THE JOURNAL OF BIOLOGICAL CHEMISTRY 0 1992 hy The American Society for Biochemistry and Molecular Biology, Inc.

Vol. 267, No , 35, Issue of December 15, PP. 25347-25351,1992 Printed in U. S. A.

Affected Paroxysmal Nocturnal Hemoglobinuria T Lymphocytes Harbor a Common Defect in Assembly of N-Acetyl-D-glucosamine Inositol Phospholipid Corresponding to That in Class A Thy-1- Murine Lymphoma Mutants*

(Received for publication, September 25, 1992)

Chris Armstrong$, Jorg Schubertg, Etsuko Ueda$, Jansen J. Knez$, Daniel Gelperin*, Shinichi HiroseSY, Robert Silber 11, Susan Hollan**, Reinhold E. Schmidt$, and M. Edward MedofS $$ From the $Institute of Pathology, Case Western Reserve University, Cleveland, Ohio 44106, the SAbteilung Irnmunologie und Transfusionsmedizin, Medizinische Hochschule, 0-3000 Hannover 61, Germany, TFukuoka University School of Medicine, Fukuoka 814-01, Japan, IINew York University Medical Center, New York, New York 10003, and the **National Institute of Haematology and Blood Transfusion, H-I 113 Budapest, Hungary

Deficient expression of glycoinositol phospholipid (GPI) anchored proteins in affected paroxysmal noc- turnal hemoglobinuria (PNH) cells has been traced to a defect in GPI anchor assembly. In a previous study (Schubert, J., Schmidt, R. E., and Medof, M. E. (1993) J. Biol. Chem., in press) we characterized the biosyn- thesis of putative Man-containing GPI anchor precur- sors in normal peripheral blood lymphocytes and in- vestigated assembly of these intracellular GPI inter- mediates in CD48- affected and CD48+ unaffected T and natural killer cell lines of PNH patients. We found that affected T cells from five patients exhibited a uniform defect in which dolichol-phosphoryl-Man was synthesized but no GPI mannolipids were expressed. In this study, membranes of patients’ affected T cells were labeled with UDP-[3H]GlcNAc to evaluate earlier steps in GPI synthesis, and intact cells were fused to Thy- 1- murine lymphoma mutants harboring different defects in early GPI assembly to test for the presence of corresponding or complementary lesions. In all cases, affected cell membranes failed to assemble GlcNAc-inositol phospholipid, the initial precursor of GPI anchor structures, and the intact cells failed to complement class A mutants while complementing other classes. Affected polymorphonuclear leukocytes from three additional patients of different origin were then labeled with [3H]Man and the labeling patterns found to correspond to those obtained with the T lym- phocytes. Taken together the data indicate that the genetic lesion in PNH cells resides in a DNA element which: 1) encodes a product required for the synthesis of GlcNAc-inositol phospholipid, 2) corresponds to that altered in class A Thy-1- murine lymphoma mutants, and 3) is commonly affected in different patients.

* This investigation was supported by National Institutes of Health Grants A123598 and POlDK38181 and Deutsche Forschungsgemein- schaft Grants Schm 596/3-2 and Schu 713/2-2. The costs of publi- cation of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “aduertise- rnent” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

$4 To whom correspondence should be addressed Inst. of Pathol- ogy, Case Western Reserve University, 2085 Adelbert Rd., Cleveland, O H 44106. Tel.: 216-368-5434; Fax: 216-844-1810.

In paroxysmal nocturnal hemoglobinuria (PNH),’ somatic mutation of one or more marrow progenitors gives rise to subpopulations of circulating blood elements which fail to express glycoinositol phospholipid (GPI) anchored proteins on their surfaces (reviewed in Ref. 1). These deficiencies include decay-accelerating factor (DAF) and CD59, which render the affected cells sensitive to autologous complement- mediated injury, as well as alkaline phosphatase and Fc,RIII receptors (CD16) on polymorphonuclear leukocytes (PMN), CD14 and urokinase-type plasminogen activator receptors on monocytes, and 5’ nucleotidase and CD48 on lymphocytes. The biochemical lesion responsible for the expression defect is unknown, but previous studies (2-6) have shown that it resides in the pathway providing for intracellular assembly of GPI-anchor moieties.

In a previous study (7) , affected GPI-anchor defective CD48- and unaffected GPI-anchor sufficient CD48+ T cell lines of patients were used to localize the site of interruption of the GPI assembly process. [3H]Man labeling of the cells showed that in five different PNH patients the affected cells failed to synthesize ethanolamine (EthN)-P-ManGlcN- acyl inositol phospholipid (PI) through EthN-P-6Man (EthN-P-G)Man(EthN-P+)ManGlcN-acyl PI (8) or any ear- lier mannolipid intermediates (see Table I of Ref. 7 or Fig. 7 of Ref. 8 for pathway). The findings contrasted with previous observations ( 5 , 6), in that: 1) no mannolipid GPIs were distinguished and 2) the five patients appeared to exhibit a uniform defect. Similar findings were obtained in studies of Epstein Barr virus (EBV)-transformed DAF- and CD59- B lymphocytes of a Japanese patient.2 In view of these obser- vations, precursors of the Man-containing GPIs were inves- tigated.

In the present study, the affected CD48- T lymphocyte lines of the different patients were assayed for their abilities to assemble GlcNAc- and GlcN-PI, the first two intermediates of the GPI fabrication sequence (4, 9-12). Based on the

’ The abbreviations used are: PNH, paroxysmal nocturnal bemo- globinuria; GPI, glycoinositol phospholipid; DAF, decay-accelerating factor; PMN, polymorphonuclear leukocyte(s); EthN, ethanolamine; PI, inositol phospholipid; EBV, Epstein-Barr virus; TLCK, N”-p- tosyl-L-lysine chloromethyl ketone; PEG, polyethylene glycol; PHA, phytohemagglutinin; mAb, monoclonal antibody; FITC, fluorescein isothiocyanate; C:M:W, ch1oroform:methanol:water; TLC, thin layer chromatography; Dol-P-Man, dolichol-phosphoryl-Man; DMEM, Dulbecco’s minimal essential medium.

Takahashi, M., Takeda, J., Hirose, S., Hyman, R., Inoue, N., Miyata, T., Ueda, E., Kitani, T., Medof, M. E., and Kinoshita, T. (1993) J. Enp. Med., in press.

25347

Site of the PNH Lesion

findings, complementation assays with Thy-1- lymphoma mutants (13) known to possess distinct genetic defects in assembly of these precursors (11, 12, 14) were performed. In common with the EBV-transformed CD59-/DAF- B cells: the CD48- T cells were found to harbor a defect in synthesis of the initial GPI anchor precursor GlcNAc-PI corresponding to that in class A cells. The affected T lymphocytes from all of the patients studied exhibited a common defect, and af- fected PMN from three additional patients of different origin exhibited a consistent GPI mannolipid labeling pattern show- ing no GPI mannolipids.

MATERIALS AND METHODS

Reagents, Antibodies, and Cell L ~ ~ ~ s - U D P - [ ~ H ] G ~ C N A C (26.8 Ci/ mmol) was purchased from Du Pont-New England Nuclear. D - [ ~ - ~ H ] Man (15 Ci/mmol) was bought from American Radiolabeled Chemi- cals, Inc. (St. Louis, MO). Polyethylene glycol (PEG) (M, 1,000) was obtained from Sigma, and phytohemagglutinin (PHA) was purchased from Burroughs-Wellcome (Research Triangle Park, NC). Phospha- tidylinositol-specific phospholipase C was provided by Dr. T. Rosen- berry (Case Western Reserve Univ., Cleveland, OH).

Murine (IA10) anti-DAF (CD55) monoclonal antibody (mAb) was prepared as described (15), murine anti-CD59 mAb IF5 provided by Dr. H. Okada (Nagoya City Univ., Nagoya, Japan), and murine anti- CD48 mAb 54.57 provided by Dr. J. Pesando (Biomembrane Institute, Seattle, WA). Rat anti-Thy-1.1/1.2 M5-41 (IgG2a) mAb was obtained from Boehringer Mannheim, murine anti-CD21 mAb from the Bind- ing Site (San Diego, CA), and fluorescein-isothiocyanate (FITC)- labeled goat anti-rat Ig F(ab')z from Organon Teknika-Cappel (Mal- vern, PA).

EL-4.G.1.4 lymphoma cells and Thy-1- mutants were obtained from Dr. R. Hyman (Salk Institute, La Jolla, CA; Ref. 13). The mutants consisted of class A, S49.1TB.2/8 5X.3.2.0UAR.1; class B, SlA.TB.4.8.2/8 10x3; class C, TlM1.7BU.2; class D, AKRl.G.1 M/ Thy 1.1- 2X.6; class E, BW5147.3/Thy l.l-.lOG.OUAR.l; class F, EL4.G.lN/thy 1.2- 2X.1.0UAR.1; and class H, S49.1 M/Thy 1-2X.4. Trypanosoma brucei were propagated as described (9, 16).

Patients and Preparation of Affected Cells-Patients E. W., H. M., and E. B. were followed at Medizinische Hochschule (Hannover, Germany), patient L. D. (of ,Dominican origin) a t New York Univer- sity Medical Center (New York, NY), and patients S. L., S. B., and C. L. at the National Institute of Haematology and Blood Transfusion (Budapest, Hungary).

CD48- and CD48+ T cell lines of patients E. W., H. M., E. B., and L. D. were prepared as described in our previous paper (7). Cells were harvested at 10-14 days of coculture and after removal from plates were treated with anti-CD21, or in the case of CD48- cells with anti- CD21 and -CD48 mAbs followed by goat anti-murine Ig-conjugated magnetic beads (Collaborative Research, Bedford, MA). Harvested cells were maintained for 1-3 days in 10% lymphocyte conditioned medium (LCM) prior to use in assays. PMN were isolated by Ficoll- Hypaque density centrifugation followed by dextran sedimentation. Flow cytometric analyses for DAF and CD59 were performed as described (2, 3).

Biosynthetic Labeling-In vitro UDP-[3H]GlcNAc labeling of T lymphocytes, EL-4 lymphomas, and 2'. brucei membranes was con- ducted as described (4). T cells (-2 X 106/ml) were pretreated at 37 "C for 3-4 h with 10 ng/ml phorbol 12-myristate 13-acetate and for 1 h with 10 pg/ml tunicamycin. Hypotonic lysates were prepared by agitation of 1-2 X lo7 cells with 1 ml of water (containing 0.1 mM N"-p-tosyl-L-lysine chloromethyl ketone (TLCK) and 0.1 pg/ml leu- peptin/lO' cells) for 5 min on ice. An equal volume of 100 mM HEPES (pH 7.4) buffer (containing 50 mM KC1,lO mM MgC12,O.l mM TLCK, 1 pg/ml leupeptin, and 20% glycerol) was added, and the product was liquid nitrogen snap-frozen. After thawing and ultracentrifugation at 189,000 X g for 20 min at 4 'C, pelleted membranes resuspended in 200 pl of 50 mM HEPES (pH 7.4) buffer supplemented with 25 mM KCI, 5 mM MgC12, 0.1 M TLCK, 1 pg/ml leupeptin, 5 mM MnClZ, 1 mM ATP, 0.5 mM dithiothreitol, and 0.2 pg/pl tunicamycin were added to tubes containing 10 pCi of dried UDP-[3H]GlcNA~ and the mixtures incubated at 30 "C for 5 or 15 min. Reactions were termi- nated by the addition of 1.3 ml of ch1oroform:methanol (C:M) 1:l (to obtain C:M:water (W), 10103) and lipids were segregated into bu- tanol.

I n vivo [3H]Man labeling of intact PMN was performed as de- scribed (6). Cells were preincuhated for l h in glucose-free RPMI- 1640 supplemented with 10% dialyzed fetal bovine serum, 10 pg/ml

tunicamycin, and 10 pg/ml glucose. 200 pCi of [3H]Man was intro- duced, the cultures incubated at 37 "C for 2 h, labeled cells were washed, cell pellets extracted with C:M:W (10103), and dried C:M:W extracts were partitioned in butano1:water as described (6).

Butanol-extracted UDP-[3H]GlcNAc- and [3H]Man-labeled prod- ucts were chromatographed on silica gel 60 thin layer chromatography (TLC) plates (Merck, Darmstadt, Germany) developed in C:M with 1 M NH3 (10:10:3) or C:M:W (10103), respectively. Separated [3H] Man-labeled lipids on TLC plates were analyzed on a EG&G Berthold LB 285 TLC scanner.

Complementation Analyses-Cell fusions were performed as de- scribed (17). Cells (2.5 X lo6 of each type) were combined, washed with L-glutamine-free Dulbecco's minimal essential medium (DMEM), and pelleted a t 1,000 X g for 10 min at 20 "C. After removal of supernatant, 1 ml of 50% PEG in L-glutamine-free DMEM (pre- viously microwaved just until boiling and returned to 20 "C) was added over 1 min with mixing. The resulting PEG/cell mixture was then diluted, first with 1 ml of L-glutamine-free DMEM over 1 min and then with 5.0 ml of DMEM/10% horse serum over 3 min. The fusion mixture was centrifuged a t 1,000 X g for 5 min at 20 "C, the cell pellet resuspended in 5.0 ml of DMEM/10% horse serum, and the suspended cells cultured for 24 h. Fused cells were stained with rat anti-Thy-1 mAb M5-41 and FITC-conjugated goat anti-rat Ig F(ab'), and after paraformaldehyde fixation were plated onto micro- scope slides and examined using a Zeiss Axiophot microscope.

RESULTS

Comparative Abilities of Affected PNH and Control Lym- phocytes to Synthesize GlcNAc- and GlcN-PI-Experiments in our previous study (7) showed that following labeling with [3H]Man, CD48- PNH T cell lines uniformly expressed doli- chol-phosphoryl-Man (Dol-P-Man) but no [3H]Man-labeled GPI products. To check for precursors of the Man-containing GPIs, synthesis of GlcNAc- and GlcN-PI by the affected cells was assessed. For this purpose, hypotonic lysates of CD48- T cells from patient E. W., control PHA blasts, GPI-positive murine EL-4 lymphomas, and T. brucei were prepared, and the respective membranes were incubated at 30 "C for 5 min with UDP-[3H]Gl~NAc (see "Materials and Methods"). Prod- ucts in butanol partitions of C:M:W extracts from each of the preparations were then compared by TLC developed in C:M:l M NH3 permitting resolution of the two [3H]GlcN-labeled GPI products. As shown in Fig. 1 (panel B ) , membranes from EL-4 lymphomas generated peaks appropriately positioned for the two species. Consistent with their identities as GlcNAc- and GlcN-PI these peaks comigrated with T. brucei standards (panel A ) and were sensitive to phosphatidylino- sitol-specific phospholipase C digestion in control studies (not shown). As seen in panels C and D, although membranes of PHA blasts supported assembly of the two species similarly to membranes of T. brucei and EL-4 lymphoma controls, no synthesis of either product was observed with membranes of the CD48- T cell line from patient E. W.

To determine if a similar defect was related to the failed mannolipid GPI production in affected T cells of the other patients and exclude possible differences in assembly of the two intermediates between patients' T cell lines and control PHA blasts, hypotonic lysates of CD48- cells from a second patient, L. D., and of paired CD48- and CD48+ lines from a third patient, H. M., were prepared. Membranes from the various samples and from PHA blasts were labeled with UDP- [3H]GlcNAc, this time for a longer period of 15 min, and GPI products were similarly isolated and analyzed. As shown in Fig. 1 (panels E-H), membranes of CD48- cells from patients L. D. and H. M. exhibited the same complete deficit in assembly of GlcNAc- and GlcN-PI as observed with mem- branes from CD48- cells of patient E. W. In contrast, mem- branes of the paired CD48+ cells from patient H. M. effected assembly of the two precursors comparably to membranes of control PHA blasts.

Complementation Analyses between Affected PNH Cells and

Site of the PNH Lesion 25349

I( G k W P I

n

+==- J"~- I . I I . I . I . , , I 0 ! A F

Migration f/cm = u J

FIG. 1. Inability of membranes of CD48- patients' T cell lines to synthesize GlcNAc- or GlcN-PI. Membrane preparations (2 X lo7 cell equivalents) of the indicated control cells and of PNH CD48- or CD48+ T cell lines were labeled at 30 "C for 5 or 15 min with UDP-['H]GlcNAc. Following CM:W extraction and butanol partitioning, ['HIMan-labeled products in the organic phase were analyzed by TLC developed with C:M:1 M NHs (10103). Although peaks corresponding to GlcNAc- and GlcN-PI were observed in all controls, neither product was detectable in any of the CD48- T cell lines. The identities of the more polar peaks are unestablished. The insets in the patient studies show flow cytometric analyses of the cells following staining with anti-CD48 antibody. The control CD48' T cells of patient H. M. were noted to contain some CD48- cells.

Thy-I- Murine Lymphomas-Thy-1- lymphoma mutants can be categorized into two groups (11, 12, IS), classes A, C, and H harboring defects in assembly of GlcNAc- and GlcN-PI, and classes E, B, and F possessing defects in subsequent GPI- anchor assembly steps involving Man and EthN-P incorpo- ration into the GPI core. Complementation studies therefore were conducted in which patient and control lymphocytes were fused with different mutants belonging to the first group and control mutants belonging to the second group. The data are summarized in Table I and selected results shown in Fig. 2. In preliminary studies, control fusions of the mutants with themselves and with each other were performed. These studies verified that self-fusions gave no complementation, whereas fusions of any two mutants of different classes complemented Thy-1 expression. Using internal self- and cross-fusion con- trols, mutant classes A, C, and E then were fused with control PHA blasts, control EBV-transformed B cells derived from a normal donor, and CD48- T cells from patients E. W. and E. B. The normal lymphocyte controls complemented Thy-1 expression by all three classes. In contrast, CD48- lympho- cytes from patients E. W. and E. B. in both cases comple- mented Thy-1 expression by classes C and E cells but showed no complementation of Thy-1 expression when fused with class A cells. To determine if CD48- T cells of the other study patients conformed to the same complementation pattern and as above exclude any artifact arising from differences in

TABLE I Thy-1 complementation after fusion of Thy-1- lymphomas

with human lymphocytes A C E

~

Exp. 1 A 0" 49 17 C -ih N D E 0 PHA blasts 40 30 5 EBV B cells 25 15 E. W.-

42 47 23

E. B.- ND A C F

Exp. 2 A 0 25 30 C ND 49 F 0 PHA blasts 20 40 16 L. D.- lol 31 29 H. M.- lol 20 30 H. M.+ 35 30 19

Percent of fusion products exhibiting Thy-I expression. In this assay, 100 fusion products identified by their larger size under light microscopy are examined for Thy-1 expression under fluorescence microscopy. Under the conditions employed when cells belonging to different complementation groups are fused, some but not all of the transient fusion products exhibit complementation.

* In other assays, control fusions of the class C mutant with self gave values of 0 and fusions of class E mutants with PHA blasts gave values >15.

ND, not done.

Light Fluor. Light Fluor. CLASS C CLASS A

CLASS F CLASS A

CLASS C CLASS A

CLASS C CLASS A

FIG. 2. Inability of CD48- patients' T cells to complement Thy-1 expression by class A murine lymphoma mutants. Af- fected CD48- T cell or controls (2.5 X IO6 cells) were fused with an equal number of Thy-1- lymphoma mutant cells of different classes. After overnight incubation, fusion products were analyzed for Thy-I expression by fluorescent microscopy of anti-Thy-1-treated and FITC-labeled anti Ig secondarily-stained cells. Representative results for the designated fusion pairs are shown. Although CD48+ T cells complemented all classes, CD48- cells failed to complement class A cells.

complementation between the mitogen-stimulated or trans- formed lymphocytes and T cell lines, additional studies were performed with CD48- T cells from patient L. D. and paired CD48- and CD48+ T cells from patient H. M., this time utilizing classes A, C, and F mutants. Control studies with PHA blasts again showed that they were able to complement Thy-1 expression by all three mutant classes. As observed with CD48- T lymphocytes from patients E. W. and E. B., CD48- cells from the two other patients, L. D. and H. M., exhibited the same specific deficit in the ability to complement Thy-1 expression by the class A line. In contrast, the paired CD48+ lymphocytes from patient H. M., grown identically to the CD48- lymphocyte line, complemented Thy-1 expression

25350 Site of the PNH Lesion

by class A cells with efficiency comparable to that observed with classes C and F cells.

To verify that the lack of complementation is specific to class A cells, additional experiments were performed with CD48- T cells from patients L. D. and E. W., this time utilizing classes A and H mutants in the former case, and classes A, H, and B mutants in the latter. Consistent with the results of the above studies, while control PHA blasts com- plemented Thy-1 expression by all three mutant classes, the CD48- T cells from both patients complemented Thy-1 expression by class H cells and those from patient E. W. also complemented Thy-1 expression by class B cells (not shown), but in neither case did the cells complement Thy-1 expression by class A cells.

Comparison of the Site of the Defect in Other P N H Pa- tients-As assessed by i n vivo [3H]Man labeling of intact CD48- T cell lines in our previous study (7), and the UDP- [3H]GlcNAc labeling and complementation analyses with Thy-1- murine lymphomas performed above in this study, affected cells of all of the patients analyzed appeared to exhibit a common biochemical defect. In view of this result, affected cells from three additional patients of different origin (Hungarian) were analyzed to ascertain if they shared the same defect. In the absence of available purified affected CD48- lymphocyte lines of these patients, affected DAF-/ CD59- PMN were utilized. After verification that the cells were >95% DAF-/CD59-, the affected PMN from the three patients and from an included healthy control (shipped and handled in an identical fashion) were labeled in vivo with [3H] Man. [3H]Man-labeled GPI products in butanol partitions of C:M:W extracts then were examined by TLC. As seen in Fig. 3, the control cells exhibited prominent peaks of ManMan(EthN-P+)ManGlcN-acyl PI (H6), EthN-P-6Man Man(EthN-P+)ManGlcN-acyl P I (H7), and EthN-P-6Man (EthN-P-G)Man(EthN-P-)ManGlcN-acyl PI (H8). In con- trast, as observed in our previous study (7) with the CD48- T cell lines of all of the other patients examined, the affected PMN from each of the three patients synthesized Dol-P-Man but assembled no [3H]Man-labeled GPIs.

DISCUSSION

In a previous study (7) it was found that, under conditions of stimulated GPI production, purified CD48- T cell lines from five PNH patients synthesized Dol-P-Man but uni- formly failed to express Man-containing GPI intermediates. Here we found that the failed mannolipid production is a consequence of inability of the cells to assemble GlcNAc-PI, the first intermediate of the GPI assembly sequence. The biochemical defect underlying this deficit corresponds to that in class A Thy-1- murine lymphoma mutants. Contrary to earlier speculations, the defect among PNH patients appears t o be homogeneous involving at least four and possibly ex- tending to as many as nine patients of diverse origin.

The absence of GlcNAc-PI assembly in the affected cells localizes the underlying biochemical lesion in the cells either to the putative synthase responsible for transfer of GlcNAc from UDP to its PI acceptor, or to assembly (or presentation) of this phospholipid acceptor presumably in the ER mem- brane. The ability of the affected PNH cells to complement Thy-1 expression by Thy-1- mutant cells other than those belonging to complementation class A indicates that the de- fect is recessive. Previous findings (11, 12, 17) that three classes of Thy-1- mutants, A, C, and H, are similarly unable t o synthesize GlcNAc-PI yet are able to complement each other in Thy-1 expression (13) indicate that at least three genetic loci are involved in GlcNAc-PI assembly. I t is possible that the multiplicity of loci could reflect a hetero-oligomeric

i"----.J c, 0

, # , , , I , , ' , , , , , , ,

Migrafion [/cm=-) F

FIG. 3. Consistency of the [3H]Man-labeling patterns of af- fected PMN from three Hungarian patients to the defect in CD48- T cells. Tunicamycin-pretreated PMN (2 X lo7) were incu- bated at 37 "C for 2 hrs with [3H]Man and organic phase products of butano1:water partitions were analyzed on TLC plates developed in C:M:W (10:10:3). The migration positions of previously characterized GPI intermediates H2-H8 (6, 8) determined from parallel in vitro GDP$'H]Man labeling of HeLa cell microsomes are indicated. While prominent peaks corresponding to H6, H7, and H8 were apparent following labeling of control PMN handled identically to the patients' cells, a peak corresponding to Dol-P-Man but no peaks corresponding to [3H]Man-containing GPIs were visualized following labeling of the patients' cells.

synthase, the requirement for a cofactor, or the need for a specific type of P I acceptor. Although some data in mamma- lian cells (6, 11) suggest involvement of alkylacyl- and diacyl- based- and in T. brucei (19) C18O-containing phospholipid, the precise nature of the initial PI acceptor is not known.

The results of UDP-[3H]GlcNAc labeling of CD48- T lym- phocytes in the present study contrast with earlier findings (4,6) in which lysates of DAF- and CD59- PMN populations of patients were found to support the in vitro assembly of GlcNAc- and GlcN-PI. Among the patients examined in this study is one (L. D.) of those previously analyzed in these earlier studies. The reason for the difference is not yet clari- fied. One possibility is that, despite flow cytometric charac- terizations, the previously studied PMN populations con- tained some residual unaffected cells, and that the activity of the missing factor provided by the residual cells is sufficiently high that the presence of even small amounts is able to complement the GlcNAc-PI assembly reaction. In this regard, it is noteworthy that although [%]Man labeling of the intact PMN showed little or no polar mannolipid products (6) in accordance with our present findings, i n vitro GDP-[3H]Man labeling of membrane fractions of the cells yielded TLC peaks corresponding to several of these products. The use in the present study of homogeneous lymphocyte lines or clones eliminates this potential complication.

Site of the PNH Lesion 25351

Our finding that affected lymphocytes of all patients ana- lyzed of both German and Dominican origin exhibited a common defect was unexpected. The results are in accordance with findings for EBV-transformed DAF- and CD59- B cell lines from two Japanese patients2 and another one of our patient^.^ Although lymphocyte complementation studies will be required to conclusively confirm their correspondence, the additional findings in the present study that DAF- and CD59- PMN of three additional patients of different (Hungarian) origin exhibited a consistent pattern of complete absence of GPI mannolipids strongly support the proposal of a uniform or predominant defect. Among Thy-1- lymphoma lines, mu- tants belonging to complementation class A were also ob- tained most frequently and noted to arise spontaneously (20). The occurrence in PNH of a common defect raises the pos- sibility that a hypermutable region exists in the genome encoding the responsible GPI anchor pathway-associated ele- ment. Similar predominant genetic defects have been found for other conditions such as chronic myelocytic leukemia (21, 22) and retinoblastoma (23, 24).

Acknowledgments-We thank Dr. K. Natonek and Dr. G. Haln (National Institute of Haematology and Blood Transfusion) for pro- viding samples from PNH patients, Dr. J. Pesando (Bio-membrane Institute) for murine anti-CD48 mAbs 54.57, Dr. R. Hyman (Salk Institute) for providing Thy-1- mutants, Dr. H. Okada (Nagoya City University) for murine anti-CD59 mAb IF5, Dr. G. Perry (Case Western Reserve University) for help with fluorescence microscopy, Drs. A. Tartakoff, T. Rosenberry, and M. Snider (Case Western Reserve) for critical review of this manuscript, L. Ravi for technical assistance, and B. McCarty for manuscript preparation.

REFERENCES 1. Yomtovian, R., Prince, G. M., and Medof, M. E. (1993) Transfusion, in

press

E. Ueda and M. E. Medof, unpublished observations.

2. Stafford, H. A,, Tykocinski, M. L., Lublin, D. M., Holers, V. M., Rosse, W. F., Atkinson, J. P., and Medof, M. E. (1988) Proc. Natl. Acad. Sci. U. S. A. 85.880-884

3. Caroihers, D: J.. Hazra. S. V.. Andresen. S. W.. and Medof. M. E. (1990) J. Clin. Invest. 86, 47-54

4. Hirose, S., Ravi, L., Hazra, S. V., and Medof, M. E. (1991) Proc. Natl. Acad. Sci. U. S. A. 88, 3762-3766

5. Mahoney, J. F., Urakaze, M., Hall, S., DeGasperi, R., Chang, H-M, Sugi- yama, E., Warren, C. D., Borowitz, M., Nicholson-Weller, A,, Rosse, W.

6. Hirose, S., Ravi, L., Prince, G. M., Rosenfeld M. G., Silber, R., Andresen, F., and Yeh, E. T. H. (1992) Blood 7 9 , 1400-1403

S. W., Hazra, S. V., and Medof, M. E. (1992) Proc. Natl. Acad. Sci. U. S. A. 89,6025-6029

7. Schubert, J., Schmidt, R. E., and Medof, M. E. (1993) J. Biol. Chem., in

, .

mess 8. H[rose, S., Prince, G., Sevlever, D., Ravi, L., Rosenberry, T., Ueda, E., and

9. Masterson, W. J., Doering, T. L., Hart, G. W., and Englund, P. T. (1989)

10. Doering, T. L., Masterson, W. J., Englund, P. T., and Hart, G. W. (1989)

Medof, M. E. (1992) J. Biol. Chem. 2 6 7 , 16968-16974

Cell 56, 793-800

J. Biol. Chem. 264.11168-11173 11. Stevens, V. L., a n d d e t z , C. R. H. (1991) J. Biol. Chem. 266,10039-10042 12. Sugiyama, E., DeGasperi, R., Urakaze, M., Chang, H. M., Thomas, L. J.,

Hyman, R., Warren, C. D., and Yeh, E. T. H. (1991) J. Biol. Chem. 2 6 6 , 12119-12122

13. Hvman. R. ( 1 9 R R I Trends &wt. 4. 5-8 14. Lemansky, P., Gupta, D. K., Mey'ae, S. , Tucker, G., and Tartakoff, A. M.

15. Kinoshita, T., Medof, M. E., Silber, R., and Nussenzweig, V. (1985) J. Exp.

~ "", ~ ~~

(1991) Mol. Cell. Bid. 11 , 3879-3885

Med. 162. 75-92 16. Cross, G. A. M. (1975) Parasitology 71,393-417 17. Hirose, S., Mohney, R. P., Mutka, S. C., Ravi L., Singleton, D. R. Perry,

G., Tartakoff, A. M., and Medof, M. E. (199i) J. Biol. Chem. 2 6 i , 5272- 5278

18. Hiiose, S., Ravi, L., Tartakoff, A., and Medof, M. E. (1991) FASEB J. 5 ,

19. Masterson W. J., Raper, J., Doering, T. L., Hart, G. W., and Englund, P. 2459 (abstr.)

T. (1990) Cell 6 2 . 73-80 20. Hyman, R.; Cunningham, K., and Stallings, V. (1980) Immunogenetics 10,

261-271 21. de Klein, A., Geurts van Kessel, A,, Grosveld, G., Bartram, C. R., Hage-

meijer, A., Bootsma, D., Spurr, N. K., Heisterkamp, N., Groffen, J., and

22. Groffen, J., Stephenson, J. R., Heisterkamp, N., de Klein, A,, Bartram, C. Stephenson, J. R. (1982) Nature 3 0 0 , 765-767

23. Cavenee, W. K., Hansen, M. F., Nordenskjold, M., Kock, E., Maumenee, R., and Grosveld, G. (1984) Cell 36,93-99

I., Squire, J. A., Phillips, R. A,, and Gallie, B. L. (1985) Science 228 , 501-503

24. Friend, S. H., Bernards, R., Rogelj, S., Weinberg, R. A., Rapaport, J. M., Albert, D. M., and Dryja, T. P. (1986) Nature 323,643-646