use of the lectin from amarant hus ... - cancer research · had a score of 224 Â±76 (p < 0.001...

[CANCER RESEARCH 51. 657-665. January 15. 1991]

Use of the Lectin from Amarant hus cautiat us as a Histochemical Probe ofProliferating Colonie Epithelial Cells1

C. R. Boland,2 Y-F. Chen, S. J. Rinderle, J. H. Resau, G. D. Luk, H. T. Lynch, and I. J. Goldstein

Department of Internal Medicine, Veterans Administration Medical Center, University of Michigan School of Medicine, Ann Arbor, Michigan 48105 fC. R. B.J;Department of Biological Chemistry, University of Michigan School of Medicine, Ann Arbor, Michigan 48109 [Y-F. C., S. J. R., I. J. G.j; Human Tissue Resource,Department of Pathology, University of Maryland School of Medicine, Baltimore, Maryland 21201 [J. H. R.]; Department of Internal Medicine, Wayne State UniversitySchool of Medicine, Detroit, Michigan 48201 [G. D. L.]; and Department of Preventative Medicine, Creighton University School of Medicine, Omaha, Nebraska 68178[H.T.L.]

ABSTRACT

A newly isolated lectin, Amaranthus caudatus agglutinin (also calledamaranthin or ACA), which binds to the Thomsen-Friedenreich antigen(T-antigen) and its sialylated variants, was used as a histochemical probe

for proliferating cells in sections of human colonie tissues. Bindinginhibition studies revealed that ACA binds to different sites on histolÃ³gica! sections when compared to peanut agglutinin, which also recognizesthe T-antigen. ACA bound selectively to the cells at the base of the

colonie crypt [46 Â±4% (SEM) of glands) which is the zone of proliferationin this tissue and preferentially labeled cytoplasmic and apical membraneglycoconjugates. Only 7 Â±2% of the upper portions of the colonie cryptswere labeled ( /' < 0.001 compared to the base), and this was largely a

result of extensive labeling in 2 of 23 samples studies. A marked increasein histochemical labeling by ACA was seen in adenomatous polyps andadenocarcinomas of the colon, in which 82 Â±7 and 97 Â±2% of theglandular units were labeled, respectively. Transitional mucosa and connective tissue adjacent to cancers were also labeled by ACA. Neuramin-

idase studies indicated that removal of sialic acid residues enhancedbinding by peanut agglutinin, but not ACA, to glycoconjugates in cancerspecimens. Specimens of colonie tissue from patients with familial adenomatous polyposis (FAP) were examined with ACA; 83 Â± 7% ofadenomatous glands and 60 Â±7% of glands in flat, normal-appearing

tissue were labeled. Colonie tissues from persons at 50% risk for hereditary nonpolyposis colorectal cancer (HNPCC), FAP, and normal colonswere studied and given "weighted average" labeling scores that ranged

from 0-400 to accommodate variable intensity and distribution of label

ing. Normal colons had a weighted average score of 65 Â±33; FAP tissueshad a score of 224 Â±76 (P < 0.001 compared to normal colon) andHNPCC tissues had a score of 74 Â±70 (P < 0.05 compared to normalcolon). A group of five HNPCC cases had scores of 203 Â±43 (P < 0.001compared to normal colon). ACA labels glycoconjugates in the prolifer-

ative region of normal human colonie epithelium and neoplastic lesionsof the colon. The results of FAP and HNPCC tissues suggest that it maybe useful for identifying foci of abnormal proliferation in familial colorectal cancer syndromes.

INTRODUCTION

A lectin has been isolated from the South American plantAmaranthus caudatus that binds to the Thomsen-Friedenreichantigen (T-antigen) and its sialylated variants (1). This lectin,amaranthin or ACA,' has a subunit molecular weight of ap-

Received 5/30/90; accepted 10/29/90.The costs of publication of this article were defrayed in part by the payment

of page charges. This article must therefore be hereby marked advertisement inaccordance with 18 U.S.C. Section 1734 solely to indicate this fact.

' Supported in part by the Research Service of the Veterans Administration

and PHS Grants CA39233 (C. R. B.). GM29470 (I. J. G.). CA 45831 (G. D. L.).CA 42705 (H. T. L.), and Contract CP-95624 (J. H. R.). NIH. Department ofHealth and Human Services. S. J. R. was a Predoctoral Fellow supported byPharmacological Sciences Training Grant 2T32-GM27767-11. NIH. Department of Health and Human Services, during the course of the work.

2To whom requests for reprints should be addressed, at GI Section (11 ID).

VA Medical Center. 2215 Fuller Road. Ann Arbor. MI 48105.3The abbreviations used are: ACA, amaranthin or Amaranthus caudatus ag

glutinin; PNA. peanut agglutinin: FAP, familial adenomatous polyposis:HNPCC, hereditary nonpolyposis colorectal cancer: T-antigen, Thomsen-Frie-

proximately 33,000 and exists as a homodimer (M, = 63,500)with one binding site per monomer.'1 The binding specificity of

ACA has been determined by the ability of defined haptenoligosaccharides to inhibit the precipitation of asialoovine sub-maxillary mucin (1). These studies defined the critical domainsin the T-antigen disaccharide necessary for lectin-carbohydrateinteraction, as illustrated in Fig. 1.

Inasmuch as ACA was isolated by its interaction with immobilized T-antigen, it was anticipated that its binding specificity would be similar to that of PNA (2, 3). A series ofsynthetically modified oligosaccharides was used to determinethe specific sites of interaction between ACA and the T-antigen(1). As illustrated in Fig. 2, ACA and PNA both recognized T-antigen but interacted with different portions of the disaccharide. These and other experiments have demonstrated that PNAwould bind T-antigen even if the C-6 hydroxyl group of GalNAcwere substituted4 (Fig. 2) and that ACA would interact with T-antigen even if both the C-6 hydroxyl group of GalNAc and theC'-3 hydroxyl group of galactose were substituted (1). This is

of some importance, since the T-antigen may be sialylated ateither of the two sugars (4), and previously it had been assumedthat sialylation of the T-antigen would make it inaccessible forrecognition by PNA (3).

The PNA lectin has been an important probe used by his-tochemists interested in colorectal neoplasia (5, 6). PNA doesnot bind readily to mucins found in the normal colon but bindsto mucin secreted by colon cancers (5-8). In addition, PNAbinds to glycoconjugates present in a variety of premalignantlesions in the colon (9-13). Cooper (6) reported that the treatment of histological sections of human colonie tissue withneuraminidase resulted in the "unmasking" of cryptic T-anti-

gens to which PNA subsequently bound (6); however, this hasnot been the experience of all investigators (5). Moreover, aseries of oligosaccharides isolated from the mucin of normalhuman colonie tissues revealed no evidence of the T-antigen orits sialylated variants (14, 15). Thus, it is not clear whether avariant form of T-antigen may be synthesized in normal colonietissues. The availability of a new probe for the T-antigen whichwas capable of recognizing variations in the T-antigen thatwould not be recognized by PNA led to our use of ACA as ahistochemical probe of human colonie tissues. Specimens ofnormal human colon, neoplastic human colon, and a series ofpotentially premalignant lesions were examined using this lectin to determine whether a variant form of T-antigen wasresponsible for some of the above described controversies. Wehave found that ACA binds selectively in the portion of thecolonie crypt where epithelial cell proliferation ordinarily oc-

denreich antigen: Gal. galactose: GalNAc. TV-acetylgalactosamine; PBS, phosphate-buffered saline: BPA. BauhÃ¬neapurpurea agglutinin; RCA,, Ricinis corn-munis agglutinin,; NeuSAc, iV-acetyl neuraminic acid (a sialic acid); DBA, Doli-chos bifloras agglutinin.

4 Unpublished data from the dissertation of S. J. R.

657

on May 22, 2020. © 1991 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

http://cancerres.aacrjournals.org/

ACÃ• HlSIOt MIMISim [N COLON

T-Antigen - Amaranthin Interaction T-Antigen - PNA Interaction

6 CH OH

Galp1.3GalNAcc. 1 â€¢-O Ser/thr

Fig. 1. T-antigen-amaranthin interaction. The T-antigen is a disaccharideconsisting of galactose (left) attached through a ii\,3 anomeric linkage to N-acetylgalactosamine(r//?/if). which is O-glycosidically bound to serine or threonine(Ser/Thr) in the peptide backbone. As indicated by the asterisks, the C'-4 hydroxylgroup on galactose and the C-4 hydroxyl group on GalNAc must be free ofsubstituent groups for the leclin ACA to interact with T-antigen. In addition,galactose is &ÃŒ.3linked to GalNAc in T-antigen and the A'-acetyl group on C-2

is important for lectin interaction. Finally. C-l must be ir-linked to the hydroxyamino acid. Of note, substituent groups at the C"-3 hydroxyl group of galactose

or the C 6 hydroxyl group of GalNAc are tolerated without destroying the abilityto interact with ACA.

curs, binds to a variety of neoplastic lesions, and recognizesglycoconjugates in certain potentially premalignant tissues.

MATERIALS AND METHODS

Lectin Preparation. ACA was purified from the seeds of A. caudatusby fractionation on DEAE-cellulose followed by affinity chromatogra-phy using Synsorb-T beads (Calbiochem, Calgary, Alberta, Canada)(Gai/il,3GalNAc-Â«-O-Synsorb) (1). The lectin was biotinylated as follows. Biotinyl-A'-hydroxysuccinimide dissolved in dimcthylformamidewas added to highly purified AC'A that had been dissolved in PBS with

0.1 M NaHCOj, pH 8.3, at a protein concentration of 5.15 mg/ml.using a 5:1, mol:mol. ratio. GalNAc, final concentration 10 mg/ml,was added during biotinylation to protect the sugar-binding sites of thelectin. The solution was gently stirred at room temperature for 6 h andthoroughly dialyzed against PBS and purified by affinity chromatog-raphy on Synsorb-T beads. The binding activity of biotinylated ACAwas compared with that of nonbiotinylated ACA by hemagglutinationusing the same amount of protein, in which the titers were 1:32 and1:16, respectively.

Biotinylated PNA was purchased from Vector Laboratories (Burlin-game. CA).

Tissue Procurement. Specimens of adenocarcinoma of the colon andadenomatous colonie polyps from resected specimens were obtainedfrom the Pathology Service at the Ann Arbor Veterans AdministrationMedical Center. These specimens were obtained during surgery orcolonoscopic polypectomy, fixed in buffered formalin, and embeddedin paraffin for routine pathological examination. Additional 4-^msections were cut for lectin histochemistry after the pathological diagnosis was confirmed (C. R. B.).

Full-thickness specimens of normal colonie epithelium were obtainedfrom the Immediate Autopsy program at the University of MarylandHuman Tissue Resource (J. H. R.). These tissues came from accidentvictims with no pathological abnormality of the colon, were immediately fixed in formalin after death, and embedded in paraffin andsectioned for lectin histochemistry. These protocols were approved bythe University of Maryland Human Volunteer Research Committee.

Full-thickness specimens of colonie tissue were obtained from colec-tomy specimens of familial adenomatous polyposis at Johns HopkinsHospital (G. D. L.), fixed in formalin, and embedded in paraffin fortissue sectioning. The tissue specimens contained normal-appearing fiatmucosa, as well as a polypoid, adenomatous tissue.

Oâ€”Ser/Thr

Gal |! 1.3GalNACtf 1 Â»â€¢O Ser/thr

Fig. 2. T-antigen-peanut agglutinin interaction. The T-antigen is a disaccharidein which galactose (left) is attached through a ii\,3 anomeric linkage to GalNAc,which is O-glycosidically bound to a serine or threoninc (Ser/Thr) in a mucin-!>pe glycoprotein. Asterisks, regions important for the lectin-sugar interaction.The C'-2, C'-3, and C'-4 hydroxyl groups of galactose and the C-4 hydroxyl of

GalNAc must be unsubstituted for lectin interaction. Therefore, any substituenton the C'-2, C'-3, C'-4, or C'-6 hydroxyl groups of galactose may interfere withrecognition by PNA; however, substituents on the C-6 hydroxyl group of GalNAcare tolerated* (1-3).

Rectal biopsy specimens were obtained from individuals who werefirst degree relatives of patients with HNPCC, as identified by Dr. H.T. Lynch. These patients were the siblings or offspring of index patientswho fit the criteria for Lynch syndrome 1 or II (16). However, sincethere is no definitive genetic or phenotypic marker for this familialsyndrome, it is not possible to be certain that these patients representa single homogeneous genetic abnormality. The specimens were fixedin formalin and embedded in paraffin for tissue sectioning.

Lectin Histochemistry. Paraffin-embedded tissue sections, 4 /jm, attached to glass slides were deparaffini/ed by serial incubations in xylene(twice), absolute ethanol (thrice), 95% ethanol, 70% ethanol, and waterand finally equilibrated in PBS, pH 7.0, containing 0.1% bovine serumalbumin (Sigma). The slides were blotted dry adjacent to the tissuesections, and biotinylated lectin was added at a concentration of 10 Â¿jg/ml (for ACA) or 20 n%/m\ (for PNA) in a quantity sufficient tocompletely cover the hydrated tissue ( 100-200 ^1).The lectin was rinsedoff after a 30-min incubation by eight 2-s dips into PBS. The slideswere again blotted, and the tissue was covered with streptavidin-horse-radish peroxidase (Zymed Laboratories, South San Francisco, CA) ata concentration of 50 ng/ml for 30 min. The slides were rinsed asmentioned above, and the pigmented reaction product was developedby the addition of a freshly prepared chromogen reagent solutionprepared within 2 h of use as follows. 3,3'-Diaminobenzidine was

aliquoted at 5 mg/ml in 0.01 M Tris saline, pH 7.5, and frozen atâ€”70Â°Cuntil use. For use, the aliquot was thawed, diluted 1:10 in 0.01

M Tris saline. pH 7.5, and 3.3 n\ 30% H2O2 was added to each 5 ml.This solution was allowed to stand for 10 min, and 100-200 ^1 wasapplied to the tissue sections that had been incubated earlier withbiotinylated lectin and streptavidin-horseradish peroxidase. After 10min, the slides were rinsed with albumin-free PBS. The tissues werebriefly counterstained for 50 s with Gill's 3 hematoxylin (Sigma), rinsed

with water, serially dehydrated using graded alcohols and xylene andembedded in Permount (Allied-Fischer Scientific, Livonia, MI).

The specificity of lectin binding was evaluated by preincubating thebiotinylated lectin with inhibitory hapten sugar for 1-2 h prior to itsapplication to the tissue. In each instance, the reagents were preparedso that the final concentration of lectin on the slide was either 10 or 20Mg/ml and the hapten sugar was as indicated in "Results." Galactose,lactose, GalNAc, and galactose-/Jl-3-yV-acetylgalactosamine-Â«-O-bcn-zyl were obtained from Sigma.

Sialidase digestions were performed on the tissue sections afterhydration. Vibrio cholerae neuraminidase (Serva, Heidelberg) was diluted in 50 mM citrate-phosphate buffer, pH 4.5, at a concentration of0.43 milliunits/ml. and 100-200 ^1 was incubated on the slide at 37Â°C

for 16 h. Newcastle disease virus neuraminidase (viral suspension;

658



ACÃ• HISTOCHEMISTRV IN COLON

United States Department of Agriculture, Agriculture Research Service,Athens, GA) was diluted in 50 HIM cacodylate buffer, pH 6.0, at aconcentration of 0.5 milliunits/ml, and 100-200 p\ was added to thetissue for a 22-h incubation at 37Â°C.Controls for these experiments

utilized the same buffers without neuraminidase.Scoring the Tissue Sections. The stained sections were examined with

a light microscope using a x25 and x40 objective lenses and xlOeyepiece. Evaluation of lectin binding required the use of several approaches. First, the binding specificity was confirmed by coincubationof lectin with hapten sugars known to bind the lectin maximally.Maximum labeling by the brown reaction product was arbitrarily assigned a score of 4+ (8). No staining of the slide compared to controlsections (i.e.. in which lectin was omitted from the staining procedure)was assigned a score of 0. Intermediate scores of 1* (trace or weakstaining definitely not seen on control slides) and 2-3* (stronger stain

ing, but definitely less than maximal) were assigned. Although subjective, the histochemical scoring system using lectins has been demonstrated to correlate well with an objective measure of the binding oflectins to glycoproteins extracted from colonie tumors (8). This methodwas used to confirm the binding specificity to known "positive" tissues

and to evaluate changes in lectin binding after neuraminidase treatment.A somewhat more simplified system (using a global 0-3* score) was

used for documenting carbohydrate-binding specificity, as describedbelow.

A second method for scoring tissues was used to survey normalcolonie tissues, adenomatous polyps, and cancers. For these studies,individual epithelial "units" were counted as either positive or negative;

i.e., each crypt of Lieberkuhn or neoplastic gland was considered anindependent scorable unit, and the tissue score was expressed in termsof the percentage of units that showed any degree of labeling (l*-4*).

The normal colonie crypt was further subdivided into the lower one-half and upper one-half for independent scoring because of the nonun-iform distribution of labeling between the proliferative region (at thebase) and the region of mature, differentiated cells (at the top).

Finally, an examination of the HNPCC tissues revealed subtle variations in labeling that required a more sophisticated method of scoring.These tissues were histologically normal but showed variations inintensity of labeling when compared to the normal colonie tissues.Therefore, each individual crypt of Lieberkuhn was given a score from0-4*, and the labeling was expressed as a "weighted average," or the

sum of scores/100 crypts, which could range from 0 (all 100 crypts notlabeled) to 400 (all 100 crypts maximally labeled). A modification ofthis approach has been used to evaluate histochemical results usingother lectins and provides reproducible scores on tissue sections (17).

For some experiments, a simplified 0-3 scoring system was used, inwhich a global estimate of absent (0). trace (1*), intermediate (2*), andintense (3*) staining was given for the whole section. Essentially, thetissues were scored globally, and the strongly labeled tissues (i.e., 3*and 4*) were grouped together under a 3* score. This method was used

for two experiments, the hapten inhibition studies (Table 1) and thecomparison of ACA and PNA (Table 4), in which only relative degreesof labeling were relevant.

RESULTS

Lectin-binding Specificity

Specimens of well-differentiated colonie cancers were incubated with biotinylated lectins. Working concentrations on the

slide were 10 Â¿Â¿g/mlfor ACA and 20 Â¿ig/mlfor PNA. Thehapten sugars Gal, GalNAc, and Gal/n,3GalNAc-Â«-0-benzyl(T-antigen) were used to inhibit lectin binding to the tissues.Using the biotinylated lectins, ACA produced intense labelingon colon cancers; for the purposes of the inhibition studies thescores were recorded from 0-3* since the cancers were uni

formly labeled, and a simplified scoring system was most useful.Glycoconjugates secreted into the malignant glands or luminawere the predominant locations of labeling; cytoplasmic labeling was less intense in all specimens. Biotinylated PNA labeledthese tissues much less intensely and produced relative scoresof l+-2* on the 15 cancer specimens as the maximal degree of

labeling. The ability of hapten sugars to inhibit labeling withthese lectins is indicated in Table 1. Labeling by ACA wasinhibited by concentrations >250 HIM GalNAc and >100 JJMT-antigen but required 400 mivi lactose or galactose for partialinhibition of binding. PNA was inhibited by >10 mivi galactoseor lactose, >100 ^M T-antigen, and >100 mM GalNAc.

Survey of Tissue Labeling with ACA

Lectin histochemistry was analyzed with biotinylated ACA,and tissue labeling was expressed in terms of the percentage ofepithelial units showing any degree of definite reaction product(l+-4*).

Normal Colon. Twenty-three specimens of normal colon wereobtained from 16 "immediate autopsy" cases and were free of

histolÃ³gica! abnormality. Full-thickness, well-oriented sectionswere available in each case. Initial examination of these tissuesindicated that ACA labeling was faint (1+ for most specimens)

and preferentially occurred at the base of the crypt of Lieberkuhn in the apical cytoplasmic region, microvillar membrane,and glycocalyx but was less frequently seen in the upper one-half of the crypt, which occurred in only 2 of 16 of the normalcolons (Fig. 3). Goblet cell mucin, connective tissue, and laminapropria were not stained. The epithelial crypt unit was evaluatedseparately for staining of the base or lower one-half and upperone-half of the crypt. Labeling was seen in the lower one-halfor base of the crypt in all colonie specimens and was present in46 Â±4 (SEM)% of all crypts in the 23 specimens studied. Theupper one-half of the crypts showed staining in 7 Â±2% of allcrypts and consisted of secreted material in the lumen of thegland in some cases but definitely labeled (l*-2+) the apical

membrane and cytoplasm in 2 of 23 of the specimens. In thesetwo cases, >30% of the epithelial units had l+-2* labeling. The

difference in labeling between upper and lower crypt regionswas statistically significant (P < 0.001). More important, however, 18 of 23 normal colonie specimens showed no labeling inthe upper portion of the crypt.

No labeling of mucosal glycoconjugates in the normal colonwas observed with PNA. Treatment with V. cholera or Newcastle disease virus neuraminidase did not produce an increase inlabeling by PNA.

Neoplastic Colon. Twenty-five specimens of colonie adeno-

Table 1 Inhibitory hapten sugarsInhibition of lectin binding by hapten sugars on well-differentiated colonie adenocarcinoma sections. Labeling was scored on a simplified scale from 0-3* for this

experiment, as described in "Materials and Methods."

LectinACA(10Mg/mL)

PNA (20 Mg/mL)No

sugar3*

1-2*T-antigen

(^M)1000

02000 04000 02.53* 1-2*GalNAc

(mM)103*

1-2*203* 1-2*1001*0250Trace02.53* TraceGal

(HIM)103*

0203* 01003* 04002* 0Lactose

(HIM)2002-3*

04001-2* 0

659



ACÃ• HISTOCHEMISTRY IN COLON

Fig. 3. l.cctin histochemistry-AOA staining of normal colon. A, low power. This photomicrograph shows three colonie crypts in which AOA selectively binds toglycoconjugatcs in the basal portion of the crypt. The reaction product appears black in this photomicrograph. H. low power. This is a photomicrograph of normalcolonie mucosa processed identically to that seen in A, except that the biotinylated lectin (ACA) was omitted from the procedure. This indicates the low level ofbackground labeling in normal tissue. (. this is a higher power view of the lower portion of a crypt of a normal colonie crypt stained with AC'A. Note the black

reaction product in the apical portion of the cytoplasm and microvillar membranes and the absence of staining in the middle and upper portions of the crypt.

carcinoma were studied with ACA, and 15 specimens werestudied with PNA. In all cases, labeling by ACA was moreintense and more diffusely distributed than that seen with PNA.which was weaker in intensity, generally localized to secretedmaterial, and focal within most of the tumors. Labeling ofcytoplasmic glycoconjugates was commonly seen with ACA(Fig. 4). Every cancer specimen showed extensive labeling withACA, and in a limited study of 13 cancer specimens, 97 Â±2%of the malignant glands showed 1+--V labeling of the secreted

or cytoplasmic glycoconjugates. This was significantly greaterthan that seen in the normal colon (P < 0.001).

Cancer specimens were digested with neuraminidases fromV. choleras [which hydrolyzes both Â«2,3-and Â«2,6-linked sialicacid (NeuSAc)] and Newcastle disease virus (which preferentially hydrolyzes Â«2.3-linked NeuSAc) (18, 19). Treatment withV. choleras neuraminidase greatly enhanced the intensity anddistribution of PNA binding, exposing binding sites in secretedand cytoplasmic glycoconjugates. However, the use of Newcastle disease virus neuraminidase had much less effect on theintensity of labeling by PNA. Both enzymes enhanced PNAstaining in connective tissue. Neither sialidase enhanced labeling with ACA.

The mucosa immediately adjacent to colonie cancer, called"transitional mucosa," is known to express abnormal glycocon

jugates in spite of its nonneoplastic pathological appearance (5.20). This mucosa demonstrated 1* labeling in the upper and

lower crypt region using ACA. Of interest, the collagen, bloodvessels, and muscle surrounding invasive cancers were variablylabeled in 22 of 25 tumors, which was never seen in surroundingnormal colonie epithelium.

A small group of six adenomatous polyps was studied usingACA. The lectin bound to 82 Â±7% of the glands in theadenomatous polyps and stained cytoplasmic as well as secretedglycoconjugates (Fig. 5). A larger group of adenomatous polypsconfirms this observation.5

1E. A. Boohaker and C. R. Boland. manuscript in preparation.

FAP. A total of 28 specimens of colonie tissue from 12patients with FAP was examined with ACA. Each specimencontained flat, normal-appearing epithelium and small adenomatous polyps. Epithelial units from flat and polypoid tissue werescored separately. Polypoid tissues demonstrated labeling in 83Â±7% of the glands, which is similar to that seen in the smallseries of sporadic adenomatous polyps, all of which were largerthan those present in the FAP sections. The flat tissue in FAPdemonstrated labeling in 60 Â±7% of the glands (P = 0.055compared with polypoid tissue), and labeling was inappropriately seen in the upper portion of the crypt as frequently asin the lower portion (Fig. 6). Labeling in FAP tissues wassignificantly greater than that seen at the apical portion of thecrypt in normal colons (P < 0.001) but was not different fromthat at the basal portion of the crypt (0.05 <P< 0.10); however,the intensity of labeling was greater in FAP sections in allportions of the crypt. Secreted glycoconjugates demonstratedan increase in staining intensity. The histochemical results aresummarized in Table 2.

HNPCC. HNPCC is a more problematic tissue to evaluate.Rectal biopsy specimens were obtained from individuals at riskfor this syndrome, but no markers have been available todefinitively identify truly affected patients. These tissues wereall histologically normal, and lectin staining was scored from0-400 using the "weighted average" described in "Material andMethods." An examination of the HNPCC tissues revealed that

a subset exhibited an inappropriate increase in labeling withACA (compared with colonie tissues from the Immediate Autopsy Program) showing both an increase in intensity andlabeling of the cytoplasm in both the upper and lower portionsof the crypt (Fig. 7). For these tissues, each crypt was consideredan independent "unit" and was scored regardless of the cellular

location within the crypt.The weighted average for normal colon was 65 Â±33. The

score for HNPCC tissues was 74 Â±70 (P < 0.05 comparedwith normal colon). However, it was apparent that the HNPCCscores were not uniformly distributed and contained a group of

660




Fig. 4. Binding of ACÃ•to adenocarcinomaof the colon. A cluster of malignant glands hasbeen labeled with ACA. and the cells appeardarker than the background stroma. Reactionproduct is darkest in the secreted glycoconju-gates.

outliers with high scores (Fig. 8). Five specimens in the grouphad a score of 203 Â±43 (P < 0.001 compared with normalcolon). When 11 of the FAP tissues were scored in this manner,a weighted average of 224 Â±76 was obtained (P < 0.001compared with normal colon), which was not different fromthat seen in the HNPCC outliers (Table 3).

Direct Comparison of Biotinylated ACA and PNA. Since priorwork with PNA on familial colon cancer had been undertakenusing the fluorescein-conjugated lectin (17), we compared bio-tinylated ACA and PNA on paired sections, using the simplified, global 0-3* scoring system as was used for hapten sugar

inhibition studies (Table 1). This experiment demonstrated thatbiotinylated PNA does not bind to normal or FAP colonietissues, rarely binds to HNPCC, and binds l-3+ in colon

cancers, as demonstrated in Table 4.

DISCUSSION

ACA is a newly isolated, T-antigen-binding lectin that weadapted for histochemical studies of colorectal neoplasia. Dataobtained during the characterization of the lectin suggested thatits recognition specificity might differ from that of PNA, another lectin that binds T-antigen (1). We found that ACA hasunique tissue-labeling characteristics compared to PNA, whichsuggests that variable forms of T-antigen may be produced inthe colon under different biological circumstances. Lectins areparticularly valuable as histochemical probes because of theirability to generate a reliable and uniform preparation of a stableagglutinin, the possibility for documentation of binding specificity with a high degree of precision (1, 21), and the facile useof lectins on formalin-fixed, paraffin-embedded tissue sections(5-13, 17,22).

The most important and novel aspect of ACA histochemistrywas our observation that it binds to epithelial cell glycoconju-gates at the basal portion of the crypt of Lieberkuhn, theproliferative zone in normal colonie mucosa. Labeling preferentially occurs in cytoplasmic, mucin-type glycoconjugates (al

though not necessarily mucins per se), on the apical membranesmembranes and in secretions at the base of the colonie cryptbut typically is not expressed as the cells mature and migratetoward the luminal surface of the epithelium. In a small numberof specimens, labeling also was seen in the upper portions ofthe colonie crypt, but this was exceptional. We have previouslyreported that ^-linked galactosyl residues are preferentiallyexpressed in goblet cell mucins from the lower one-half of the

colonie crypt using BPA and RCA, (5). A major limitation ofthese two lectins is that their binding specificity is relatively-

broad, and epithelial cell glycoconjugates are labeled by bothlectins throughout the colonie crypt, limiting their utility tofurther identify proliferation-associated structures. ACA has amore restricted range of labeling in the colon but shares withBPA and RCA, preferential binding to glycoconjugates in thelower one-half or proliferative region of the colonie crypt. Thisoccurs in spite of the apparent paradox that BPA and RCA,bind to cytoplasmic and secreted glycoconjugates possessingterminal galactosyl residues (21), whereas ACA is a T-antigen-binding lectin that has a critical interaction with GalNAc (1).BPA, RCA,, and ACA have a labeling pattern in the humancolon that is complementary to DBA and soybean agglutinin;these latter lectins label goblet cell mucin in the upper one-halfof the crypt in those epithelial cells that have migrated into thezone of differentiation (5). DBA and soybean agglutinin appeared histochemically to bind preferentially to goblet cellmucin. BPA and RCA, label mucins in the lower crypt regionbut, as previously mentioned, also labeled cytoplasmic structures throughout the crypt region (5). In this study, we foundthat ACA labels cytoplasmic and membrane-bound structuresand tends not to label goblet cell mucin in the normal colon.Secreted glycoconjugates are prominently labeled in neoplasticlesions. Therefore, entirely different molecular species mayaccount for the labeling differences between ACA and BPA/RCA,.

We observed ACA binding in virtually all neoplastic lesionsof the colon, including small adenomas. This is of particular

661




A** A, S >.

B

Fig. 5. Binding of ACÃ• Io adenomalous polyps. A, this photomicrographdemonstrates the binding of ACA to secreted and cytoplasmic glycoconjugates ina benign adenomatous polyp. B, this tissue section of an adenomatous polyp wascut immediately adjacent to that seen in A', however, the biotinylatcd lectin wasomitted from the histochemical procedure, indicating the background labeling inadenomatous tissue.

Fig. 6. Labeling of ACA on FAR tissue. A, this full-thickness tissue sectiondemonstrates the avid binding of ACA to glycoconjugates throughout the cryptin flat, normal-appearing colonie epithelium from a patient with FAP. includingthe goblet cell mucin. //. this tissue section was cut immediately adjacent to thatin A; however, the biotinylated lectin was omitted from the histochemical procedure.

interest since PNA binds to colonie neoplastic lesions at a timelater in their natural history. PNA tends not to label adenomatous polyps <1 cm in diameter and binds to an increasingpercentage of neoplastic glands in polyps that are larger, morevillous, or more dysplastic (9, 22). The appearance of glycoconjugates that bind PNA is therefore a relatively late phenomenonin colorectal neoplasia, occurring in the same time frame asgenetic mutations in the ras oncogene and deletions on chromosomes 5, 17, and 18 (23, 24). On the other hand, glycoconjugates that bind ACA occur earlier in the natural history ofneoplasia, in the time frame reported for gene hypomethylation(25).

ACA-binding glycoconjugates are present in the proliferativeregion of the normal colon. It is not known whether the struc-ture(s) recognized at the bottom of the colonie crypt is identicalto those found in neoplastic tissue, which is a critical issue inneed of clarification. ACA prominently labels secreted glycoconjugates in cancers (whereas cytoplasmic structures are theprimary sites of labeling in the normal colon), suggesting thatmodifications in glycosylation may be associated with changesin glycoprotein trafficking.

The identification of T-antigen-like structures is highly dependent upon the methods used for their detection. Fluoresceinisothiocyanate-conjugated PNA does not bind to glycoconju-

662



ACÃ• HISTOCHEMISTRV IN COLON

Table 2 Summary of ACÃ•histocheniistry IThe percentage of epithelial units showing any definite lectin labeling is

expressed as the mean Â±SEM.

Tissue^r of epithelialunits labeled

Normal colonie epithelium (N = 23)Upper one-half of crypt 7 Â±2Lower one-half of crypt 46 Â±4

Well-differentiated colonie cancer (N = 25) 97 Â±2Adenomatous colonie polyps (,V = 6) 82 Â±7Familial adenomatous polyposis (N = 28)

Flat tissue 60 Â±7Polypoid tissue 83 Â±7

Fig. 7. Binding of ACA to normal-appearing colonie tissues from patients withHNPCC. This is a section of normal-appearing tissue from a patient at risk forHNPCC. This tissue was selected from the group outliers who had an increase inACA binding compared to the rest of the group. The labeling of epithelial cellglycoconjugates in this section is similar to that seen in FAP; omission of lectingave minimal background labeling, as shown in Fig. 6Ã„.

colon, although a cryptic variant of T-antigen may be presentwith chain extension or substituents on the C'-2, C'-3, C'-4, orC'-6 positions of the galactose residue that would interfere with

recognition by PNA (Fig. 2). A candidate variation of the T-antigen that may be present in the normal colon is the disialy-lated tetrasaccharide: Neu5AcÂ«2,3-Gal/:il,3(Neu5Ac-rt2,6)GalNAc, which has not yet been identified in the colon. However, the failure of neuraminidase to uncover PNA-binding sitesin the normal colon would argue against this. Additional formssuch as a monosialylated trisaccharide (in which only the Â«2,3-linked sialic acid remains) or the nonsialylated native T-antigenappear to be expressed in neoplastic epithelium. This speculation was supported by the experiments in which cancer specimens were treated with neuraminidase. Removal of Â«2,3-linkedsialic acids with Newcastle disease virus neuraminidase enhanced labeling by PNA but had no effect on labeling by ACA.Additional characterization of the mucin-type oligosaccharidesfrom colonie neoplasia is necessary to confirm this hypothesis.

FAP is a genetic disease in which the normal restraints onepithelial cell proliferation eventually fail, the colon develops alarge number of adenomatous polyps, and colonie cancer inevitably ensues. Prior investigators have used thymidine incorporation studies (27) and measurement of ornithine decarbox-ylase (28) to demonstrate expression of the proliferative zoneand defective regulation of cell division in the normal-appearing, preadenomatous tissue. Both these techniques require the

300 -

100 -

T

1

Normal

Tissue

Fig. 8. Weighted averages. The tissues were given a weighted average derivedby adding the individual scores (ranging from 0-4) from 100 crypts or epithelialunits, which could range from 0 (all 100 = 0) to 400 (all 100 = 4). Circles, meansfor each group; bars, SEM. The mean score for HNPCC (74 Â±70) is significantlyhigher than that for the normal colon (65 Â±33, P< 0.05). but there is considerableoverlap in the scores. HNPCC tissues have a group of outliers with high scores,which may differ from the rest of the group. The weighted average for FAP is224 Â±76. which is significantly higher than that for the normal colon (P< 0.001).

gates in the normal colon. However, using a lectin and antibodytechnique (6) or a polyclonal antibody to T-antigen for immu-nohistochemical studies (22), labeling at the base of the crypthas been reported in a small proportion (approximately 5%) ofnormal colons (reviewed in Ref. 26). This has created confusionregarding the synthesis of T-antigen by normal colonie epithelium, which has been exacerbated by a detailed study of mucinoligosaccharides from the normal colon that failed to detect theT-antigen or any "cryptic" variations (14, 15). It is now apparent that T-antigen is not abundantly expressed in the normal

663

Table 3 Summary of ACA histocheniistry IIThe weighted average is a sum of labeling scores, each of which ranges from

0-4, for 100 epithelial units in the tissue. The scores, which can range from 0 (allunits = 0) to 400 (all units = 4*), are expressed as mean Â±SEM.

TissueNormal

colon (,V= 14)HNPCC (N = 32)HNPCC outliers (N = 5)FAP(.V= 11)Weighted

average(0-400)65

Â±3374 Â±70Â°

203 Â±43*224Â±76*

' Significant difference from normal colon, P < 0.05.*Significant difference from normal colon. P < 0.001.




Table 4 Comparison of biotinylatetl ACÃ•and PNA on human colonie ÃissitesTissues stained with biotinylated ACA (10 (jg/mL) or PNA (20 /<g/mL) and

scored from 0-3* using the global assessment as described in "Materials andMethods."

TissueNormal

colonHNPCCFARColon

cancerN619715ACA

stainingI*i*

(i?)2-3*(2)2-3*3*PNA

staining00(17)r

(2)01-2*(14)3*(1)

availability of fresh colonie epithelium for their application.Lectin histochemistry may be performed on fixed tissues, including previously obtained archival specimens, because of thestability of carbohydrate structures during the fixation andembeddng procedures. The labeling by ACA of small adenomasand fiat, normal-appearing colonie tissue in FAP is consistentwith the previously described defects in proliferation, whichdistinguishes ACA from PNA. We have recently reported thatPNA does not label the flat mucosa or small adenomas in FAP(17). Collectively, these data suggest that early events in colo-rectal neoplasia, i.e., those associated with defective control ofproliferation, may be associated with a form of T-antigen recognized by ACA but that later events, i.e., those seen in largerpolyps and cancers, are associated with forms recognized byboth ACA and PNA. In any event, ACA and PNA can bind toslightly different carbohydrate structures, which may differprimarily by their state of sialylation.

Potentially the most important histochemical finding withACA occurred with the tissues obtained from individuals at riskfor HNPCC. A subset of these tissues showed labeling withincreased intensity and in inappropriate locations compared tonormal colons. When studied with PNA, we have found nolabeling in HNPCC tissues (17). However, a subset of patientsat risk for HNPCC was found who had a significant reductionin the weighted average for labeling with DBA, the lectin thatselectively binds to goblet cell mucin in the upper one-half, orone of differentiation, in the normal colon. When ACA wasused, a subset of outliers had labeling scores similar to thoseseen in FAP specimens. If our ascertainment were perfect, onlyone-half of our group of patients at risk for HNPCC would becarriers of the putative gene for this autosomal dominantlyinherited disease (16). Our results suggest that a subset of theseindividuals may have a defect in the control of proliferationthat is in some way similar to that in FAP. The fact that only5 of 32 had very high ACA-labeling scores could indicate poorascertainment of the HNPCC trait, heterogeneity in the underlying defect that results in HNPCC, or the failure to samplethe appropriate anatomic location in the colon. Furthermore,an abnormality in proliferation or glycoconjugate expressionmay occur in a limited time frame in the natural history offamilial colonie cancer. This promising lead will require additional follow-up to determine the clinical outcome in thoseindividuals with abnormal labeling characteristics.

In summary, a new lectin has been used to localize T-antigenvariants in samples of human colon. ACA labels the base of thenormal colonie crypt of Lieberkuhn and is abundantly expressedin colonie adenomas and carcinomas. Our data suggest thatACA recognizes sialylated variants of the T-antigen that arenot recognized by the closely related T-antigen-binding PNAlectin. ACA labels epithelial cells not located in the normal

zone of proliferation in a small number of normal tissues.Increased intensity of labeling and labeling outside of the pro-liferative region are found in normal-appearing FAP tissuesand in the biopsy specimens of a subset of people at risk forHNPCC. ACA may permit the identification of the proliferativepool of cells in the colon and provides information regardingthe oligosaccharide structures associated with this behavior.However, the most valuable potential use of ACA may be inthe identification of zones of inappropriate proliferation, suchas those present in FAP. Since there are no markers availableto identify patients at risk for HNPCC, this potential usedeserves additional investigation.

ACKNOWLEDGMENTS

The authors gratefully acknowledge the manuscript preparation ofDiana Drescher and the assistance of Linda Lee Austin, Eugene Ang,and William O. Dobbins III, M.D., with the histochemistry.

REFERENCES

1. Rinderle, S. ]., Goldstein. I. !.. Matta. K. I... and Ratcliffc, R. M. Isolationand characterization of amaranthin. a lectin present in the seeds of Amar-anthus caudalus. that recognizes the T- (or cryptic T-) antigen. J. Biol.Chem., 264: 16123-16131, 1989.

2. Lotan. R.. Skutelsky, E.. Danon. D., and Sharon, N. The purification,composition, and specificity of the anti-T lectin from peanut (Arachis h\po-gaea). J. Biol. Chem.. 250: 8518-8523, 1975.

3. Pereira, M. E. A.. Rabat. E. A.. Lotan. R.. and Sharon. N. Imnumochcmicalstudies on the specificity of the peanut (Arachis hvpogaea) agglutinin. Car-bohydr. Res.. 51: 107-118. 1976.

4. Uh.i\.m.imi.in. V. P., Umemoto, J., Banks. J. R.. and Davidson, E. A.Isolation and partial characterization of sialoglycopeptides produced by amurine melanoma. Biochemistry. 7(5:4426-4437, 1977.

5. Boland. C. R.. Montgomery. C. K., and Kim. V. S. Alterations in humancolonie mucin occurring with cellular differentiation and malignant transformation. Proc. Nati. Acad. Sci. USA. 79: 2051-2055. 1982.

6. Cooper, H. S. Peanut lectin-binding sites in large bowel carcinoma. Lab.Invest.. 47: 383-390, 1982.

7. Boland. C. R.. Roberts. J. A., Siddiqui. B.. Byrd. J. C.. and Kim, Y. S.Cancer-associated colonie mucins in cultured tumor cells and nude mousexenografls. Cancer Res.. 46: 5724-5729. 1986.

8. Boland. C. R.. Bresalier. R. S., and Roberts, J. A. Assay of lectin bindingsites in colonie cancer-associated mucins: comparison of fluorescence microscopy with a quantitative Chromatographie technique. J. Histochem.Cytochem., 36: 1329-1334. 1988.

9. Boland. C. R., Montgomery, C. K., and Kim. V. S. Cancer associated mucinalterations in benign colonie polyps. Gastroenterology. 82: 664-672, 1982.

10. Boland. C. R.. Lance. P.. Levin. B.. Riddell. R. H.. and Kim. Y. S. Abnormalgoblet cell glycoconjugates in rectal biopsies associated with an increasedrisk of neoplasia in patients with ulcerative colitis: early results of a prospective study. Gut, 25: 1364-1371, 1984.

11. Boland. C. R.. and Ahnen. D. J. The binding of lectins to goblet cell mucinin premalignant colonie epithelium of the CF-1 mouse. Gastroenterology.89: 127-137, 1985.

12. Boland. C. R.. and Clapp. N. K. Glycoconjugates in the colons of new worldmonkeys with spontaneous colitis and cancer: the association between inflammation and neoplasia. Gstroenterology. 92: 625-634, 1987.

13. Boland. C. R. Mucin histochemistry in colonie polyps and cancer. Semin.Surg. Oncol., 3: 183-189. 1987.

14. Podolsky. D. K. Oligosaccharide structures of human colonie mucin. J. Biol.Chem.. 260: 8262-8271. 1985.

15. Podolsky. D. K. Oligosaccharidc structures of isolated human colonie mucinspecies. J. Biol. Chem.. 260: 15510-15515. 1985.

16. Boland. R. R.. and Troncale. F. J. Familial colonie cancer without antecedentpolyposis. Ann. Intern. Med.. 100: 700-701, 1984.

17. Sams. J. S.. Lynch, H. T., Burt, R. W., Lanspa. S.. and Boland, C. R.Abnormalities of Icctin histochemistry in familial polyposis coli and hereditary nonpolyposis colorectal cancer. Cancer (Phila.). 66: 502-508. 1990.

18. Corfield, P., Wember. M.. Schauer. R., and Rott. R. The specificity of viralsialidasesâ€”the use of oligosaccharide substrates to probe cnz.ymic characteristics and strain-specific differences. Eur. J. Biochem.. 124: 521-525. 1982.

19. Schauer. R. Analysis of sialic acids. Methods Enzymol.. 138: 132-161. 1987.20. Boland, C. R.. and Kim. Y. S. Transitional mucosa of the colon and tumor

growth factors. Med. Hypotheses, 22: 237-243, 1987.21. Goldstein. I. J., and Poretz, R. D. Isolation, physicochemical characteriza

tion, and carbohydrate-binding specificity of lectins. In: I. E. Licncr. N.

664




Sharon, and I. J. Goldstein (eds.). The Lectins. Properties. Functions andApplications in Biology and Medicine, Chap. 2. pp. 33-247. Orlando. FL:Academic Press. Inc.. 1984.

22. Yuan. M., Itzkowitz. S. H.. Boland. C. R.. Kim, Y. D., Tornita, J. T.. Palekar.A.. Bennington, J. L., Trump, B. F.. and Kim, Y. S. Comparison of T-antigcn expression in normal, premalignant and ni.ihMi.nn human colonietissue using lectin and antibody immunohistochcmistry. C'ancer Res.. 46:4841-4847. 1986.

23. Vogelstein. B.. Fearon, E. R., Hamilton. S. R., Kern, S. E., Preisinger, A.C.. Leppert. M.. Nakamura. Y.. White. R.. Smits. A. M. M., and BÃ¶s,J. L.Genetic alterations during colorectal-tumor development. N. Engl. J. Med..319: 525-532. 1988.

24. Law. D. J.. Olschwang, S.. Monpezat. J-P. Lefrancois, D.. Jagclman. D..Petrelli. N.J.. Thomas. G., and Feinberg. A. P. Concerted nonsyntenic allclic

loss in human colorectal carcinoma. Science (Washington DC). 241: 961-965. 1988.

25. Goelz. S. E.. Vogelstein, B., Hamilton, S. R., and Feinberg. A. P. Hypome-thylation of DNA from benign and malignant human colon neoplasms.Science (Washington DC), 22Â«:187-190. 1985.

26. Boland. C. R. Lectin histochemistry in colorectal polyps. In: G. Steele. R.\\ . Burt. S. J. Winawer, and J. P. Karr (eds.). Basic and Clinical Perspectivesof Colorectal Polyps and Cancer, pp. 277-287. New York: A. L. Liss. Inc.,1988.

27. Deschner. E. E., Lewis. C. M., and Lipkin. M. In vitro study of humanepithelial cells. I. Atypical zone of 3H-thymidine incorporation in mucosa ofmultiple polyposis. J. Clin. Invest.. 42: 1922-1928. 1963.

28. Luk. G. D., Baylin. S. B. Ornithine decarboxylase as a biologic marker infamilial colonie polyposis. N. Engl. J. Med.. ill: 80-83. 1984.

665



1991;51:657-665. Cancer Res C. R. Boland, Y-F. Chen, S. J. Rinderle, et al. Probe of Proliferating Colonic Epithelial Cells

as a HistochemicalAmaranthus caudatusUse of the Lectin from

Updated version

http://cancerres.aacrjournals.org/content/51/2/657

Access the most recent version of this article at:

E-mail alerts related to this article or journal.Sign up to receive free email-alerts

Subscriptions

Reprints and

[email protected] at

To order reprints of this article or to subscribe to the journal, contact the AACR Publications

Permissions

Rightslink site. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC)

.http://cancerres.aacrjournals.org/content/51/2/657To request permission to re-use all or part of this article, use this link



http://cancerres.aacrjournals.org/cgi/alerts

mailto:[email protected]



use of the lectin from amarant hus ... - cancer research · had a score of 224 Â±76 (p < 0.001...

Documents