Proc. Nail. Acad. Sci. USAVol. 86, pp. 5557-5561, July 1989Immunology

Loss of HLA-A,B,C allele products and lymphocytefunction-associated antigen 3 in colorectal neoplasia

(major histocompatibility antigens/colorectal adenoma/colorectal adenocarinoma/imohistochemUitry)

MARK E. F. SMITH*t, STEVEN G. E. MARSH*, JULIA G. BODMER*, KEITH GELSTHORPEX,AND WALTER F. BODMER**Imperial Cancer Research Fund, 44 Lincoln's Inn Fields, London WC2A 3PX, United Kingdom; and SNational Blood Transfusion Service, Longley Lane,Sheffield S5 7JN, United Kingdom

Contributed by Walter F. Bodmer, April 17, 1989

ABSTRACT The expression of HLA-A,B,C antigens andlymphocyte function-associated antigen 3 in human colorectaladenomas and adenocarcinomas was studied by immunohis-tochemistry. None of 10 adenomas and only 1 of 30 carcinomashad lost expression of all HLA-A,B,C molecules. On the otherhand, focal loss of an HLA-B product was seen in 2 of theadenomas, and complete losses of tumor cell HLA-A2 (in 7 of13 cases), HLA-Bw4 (in 4 of 13 cases), and HLA-A3 (in 1 of 6cases) were seen in the carcinomas. No complete losses ofHLA-A1 (in 6 cases) or HLA-Bw6 (in 22 cases) occurred in thecarcinomas. In addition, 1 of 20 adenocarcinomas totallylacked lymphocyte function-associated antigen 3. Because aloss of tumor cell HLA-A,B,C antigen or lymphocyte function-associated antigen 3 could be selected for through an advantagein escape from cytotoxic T-lymphocyte attack, our resultssuggest that immunoselection may be a more important mech-anism in tumor progression than has previously been assumed.

HLA-A,B,C molecules are membrane glycoproteins com-posed of a heavy chain, encoded at one of three highlypolymorphic loci (A, B, and C) in the major histocompati-bility (HLA) region of chromosome 6, and a light chain,f32-microglobulin (J2m), whose association with the heavychain is required for functional surface membrane expressionof the complete product. These molecules control the rec-ognition of antigen by cytotoxic T lymphocytes, differentallele products varying in their ability to present alternativeantigen epitopes (1, 2). The antigen specificity of target-cellkilling by cytotoxic T lymphocytes is thought to be mediatedby the binding of the T-cell antigen receptor to processedantigen complexed with target-cell HLA-A,B,C molecules.T-lymphocyte killing of target cells also requires lymphocyteCD2 antigen to bind target-cell lymphocyte function-associ-ated antigen 3 (LFA-3) (3, 4). This receptor-ligand interactionincreases the adhesion between cells in a non-antigen-spe-cific manner.The observation that some tumor-derived cell lines lack

surface expression of HLA-A,B,C molecules (5-7) led to thesuggestion that tumors expressing tumor-specific antigenscould escape from T-cell immune attack through the selectiveoutgrowth of HLA-A,B,C loss variants. This idea has sub-sequently been supported by many examples of loss ofHLA-A,B,C expression from tumors, by using immunohis-tochemical techniques (for review, see ref. 8). Approximately1%o of colorectal adenocarcinomas completely lack expres-sion of HLA-A,B,C antigens (refs. 8 and 9 and P. Richmanand W.F.B., unpublished data), which are consistentlystrongly expressed in normal colorectal epithelium. The lossof some HLA-A,B,C allele products with the retention of

others within single colorectal adenocarcinomas has beendescribed (10-12), and this selective loss of allele products isalso most probably important in escape from tumor immu-nity. This is also likely to be the case for a loss of tumor-cellLFA-3.

This study investigates the expression of HLA-A,B,Callele products and LFA-3 in a series of 10 colorectal ade-nomas and 30 adenocarcinomas, extending the results of apreliminary communication (10).

MATERIALS AND METHODSTissues. Samples from 10 colonic adenomas (cases 1-10,

Table 1) and 30 colonic adenocarcinomas (cases 11-40) froma total of 40 patients were kindly supplied by the ImperialCancer Research Fund Colorectal Unit and Pathology De-partment, Saint Mark's Hospital, London. Nine of the ade-nomas (cases 1-9) arose in patients suffering from familialadenomatous polyposis. All 9 measured 1 cm or less inmaximum diameter and were of tubular histology. One ade-noma (case 10) was solitary, of tubulo-villous histology andmeasured 6.5 cm in maximum diameter. One adenocarci-noma (case 12) arose in a patient with ulcerative colitis.Non-neoplastic colonic epithelium was present in the sametissue block as 4 adenocarcinomas and histologically normalcolonic mucosa was also separately sampled in 7 adenomasand 15 adenocarcinomas. In adenocarcinoma patients, thesample of normal colon was taken approximately 10 cm fromthe edge of the tumor. Tissues were snap frozen in liquidnitrogen-cooled isopentane and subsequently stored underliquid nitrogen.Monoclonal Antibodies (mAbs) and HLA Determinants. The

mAbs, used as primary reagents in the form of tissue culturesupernatants, and their specificities are as follows: BBM.1detects f32m (13); HC-10 detects HLA-A,B,C free heavy chainwith a preference for HLA-B products (14); PA2.6 and W6/32detect all f82m-associated HLA-A,B,C antigens (6, 15); 142.2detects HLA-A1 and Aw36 (S.G.E.M., unpublished data);MA2.1 detects HLA-A2 and B17 (16); BB7.2 detects HLA-A2and Aw69 (6); GAP A3 and 129.2.47 detect HLA-A3 (ref. 17and K.G., unpublished data); 116.5.28 detects HLA-Bw4(K.G., unpublished data); 126.39 detects HLA-Bw6 (K.G.,unpublished data); RR1/1 detects intercellular adhesion mol-ecule 1 (ICAM-1) (18); and TS2/9 detects LFA-3 (19). ThemAbs were used at concentrations, ranging from undiluted todiluted 1:5 in phosphate-buffered saline (PBS; 0.145 M NaCl/2.5 mM NaH2PO4'2H20/7.5 mM Na2HPO4, pH 7.2) for dif-ferent antibodies, derived from titration studies performed oncryostat sections of tonsil and thymus. The immunohis-tochemical specificity of the mAbs directed against HLA-

Abbreviations: j2m, f32-microglobulin; LFA-3, lymphocyte function-associated antigen 3; mAb, monoclonal antibody; ICAM-1, intercel-lular adhesion molecule 1.tTo whom reprint requests should be addressed.

5557

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

Dow

nloa

ded

by g

uest

on

Janu

ary

16, 2

021

Proc. Natl. Acad. Sci. USA 86 (1989)

Table 1. HLA-A,B,C antigen expression in immunohistochemical sections of colorectal tumors(2m-associated Free heavy A3

Tumor heavy chain chain Al/Aw36 A2/B17 A2/Aw69 (GAP A3/ Bw4 Bw6(case number) (BBM.1/PA2.6/W6/32) (HC-10) (142.2) (MA2.1) (BB7.2) 129.2.47) (116.5.28) (126.39)

Adenomas12345678910

Adenocarcinomas111213141516171819202122232425262728293031323334353637383940

4444444444

4444444444

0*4*4*444444444444444444444444444t

312444444444444444444444414443

444

4

44

34

4

43

4

4

4444444

44

3

02

000

0

4

0

0440

44

4

0

000

0

2

0

0440

44

4

4

00

40034

40

0344444

44244

44444

4

433

4443

0

44

4

33

0

424

3

2

4

4

0

4

3

4

0

4

4

Immunohistochemical sections were assessed semi-quantitatively as follows: -, no tumor cell staining; 0, stromal staining present, but withno tumor cell staining; grades 1, 2, 3, and 4, stromal and tumor cell staining present. Grades: 1, <10%1; 2, .10%o but <50%o; 3, .50%~b but <95%;and 4, .95% of tumor cells stain positively.*mAbs BBM.1, PA2.6, and W6/32 detected either no or only very weak expression of tumor HLA-A,B,C antigens. The data on allele productexpression in these cases are, therefore, not analyzed in the text.tHLA-A,B,C antigens were in all tumor cells of case 40 but staining intensity was very variable between different parts of the tumor.

A,B,C allele products was demonstrated by their reactivity ontumor stroma in a series of 12 breast carcinomas that had beenHLA-typed (J.G.B., unpublished data) by standard tech-niques (20) using peripheral blood. Though HLA-A23 andHLA-A24 carry the HLA-Bw4 determinant (21), mAb116.5.28 did not detect an A24 antigen in an immunohis-tochemical section of an A24-positive individual.

Either the HLA-Bw4 or the HLA-Bw6 determinants areexpressed on essentially all HLA-B products. Approximategene frequencies in a European Caucasoid population are asfollows: HLA-A1, 14%; A2, 29%; A3, 13%; Aw36, 0.1%;B17, 3% (22); Aw69, <1% (J.G.B., unpublished data).

Immunohistochemistry. A peroxidase-anti-peroxidase im-munohistochemical method was employed. Cryostat tissuesections (7 ,um thick) were fixed in a chloroform/acetone'mixture [1:1 (vol/vol)] at 4°C, briefly immersed in acetone at

40C, rehydrated in PBS, and then incubated in 90%o (vol/vol)PBS plus 10%o (vol/vol) normal rabbit serum for 20 min.Excess fluid was drained from the tissue sections beforeincubating with mAbs for 60 min. This was followed sequen-tially by rabbit anti-mouse immunoglobulin [Dakopatts, Co-penhagen; 1:25 dilution for 45 min] in PBS containing 4%(vol/vol) normal human serum and then by peroxidase-anti-peroxidase complexes (Dakopatts; 1:100 dilution for 45min) in PBS. All antibody incubations were followed by two5-min washes in PBS. The peroxidase was developed in3,3'-diaminobenzidine (Sigma) at 1 mg/ml in PBS containing0.3% hydrogen peroxide. Sections were washed in tap waterfor 10 minm counterstained with hematoxylin, cleared, andmounted in D.P.X. mounting medium (BDH).

Assessment. Presence of immunohistochemical staining inepithelium and stroma was assessed in normal and neoplastic

5558 Immunology: Smith et al.

Dow

nloa

ded

by g

uest

on

Janu

ary

16, 2

021

Proc. Natl. Acad. Sci. USA 86 (1989) 5559

colonic tissue. A limited patient HLA-A,B,C type was de-duced from the reaction of mAbs directed against polymor-phic HLA-A,B,C determinants with tumor stroma and,where available, normal colorectal mucosa.An HLA-B17 phenotype was deduced from stromal stain-

ing with mAb MA2.1 and lack of staining with mAb BB7.2.The proportion of tumor cells stained by each mAb wasassessed semi-quantitatively (see Table 1).

RESULTSHLA-A,B,C Expression. Strong and fairly uniform HLA-

A,B,C expression, as detected by reactivity with mAbsPA2.6 and W6/32, was present in all normal colorectalepithelium, all 10 adenomas, and 26 of 30 adenocarcinomas.In 1 of 30 adenocarcinomas (case 11, Table 1) no tumor cellHLA-A,B,C was detected, in 2 of30 adenocarcinomas (cases12 and 13) tumor HLA-A,B,C antigen was uniformly veryweak, whereas in 1 of 30 adenocarcinomas (case 40) theintensity of tumor HLA-A,B,C antigen varied from weak tostrong in different areas of the tumor. Strong cytoplasmictumor staining with mAb HC-10 and lack of staining withmAb BBM.1 in the single adenocarcinoma whose tumor cellsshowed no staining with mAb W6/32 (case 11) demonstratedthat a loss of 82m underlies the absence of functional (832m-associated) HLA-A,B,C molecules. In the 3 adenocarcino-mas showing areas of weak HLA-A,B,C expression (cases12, 13, and 40), the staining of smaller numbers oftumor cellsby mAb HC-10 than by mAb BBM.1 suggests a coordinatereduction of all HLA-A,B,C products. The 3 adenocarcino-mas (cases 11, 12, and 13) that showed either no detectableor uniformly very weak tumor cell HLA-A,B,C antigenexpression were all graded as poorly differentiated tumors.The frequencies of the various HLA-A allele products in

this series, as demonstrated by staining of tumor stroma andnormal colorectal mucosa with the allele-product-specificmAbs, were similar to those derived from standard micro-lymphocytotoxicity testing of European Caucasoid popula-tions. Out of 40 individuals there were 7 HLA-Bw4 homozy-gotes, 21 Bw6 homozygotes, and 12 Bw4,6 heterozygotes. Byassuming that the distribution of HLA-Bw4 and Bw6 alleleproducts obeys the Hardy-Weinberg law, this series showsan excess of HLA-Bw6 homozygotes and a deficiency ofBw4,6 heterozygotes (X2 = 4, giving 0.05 > P > 0.02). Thissuggests that the mAb 116.5.28 does not recognize all HLA-Bw4-bearing allele products in immunohistochemical prepa-rations. The lack of immunohistochemical reactivity ofmAb116.5.28 in an HLA-A24-typed individual (HLA-A24 carriesthe HLA-Bw4 determinant) (21) also supports this sugges-tion. No loss of allele products from normal colorectalepithelium was detected.The intensity of tissue staining was often less for mAb

BB7.2 than for mAb MA2.1, even though both mAbs detectHLA-A2 antigen and, occasionally, BB7.2 stained fewertumor cells. mAbs 129.2.47 and GAP A3 both detect HLA-A3antigen and gave very similar staining results.Though none of the 10 adenomas studied contained areas

that had lost all HLA-A,B,C antigens from tumor cells, twoshowed focal loss of an HLA-B product. One adenoma (case2) contained a glandular focus comprising some 1o of thesection in which, in contrast to surrounding glands, HLA-Bw4 was undetectable. HLA-Bw4-positive and -negativeareas were morphologically indistinguishable. A second ad-enoma (case 10), the solitary (nonpolyposis), large tubulo-villous adenoma, showed an area of HLA-B17 loss compris-ing 50% of the tumor (Fig. 1). This case was HLA-Bw4homozygous (B17 carries the Bw4 determinant) and a prom-inent reduction in the intensity of staining with mAb 116.5.28(anti-Bw4) in those areas not expressing B17 was consistentwith a loss of one HLA-B product and the retention of the

other. The tumor glands not expressing the HLA-B17 deter-minant were morphologically different from those expressingit, having smoother glandular profiles, a greater degree ofnuclear crowding and stratification, and less interglandularstroma.The 3 adenocarcinomas in which all HLA-A,B,C antigens

were either undetectable or uniformly very weakly expressed(cases 11-13) were excluded from the analysis of individualallele product loss because in these cases allele product losswould simply reflect the loss of, or reduction in, all HLA-A,B,C antigens. Complete loss of individual HLA-A,B,Callele products from adenocarcinoma cells was seen in 8 of atotal of 27 cases, occurring in 7 of 13 HLA-A2 (Table 1 andFig. 1) and 4 (1 an apparent homozygote) of 13 Bw4 patients(of which 8 were heterozygotes) (Table 1). Complete loss ofHLA-A3 occurred in only 1 of 6 cases, and complete lossesof Al and Bw6 were not seen in any of 6 and 22 (of which 8were heterozygotes) cases, respectively. The complete lossof tumor HLA-A2 antigen (in 7 of 13 cases) was significantlymore common than the complete loss ofHLA-A1 and HLA-A3 (in 1 of 12 cases) (V2 = 4, giving 0.05 > P > 0.02). All 4cases that showed complete loss of HLA-Bw4 had also lostHLA-A2, suggesting the possibility of loss of a completehaplotype. However, in at least 1 case, case 14, tumor cellshad completely lost HLA-A2 but retained both their Bw4 andBw6 products. Our panel of mAbs was informative in iden-tifying both HLA-A products in a single adenocarcinoma thatshowed a selective allele product loss (case 16). In this case,tumor cells had lost HLA-A2 and Bw4 products but retainedAl and Bw6 products. Three further cases also showed lossof individual allele products in a majority of tumor cells(HLA-A1 in cases 17 and 31; Bw6 in case 21). The multipletumor foci that expressed these allele products within each ofthese 3 cases were not obviously related to an increasedinfiltrate of inflammatory cells. There was no obvious cor-relation of grade or stage with loss of HLA-A2.LFA-3 and ICAM-1 Expression. LFA-3 was strongly ex-

pressed on epithelial cells in all of 17 cases of normalcolorectal mucosa and 4 adenomas but not in 1 of 20 adeno-carcinomas. This was case 35 (Fig. 1), which also showedselective loss of HLA-A3, though LFA-3 was strongly ex-pressed on tumor stromal cells and adjacent normal colorec-tal mucosal epithelium.Adhesion molecule ICAM-1 was not detected on the epi-

thelial cells of any of the 17 cases of normal mucosa, 4adenomas, or 18 of the 20 adenocarcinomas. In 2 adenocar-cinomas, both of which had lost HLA-A2 expression, ICAM-1 positivity was present in approximately 1% (case 30) and75% (case 24) of tumor cells. In case 24 staining was pre-dominately associated with the tumor-cell luminal mem-brane.

DISCUSSIONOur finding that 10% of colorectal adenocarcinomas (3 of 30cases) express little or no HLA-A,B,C antigens compareswith results of a large study (9) in which 31% of adenocar-cinomas showed either reduced HLA-A,B,C expression orloss of HLA-A,B,C antigens from a proportion or all of theneoplastic cells. Loss of tumor-cell HLA-A,B,C antigens incolorectal adenocarcinoma has been associated (9) with apoor degree of tumor differentiation and, in agreement withthis finding, all 3 adenocarcinomas in this study with weak orundetectable expression of HLA-A,B,C antigens werepoorly differentiated. Because the parent epithelium of thesetumors, that of the colorectal mucosa, shows consistentlystrong expression of HLA-A,B,C antigens, tumors not ex-pressing these molecules probably arise through the selectionof mutants lacking expression of HLA-A,B,C antigens. Thisselection presumably results from the ability of tumor cells

Immunology: Smith et al.

Dow

nloa

ded

by g

uest

on

Janu

ary

16, 2

021

Proc. Natl. Acad. Sci. USA 86 (1989)

B *..- *.~a<..* by..

tX rm ~ be<>- X i

* s mmm...!G 1s

FIG. 1. Immunoperoxidase-stained hematoxylin-counterstained cryostat sections of colorectal tumors. (A and B) An adenoma (case 10,Table 1) strongly expresses HLA-A,B,C antigens (mAb PA2.6) on both stromal (S) and tumor (T) cells (A) but expresses HLA-B17 antigen (mAbMA2.1) on stromal but not on tumor cells (tumor HLA-B17 loss was focal) (B). (C and D) An adenocarcinoma (case 19, Table 1) expressesHLA-A,B,C antigens (mAb W6/32) on tumor and stromal cells (C) but has lost HLA-A2 antigen (mAb MA2.1) from tumor cells (D). (E-G)LFA-3 (mAb TS2/9) is expressed on normal colorectal epithelium (E) but not on adenocarcinoma cells (F) from the same patient (case 35, Table1), which retain HLA-A,B,C antigens (mAb W6/32) (G). (A-D, x300; E, X130; F and G, X200.)

bearing tumor-specific antigens but not expressing HLA-A,B,C antigens to escape destruction by cytotoxic T lym-phocytes.The underlying defect in the single case failing to express

any HLA-A,B,C antigen was a failure to produce f32m, whichis essential for the surface expression of these molecules (5,23). The same defect has been identified in the Burkittlymphoma cell line Daudi (5, 24) and the colorectal adeno-carcinoma cell line LoVo (7).The loss or down regulation of some HLA-A,B,C allele

products, with the retention ofothers, has been demonstratedin Burkitt lymphoma (25) and colorectal adenocarcinoma(10-12) and similar losses of the equivalent molecules inmurine tumors have been documented (26). Extending ourearlier observations, we have found such selective allele

product loss to be very common in colorectal adenocarcino-mas, complete tumor HLA-A2 loss occurring in 7 of 13HLA-A2-positive patients. These selective losses may haveimportant functional consequences because of variation inthe ability of various allele products to present differentantigen epitopes to cytotoxic T lymphocytes (1, 2). Thus, insome cases, the loss of single allele products may be func-tionally equivalent, in terms of escape from T-cell attack, tothe loss of all HLA-A,B,C molecules. Complete tumor HLA-A2 loss (in 7 of 13 cases) occurred significantly more fre-quently than complete losses of HLA-A1 (0 of 6 cases) andHLA-A3 (1 of6 cases) (P < 0.05). This suggests that HLA-A2antigen may be the most effective of the three allele productsstudied here in presenting a common tumor antigen tocytotoxic T-cell attack.

5560 Immunology: Smith et al.

i'A* vr

Dow

nloa

ded

by g

uest

on

Janu

ary

16, 2

021

Proc. Natl. Acad. Sci. USA 86 (1989) 5561

None of the 10 adenomas examined contained areas lack-ing all HLA-A,B,C antigens, but 2 showed well demarcatedfoci, demonstrating selective loss of HLA-B products. Suchareas of allele product loss can, therefore, arise relativelyearly in tumor development and may, ofcourse, represent theprecursors of the more extensive areas of allele product lossseen in the adenocarcinomas.

Selective HLA-A,B,C allele product loss in colorectaladenocarcinoma may occur with such high frequency be-cause it can result from a single genetic mutation. In contrast,loss of all HLA-A,B,C products usually requires two muta-tions. However, loss of HLA-A,B,C allele products couldsimply be a marker of chromosomal loss, the gene whose lossis selected for lying elsewhere on chromosome 6. The com-bined loss of HLA-A2 and Bw4 specificities in several casesis consistent with the deletion of at least a considerableportion of one HLA region. The apparently nonrandom lossof tumor HLA-A products (HLA-A2 loss occurred signifi-cantly more than Al and A3 loss) and, in case 14, the loss oftumor A2 antigen with the retention of the Bw4 and Bw6antigens, which would be consistent with a small geneticdefect involving only the HLA-A gene, argue that it is the lossof HLA allele products themselves that is selected for.LFA-3 was strongly expressed on all cases of normal

colorectal mucosal epithelium examined but was completelyabsent from tumor cells in 1 of 20 adenocarcinomas. BecauseLFA-3 is thought to increase the strength of adhesion be-tween cytotoxic T lymphocytes and their targets, its loss, likethat of HLA-A,B,C antigens, could allow tumor cells toescape T-cell killing. Loss of LFA-3 has also been describedin some cases of Burkitt lymphoma (27), a tumor expressingforeign Epstein-Barr virus antigens. Though ICAM-1 wasnot detected on normal colorectal mucosal epithelium, it wasexpressed focally in 2 of 20 adenocarcinomas examined. Theappearance of tumor ICAM-1 may reflect its ease of induc-ibility by a number of agents, such as interferon y (28).The loss of LFA-3 and of individual HLA-A,B,C allele

products from tumor cells may be selected for through anadvantage in escape from T-cell killing. However, tumor-specific antigens have not been well characterized. Thedemonstration that cytotoxic T lymphocytes can recognizeshort peptides that arise from the intracellular processing ofantigen (29, 30) extends the range ofcandidate tumor-specificantigens to the intracellular products of mutated oncogenes.For example, products of the RAS gene family, which havean intracellular localization, could act as tumor antigens,detectable by host cytotoxic T lymphocytes, in the approx-imately 50%o of colorectal adenocarcinomas possessing amutated RAS gene (31). Because HLA-A,B,C antigen losspresumably results from immune selection directed againstneoplastic cells bearing tumor-specific antigens, HLA-A,B,C-lacking tumor cell lines will probably be the best celltypes in which to characterize such tumor antigens. Theclinical detection of tumor antigens at an early stage in tumorprogression, before the outgrowth of mutants that can evadean immune response, may become an important indicationfor patient immunotherapy.

We are grateful to the Imperial Cancer Research Fund ColorectalUnit and Pathology Department, Saint Mark's Hospital, London, forproviding tumor tissue and data on tumor grade and stage and to thefollowing for the gift of mAbs: Dr. H. Ploegh (HC-10), Dr. T. A.Springer (RR1/1 and TS2/9), and Genetic Systems (142.2).

1. McMichael, A. J., Gotch, F. M. & Rothbard, J. (1986) J. Exp.Med. 164, 1397-1406.

2. Gotch, F., Rothbard, J., Howland, K., Townsend, A. &McMichael, A. (1987) Nature (London) 326, 881-882.

3. Krensky, A. M., Robbins, E., Springer, T. A. & Burakoff,S. J. (1984) J. Immunol. 132, 2180-2182.

4. Shaw, S. & Ginther Luce, G. E. (1987) J. Immunol. 139,1037-1045.

5. Arce-Gomez, B., Jones, E. A., Barnstable, C. J., Solomon, E.& Bodmer, W. F. (1978) Tissue Antigens 11, 96-112.

6. Brodsky, F. M., Parham, P., Barnstable, C. J., Crumpton,M. J. & Bodmer, W. F. (1979) Immunol. Rev. 47, 3-61.

7. Travers, P. J., Arklie, J. L., Trowsdale, J., Patillo, R. A. &Bodmer, W. F. (1980) Nat!. Cancer Inst. Monogr. 60, 175-180.

8. Bodmer, W. F. (1987) J. Clin. Pathol. 40, 948-958.9. Stein, B., Momburg, F., Schwarz, V., Schlag, P., Molden-

hauer, G. & Moller, P. (1988) Br. J. Cancer 57, 364-368.10. Smith, M. E. F., Bodmer, W. F. & Bodmer, J. G. (1988)

Lancet i, 823-824.11. Rees, R. C., Buckle, A.-M., Gelsthorpe, K., James, V., Potter,

C. W., Rogers, K. & Jacob, G. (1988) Br. J. Cancer 57,374-377.

12. Momburg, F., Ziegler, A., Harpprecht, J., Moller, P., Mold-enhaur, G. & HAmmerling, G. J. (1989) J. Immunol. 142,352-358.

13. Brodsky, F. M., Bodmer, W. F. & Parham, P. (1979) Eur. J.Immunol. 9, 536-545.

14. Stam, N. J., Spits, H. & Ploegh, H. L. (1986) J. Immunol. 137,2299-2306.

15. Barnstable, C. J., Bodmer, W. F., Brown, G., Galfre, G.,Milstein, C., Williams, A. F. & Ziegler, A. (1978) Cell 14,9-20.

16. McMichael, A. J., Parham, P., Rust, N. & Brodsky, F. (1980)Hum. Immunol. 1, 121-129.

17. Berger, A. E., Davis, J. E. & Cresswell, P. (1982) Hybridoma1, 87-90.

18. Rothlein, R., Dustin, M. L., Marlin, S. D. & Springer, T. A.(1986) J. Immunol. 137, 1270-1274.

19. Krensky, A. M., Sanchez-Madrid, F., Robbins, E., Nagy,J. A., Springer, T. A. & Burakoff, S. J. (1983) J. Immunol. 131,611-616.

20. Bodmer, W. F. & Bodmer, J. G. (1979) in NIAID Manual ofTissue Typing Techniques, NIH Publication 80-545, ed. Ray,J. G. (Natl. Inst. Health, Washington, DC), pp. 46-54.

21. Richiardi, P., Cambon-Thomsen, A. & Menicucci, A. (1984) inHistocompatibility Testing-1984, eds. Albert, E. D., Baur,M. P. & Mayr, W. R. (Springer, Berlin), pp. 237-238.

22. Baur, M. P., Neugebauer, M., Deppe, H., Sigmund, M., Lu-ton, T., Mayr, W. R. & Albert, E. D. (1984) in Histocompat-ibility Testing-1984, eds. Albert, E. D., Baur, M. P. & Mayr,W. R. (Springer, Berlin), pp. 333-341.

23. Seong, R. H., Clayberger, C. A., Krensky, A. M. & Parnes,J. R. (1988) J. Exp. Med. 167, 228-299.

24. Ploegh, H. L., Cannon, L. E. & Strominger, J. L. (1979) Proc.Natl. Acad. Sci. USA 76, 2273-2277.

25. Masucci, M. G., Torsteinsdottir, S., Colombani, J., Brautbar,C., Klein, E. & Klein, G. (1987) Proc. Nat!. Acad. Sci. USA 84,4567-4571.

26. Festenstein, H. & Schmidt, W. (1981) Immunol. Rev. 60,85-127.

27. Gregory, C. D., Murray, R. J., Edwards, C. F. & Rickinson,A. B. (1988) J. Exp. Med. 167, 1811-1824.

28. Dustin, M. L., Staunton, D. E. & Springer, T. A. (1988) Im-munol. Today 9, 213-215.

29. Townsend, A. R. M., Gotch, F. M. & Davey, J. (1985) Cell 42,457-467.

30. Townsend, A. R. M., Rothbard, J., Gotch, F. M., Bahadur,G., Wraith, D. & McMichael, A. J. (1986) Cell 44, 959-968.

31. Vogelstein, B., Fearon, E. R., Hamilton, S. R., Kern, S. E.,Preisinger, A. C., Leppert, M., Nakamura, Y., White, R.,Smits, A. M. M. & Bos, J. L. (1988) N. Engl. J. Med. 319,525-532.

Immunology: Smith et al.

Dow

nloa

ded

by g

uest

on

Janu

ary

16, 2

021