Laboratories, San Ramon, CA)]. Briefly, sections werefixed in acetone, blocked with 3% H202 for 7 min at roomtemperature, and incubated for 20 min at room temperature inphosphate-buffered saline (PBS) containing 3% lamb serum.Sections were then incubated with Pro 1.5 (1:100) or PSA(1:200) (Dako) for 45 min at room temperature. Samples wereincubated with biotinylated secondary antibody and horserad-ish peroxidase conjugated streptavidin in PBS. Prior to eachincubation step, sections were washed three to five with PBS.Reactivity was detected by adding diaminobenzidine andcounterstaining with Hematoxylin.

Immunoprecipitation Analysis. Cells were labeled with[35S]methionine and cell lysates were analyzed for PCTA-1protein levels by immunoprecipitation analysis using Pro 1.5mAbs as described (3, 17). Secreted PCTA-1 was detected bylabeling cells for 4 hr with [35S]methionine, growing cells foran additional 24 hr in the absence of label, collecting themedium, concentrating the medium, and performing immu-noprecipitation analysis with Pro 1.5 mAbs.RNA Preparation and RT-PCR. Total cytoplasmic RNA

was isolated from logarithmically growing cell cultures asdescribed (6, 14). Tissue samples from normal prostates andpatients with prostatic carcinomas or BPH were frozen inliquid nitrogen, and RNA was isolated using the TRIzolreagent as described by GIBCO/BRL. Tissue samples weresupplied by the Cooperative Human Tumor Network. Threesamples of normal prostate were obtained from autopsies ofmales <40 years of age. All tissues were histologically con-firmed as normal, BPH, or carcinoma of the prostate (6).RT-PCR using primer pairs for PCTA-1, PSA and glyceral-dehyde-3-phosphate dehydrogenase (GAPDH) was per-formed as described (6, 17).

RESULTSProduction of Pro 1.5 mAbs by SEM and Reactivity of Pro

1.5 and PSA with Normal Prostate, BPH, and ProstateCarcinomas. The SEM approach of blocking antigenicepitopes on nude mouse tumor-derived LNCaP DNA-transfected CREF-Trans 6 cells with CREF-Trans 6 poly-clonal antibody before injection into mice was used to producehybridomas secreting the Pro (prostate carcinoma) series ofmAbs (3). The Pro mAbs can detect by in situ fluorescencemicroscopy and fluorescence-activated cell sorter (FACS)analysis the surface expression of tumor-associated antigenson LNCaP-transfected primary (CREF-Trans 6:4 NMT) andsecondary tumor-derived CREF-Trans 6 cells and LNCaP,DU-145, and PC-3 human prostate cancer cell lines (3) (Fig.1A-C and data not shown). In contrast, Pro 1.5 does not reactusing fluorescence microscopy or FACS with CREF-Trans 6cells (3). The staining pattern with Pro 1.5 in human prostatecarcinoma cells is irregular with microclusters, as previouslyobserved with mAbs reacting with specific galectins (10, 17, 18)(Fig. 1 A-C). Immunoprecipitation analysis identifies an -35-to 42-kDa protein in lysates from LNCaP-transfected primary

Medical Sciences: Su et aL

Dow

nloa

ded

by g

uest

on

Nov

embe

r 17

, 202

0

Proc. Natl. Acad. Sci. USA 93 (1996)

A LNCaP C PC-3

D Secreted and Cellular PCTA-1

.A~ ~ ~ ~ V

_ Secreted PCTA-*!':.' ...,,WCellular PCTA-1

FIG. 1. Immunofluroescence staining and immunoprecipitationanalysis of secreted and cellular PCTA-1. Live cells were incubatedwith mAb Pro 1.5 and analyzed by fluorescence microscopy (A, B, andC). Cells were labeled with [35S]methionine, and secreted and cellularPCTA-1 levels were determined by immunoprecipitation with the Pro1.5 mAb (D).

and secondary tumor-derived CREF-Trans 6, LNCaP, DU-145, and PC-3 cells (3) (Fig. 1D and data not shown). Of theprostate cancer cell lines studied, the PC-3 clone contains thelowest quantities of cell-associated, surface-expressed (FACS)and secreted PCTA-1 protein.To determine if Pro 1.5 mAbs can react with patient-derived

prostate cancer specimens, frozen sections were preparedfrom normal (males <40 years of age), BPH, and carcinomasof the prostate. Normal prostate displays limited reactivitywith Pro 1.5, whereas PSA readily stains normal prostate

(A) .A~ (B) -

44 jr,im

kw ?' 1 ' XCJ

epithelial cells (Fig. 2). As expected, PSA also stains prostatecells in tissue sections of BPH and prostate carcinomas. mAbPro 1.5 reacts strongly with prostate carcinoma cells present infrozen tissue sections, but not with adjacent benign glands ortissue sections containing normal prostate or BPH epithelium.However, some reactivity with Pro 1.5 is found in PIN. Thesestudies indicate that the Pro 1.5 mAb can distinguish betweenprostate carcinoma and PIN versus normal prostate epithelialcells and BPH. In this context, Pro 1.5 provides a discrimina-tory capacity for detection of cancer of the prostate thatexceeds that of the nonspecific prostate epithelial cell markerPSA.

Expression Cloning of PCTA-1 Using SEM-Derived Pro 1.5mAbs. To identify the gene encoding the tumor-associatedantigens identified on human prostate cancer cells by mAb Pro1.5, an antibody expression cloning strategy was used. AnLNCaP cDNA library was constructed in the Uni-ZAP XRvector (Stratagene) and screened with the Pro 1.5 mAb (12,13). This approach resulted in the identification of a 3.85-kbcDNA clone referred to as PCTA-1. In vitro protein translationof the PCTA-I cDNA results in a 317-aa protein of -35 kDa(data not shown). The DNA sequence ofPCTA-1 displays 83%homology, and the protein sequence of PCTA-1 displays 81%sequence homology with rat galectin-8 (7), a member of theS-lectin family of galactose-binding lectin proteins. The full-length rat galectin-8 cDNA is 1247 bp, whereas the full-lengthPCTA-1 cDNA is 3850 bp. This result suggests that PCTA-1encodes a human tumor homologue of galectin-8, galectin-8HT. Amino acid comparison of PCTA-1 with other membersof the galectin gene family indicates that this protein is -40%homologous to specific regions of these galectins (Fig. 3).Several amino acid sequence homologies are found betweenPCTA-1, rat galectin-8, and the other galectins. Stretches ofidentical amino acids are found in PCTA-1, rat galectin-8, andboth the small '14-kDa and larger -29- to 34-kDa galectins,e.g., HFNPRF, IVCN, and WG. These amino acids are wellconserved and are found in galectins isolated from diversespecies, including eel, chicken, mouse, rat, bovine, and human(Fig. 3B) (11, 19, 20). They are important structural compo-nents of the galectins and may mediate galectin binding to itsputative ligands (12). Of potential interest is the replacementin PCTA-1 of two amino acids normally present in theconserved regions of most of the galectins-i.e., the substitu-

FIG. 2. Immunohistochemical detection of PCTA-1 and PSA in prostate tissue sections. Benign prostate [PCTA-1 (A) and PSA (E)], prostaticintraepithelial neoplasia (PIN) [PCTA-1 (B) and PSA (F)], and invasive prostate carcinoma [PCTA-1 (C and D) and PSA (G and H)]. (A, B, D,E, F, and H, x250; C and G, x100.)

7254 Medical Sciences: Su et al.

1

Dow

nloa

ded

by g

uest

on

Nov

embe

r 17

, 202

0

Medical Sciences: Su et al. Proc. Natl. Acad. Sci. USA 93 (1996) 7255

A 54 ATG ATG TTG TCC TTA AAC AAC CTA CAG AAT ATC ATC TAT AAC CCG GTA ATC CCG TTT GTTM M L S L N N L Q N I I Y N P V I P F V

114 GGC ACC ATT CCT GAT CAG CTG GAT CCT GGA ACT TTG ATT GTG ATA CGT 000 CAT OTT CCTG T I P D Q L D P G T L I V I R G H V P

174 AGT GAC OCA GAC AGA TTC CAG OTG GAT CTG CAG AAT GGC AGC AGC GTG AAA CCT CGA GCCS D A D R F Q V D L Q N G S S V K P R A

234 GAT GTG GCC TTT CAT TTC ART CCT CGT TTC AAA AGG GCC GGC TGC ATT GTT TGC ART ACTD V A F H P N P R F K R A G C I V C N T

294 TTG ATA AAT GAA AAA TGG GGA CGG GAA GAG ATC ACC TAT GAC ACG CCT TTC AAA AGA GAAL I N E K W G R E E I T Y D T P F K R .H

354 AAG TCT TTT GAG ATC GTO ATT ATG GTG CTG AAG GAC AAA TTC CAG GTG GCT GTA AAT GGAK S F H I V I M V L K D K F Q V A V N G

414 AR CAT ACT CTG CTC TAT GGC CAC AGG ATC GGC CCA GAG AAA ATA GAC ACT CTG GGC ATTK H T L L Y G H R I G P E K I D T L G I

474 TAT GGC AAA GTG AAT ATT CAC TCA ATT GGT TTT AGC TTC AGC TCG GAC TTA CAA AGT ACCY G K V N I H S I G F S F S S D L Q S T

534 CAA GCA TCT AGT CTG GAA CTG ACA GAG ATA GTT AGA GAA AAT GTT CCA AAG TCT GGC ACGQ A S S L E L T H I V R E N V P K S G T

594 CCC CAG CTT AGC CTG CCA TTC GCT GCA AGG TTG AAC ACC CCC ATG GGC CCT GGA CGA ACTP Q L S L P F A A R L N T P M G P G R T

654 GTC GTC GTT CAA GGA GAA GTG AAT GCA AAT GCC AAA AGC TTT AAT GTT GAC CTA CTA GCAV V V Q G H V N A N A K S F N V D L L A

714 GGA AAA TCA AAG GAT ATT GCT CTA CAC TTG AAC CCA CGC CTG AAT ATT AAA GCA TTT GTAG X S K D I A L H L N P R L N I K A F V

774 AGA AAT TCT TTT CTT CAG GAG TCC TGG GGA GAA GAA GAG AGA AAT ATT ACC TCT TTC CCAR N S F L Q H S W G E E E R N I T S F P

834 TTT AGT CCT GGG ATG TAC TTT GAG ATG ATA ATT TAT TGT GAT GTT AGA GAA TTC AAG GTTF S P G N Y F E M I I Y C D V R E F X V

894 GCA GTA AAT GOC GTA CAC AGC CTG GAG TAC AAA CAC ARA TTT AAA GAG CTC AGC AGT ATTA V N G V H S L E Y K H R F K H L S S I

954 GAC ACG CTG GAA ATT AAT GGA GAC ATC CAC TTA CTO GOA GTA AGG AGC TO TAOD T L H I N 0 D I H L L E V R S W

B

Eel-L14Chicken-L14Mouse-L14

Rat-L14Bovine-L14

Human Lung-L14Human Placenta-L14

Human Hepatoma-1-L14Human Hepatoma-2-L14

Mouse-L34Human Galectin-3-L29

Human-L31Rat-Galectin-8

PCTA-1

A EfL ILS T H L G LS N N L C LS N N L C L

S N N L C I

S N N L - LS G K - - -

T D K L N L

G N FV A F

GNDIVAIF

RADJVAFR A D VLAF* * A AA

H INPRFH F N P R F

LH F N P R FLH F N P R FLH F N P R F

-F_

F N P RF

L H F N P R FP H F N P R F

P H F N P RF

P H F N P R F?H F N P R F?H F NP RF

D A H G D Q Q ANAHGDANL IVCNNAHGDVNL IVCNN A H G D A N T I V C NN A H G D A N T I V C N

NAH -----S G S T - - - - I V C NSGST----IVCNS E N N F F V V I V C NN E N N R R - V I V C NNENNRRV- IVCNK R S N - - - C I V C NKKRAG --- C IVCN*RAGA--C

Eel-L14Chicken-L14Mouse-L14

Rat-L14Bovine-L14

Human Lung-L14Human Placenta-L14

Human Hepatoma-1-L14Human Hepatoma-2-L14

Mouse-L34Human Galectin-3 -L2 9

Human-L3 1Rat-Galectin-8

PCTA-1

SFQGGNWGT-EQaK K M E E W G T E QUKEDG T W G TE H

S K D D G T W G T E QS K D G G A W G T E Q---G G A W G - - Q--S G G A W G T E QS L D G S N W G Q E Q- - - T W G T E QTKQDNNWGKEET K L D N N W G R E ET K L H N N W G R E ElT L T N E K W G W E ET L I N E K W g RLE E

*A*** A

R E G G F P F L Q G ER E T V F P F Q Q G QR E P A F P F Q P F QR E T A F P F Q P G SR E A V F P F Q P G S

E A V F0 F PR E A V F PF QPGSR E D H L 5JF S P G SR E D H ILPF Q P G SR K S A FP F E C G NR Q S V F P F E S G KR Q S V FP F E S G KI T H D M P F R K E KI T Y D T I FK R E KR GG FPFLQG

FiG. 3. (A) Predicted amino acid sequence of PCTA-1 and alignment of PCTA-1 with other galactose-binding lectin (galectin) proteins. Thepredicted amino acid sequence is shown below the nucleotide sequence. (B) Comparison of PCTA-1 with sequences of the Mr 14,000 and Mr29,000-35,000 galectins from eel, chicken, mouse, rat, bovine, and human galectins. *, Amino acids unique to PCTA-1 and rat galectin-8; *A,, aminoacid differences between PCTA-1 and rat galectin-8; A, amino acids shared by PCTA-1, rat galectin-8, mouse-L34, human galectin-3-L29, andhuman-L31; AE, amino acid difference between PCTA-1, human galectin-3-L29, and human-L31 versus rat galectin-8; 0, amino acids shared byPCTA-1, rat galectin-8, and the Mr 14,000 and Mr 29,000-31,000 galectins from different species.

tion of an I for an R and a T for an F (Fig. 3B). The I for anR substitution is also found in rat galectin-8, whereas an M forF substitution is found in rat galectin-8. A number of similarpositioned amino acids are found in PCTA-1, rat galectin-8,and the larger -29- to 34-kDa galectins that are not present inthe 14-kDa galectins isolated from most other species-e.g.,DVAF (Fig. 3B). The amino acid sequence WGREE is con-served in PCTA-1 and the human -29- to 34-kDa humangalectins but is not present in the L34 mouse galectin (WG-KEE) or rat galectin-8 (WGWEE). In addition, a number ofamino acids are distinct for PCTA-1 and rat galectin-8 versusthe other similarly positioned amino acids that are found in thehuman galectin-3 L29 and human L-31 proteins, including RA,RFKR, CIVC, and EEIT. A number of amino acid changesalso distinguish PCTA-1 from rat galectin-8, including KRAG,LINE, and TYDTPFKRE. The similarities in regions of con-served structure between PCTA-1, rat galectin-8, and theother galectins suggest that they will share a number of

overlapping properties. However, the numerous amino acidchanges observed in PCTA-1, in both the conserved regionsand throughout the remainder of this protein, can be expectedto impact on the structure and affect the properties of thePCTA-1 protein.

Expression of PCTA-1 in Cell Lines and Normal Prostate,BPH, and Prostate Carcinomas. After obtaining the sequenceof PCTA-1, studies were performed to determine if primers forspecific regions of this gene could be identified that wouldpermit detection of RNA expression by RT-PCR. Usingprimers located between bp 3010 and 3423 in the 3' untrans-lated region, PCTA-1 expression is apparent in LNCaP DNA-transfected tumor-derived CREF-Trans 6, LNCaP, and DU-145 cells but not in untransfected CREF-Trans 6 cells (Fig. 4).PCTA-1 expression also occurs in seven of seven patient-derived prostate carcinomas, one of four BPH, and one of fourputative normal prostate tissue samples (Fig. 4). In the oneBPH sample displaying PCTA-1 expression, histological anal-

Dow

nloa

ded

by g

uest

on

Nov

embe

r 17

, 202

0

7256 Medical Sciences: Su et al.

BenignProttc

Normal Prostate Hypertrophy Fa. ~ op4~ c,) c ir cc1a

X cc r O- m ::z c cc 0 0 00 C oL_ o o a z z z z ax a] m J0O C1 ''W ';''' ''''" '' '' m"''

Prostatic Carcnoma

0 0

FIG. 4. RT-PCR analysis of PCTA-1, PSA, and GAPDH expression in cell lines and tissue samples of normal prostate, BPH, and prostatecarcinoma. RT-PCR of PCTA-1 uses two 20-mers with the sequences 5'-AAGCTGACGCCTCATTTGCA-3' and 5'-AACCACCAATGGAACT-GGGT-3'. RT-PCR of PSA uses two primers: primer A, 5'-AGACACAGGCCAGGTATTTCAGGTC-3' and primer B, 5'-CACGATGGTGTCCTTGATCCACTTC-3'. RT-PCR of GAPDH uses a pair of primers with the sequences 5'-TCTTACTCCTTGGAGGC-CATG-3' and 5'-CGTCTTCACCACCATGGAGAA-3'. The PCR-amplified products were blotted on nylon membranes and probed with a32P-labeled DNA fragment of PCTA-1, PSA, or GAPDH, respectively.

ysis indicates the presence of epithelial atypia consistent withPIN. In the one putative normal prostate tissue sample ex-pressing PCTA-1, no tissue was available for histologicalanalysis. However, because this tissue was obtained from a60-year-old male, it is possible that this tissue may havecontained clinically unsuspected prostate disease. The sameRNA samples were tested for PSA expression (Fig. 4). PSA wasdetected by RT-PCR in LNCaP cells and all of the prostatetissue samples, but not in CREF-Trans 6, LNCaP DNA-transfected tumor-derived CREF-Trans 6, or DU-145 cells. Aspredicted, all samples were positive for expression of thehousekeeping gene GAPDH (Fig. 4). Although the samplingsize is small, these results indicate that PCTA-1 expression maybe restricted to prostate carcinomas and subsets of BPHdisplaying early stages of cancer (i.e., PIN). In contrast,PCTA-1 expression is not evident in normal prostate or BPH.

Secretion of PCTA-1 by Human Prostate Carcinoma CellLines. Many galectins are externalized by nonclassical secre-tory mechanisms that do not involve a typical secretion signalpeptide (11, 18, 21). To determine if PCTA-1 is secreted by Pro1.5 mAb positive cells, target cells were labeled for 4 hr with[35S]methionine, the label was removed, and cells were washedthree times in methionine-free medium and then incubated foran additional 18-24 hr in complete medium without label. Theconditioned medium (CM) was collected, contaminating cellswere removed, and the CM was concentrated and immuno-precipitated using the Pro 1.5 mAb (Fig. 1C). Under theseexperimental conditions, no reduction in cell viability wasevident. A protein of -35-42 kDa was present in CM fromCREF-Trans 6:4 NMT, LNCaP, DU-145, and PC-3, but notfrom CM obtained from CREF-Trans 6 cells (Fig. 1C).Immunoprecipitation analysis of [35S]methionine labeled celllysates from the same cell types indicate a similar pattern ofPCTA-1 expression-i.e., present in CREF-Trans 6:4 NMTand human prostate carcinoma cell lines but not in untrans-fected CREF-Trans 6 cells (Fig. 1C). As previously foundusing FACS analysis with the Pro mAbs, PC-3 cells producedthe lowest levels of cell-associated and secreted PCTA-1.These results demonstrate that PCTA-1 is shed from prostatecancer cells. In this context, monitoring PCTA-1 protein in the

circulation might prove beneficial as a diagnostic marker forprostate cancer.

DISCUSSIONWe describe the cloning of PCTA-1, a gene encoding TAAsthat are present on the surface of invasive human prostatecarcinoma and early prostate cancer, PIN, but not on histo-logically confirmed normal prostate or BPH tissue. ThePCTA-1-encoded TAAs are detected on the surface of pros-tate cancers using the SEM-derived Pro series of mAbs, andthese TAAs are shed by prostate carcinoma cells. Using a pairof primers hybridizing with sequences in the 3'-untranslatedregion of PCTA-1, RT-PCR detects PCTA-1 expression inprostatic carcinomas and PIN, but not in histologically con-firmed normal prostate or BPH. These attributes should allowthe direct use of the Pro series of mAbs and the PCTA-1 genefor cancer diagnostic applications. If the PCTA-1 gene displaysappropriate specificity using a larger tissue sampling, this genemay also prove of value for designing gene-based strategies forthe therapy of prostate cancer. Moreover, the innovativeprocedures used, including rapid expression cloning, SEM, andantibody expression cloning, represent integrated approachesfor the efficient identification and cloning of molecules (in-cluding TAAs) of potential clinical interest that are expressedon the surface of diverse human cancer cells.The PCTA-1 protein retains a number of conserved struc-

tural motifs that are found in most members of the galectingene family (7, 11, 19, 22). These conserved regions are presentin species as diverse as eel, mouse, rat, and human (11). On thebasis of amino acid sequence, PCTA-1 is a human tumorhomologue of rat galectin-8 (7), galectin-8HT, that may con-tribute to the cancer phenotype of human prostate carcinomas.The galectins display wide tissue distribution, clear develop-mental regulation, and differential levels in specific tissues,supporting the hypothesis that they contribute to many phys-iologically important processes in mammalian cells (11). Ofdirect relevance to cancer is the finding that the galectins, aswell as the selectin subgroup of C-type lectins (20, 23), canmediate both cell-cell and cell-matrix interactions (11, 18, 24).These associations are critical elements in mediating the

PCTA-1

PSA

GAPDH

Proc. Natl. Acad. Sci. USA 93 (1996)

Dow

nloa

ded

by g

uest

on

Nov

embe

r 17

, 202

0

Proc. Natl. Acad. Sci. USA 93 (1996) 7257

metastatic spread of tumor cells (25, 26). Moreover, experi-mental evidence has accumulated indicating that galectin-3may play an important role in the metastatic process (18,27-31). Galectin-3 is overexpressed in human colon and gastriccarcinomas versus normal and benign tissue, and elevatedexpression of recombinant L-34 in a weakly metastatic UV-2237-cl-15 mouse fibrosarcoma cell line increases lung metas-tases in syngeneic and nude mice (18, 27-31). Moreover,anti-galectin mAbs inhibit homotypic aggregation, anchorage-independent growth, and experimental metastases in UV-2237subclones (32-34). Studies are in progress to determine if thePCTA-1 gene and the Pro mAbs display similar properties asthe cloned galectin-3 gene and the anti-galectin mAbs, respec-tively. In preliminary studies, the effect of mAb Pro 1.5 ontumor growth in athymic nude mice containing establishedDU-145 tumors ("200 mm3) has been determined. In thismodel system, injection of Pro 1.5 mAbs (200 ,ug/injection; 11injections) reduced tumor size in seven of nine animals (un-published studies). In contrast, 10 of 12 animals receiving anirrelevant mAb developed actively growing tumors (unpub-lished data). These preliminary findings demonstrate a directeffect of Pro 1.5 mAbs on the growth and progression ofhuman prostate cancers in vivo. It will also be important toascertain if inhibition in PCTA-1 expression, using antisenseoligonucleotides, antisense expression vectors, or ribozymeapproaches, alters the tumorigenic or metastatic properties ofhuman prostate cancers.

Prior to SEM, no simple and direct procedure was availablefor efficiently generating mAbs reacting with differentiallyexpressed surface molecules. Conceptually, SEM involves im-munological subtraction that induces the immune system ofmice to target antibody production toward surface moleculesexpressed on a genetically modified tester cell line, but notexpressed or expressed in lower abundance on its' cognateunmodified driver cell line (3, 4). By using polyclonal antibod-ies produced against the driver cell line to coat (mask) epitopeson the tester cell line, enriched production of mAbs targetedtoward epitopes expressed on the surface of the tester cell typeis achieved (3, 4). Confirmation of the SEM approach hascome from studies in which this process was used to target thedevelopment of mAbs reacting with the Mr 170,000 humanmultidrug resistance protein (P-glycoprotein) and the humaninterferon y receptor (3, 4). The SEM approach has also beenused to produce mAbs that react with unknown TAAs, in-cluding those presently identified as the gene product ofPCTA-1 and undefined TAAs transferred by rapid expressioncloning from the T47D human breast carcinoma cell line toCREF-Trans 6 (data not shown). These results indicate thatrapid expression cloning combined with SEM may representefficient strategies for identifying antigenic epitopes on humancancers. Additional applications of SEM may also result in thetargeted production of mAbs and the identification of genesassociated with important physiological processes, includingcellular growth and differentiation, immunological recogni-tion, tumorigenesis, metastasis, cellular senescence, atypicalmultiple drug resistance, and autoimmune disease.

In summary, we presently demonstrate direct applications ofthe rapid expression cloning and SEM technologies for thedevelopment of mAbs and the cloning of the PCTA-1 geneassociated with human prostate cancer. The PCTA-1 gene andthe recently identified prostate tumor inducing gene PTI-1 (6)are genetic elements that can distinguish prostate cancer fromnormal prostate and BPH. Although further studies are re-quired, it appears that PCTA-1 may represent an earliergenetic change in human prostate cancer development thanPTI-1. In this context, both the Pro mAbs and the PCTA-1gene should find direct applications for prostate cancer diag-

nosis and staging, and they may also represent importanttherapeutic reagents for intervention in this pervasive andoften fatal neoplastic disease.

This research was supported in part by National Institutes of HealthGrant CA35675, the Chernow Endowment, the Samuel WaxmanCancer Foundation, an award from the CaP CURE Foundation, anda gift from S. Leslie Misrock, Esq. P.B.F. is a Chernow ResearchScientist in the Departments of Pathology and Urology.

1. Waldmann, T. A. (1991) Science 252, 1657-1662.2. Leon, J. A., Goldstein, N. I. & Fisher, P. B. (1994) Pharmacol.

Ther. 61, 237-278.3. Shen, R., Su, Z.-z., Olsson, C. A., Goldstein, N. I. & Fisher, P. B.

(1994) J. Natl. Cancer Inst. 86, 91-98.4. Fisher, P. B. (1995) Pharm. Technol. 19, 42-48.5. Su, Z.-z., Olsson, C. A., Zimmer, S. G. & Fisher, P. B. (1992)

Anticancer Res. 12, 297-304.6. Shen, R., Su, Z.-z., Olsson, C. A. & Fisher, P. B. (1995) Proc.

Natl. Acad. Sci. USA 92, 6778-6782.7. Hadari, Y. R., Paz, K., Dekel, R., Mestrovic, T., Accili, D. & Zick,

Y. (1995). 1. Biol. Chem. 270, 3447-3453.8. Adams, M. D., Dubnick, M., Kerlavage, A. R., Moreno, R.,

Kelley, J. M.,Utterback, T. R., Nagle, J. W., Fields, C. & Venter,J. C. (1992) Nature (London) 355, 632-634.

9. Hughes, R. C. (1992) Curr. Opin Struct. Biol. 2, 687-692.10. Hirabayashi, J. & Kasai, K-i. (1993) Glycobiology 3, 297-304.11. Barondes, S. H., Cooper, D. N. W., Gitt, M. A. & Leffler, H.

(1994) J. Biol. Chem. 269, 20807-20810.12. Horoszewicz, J. S., Leong, S. S., Kawinski, E., Karr, J. P.,

Rosenthal, H., Ming Chu, T., Mirand, E. A. & Murphy, G. G.(1983) Cancer Res. 43, 1809-1818.

13. Reddy, P. G., Su, Z.-z. & Fisher, P. B. (1993) in Chromosome andGenetic Analysis: Methods in Molecular Genetics, ed. Adolph,K. W. (Academic, Orlando, FL), Vol. 1, pp. 68-102.

14. Jiang, H. & Fisher, P. B. (1993) Mol. Cell. Differ. 1, 285-299.15. Sanger, F., Nicklen, S. & Coulson, A. R. (1977) Proc. Natl. Acad.

Sci. USA 74, 5463-5467.16. Miranda, A., Duigou, G. J., Hernandez, E. & Fisher, P. B. (1988)

J. Cell Sci. 89, 481-493.17. Jiang, H., Lin, J., Su, Z.-z., Herlyn, M., Kerbel, R. S., Weissman,

B. E., Welch, D. R. & Fisher, P. B. (1995) Oncogene 10, 1855-1864, 1995.

18. Raz, A. & Lotan, R. (1987) Cancer Metastasis Rev. 6, 433-452.19. Raz, A., Carmi, P., Raz, T., Hogan, V., Mohamed, A. & Wolman,

S. R. (1991) Cancer Res. 51, 2173-2178.20. Drickamer, K. & Taylor, M. E. (1993) Annu. Rev. Cell Biol. 9,

237-264.21. Lindstedt, R., Apodaca, G., Barondes, S. H., Mostov, K. E. &

Leffler, H. (1993) J. Biol. Chem. 268, 11750-11757.22. Raz, A., Carmi, P. & Pazerini, G. (1987) CancerRes. 48,645-649.23. Rosen, S. D. (1993) Semin. Immunol. 5, 237-247.24. Barondes, S. H. (1984) Science 223, 1259-1264.25. Liotta, L. A., Steeg, P. G. & Stetler-Stevenson, W. G. (1991) Cell

64, 327-336.26. Su, Z.-z., Yemul, S., Estabrook, A., Friedman, R. M., Zimmer,

S. G. & Fisher, P. B. (1995) Int. J. Oncol. 7, 1279-1284.27. Raz, A., Zhu, D., Hogan, V., Shah, N., Raz, T., Karkash, R.,

Pazerini, G. & Carmi, P. (1990) Int. J. Cancer 46, 871-877.28. Irimura, T., Matsushita, Y., Sutton, R. C., Carralero, D., Ohan-

nesian, D. W., Cleary, K. R., Ota, D. M., Nicolson, G. L. &Lotan, R. (1991) Cancer Res. 51, 387-393.

29. Lotan, R., Ito, H., Yasui, W., Yokoaki, H., Lotan, D. & Tahara,E. (1994) Int. J. Cancer 56, 474-480.

30. Schoeppner, H. L., Raz, A., Ho, S. B. & Bresalier, R. S. (1995)Cancer 75, 2818-2826.

31. Ohannesian, D. W., Lotan, D., Jessup, P. T. J. M., Fukuda, M.,Gabius, H.-J. & Lotan, R. (1995) Cancer Res. 55, 2191-2199.

32. Lotan, R., Lotan, D. & Raz, A. (1985) CancerRes. 45,4349-4353.33. Raz, A., Meromsky, L. & Lotan, R. (1986) Cancer Res. 46,

3667-3672.34. Meromsky, L., Lotan, R. & Raz, A. (1986) Cancer Res. 46,

5270-5276.

Medical Sciences: Su et aL

Dow

nloa

ded

by g

uest

on

Nov

embe

r 17

, 202

0