Proc. Nati. Acad. Sci. USA Vol. 89, pp. 6025-6029, July 1992 Medical Sciences Synthesis of mannosylglucosaminylinositol phospholipids in normal but not paroxysmal nocturnal hemoglobinuria cells (complement/decay-accelerating factor/CD59) SHINICHI HIROSE*, LAKSHMESWARI RAVI*, GREGORY M. PRINCE*, MELVIN G. ROSENFELDt, ROBERT SILBERf, STEVEN W. ANDRESEN§, SANDRA V. HAZRA¶, AND M. EDWARD MEDOF*II *Institute of Pathology, Case Western Reserve University, Cleveland, OH 44106; Departments of tCell Biology and *Medicine, New York University Medical Center, New York, NY 10016; §Cleveland Clinic Foundation, Cleveland, OH 44106; and lAkron General Medical Center, Akron, OH 44307 Communicated by Frederick C. Robbins, November 29, 1991 (received for review September 12, 1991)

ABSTRACT To identify mannosyl (Man)-containing inter- have been identified, information concerning the biochemical mediates ofthe human glycoinositol phospholipid (GPI) anchor pathway(s) responsible for their intracellular GPI-anchor pathway and examine their expression in paroxysmal nocturnal assembly and protein incorporation is only beginning to hemoglobinuria (PNH), mannolipid products deriving from in emerge. vitro guanosine diphosphate [3HJMan labeling of HeLa cell Among human GPI-anchored proteins are the decay- microsomes were characterized. The defined GPI species were accelerating factor (DAF) and CD59, membrane-associated correlated with products deriving from in vivo [3HJMan label- regulators of the complement cascade which protect host ing of normal and (GPI-anchor defective) affected leukocytes. tissues from autologous complement attack (reviewed in refs. In vitro analyses in HeLa cells showed dolichol-phosphoryl 2-4). Deficient expression of these proteins underlies an (Dol-P)-[3H]Man and a spectrum of [3H]Man lipids exhibiting acquired hemolytic disorder termed paroxysmal nocturnal TLC mobilities approximating those of Trypanosoma brucei hemoglobinuria (PNH), and the lesion responsible for the (Tryp) GPI precursors. Iatrobead HPLC separations and deficit has been traced to defective GPI-anchor pathway partial characterizations ofthe major isolated [3H]Man species function in a bone marrow stem cell(s) (5). The precise site(s) (designated H1-H8) showed that all but Hi (Dol-P-Man) were of interruption in the pathway, however, remains unidenti- sensitive to HNO2 deamination and serum GPI-specific phos- fied. pholipase D digestion but were resistant to phosphatidylinosi- In a previous communication (6), we reported on the use of tol-specific phospholipase C digestion unless previously deac- an in vitro experimental system employing hypotonic cell ylated with mild alkali. [3H]Man label in H3, H4, and H6 but lysates to study mammalian GPI-anchor biosynthesis. The not in H5 or H7 was efficiently released into the aqueous phase experimental system parallels that used for analyses of GPI- by jack bean a-mannosidase digestion. BioGel P-4 and AX-5 anchor structures synthesized in Trypanosoma brucei sizing ofthe dephosphorylated core glycan fragments ofH6 and (Tryps) (7, 8). We previously demonstrated that mammalian H7 gave values that coincided precisely with the corresponding cells transfer GlcNAc from UDP to a PI acceptor and rapidly Tryp anchor donor deacetylate the GlcNAc-PI product to generate GlcN-PI, two glycan fragments from the fully assembled species (designated GPI-A and -B) that correspond to the two A' (P2). Affected leukocytes from four patients with PNH initial intermediates in the Tryp GPI-anchor pathway (7, 8). supported formation of GkcNAc- and GlcN-PI but all failed to We were further able to show that cell lysates ofGPI-anchor- express H6 and H7 as well as H8 and two showed complete defective leukocytes from two PNH patients were able to absence of earlier Man-containing intermediates. These rind- support these biochemical conversions with normal efficien- ings argue that human intracellular GPI mannolipids are built cies. on acylated inositol phospholipids, that H6 and H7 contain In the present communication, we report on the extension differentially phosphoethanolamine-substituted Man3-GIcN- of this experimental system to characterize further steps in inositol cores, and that PNH cells are defective in conversion of the human GPI-anchor pathway. We also report on the partial GlcN-PI into these more mature mannolipid structures. localization within these further steps of the block(s) in GPI-anchor synthesis that undermines expression of DAF, Cell-surface proteins that are membrane anchored by cova- CD59, and other GPI-anchored proteins in affected PNH lently attached glycoinositol phospholipid (GPI) structures cells. are widely distributed on eukaryotic cells. Experimental evidence to date (reviewed in ref. 1) indicates that, despite individual variations, the GPI-anchoring units of these pro- EXPERIMENTAL PROCEDURES teins share a common core structure (Fig. 1) consisting of a Reagents. GDP-[3H]Man (24.3 Ci/mmol; 1 Ci = 37 GBq), phosphoethanolamine (P-Etn), three mannosyl (Man) resi- UDP-[3H]GlcNAc (26.8 Ci/mmol), D-[2-3H]Man (15 Ci/ dues, and a nonacetylated GlcN linked to an inositol phos- mmol), and EN3HANCE spray were obtained from DuPont/ pholipid. This inositol phospholipid is characteristically New England Nuclear. [3H]Glucose oligomer standards were phosphatidylinositol-specific phospholipase C (PI-PLC) sen- provided by J. Baenziger (Washington University, St. Louis, sitive but it may be PI-PLC resistant due to inositol acylation. MO). Tunicamycin, ATP, sphingomyelinase, and all protease In all cases studied, GPI-anchoring moieties appear to be preassembled in the endoplasmic reticulum (ER) and, once Abbreviations: GPI, glycoinositol phospholipid; P-Etn, phosphoeth- synthesized, transferred en bloc to primary translation prod- anolamine; ER, endoplasmic reticulum; DAF, decay-accelerating ucts immediately upon their emergence from ribosomes. factor; PNH, paroxysmal nocturnal hemoglobinuria; Tryps, Try- Although more than 50 proteins containing these structures panosoma brucei; PI-PLC, phosphatidylinositol-specific phospholi- pase C; GPI-PLD, GPI-specific phospholipase D; PMN, polymor- phonuclear leukocyte; Chl, chloroform; Dol-P, dolichol-phosphoryl. The publication costs of this article were defrayed in part by page charge 'To whom correspondence should be addressed at: Institute of payment. This article must therefore be hereby marked "advertisement" Pathology, Case Western Reserve University, 2085 Adelbert Road, in accordance with 18 U.S.C. §1734 solely to indicate this fact. Cleveland, OH 44106. 6025 Downloaded by guest on September 24, 2021 6026 Medical Sciences: Hirose et al. Proc. Natl. Acad. Sci. USA 89 (1992) GPI-PLD - - 'AL E-P EP G PTIp---. E-P-..oK-P--- BUTANOL ALK *atn HN02/NoBH4 P1-PLC *-POO;t E-P-..O E-PmO pO AOUEOUS FIG. 1. A human GPI-anchor core structure characterized by an acylated inositol phospholipid is boxed in the center. Structures and partitioning behaviors of predicted products of various treatments are indicated. *, Effective only in the absence of P-Etn (E-P). inhibitors were purchased from Sigma. Bacillus thuringiensis Mannolipid Purifications and Glycan Analyses. [3H]Man- PI-PLC was provided by T. Rosenberry (Case Western labeled lipid products were separated on a 6RSP-8010 (4.6 x Reserve University, Cleveland, OH). GPI-specific phospho- 250 mm) latrobead HPLC column (latron Laboratories, lipase D (GPI-PLD) was obtained from K.-S. Huang (Hoff- Tokyo) in a Dionex DX-300 apparatus using a linear gradient mann-La Roche). Jack bean a-mannosidase was purchased starting with Chl/methanol (80:20) extending to Chl/ from Oxford. methanol (55:45) followed by water addition to Chl/ Murine anti-human DAF monoclonal antibody (mAb) IA10 methanol/water (41:50:9). One-milliliter fractions were col- (IgG2a) was prepared as described (9) and rat anti-human lected. Reductions of HNO2 deamination fragments and CD59 mAb YTH53.1 (IgG2a) was provided by P. Lachmann dephosphorylations with HF were performed as described (Cambridge University, Cambridge, England). Appropriate (6). P4 (7) and AX-5 (12) chromatography were performed as murine and rat nonrelevant mAbs and fluorescein isothiocy- described. anate-conjugated sheep anti-murine and goat anti-rat immu- noglobulins were purchased from Organon Teknika-Cappel. RESULTS Cell Preparations. Tryps were expanded, human cells were Comparison of Mannoipids Synthesized in Vitro by HeLa cultured, and hypotonic lysates of each were prepared as Cells and Tryps. In initial analyses, HeLa cell lysates were described (6). Subcellular fractionation of HeLa cells was labeled with GDP-[3H]Man. Products were then compared by conducted as described (10). The rough microsome fraction TLC to those generated by Tryps. As shown in Fig. 2B, a was recovered at the 1.55/1.8 M sucrose interface. PNH cells spectrum of HeLa cell-derived [3H]Man-labeled lipids was stained with IA10 and YTH53.1 were analyzed on a Becton identified, the first corresponding to human dolichol- Dickinson FACStarPLUS flow cytometer. In patients (M.W. phosphoryl (Dol-P)-Man, and the others coinciding roughly and S.T.) with DAF/CD59-positive polymorphonuclear leu- in mobility (Fig. 2A) with Tryp GPI intermediates P, y, 6, {', kocytes (PMNs), affected cell subsets were enriched by A', and 0 (see legend for structures). anti-DAF cell-affinity chromatography (5). Next, the HeLa-derived products were tested for proper- Biosynthetic Labeling. Human cell lysates were centrifuged ties of GPI mannolipids. First, the presence in the products and particulates were washed as described (6). Tryp lysates of GlcN as well as Man was assessed by labeling with and purified microsomes were used unwashed with the buffer UDP-[3H]GlcNAc [in the presence of excess (0.1 mM) un- adjusted to contain the same constituents. Cell preparations labeled GDP-Man]. The formation of polar [3H]GlcN-labeled were added to 10 ,uCi of either GDP-[3H]Man or UDP- products was markedly less efficient than in Tryps. In most [3H]GlcNAc in a final vol of 200 IlI containing 5 mM MnCI2, experiments, no incorporation of label into these species 1 mM ATP, 0.5 mM dithiothreitol, and tunicamycin (0.2 occurred, but in some, [3H]GlcN-labeled species comigrating ,ug/ml). Intact cells (2 x 107) were preincubated for 60 min in with several of the [3H]Man-labeled species were observed. 10 ml ofglucose-free RPMI medium containing 10%o dialyzed Next, labeling was repeated by substituting purified HeLa fetal bovine serum, glucose (10 ,ug/ml), and tunicamycin (10 rough microsomes to verify that the products do not derive ,ug/ml) and then labeled for 45 min with 250 ,uCi of [3H]Man. from other components in cell lysates. The same spectrum of Reaction mixtures were extracted with chloroform (Chl)/ labeled species was recovered (Fig. 2C). The yield increased in as a function of microsome dose with 360 I&g giving radio- methanol, and dried extracts were partitioned butanol/ activity approximately equal to that using 3 x 107 cell water (6). equivalents (56 ,ug) of Tryp lysate. Kinetic analyses showed Enzymatic and Chemical Treatments. Sphingomyelinase that the HeLa species formed less rapidly than the Tryp digestions were performed as described (11) using enzyme species, with the most polar species requiring 30-60 min. pretested for activity with sphingomyelin. HNO2 deamina- Finally, the dependence of formation of the HeLa [3H]Man- tions and GPI-PLD and PI-PLC digestions were performed as labeled products on added UDP-GlcNAc was assessed, In described (6) except that with PI-PLC buffer containing 2 mM contrast to findings with Tryps (7) where UDP-[3H]GlcNAc deheptanoyl lecithin instead of 0.2% Triton X-100 was used. was essential, synthesis of the HeLa cell products was For a-mannosidase digestions, dried lipids were incubated at enhanced [Fig. 2C (Inset)] but not dependent on UDP- 37°C for 18 hr with 10 units of enzyme per ml in 10 .ul of the GlcNAc supplementation. manufacturer's buffer supplemented with 0.2% Triton X-100. Partial Structural Characterization of HeLa Cell Manno- With all analyses, control samples subjected to identical lipid Products. Initially, the putative GPI species were ana- conditions in the absence ofa modifying agent were included. lyzed unseparated with products obtained from labeling Reactions were stopped by extracting lipids with water- HeLa microsomes (360 ,g) for 90 min. Similar to Tryp y, 8, saturated 1-butanol and derivatives recovered in the 1-buta- C, and A', HNO2 deamination (see Fig. 1) quantitatively nol and/or water phase were analyzed. released [3H]Man label from all species except Dol-P-Man TLC Analyses. Radioactive species were chromatographed into the aqueous phase. on Silica gel 60 TLC plates developed, for products of To permit further structural analyses, 25-fold increased GDP-[3H]Man labeling, with Chl/methanol/water (10:10:3) amounts of the [3H]Man-labeled products were prepared and and, for products of UDP-[3H]GlcNAc labeling, with Chl/ products were separated by latrobead HPLC. The gradient methanol/1 M NH3 (10:10:3). TLC plates were analyzed (Fig. 3A) resolved Dol-P-Man (H1) and seven major species directly on a Berthold LB 285 scanner, or after spraying with designated H2-H8, which were individually collected. After EN3HANCE they were examined by fluorography. verification of homogeneities by TLC, aliquots (2000-5000 Downloaded by guest on September 24, 2021 Medical Sciences: Hirose et al. Proc. Nati. Acad. Sci. USA 89 (1992) 6027

A A A 5 Tryps. 0 H6 I1- 60(0- H7

%4o o- IH H5 201 I HS a ; 0 i i\--jI ~~, ~~~Do/-P- Mon _e|___ 0 20 40 60 80 100 120 Froction H/eLa B N Whole Ce/I Lysate I Dol-P (III - Mon 11

01 ,.I i II I. h11 I i

Jut a' tt,: 1 \,r~ If I " I,

C 5 HeLa I C Microsomes

F- WV I

.. ,. Ha N ,1 0 I.__ Migration (/cm= I-4) 0- A I FIG. 3. (A) Lipids deriving from 9 mg of HeLa rough microsomes 0 F Migra/ion (/cm= -) and 250 ,uCi of GDP-[3H]Man were separated by latrobead HPLC with a linear Chl/methanol/water program. The fractions pooled from each of the peaks H1-H8 for chemical analyses are indicated. FIG. 2. (A and B) Tryp (6 x 107 cell equivalents) or HeLa cell (1 (B) Tryp C, H2, and H6 were treated with GPI-PLD or buffer control x 107 cell equivalents) lysates were labeled at 3TC for 30 min with (BUF), and butanol partitions were compared. (C) Products (minor GDP-[3H]Man in the presence of 2.0 mM unlabeled UDP-GlcNAc. products designated by v) of Tryp H2, and H6 remaining in the (C) HeLa rough microsomes (360 were substituted for HeLa C, g.g) butanol phase after treatment with alkali at 20°( for 1 hr were treated lysate and the effect of labeling 90 in the presence Qane 1) or I&g with PI-PLC (1.6 /g/ml at 370C for 16 hr) or buffer control, and absence (lane 2) of excess unlabeled UDP-GlcNAc (Inset) was butanol partitions were compared. assessed. Lipids were analyzed on TLC developed in Chl/methanol/ water (10:10:3). y/8, Man1/2-GlcN-PI; 8/{, Man2/P-Etn-Man3- GlcN-acyl-PI; A'/O, P-Etn-Man3-GlcN-PI/lyso-PI (7). *, Dol-P- ited alkali treatment (20CC for 1 hr) of H2, H6, and H7 were As in for and Man; 0 and F, origin and front, respectively. analyzed. shown Fig. 3C Tryp C, H2, H6, incubation with PI-PLC had no effect on the residual unal- cpm) of each species were analyzed. Treatment with sphin- tered substrates but differentially cleaved their partially gomyelinase (which cleaves phosphorylated head groups deacylated products. Moreover, GPI-PLD induced compa- from ceramide in sphingomyelin glycolipids) had no effect on rable release (25% of counts). To exclude heterogeneity of any ofthe species. In contrast, as shown in Fig. 3B for H2 and glycans within individual species, the aqueous phase- H6 (and as found for each of the other species), incubation released and butanol phase-retained products of alkali treat- with GPI-PLD [which cleaves phosphatidic acid from inositol ment ofH6 were HN02 deaminated. Analyses ofthe reduced (see Fig. 1) irrespective of inositol acylation] generated and dephosphorylated [3H]Man glycan derivatives on P4 butanol phase-retained products with mobilities characteris- columns gave identical elution times. tic of acyl inositol glycans (paralleling that of Tryp {). In Next, H2-H7 were subjected to further glycan analyses. accordance with this interpretation, mild alkali treatment Incubation with jack bean a-mannosidase released substan- [which cleaves acyl linkages (see Fig. 1)] released 88-96% of tial [3H]Man label from H3, H4, and H6 but none from H5 and the label to the aqueous phase from each of the GPI-PLD- only small amounts from H7 (Table 1). Dephosphorylation of treated species as compared to 76% from the treated r the glycan fragments deriving from (i) alkali pretreatment and control. In further accordance, incubation of untreated PI-PLC digestion and from (ii) HNO2 deamination and re- H2-H8 with PI-PLC [which cleaves diradylglycerol from duction of H7, and sizing of the products on both P4 and inositol monophosphate (see Fig. 1) but is constrained by AX-S columns gave values coinciding with those of the inositol acylation] released <20%o of the [3H]Man label into corresponding glycan fragments of Tryp A' (P2) (Fig. 4). the aqueous phase. However, incubation ofuntreated H2-H8 Similar analyses of the fragments deriving from the same with mild alkali (at 370C for 30 min) released variable pro- treatments ofH6 gave virtually identical values (Table 1). H6 portions of the [3H]Man label from each species, and each of thus was identified as the Man3-GlcN-containing precursor the species exhibited augmented aqueous phase [3H]Man and H7 was identified as its P-Etn-substituted product. release upon PI-PLC exposure when examined in conjunc- Consistent with this interpretation, while the ratio of label in tion with alkali pretreatment. To clarify this observation, H7 to that in H6 was 0.42 after 8 min of pulse labeling, it was [3H]Man products remaining in the butanol phase after lim- 1.2 after 30 min of chase. Downloaded by guest on September 24, 2021 6028 Medical Sciences: Hirose et al. Proc. Natl. Acad Sci. USA 89 (1992) Table 1. Properties of [3H]Man-labeled GPI glycans Size, glucose units Jack bean NH3/PI-PLC a-Manno- HN02 Species sidase* Pa AX-5 Pa AX-5 H6 34.0 6.0 4.0 4.0 2.9 H7 8.0 6.2 3.5 4.0 3.4 A' 1.5 6.2 3.7 4.0 3.5 3.9 8 37.0 *Percentage release into aqueous phase of butanol/water partitions. Block in GPI Assembly in Affected PNH Leukocytes. As a prerequisite to studies of PNH leukocytes, normal PMNs were labeled with [3H]Man and the in vivo products corre- lated with H2-H8 deriving from GDP-[3H]Man labeling of HeLa cell microsomes. In studies of six different donors, prominent products (Fig. 5B) aligning precisely with H6, H7, and H8 (Fig. 5A) were uniformly observed. Based on these findings, affected PMNs from two patients (J.V. and L.D.) with >95% DAF- and CD59-deficient PMNs were similarly analyzed. As shown in Fig. 5 C and D, the cells from both patients synthesized Dol-P-Man but failed to support assem- bly of H6-H8. No accumulation of H5 or less polar species was observed, although J.V. showed minor products corre- sponding in mobilities to H2 and H4. Affected PMNs from two other patients (M.W. and S.T.) next were analyzed for the same or different defects. While studies of three addi- tional controls showed prominent H6-H8 peaks paralleling those of the other normal subjects, studies of both patients showed the complete absence of all three products, with that of M.W. exhibiting minor products with mobilities corre- Migrai/km (/cmzn sponding to H2 and H4 as seen in J.V. UDP-[3H]GlcNAc labeling demonstrated that as reported for J.V. and P.K. (6), FIG. 5. HeLa cell microsomes (360 /ug) were labeled for 90 min with GDP-[3H]Man (A), PMNs from normal control S.H. (B), or patients L.D., M.W., and S.T. all synthesized GlcNAc- and PNH patients J.V. and L.D. (C and D) were labeled with [3H]Man, GlcN-PI (GPI-A and -B) with normal efficiency (data not and products were compared by TLC. (Insets) Flow cytograms of shown). anti-CD59-stained cells from each donor. Positions of in vitro HeLa cell standards run internally are indicated. D, Dol-P-Man; 2-8, DISCUSSION H2-H8.

The main findings of the present study are (i) the identifica- istic of differential P-Etn substitution, and (iii) the finding tion ofintracellular GPI in human cells that uniformly exhibit that both ofthese species, as well as H8 are uniformly absent properties of acylated inositol phospholipiids, (ii) the dem- in affected cells offour PNH patients. In conjunction with our onstration that two ofthe most polar ofthesse species, H6 and previous characterization in human cells of GlcNAc- and H7, contain core glycans corresponding in s;ize to that ofTryp GlcN-PI, and our demonstration that assembly ofthese initial A' (P2) and exhibiting a-mannosidase sensiitivities character- two intermediates is normal in affected cells of two PNH patients, the data in this report further contribute to a AX 5 P4 coherent characterization of human GPI synthesis and more precisely localize the biochemical defect(s) underlying the PNH phenotype. Our observation that assembly of mannolipids H2-H7 was accomplished with microsomes establishes that the machin- ery for synthesis of GPI-anchor core structures in human cells, as in Tryps, is contained within the ER. The lower 200- efficiency in the mammalian system ofconversion ofGlcN-PI to subsequent products may reflect a more complicated process of ER intraluminal translocation-e.g., involving inositol acylation. The lesser dependency of GPI mannolipid assembly on added UDP-GlcNAc suggests that mammalian L__ but not Tryp ER contains a substantial pool of GlcN- 100 0 60 120 containing intermediates, which are available for mannosy- Fraction lation. The conversion by GPI-PLD of H2-H7 into products FIG. 4. latrobead-purified H7 (Lower) and Ti A' (Upper) were characteristic of acyl inositol glycans and the aqueous-phase treated with alkali followed by PI-PLC. After doeryp the released glycans with HF, the products were mixed with [14C]glu- release of [3H]Man label in these species by PI-PLC only in cose monosaccharide and compared on AX-5 (L,eft) and P4 (Right). conjunction with alkali pretreatment suggests that inositol in V0, void volume; numbers correspond to posiitions of mono- and these intermediates is acylated. This interpretation is sup- oligoglucose standards prepared from dextran/lborohydrate. ported by the findings that butanol phase-retained products Downloaded by guest on September 24, 2021 Medical Sciences: Hirose et al. Proc. Natl. Acad. Sci. USA 89 (1992) 6029 (i) following GPI-PLD treatment were converted to aqueous- equal to normal cells predicted that the defect in these phase products by alkali and (ii) following alkali pretreatment patients must reside in subsequent steps in GPI-anchor exhibited TLC mobilities characteristic of partially deacy- assembly or incorporation into protein. Our observation in lated derivatives and were cleaved by PI-PLC. The human the present study, that cells from three other patients simi- GPIs thus correspond to the acylated Tryp intermediates f3 larly synthesize GPI-A and -B but that cells from all of four and ; (P3) (1). Our finding that human GPI mannolipids, that patients studied lack H6-H8 and at least two show complete uniformly contain acylated inositol, are present in cells absence of earlier intermediates localizes the biochemical bearing surface proteins with GPI-anchor structures that block to (i) GlcN-PI acylation or luminal translocation, (ii) contain predominantly unsubstituted inositol [as described in P-Etn or Man incorporation, or, conceivably, (iii) altered HeLa cell DAF molecules (13)], implies the existence of a regulation. Further analyses are necessary to permit delin- terminal deacylase. eation of the precise alteration(s) and confirmation of The cleavage and partitioning of H2-H7 into the aqueous whether the disease is heterogeneous as previously postu- phase following alkali treatment alone suggests that they lated (2, 5). contain diacylglycerol as do Tryp intermediates (1). How- ever, further studies will be required to allow distinction We thank Drs. T. Rosenberry and A. Tartakoff (Case Western between exclusively diacyl and mixed diacyl and alkylacyl Reserve University) and Dr. S. Ogata (Fukuoka University) for phospholipids and clarification of the relationship between helpful discussion, Dr. P. J. Lachmann (Cambridge University) for phospholipids in these mannolipid GPI precursors YTH53.1 anti-CD59 antibody, and Beth McCarty for manuscript inositol preparation. This investigation was supported by grants from the and those in mature human GPI-anchor structures (13). National Institutes of Health (AI23598 and P01DK38181). M.E.M. Comparative analyses of the HNO2-generated glycan frag- is an Established Investigator of the American Heart Association. ments of the aqueous phase-released and butanol phase- retained products indicated that the incomplete alkali release 1. Thomas, J. R., Dwek, R. A. & Rademacher, T. W. (1990) did not reflect heterogeneity of glycans. Biochemistry 29, 5413-5422. The observation of only 30-40%o release of [3H]Man label 2. Medof, M. E. (1988) in Cytolytic Lymphocytes and Comple- from H3, H4, and H6 following jack bean a-mannosidase ment: Effectors of the Immune System, ed. Podack, E. R. digestion is most likely due to reduced efficiency of enzy- (CRC, Boca Raton, FL), pp. 57-86. matic cleavage of glycolipid as compared to glycan since 3. Davies, A., Simmons, D. L., Hale, G., Harrison, R. A., Tighe, H., Lachmann, P. J. & Waldmann, H. (1989)J. Exp. Med. 170, similar release was measured with Tryp S. Another factor, 637-654. however, may be that during GPI-anchor core glycan assem- 4. Okada, N., Harada, R., Fujita, T. & Okada, H. (1989) J. bly, Man 1 in human GPI precursors is substituted with Immunol. 143, 2262-2266. P-Etn, in accordance with the presence of this substituent in 5. Carothers, D. J., Hazra, S. V., Andreson, S. W. & Medof, mature human GPI-anchor structures (14, 15). Such substi- M. E. (1990) J. Clin. Invest. 85, 47-54. tution would account for similarities in TLC mobilities be- 6. Hirose, S., Ravi, L., Hazra, S. V. & Medof, M. E. (1991) Proc. tween the human and Tryp species despite the presence ofthe Natl. Acad. Sci. USA 88, 3762-3766. third inositol-associated acyl substituent in the former. Ob- 7. Masterson, W. J., Doering, T. L., Hart, G. W. & Englund, servations of shifts in mobilities of the HF and non-HF P. T. (1989) Cell 56, 793-800. 8. Menon, A. K., Schwarz, R. T., Mayor, S. & Cross, G. A. M. reduced HNO2 fragments of H5 and H6 on Dionex chroma- (1990) J. Biol. Chem. 265, 9033-9042. tography (S.H., G.M.P., L.R., and M.E.M., unpublished 9. Kinoshita, T., Medof, M. E., Silber, R. & Nussenzweig, V. observations) are consistent with this interpretation. (1985) J. Exp. Med. 162, 75-92. The greater sensitivity ofH6 than the more polar H7 tojack 10. Feldman, R. A., Gaetani, S. & Morimoto, T. (1982) in Cancer bean a-mannosidase suggested that a distal Man residue in Cell Organelles, eds. Reid, E., Cook, G. M. & Morre, D. J. H7 was substituted. The findings that the (dephosphorylated) (Wiley, New York), Vol. 2, pp. 263-292. glycan derivatives of HNO2 deamination as well as alkali 11. Lemansky, P., Gupta, D. K., Tucker, M. G. & Tartakoff, pretreatment and PI-PLC digestion of both H6 and H7 A. M. (1991) Mol. Cell. Biol. 11, 3879-3885. exhibited virtually identical elution behavior on both P-4 and 12. Ogata, S., Misumi, Y., Miki, K. & Ikehara, Y. (1986) Eur. J. Biochem. 161, 315-320. AX-5 columns, which coincided in all cases with the corre- 13. Walter, E. I., Roberts, W. L., Rosenberry, T. L., Ratnoff, sponding fragments (Man3-anhydromannitol and Man3- W. D. & Medof, M. E. (1990) J. Immunol. 144, 1030-1036. GlcN-inositol) from Tryp A' (P2), argue that human and Tryp 14. Roberts, W. L., Santikern, S., Reinhold, U. N. & Rosenberry, intracellular GPIs share common core glycans. T. L. (1988) J. Biol. Chem. 263, 18776-18784. Our previous findings that affected PMNs of two PNH 15. Medof, M. E., Walter, E. I., Roberts, W. L., Haas, R. & patients assemble GlcNAc-PI and GlcN-PI with efficiency Rosenberry, T. L. (1986) Biochemistry 25, 6740-6747. Downloaded by guest on September 24, 2021