Cell surface human Lfucosidase - Wiley Online Library

Eur. J. Biochem. 268, 3321±3331 (2001) q FEBS 2001

Cell surface human a-L-fucosidase Oscar J. Cordero1, Ana Merino2, MarõÂa PaÂez de la Cadena2, BegonÄa BugõÂa1, Montserrat Nogueira1, Juan E. VinÄuela3, Vicenta S. MartõÂnez-Zorzano2, Alejandro de Carlos2 and F. Javier RodrõÂguez-Berrocal2 1 Department of Biochemistry and Molecular Biology, University of Santiago de Compostela, Faculty of Biology, Santiago de Compostela, Spain; 2Department of Biochemistry, Genetics and Immunology, University of Vigo, Faculty of Sciences, Vigo, Spain; 3Service of Immunology, Clinical University Hospital of Santiago de Compostela, Spain

The acid a-l-fucosidase is usually found as a soluble component of lysosomes where fucoglycoconjugates are degraded. In the present investigation, we have demonstrated the existence of a cell surface protein with enzymatic a-l-fucosidase activity that crossreacts specifically with a rabbit anti-(a-l-fucosidase) Ig. By different approaches, this a-l-fucosidase, which represents 10±20% of the total cellular fucosidase activity, was detected in all the tested human cells (hemopoietic, epithelial,

The lysosomal a-l-fucosidase (EC 3.2.1.51) is involved in the hydrolytic degradation of fucose-containing molecules. Mammalian a-l-fucosidases are multimeric forms of glycoproteins of < 53 kDa exhibiting optimal activity between pH values of 4 and 6.5 [1]. There appears to be a considerable degree of structural heterogeneity, both tissue-specific and within the same tissue. This is probably due to a different content in sialic acid residues [2,3], in addition to the two different alleles and genetic polymorphism encoded by the FUCA1 gene [4,5]. The significance of the FUCA2 gene is currently unknown [4±6]. The absence or gross deficiency of a-l-fucosidase activity causes accumulation of fucoglycoconjugates, which leads to the genetic neurovisceral storage disease fucosidosis in mammals [7,8]. A new syndrome, previously known as leukocyte adhesion deficiency II, has been recently characterized as a generalized metabolic disease consisting of severe hypofucosylation of glycoconjugates [9]. Moreover, an abnormal intracellular and extracellular distribution of a-l-fucosidase is found, for example, in cystic fibrosis (decreased levels in antiserum and higher levels in skin fibroblasts) or in colon carcinomas (decreased levels in antiserum and in tumor tissue) [10±13]. In general, tissue differentiation and development through cell±cell Correspondence to O. J. Cordero, Departamento BioquõÂmica e BioloxõÂa Molecular, Universidade de Santiago de Compostela, Facultade de BioloxõÂa. Campus Sur, 15706 Santiago de Compostela, Spain. Fax: 1 34 981 596904, Tel.: 1 34 981 563100, E-mail: [email protected] Abbreviations: PM, plasma membrane; PBMC, peripheral blood mononuclear cells; PMN, polymorphonuclear cells; PFA, paraformaldehyde; BS3, Bis(sulfosuccinimidyl) suberate; At, total activity; IP, immunoprecipitation; MPR, mannose-6-phosphate receptors; HLB, hypotonic lysis buffer; INT, p-iodonitrotetrazolium violet; 4-MU-fucoside, 4-methyl-umbelliferyl-a-l-fucopyranoside. Enzyme: a-l-fucosidase (EC 3.2.1.51). (Received 6 February 2001, accepted 17 April 2001)

mesenchymal). Two bands of < 43±49 kDa were observed, although theoretical data support the possibility of having the same genetic origin that the known 50 to 55-kDa Mr a-l-fucosidase. We speculate about an alternative traffic pathway for the plasma membrane a-l-fucosidase to work on the rapid turnover of glycoproteins. Keywords: a-l-fucosidase; plasma membrane; protein traffic; CD26; glycoprotein turnover.

recognition are modulated by sequential changes of the sugar chains of cell surface glycoproteins. As the expression or deletion of a-fucosyl residues linked at various positions of sugar chains of glycoproteins is one of these alterations, the role of a-l-fucosidase in these processes is of considerable interest. The majority (90±100%) of the a-l-fucosidase activity in almost all mammalian tissues investigated is in the soluble fraction [1], the human brain being the exception [14]. A serum form of a-l-fucosidase, which has attracted interest as a diagnostic marker in many clinical studies [15], has turned out to be similar to the tissue forms [4±6,16,17]. Very recently, a-l-fucosidase from rat testis and epididymal spermatozoa was found by immunocytochemistry to be associated with the outer plasma membrane and a-lfucosidase active site to have a neutral pH optimum and also to be accessible to fucoconjugates [18,19]. Using a newly developed polyclonal rabbit anti-(a-l-fucosidase) Ig against the human placenta protein, we have investigated whether the a-l-fucosidase is universally present on the cell surface.

M AT E R I A L S A N D M E T H O D S Polyclonal Ig preparation In order to obtain a polyclonal anti-(a-l-fucosidase) Ig, two female New Zealand White rabbits were immunized with purified a-l-fucosidase from human placenta (SigmaAldrich, Madrid, Spain) (containing less than 5% a-dmannosidase and less than 2.5% b-d-galactosidase and N-acetyl-b-d-glucosaminidase) by several subcutaneous injections of 0.5 mg of protein mixed in 1.5 vol. of complete Freund's adjuvant. The second immunization was performed intramuscularly on day 55 with 0.25 mg of protein. Serum was obtained by ear vein bleeding prior to the first immunization and 15 days after the second immunization. The purification of the whole serum Ig

3322 O. J. Cordero et al. (Eur. J. Biochem. 268)

fraction was achieved by precipitation with 45% ammonium sulfate, exhaustive dialyzation against NaCl/Pi pH 7.4, and affinity chromatography using protein A coupled to Trisacril GF-2000 (Pierce, Rockford, IL, USA). After the antisera had recirculated through the column, it was firstly washed with 3m NaCl, 0.05 m sodium borate, pH 8.9, then washed with the same buffer containing 0.01 m sodium borate, and eluted by addition of 0.1 m glycine/HCl, pH 3. Fractions (0.5 mL) were collected in tubes containing 50 mL of 1 m Tris, to give a final pH of 7.4. The most enriched protein-containing fraction was dialyzed and aliquoted for storage. The antibody was initially characterized by Western blotting and immunoprecipitation. Cell preparation and culture Human peripheral blood mononuclear cells (PBMC) and polymorphonuclear cells (PMN; primarily neutrophils) were obtained from healthy buffy coats (kindly provided by the Centro de Transfusiones de Galicia, Santiago, Spain) as described previously [20]. PBMC were isolated by Ficoll-Paque PLUS (Pharmacia Biotech, Uppsala, Sweden), from which monocytes were separated by a One-Step Monocytes (Accurate Chem. & Sci. Corp., Westbury, NY, USA) gradient. PMN were obtained from the Ficoll gradient pellet by One-Step PMN (Accurate Chem. & Sci. Corp.). Lymphocytes and cell lines from lymphoid origin were cultured in suspension in RPMI 1640 medium supplemented with 10% inactivated (56 8C, 1 h) fetal bovine serum, 100 mg´mL21 streptomycin, and 100 IU´mL21 penicillin (all from Sigma, except fetal bovine serum from Gibco, Grand Island, NY, USA), in a humidified atmosphere with 5% of CO2 in air. PBMC at 106 cells´mL21 were activated in the presence of 1 mg´mL21 of phytohemagglutinin (Sigma) for 3 or 5 days. Epithelial cell lines were grown similarly using Dulbecco's modified Eagle's medium (Biochrom, Hamburg, Germany) instead of RPMI. The mesenchymal MG63 cell line was grown in complete Eagle's minimal essential medium supplemented additionally with nonessential amino acids (Biochrom). Many of these cell lines are of common use in our laboratory; the others were purchased from the ATCC (Rockville, MA, USA) or from the ECACC (Salisbury, UK). Adherent cells were trypsinized every 3±4 days. Flow cytometry Indirect immunofluorescence staining of resuspended cells was performed in NaCl/Pi containing 1% BSA and 0.05% sodium azide, as previously described [21]. After incubation with the rabbit anti-(a-l-fucosidase) Ig (30 min on ice), cells were washed twice and FITC-labeled goat anti(rabbit Ig) secondary Ig (Sigma) was added for other 30 min on ice. When indicated, cells were fixed in NaCl/Pi, 1.5% paraformaldehyde (PFA). The samples were analyzed on a Becton Dickinson Excalibur cytometer. The preimmune rabbit serum was used as control, and propidium iodide (Sigma) for cell viability determinations. The winmdi software (a kind gift from J. Trotter, Scripps Institute, LaJolla, CA, USA) was used to analyze the data.

q FEBS 2001

Membrane preparation A protocol designed for a high quality plasma membrane (PM) preparation was carried out with minor variations [22]. Briefly, the cells were resuspended in hypotonic lysis buffer (HLB; 20 mm Tris HCl, pH 7.5, 25 mm sucrose, 1 mm EDTA, 0.15 m NaCl, 5 mm dithiothreitol, 10 mg´mL21 leupeptin, 5 mg´mL21 aprotinin, all from Sigma, and 1 mm Pefabloc, Boehringer Mannheim) at 3 106 cells´mL21, and sonicated. Nuclei were removed by centrifugation at low speed (1000 g for 10 min at 4 8C). The turbid supernatant was collected and loaded onto a 41% solution of sucrose in HLB. Membranes were pelleted at 95 000 g at 4 8C for 60 min using a Beckman L8-M ultracentrifuge. Finally, the supernatant fluid representing the soluble or cytoplasmic fraction was discarded, and the resulting pellet was disrupted in HLB, 1% Triton X-100, by sonication. The protein content was determined by the BCA Kit assay (Pierce, Rockford, IL, USA), using BSA as the standard. Aliquots of the HT-29 crude cell extract and the pellet were also assayed for enzymes 5 0 -nucleotidase, a plasma membrane marker, fucosidase and the lysosomal markers b-d-galactosidase and acid phosphatase, so that specific activities and enzyme-enrichment factors of the protocol could be determined. The purity of the PM fraction obtained in this way from human colon tissue (removed from colorectal carcinoma patients [23]) had been also evaluated previously in our laboratory by the measurement of the following enzyme markers in the crude homogenate and the membrane preparation: a-d-mannosidase and N-acetyl-b-d-glucosaminidase (internal membranes), succinate dehydrogenase (mitochondrial) and lactate dehydrogenase (cytoplasmic), in addition to a-l-fucosidase and 5 0 -nucleotidase. Glycosidase activities were measured as described in the following section. 5 0 -Nucleotidase was determined with the Enzyline 5 0 NU kit (bioMeÂrieux, Marcy-l'Etoile, France) following the manufacturer's protocol (3 min of preincubation at 25 8C). For the determination of acid phosphatase activity, 50 mL of the preparations were mixed with 200 mL of 8 mm of p-nytrophenil phosphate (Merck) in 90 mm acetate buffer pH 5.0 without MgCl2, and incubated at 37 8C. After 25 min, the reaction was stopped by addition of 0.6 mL of 0.25 m NaOH. The phosphatase activity was recorded at 410 nm after centrifugation of the samples at 700 g for 5 min. For the determination of lactate dehydrogenase activity, the reaction mixture contained 2 mm sodium pyruvate and 0.2 mm NADH in Tris/HCl pH 7.5, and the NADH extinction at 30 8C was followed at 339 nm. The preparations were also incubated with 0.05 m sacarose and 0.1 m sodium succinate in 0.1 m sodium phosphate buffer, pH 7.4 at 37 8C for 20 min in the presence of 0.1% p-iodonitrotetrazolium violet (INT; Sigma) in ethanol. The reaction was stopped with 10% trichloroacetic acid. Ethyl acetate was used for colour formation, and the succinate dehydrogenase activity recorded at 490 nm in the supernatant after agitation and 5 min sample centrifugation at 900 g. Table 1 shows that the final plasma membrane pellet, obtained from the two kinds of cells, was enriched only for the 5 0 -nucleotidase activity and by approximately fivefold to sevenfold, as determined by the specific activity ratios. Moreover, the overall recovery of 5 0 -nucleotidase activity

q FEBS 2001

Cell surface human a-l-fucosidase (Eur. J. Biochem. 268) 3323

SDS/PAGE (10% in the first case, 12% in the others) as described [21,24] before processing for immunoblot assay on Hybond-P poly(vinylidene difluoride) membranes (Amersham PharmaciaBiotech, Rainham, UK) by using rabbit anti-(a-l-fucosidase) Ig, 1 : 100 (v/v) from the antisera Ig fraction (1 : 500 in the Ig characterization experiments). The specific Ig was revealed with goat anti-(rabbit Ig) Ig labeled with horseradish peroxidase and ECL (Amersham PharmaciaBiotech), or with goat anti-(rabbit Ig) Ig coupled to alkaline phosphatase (1 : 1000 dilution) and 5-bromo-4-chloro-3-indolyl phosphate (BCIP) and Nitro Blue Tetrazolium (NBT) substrates.

from the original homogenate was 29 ^ 2% in tissue and 59 ^ 5% in HT-29, while that of the other activities ranged between 3.2±8.7% (results from four experiments). Immunoprecipitation (IP) and enzymatic assays IP was essentially carried out as follows [23]. Different volumes (0±50 mL) of preimmune or rabbit anti-(a-lfucosidase) Ig were incubated at 4 8C in 10 mm NaCl/Pi, pH 6.5, with 50±100 mL Protein A (Sigma). One hour later, the pellets obtained after centrifugation at 400 g for 10 min, were mixed with the membrane or the whole lysate preparation and incubated at 4 8C for 1 h with occasional shaking. After centrifugation as above, the resulting supernatant and pellet were assayed for a-l-fucosidase activity as described previously [12,23] with minor modifications. Synthetic substrate 4-methyl-umbelliferyl-a-l-fucopyranoside (4-MU-fucoside; Sigma) at a final concentration (under heat and agitation) of 1.46 mm in 0.1 m NaCl/Pi, pH 6.5, was mixed with an aliquot of supernatant or the resuspended pellet and incubated at 37 8C for 45 min (under linear conditions) before fluorescence reading on a Hitachi F-2500 spectrophotometer (450 nm). One unit of a-l-fucosidase activity was defined as the amount of enzyme required to hydrolyze 1 nmol of substrate per min at 37 8C. In the Ig characterization experiments, a-lfucosidase, a-d-mannosidase, b-d-galactosidase, b-dglucuronidase and N-acetyl-b-d-glucosaminidase activities, in total lysates or in the supernatants and pellets after IP, were assayed with the respective 4-MU-derivatives at a final concentration of 0.35 mm in 0.1 m K2PO4H citric acid buffer, pH 4.

Rabbit anti-(a-l-fucosidase) Ig was radiolabeled with Na125I (DuPont NEN, Germany, or Amersham PharmaciaBiotech) by using Iodobeads (Pierce, Rockford, IL), a modification of the Chloramine-T method, following the method previously described [24]. Fresh or cultured cells were resuspended in isotonic NaCl/Pi and stained with 125I-labeled rabbit anti-(a-l-fucosidase) Ig for 30 min as described above. After extensive washing, cells were incubated for 2 h with or without 1 mm Bis(sulfosuccinimidyl) suberate (BS3; Pierce), a water-soluble membraneimpermeable crosslinker, according to the manufacturer's instructions. Cells were washed and then lysed as described above before SDS/PAGE. Dried gels were autoradiographed at different timepoints. For control, rabbit anti(a-l-fucosidase) Ig was incubated with purified human placenta a-l-fucosidase in the presence or absence of 1 mm BS3 prior to denaturalization and SDS/PAGE.

Immunoblot assay

Transmembrane regions detection

The solubilized membrane proteins, lyophilized if necessary for protein concentration, as well as whole-cell lysate samples or commercial a-l-fucosidase were diluted in the electrophoresis sample buffer and analyzed on an

The analyses of protein sequences to detect transmembrane regions were performed using the expasy Molecular Biology Server (http://www.expasy.ch) software package available from the Swiss Institute of Bioinformatics, and

Radiolabeling and crosslinking

Table 1. Marker enzyme analysis of the plasma membrane preparations. Human colon mucosa tissue was homogenized as reported [23] and HT-29 cells were cultured as described in Materials and methods. The plasma membrane preparation and enzyme activity analysis are also described in the text. Marker enzymes were analysed in homogenates of both crude extracts and the PM-enriched pellet. Lactate DH, lactate dehydrogenase (EC 1.1.1.27); N-acetyl-b-d-glu, N-acetyl-b-d-glucosaminidase (EC 3.2.1.50); a-d-mannosidase (EC 3.2.1.24); succinate DH, succinate dehydrogenase (EC 1.3.99.1); 5 0 NT, 5 0 nucleotidase (EC 3.1.3.5); acid phosphatase (EC 3.1.3.2). Data are presented as mean ^ SD of triplicate determinations from one representative experiment out of four. SD from the means of the four experiments were around the 15%, this fact pending basically on the number of cells obtained in each preparation. Data are expressed in nmol per min per U (at the respective optimal pH and temperature) per mg protein, except *, U mmol substrate´min21. HT-29 cell line Specific enzyme activity (^ SD) a-l-Fucosidase Acid phosphatase * b-d-Galactosidase 5 0 NT Lactate DH * N-acetyl-b-d-glu a-d-Mannosidase Succinate DH *

Colon mucosa

Crude extract

Plasma membrane

Crude extract

Plasma membrane

4.304 0.126 0.351 0.773

1.206 0.087 0.221 3.580

5.729 ^ 0.438

6.752 ^ 0.578

0.658 2.385 2.340 0.408 0.006

4.258 1.332 3.501 0.575 0.004

^ ^ ^ ^

0.375 0.009 0.023 0.045

^ ^ ^ ^

0.098 0.005 0.016 0.269

^ ^ ^ ^ ^

0.065 0.192 1.425 0.024 0.001

^ ^ ^ ^ ^

0.487 0.086 0.290 0.030 0.001


q FEBS 2001

the SPLIT server (htpp://pref.etfos.hr/split) from the Faculty of Electronics, University of Osijek, Croatia.

R E S U LT S Specificity of rabbit anti-(a-l-fucosidase) Ig and indirect immunofluorescence Rabbit anti-(a-l-fucosidase) Ig identified a main band of 51 kDa and a minor band of 47 kDa on blotted SDS/PAGE gels of purified human placenta a-l-fucosidase. A main band of about 49 kDa was identified on blotted SDS/PAGE gels of total lysate extracts of human enterocytic HT-29 and macrophage RAW cell lines (Fig. 1). Moreover, rabbit anti-(a-l-fucosidase) Ig, but not preimmune serum, immunoprecipitated specifically more than 80% of the a-l-fucosidase enzymatic activity from lysates of human colon adenocarcinoma HT-29 cells. On the contrary, they did not precipitate any other glycosidase (including those contaminating the commercial a-l-fucosidase preparation) activity (A. Merino, J. Fernandez, A. de Carlos, V. Martinez-Zorzano, M. Paez de la Cadena & F. J. Rodriguez-Berrocal, unpublished results) the wide range of Ig concentrations used. Once a specific polyclonal anti-(a-l-fucosidase) purified Ig was obtained, we sought to determine by flow cytometry whether a-l-fucosidase might be localized on the surface of cells available in our laboratory. Table 2 demonstrates the existence of a cell surface protein that crossreacts with anti(a-l-fucosidase) serum. Several facts point towards the specificity of this reaction. Firstly, the results from Table 2 show staining of living (propidium iodide free) cells. In some cultures where a percentage of cells died the same results were found, thus suggesting that the a-l-fucosidase did not come from dead cells [bearing in mind the existence of mannose-6-phosphate receptors (MPR) on the cell surface] [25]. Secondly, the cells of the same origin were stained differently, e.g. colon adenocarcinoma epithelial HT-29, Caco-2 and COLO 205 cells, or hemopoietic histiocytic HL-60 and K562 cells (Fig. 2). Thirdly, these results were reproducible in at least three experiments with three exceptions: (a) monocytes (purified and from fresh blood) from Table 2 show the results from three donors, monocytes from two additional donors were about 20% positive for a-l-fucosidase; (b) Jurkat cells (Fig. 1); and (c) MG-63 cells, when fixed, showed in some experiments two subpopulations of positive and negative cells for a-lfucosidase. The fixation of cells with 1.5% PFA is a common technique used in flow cytometry. We observed that some although not all of the cell lines treated with PFA changed their staining pattern with rabbit anti-(a-lfucosidase) Ig indicating again the specificity of the staining. In our conditions, PM was not permeabilized with PFA suggesting another mechanism [e.g. conformational change of the epitope recognized by the rabbit anti-(a-l-fucosidase) Ig]. This being true, a-l-fucosidase may be present on the surface of every kind of cells but showing a low accessibility to this Ig, being the neutrophils (Fig. 2) and Caco-2 0 s a-l-fucosidase the most accessible in hemopoietic and epithelial cells, respectively. The fact that rat SP2 and monkey Cos1 cells were stained by the Ig suggests that our finding could be extended at least to mammals. At least in T cells, cell surface a-l-fucosidase is

Fig. 1. SDS/PAGE and a-l-fucosidase immunodetection. Mr markers (lane 1) and 5 mg of commercial purified human placenta a-l-fucosidase (lane 2) were resolved on a 12% polyacrylamide gel and stained with Coomassie Blue. In parallel, in lane 3, the protein was transferred to nitrocellulose and probed with a 1 : 500 dilution of anti-(a-l-fucosidase) Ig. In lanes 4 and 5, 20 mg of protein from total lysates of HT-29 and RAW human cell lines were treated under the same protocol. Blot detection was performed with goat anti-(rabbit Ig) Ig coupled to alkaline phosphatase (1 : 1000 dilution) and 5-bromo4chloro-3-indolyl phosphate (BCIP) and Nitro Blue Tetrazolium (NBT) substrates. An experiment representative of several with the same results.

not regulated by activation (Table 2), which suggests it is a housekeeping protein. Western blot As rabbit anti-(a-l-fucosidase) Ig recognized the bands of purified placenta a-l-fucosidase by Western blot, a PM purification from the cell lysates was performed in order to solubilize the membrane proteins. Figure 3 confirms the presence of a plasma membrane crossreactive protein band of about 47±49 kDa, being identified two close immunoreactive bands in some preparations. There are some discrepancies between data from Table 2 and Fig. 3 which represents a blot from SDS/PAGE of 25 mg of 1% Triton X-100 solubilized proteins. In a second blot in which 6 mg of protein per lane was loaded, the band(s) were observed only in the lanes of lymphocytes, neutrophils, Daudi and Jurkat cells, suggesting a low quantity of anti-(al-fucosidase) Ig crossreactive protein in the PM preparations of many cell types. Curiously, a band of the same Mr was also observed in the K562 lane of this second blot. Immunoprecipitation and enzymatic activity The discrepancies between data from Table 2 and from Fig. 3 led us to assay the presence of a-l-fucosidase activity in the same preparations used above. We chose four preparations, HL-60 and Jurkat where antisera crossreactive PM protein was recognized by Western blot and U937 and COLO 205, where it was not recognized in spite of the weak positive immunofluorescence signal. A preparation from total lysate of HT-29 cells was also measured for comparison. Table 3 shows that a-l-fucosidase activity was

q FEBS 2001


Table 2. Native cells and cell lines screened for the presence of a-l-fucosidase on cell surface. Reactivity was assessed by flow cytometry of duplicates and data represent the mean values (SD were lesser than 10% of the values if relativised) of one representative experiment (out of at least three, values closely agreeing except those marked with *., one representative experiment out of various with different results, see text). ±, Less than 5% of positive cells. MFI, mean fluorescence intensity. Fresh

PFA treated

Origin

%

MFI

Human hemopoietic cells and lines Resting lymphocytes T lymphoblasts, phytohemagglutinin day 3 Monocytes PMN HL-60 (promyelocytic leukaemia) K562 (chronic myelogenous granulocytic leukaemia) Daudi (Burkitt's lymphoma, B lymphoblasts) Jurkat (leukaemic T cell lymphoblasts) U937 (monocyte like histiocytic lymphoma) NC-37 (peripheral blood B lymphoblasts) YT (NK-like lymphoblasts)

19 13 71 83 11 21 42 24 20 55 ±

(21.6) (25.0) (19.6) * (105.4) (16.3) (14.1) (9.9) (11.6) (12.0) (44.3)

Human epithelial cell lines HeLa (cervix carcinoma) HT-29 (colon adenocarcinoma) Caco-2 (colon adenocarcinoma) COLO 205 (colon adenocarcinoma, suspension)

± 17 93 8

Human mesenchymal cell lines MG63 (osteosarcoma)

±

Animal cell lines Cos1 (monkey african green kidney, SV40 transformed) Sp-2/0-Ag 14 (mouse myeloma)

20 23

detected in the four digested PM preparations, confirming the data from Table 2. From our protein measurements (data not shown), we estimated the PM a-l-fucosidase at about 2±4% of total cellular a-l-fucosidase. We also performed immunoprecipitation (IP) to check the specificity of the PM protein which crossreacts with rabbit anti-(a-l-fucosidase) Ig. Figure 4 shows that anti-(a-l-fucosidase) serum but not preimmune serum immunoprecipitates enzymatic activity. Although 50 mL of antiserum were enough to precipitate nearly 90% of enzymatic activity from 300 mg of HT-29 total extracts at 37 8C (A. Merino, J. Fernandez, A. de Carlos, V. MartinezZorzano, M. Paez de la Cadena & F. J. Rodriguez-Berrocal, unpublished results), under set conditions used in this experiment we found an impairment of the IP yield (30.8%) due to both sample quantity (30 mg) and temperature [4 8C, chosen to avoid total activity (At) loss] but not to the presence of detergent (Triton) in HT-29 total extracts (Fig. 4A). These results show four points: (a) At recovered after immunoprecipitation was found to be higher (in some of the PM preparations several-fold) than before IP (Table 3). (b) Longer incubation times improved the precipitate yield, as seen when supernatants were reimmunoprecipitated (Fig. 4B). Moreover, it should be taken into account that we were playing against a loss of enzymatic activity with time, see HT-29 total lysates in Table 3, so the IP yield may be underestimated. (c) The recovery of a-l-fucosidase At after IP was similar in the four PM preparations (Table 3). However, the percentage of

(47.0) (55.7) (13.7)

(13.1) (13.2)

%

MFI

100

(53.2) *

86

(11.4)

12

(37.1)

61

(23.7) *

13

(9.0) *

immunoprecipitated activity was higher in the preparations which showed higher activity before IP (U937 and HL-60) (Fig. 4B). (d) One immunoprecipitation from the same concentration of protein of total lysates from HT-29 cells yielded from two- to fourfold more fucosidase activity (in percentage) than PM preparations (compare Fig. 4A,B). Together, these results and data from Western blots suggest that membrane a-l-fucosidase is difficult to solubilize and this may be the reason why its detection has not been possible so far. Moreover, taking into account these points, we should reestimate the content of PM a-l-fucosidase at a minimum of 10±20% of the total cellular a-l-fucosidase. Cross-linking experiments Although in our PM preparations it was not found contamination with membranes from other organella, crosslinking experiments were carried to confirm that the immunofluorescence studies were actually detecting cell surface anti-(a-l-fucosidase) Ig crossreactive protein. The binding of 125I-labeled rabbit anti-(a-l-fucosidase) Ig to intact cells was studied with the presence or absence of water-soluble BS3 crosslinker, which is membrane impermeable, before cell lysis, membrane solubilization, SDS/ PAGE and autoradiography. This approach allowed us to clearly observe in Fig. 5A,B two close bands of about 92 and 86 kDa which were not present in the controls, as well as two other considerable bands of 43 and 26 kDa


q FEBS 2001

Fig. 2. Cell surface immunofluorescent staining patterns. Cells (106´mL21) were stained by indirect immunofluorescence as described in Materials and methods and were analyzed in a Becton Dickinson Excalibur flow cytometer. Controls (thin dotted lines) were stained with the F(ab 0 )2 FITC-labeled goat anti-(rabbit IgG) Ig alone. Cells were stained with the anti-(a-l-fucosidase) serum (bold solid lines) prior to the FITClabeled goat anti-(rabbit Ig) secondary Ig as second reagent. (A) K562 staining. The same pattern showing a slight displacement of the fluorescent intensity to the right, which usually represents the presence of small amounts of antigen on the cell surface, is also observed for example in the COLO 205 cell line, Cos1 cell line or fresh lymphocytes. (B) PMN staining. The same pattern as in (A) but with the greater displacement of intensity to the right observed in the cells not treated with PFA. (C) Jurkat staining in the presence of PFA. The same pattern was also observed for the MG-63 cell line. In both cases, this result showing two different populations for the a-l-fucosidase was not always observed. (D) HT-29 staining also shows two different populations, a minor one expressing high levels of a-l-fucosidase and the major one negative for a-l-fucosidase. The same pattern was observed for the HL-60 cells or the HeLa cell line (in this case, below the 5% necessary to be considered positive). Data are representative of at least (see text) three independent measurements of the different cells.

Fig. 3. Plasma membrane protein Western blotting. PM protein (25 mg) from the indicated kind of cell was resolved by SDS/PAGE on a 10% polyacrylamide gel, transferred to PVDF membrane and probed with 1 : 100 (v/v) anti-(a-l-fucosidase) Ig.

corresponding to the heavy and light chain of the IgG. By subtracting the heavy chain 43 kDa of the crosslinked bands, we identified two cell surface proteins of 49 and 43 kDa crossreactive with the anti-(a-l-fucosidase) Ig, which well corresponded to the values obtained from the Western blot. Four kinds of cells were used. As expected, the crosslinked bands were detected in the Jurkat and Daudi lysates but not in the HT-29 when BS3 was present. Interestingly, as shown in Fig. 5B, the bands were also detected by this approach in MG63 cells. In fact, although hardly detected in Fig. 5A,B, the bands were also present in the absence of BS3 with longer exposures, in which these bands were also found in the HT-29 lysates (Fig. 5C). In the controls, the commercial human placenta a-lfucosidase was incubated with 125I-labeled rabbit anti-(a-lfucosidase) Ig with the presence or absence of BS3. A 105-kDa band was enhanced when BS3 was present and may correspond to the crosslinked ligand±Ig complex. But when the Ig was crosslinked in the absence of ligand, a similar band was observed (data not shown). Thus, this

q FEBS 2001


Fig. 4. a-l-Fucosidase enzymatic activity immunoprecipitation. The same membrane extracts of Table 3 were immunoprecipitated with human anti-(rabbit a-l-fucosidase) serum (AFS, clear bars) and preimmune rabbit serum (PRS, shadowed bars). Enzymatic activity of aliquots from supernatants and resuspended pellets was measured as described in Materials and methods. In A, immunoprecipitation in the absence of Triton from total extracts (30 mg, 10 mL antiserum) of HT-29 cell line, i.e. the recovery of classical immunoprecipitated a-l-fucosidase, is shown. The immunoprecipitation yield (% of precipitated activity) is impaired with small samples of protein (see text) and temperature. In B, PM extracts from four cell lines (60 mg each): two in which a-l-fucosidase was detected in Fig. 2 (HL-60 and Jurkat); and two in which was not (U937 and COLO 205). They were immunoprecipitated with 50 mL of antiserum, at 4 8C to avoid loss of a-l-fucosidase At. Supernatants from this immunoprecipitation (IP I) were reimmunoprecipitated (IP II) in the same conditions. Nonspecific IP with PRS was subtracted from the values shown in B. In the first IP (IP I), a higher yield was found in U937 and HL-60 cell PM which showed a higher At before immunoprecipitation (Table 3). After IPs I and II, both At (Table 3) and Immunoprecipitation yield were similar in the four cell lines. One experiment out of at least two (SD of the calculated means were all lesser than 10% of the original activity determinations, data not shown).

band at least in part, as well as the 77-kDa band, also observed in the lysates lanes, would correspond to crosslinking between heavy and light chains (Fig. 5D). The presence of high Mr bands (which correspond to the Ig fraction) in the control without crosslinker, as well as the disappearance of these bands in the control with BS3 due to the formation of high Mr complexes with the placenta a-lfucosidase subunits, can be explained as we were not able to completely denature the Ig molecules in these conditions. The presence of a faint band of 122 kDa in three out of the four cell lines might correspond to an additional crosslinking to the complex of the 20 to 30-kDa a-lfucosidase subunit sometimes described [2,3,26], also detected by our Ig in the Western blot of total lysates from HT-29 cells. Detection of transmembrane regions All data suggest the presence of a membrane protein in many of the cells but a-l-fucosidase is described as a soluble protein and it is traditionally detected in lysosomes and antiserum. We performed a theoretical analysis of the a-l-fucosidase sequence in order to confirm the physicochemical possibilities of our finding. Looking for transmembrane regions, the program hmmtop (Institute of Enzymology, Hungarian Academy of Sciences) [27] and the split server (University of Osijek, Croatia) identified one transmembrane helix from amino acids 5±23, which corresponded well to the previously described a-l-fucosidase signal peptide [1,4]. In the same way, the programs sosui (Tokyo University of Agriculture and Technology) and tmhmm (Center for Biological Sequencing, Denmark) identified a-l-fucosidase as a soluble protein, although they

did not identify the above signal peptide. However, tmpred (EMBnet, Switzerland) and toppred (Stockholm University) [28,29] identified four possible transmembrane helices: two of which were considered insignificant but the other two were identified as strong transmembrane helices with certainty. The strongly preferred model (Fig. 6) suggested that the first transmembrane helix from amino acids 5±23 (or 3±23 with the second program) should be oriented outside-inside, and that the second one from amino acid 53±71 (or 51±71) should be oriented inside±outside. The first helix corresponds again to the signal peptide described. When the sequence of a-lfucosidase was submitted without the N-terminal signal peptide, the second transmembrane helix was similarly identified. Table 3. a-l-Fucosidase enzymatic activity in PM or total HT-29 cellular extracts before and after immunoprecipitation. ATs (U) are estimated from the enzymatic activity (U´mL21) measured as described in Materials and methods, before and after IP of 60 mg of protein preparations. HT-29 cellular extracts were obtained in the absence of Triton. Values of ATs after IP were obtained from the addition of supernatant plus precipitate ATs. Data are means ^ SD from at least two experiments, performed at 4 8C to avoid loss of a-l-fucosidase At. Cell line

At before IP

(SD)

At after IP

(SD)

U937 Jurkat COLO 205 HL-60 HT-29a

0.105 0.026 0.048 0.060 0.620

0.008 0.002 0.003 0.005 0.060

0.208 0.194 0.152 0.231 0.476

0.019 0.012 0.009 0.020 0.032


q FEBS 2001

Fig. 5. Effect of chemical crosslinking on SDS/PAGE analysis of 125I-labeled rabbit anti-(a-l-fucosidase) Ig binding to cell surface PM. Cells (107 mL21) were incubated with 125I-labeled rabbit anti-(a-l-fucosidase) Ig (400 mg´mL21) at 4 8C for 1 h and after extensive washing, they were subjected to mock crosslinking with NaCl/Pi (lanes marked with ±) or chemically crosslinked with BS3 (1 mm; lanes marked with 1) for 2 h at 4 8C. Cell lysates void of nucleus were analyzed on 10% gels by SDS/PAGE under reducing conditions. As controls (D), 125I-labeled rabbit anti-(al-fucosidase) Ig (2 mg) was incubated with human placenta a-l-fucosidase (0.2 mg) in the absence (±) or presence (1) of BS3 under the same conditions we can see above, before SDS/PAGE under reducing conditions. Molecular masses in A, B and C denote the bands observed in cell lysates but not in Ig controls before or after binding and crosslinking. In C, the same gel as in B but exposed longer to X-ray film (10 d) shows the same bands observed previously in the other cell lysates (Daudi, Jurkat and MG-63 cells) with BS3 in a 2-day exposition. A and D are from one experiment, and B and C from another. Jurkat and HT-29 cells were used in both giving the same results, as well as controls. HT-29 cells were used only with BS3 because we thought at that moment that they were a negative control.

DISCUSSION

Fig. 6. a-l-Fucosidase hydropathy profiles from the tmpred and toppred 2 predictions.

The acid hydrolase a-l-fucosidase is usually found as a soluble component of lysosomes and antiserum. However, a-l-fucosidase was also found as a membrane fractionassociated in brain [14] and colon adenocarcinoma HT-29 cells (including plasma membrane fraction, manuscript in preparation), and as plasma membrane-associated in rat spermatozoa [18,30]. In this paper we demonstrate the existence of a cell surface protein which crossreacts with anti-(a-l-fucosidase) Ig. By the different approaches used, all the cells tested (hematopoietic, epithelial, mesenchimal) were found to express this cell surface protein with a-lfucosidase activity. It has been described that most, if not all, mammalian a-l-fucosidase share some sequence identity and that they have epitopes in common [1]. The data from animal cell lines suggest that this result can be probably extended at least to mammals. Immunoprecipitation and Western blot data from enriched plasma membrane solubilized with Triton X-100 suggest that in many cases, if not all, the crossreactive protein is an integral membrane protein. The data shown in Western blots are from 25 mg of membrane protein per lane of SDS/PAGE. The fact that some bands are not observed when 6 mg of protein per channel was analyzed, may indicate the low quantity of membrane a-l-fucosidase in many cell lines, which was also supported by the immunofluorescence profiles of these cell lines. However, the exceptions of the K562 and in particular MG63 cell lines, and the presence of a-l-fucosidase activity in PM, poorly but specifically immunoprecipitable, should be kept in mind as a possible presence of PM-associated and/or nonsolubilized a-l-fucosidase (in Triton). In this sense, it has been very recently described that a-l-fucosidase (as

q FEBS 2001

well as other glycohydrolases) is anchored/associated to human erythrocyte membranes by all these different patterns [31]. In the extensive characterization performed for the polyclonal anti-(a-l-fucosidase) Ig, the main recognition of the 51-kDa form (with respect to the 47-kDa form) from purified human placenta a-l-fucosidase of a commercial source as well as one 49-kDa band from HT-29 cell line extracts (although in certain circumstances of HT-29 differentiation the minor band has been observed; data not shown) point out to a limited performance of this Ig in Western blotting as usually two isoforms of about 50 and 55 kDa are equally recognized in human tissues. These are the main two sialylated molecular forms of a-l-fucosidase which are maintained after deglycosylation [1]. The fact that the bands observed in our work maintain this pattern as well as the specific precipitation of fucosidase activity from solubilized plasma membrane by antiserum support the identity of cell surface a-l-fucosidase. A number of intriguing questions remain to be answered. What is the source of membrane a-l-fucosidase? Is the enzyme different from soluble a-l-fucosidase? If membrane a-lfucosidase was originated by a deficient cleavage of the signal peptide as it was shown for other transmembrane peptides which are also found as soluble forms, e.g. CD26 [32], bands of at least 60 kDa (as was described for the precursor) [1] should be observed. However, the bands observed in this work (43±49 kDa) have a lower Mr. They may correspond to deglycosylated soluble a-l-fucosidase, both bands corresponding to normal allelic variation or, alternatively, to other modifications (not the glycosylation pattern) because of the presence in the a-l-fucosidase amino-acid sequence of a second transmembrane helix different to the putative signal peptide. This region might permit a membrane insertion of the mature form and the extracellular presence of the active site or, alternatively, a second signal peptide processing. Interestingly, it has been reported that the loss of carbohydrate led to decrease the activity at acidic pH values and a shift to a more neutral pH optimum [1], which is supposed to need a nonlysosomic enzyme. It should be pointed out that a-l-fucosidase molecules of lower Mr (45±47 kDa) and neutral pH activity were sometimes identified [1,5,26,30] including human placenta and rat epididymis. Also, if both transmembrane regions were present in a-l-fucosidase, the probabilities of the two models of insertion (with the catalytic center inside or outside) are similar. The cells with a high a-l-fucosidase intensity and percentage were fresh monocytes and PMN (neutrophils), or the plasmatic NC-37 cell line, i.e. with an intense secretory pathway (may also be the case for Caco-2). Moreover, although in a nonrepetitive form and not in all cells, PFA, which is a fixative molecule, induced the expression of surface a-l-fucosidase by an unknown process. We speculate that a-l-fucosidase in an early stage (absence of glycosylation), previous to the sorting to lysosomes or secretory vesicles, can be inserted into the internal membrane. a-l-Fucosidase will reach the plasma membrane and cell surface through this presence in the vesicular network, PFA inducing membrane fusion of vesicles and endosomes near the plasma membrane. Alternatively, PFA could induce an inside±outside redistribution of the a-l-fucosidase [33]. In this sense it has been


documented that a serine phosphorylation mechanism is involved in intracellular retention of a-l-fucosidase in lymphoid cells lacking the mannose 6-phosphate that targets newly made hydrolases to lysosomes [34]. Another lysosomal enzyme, cathepsin D, shows a very complex intracellular trafficking, including a MPR±independent association to the membrane, and in fact it can be found at the periphery and processing some secreted proteins [35,36]. A different regulation for this enzyme was described for HT-29 and Caco-2 colorectal cells [35]. HT-29 are about 95% undifferentiated whereas Caco-2 cells differentiate spontaneously after confluence. It is very probable that a similar mechanism can explain the differences in PM a-l-fucosidase we found between both cell lines. These cases are not an exception. In particular circumstances, membrane lysosomal proteins such as lgp120 and LEP100 have been described to appear on the outer plasma membrane, and which is most important, the lysosomal enzyme acid phosphatase is synthesized as a transmembrane precursor that is transported to lysosomes via the cell surface in BHK cells [37,38]. Whatever the source and the origin of membrane a-lfucosidase, our finding must be emphasized in the context of the well known occurrence of glycosydases mainly in lysosomes, but also in other subcellular fractions [39]. For integral membrane proteins, but not for serum glycoproteins, a new mechanism termed `reprocessing' or intramolecular heterogeneous turnover has been described [40,41]. In this mechanism a rapid turnover of N-glycans (including l-fucose) followed by a reglycosylation and reinsertion of these proteins into the plasma membrane has been shown [40,42]. The presence of specific glycosidases is necessary for this mechanism. There is an open question left, in which subcellular compartment, endosomes, transGolgi network or plasma membrane, the reprocessing occurs. The endosomal pathway has been suggested for the glycoprotein CD26 in BHK cells [42]. Alternatively, extracellular a-l-fucosidase has been shown to hydrolyze l-fucose from glycosydic linkages occurring in PM glycoproteins [16]. From the whole data, we hypothesize a sorting of a-l-fucosidase into a membrane compartment in order to work in `reprocessing', being the biological significance of reprocessing related to the regulation of cell surface carbohydrate information-bearing function in cell±cell recognition and adhesion [43,44].

ACKNOWLEDGEMENTS We thank Dr Federico Mallo (Departamento de FisioloxõÂa da Universidade de Vigo) for technical help and the Centro de TransfusioÂn de Galicia for providing the blood samples. This work was supported by grants from Xunta de Galicia (XUGA 30110B97 and XUGA20007B96 and from Universidade de Vigo. A. Merino was supported by a predoctoral fellowship from the Ministerio de EducacioÂn y Cultura, Spain.

REFERENCES 1. Alhadeff, J.A. (1998) Structure and function of mammalian a-lfucosidases. Trends Comp. Biochem. Phys. 4, 105±118. 2. DiMatteo, G., Orfeo, M.A. & Romeo, G. (1976) Human a-fucosidase. Purification and properties. Biochim. Biophys. Acta 429, 529±537.

3330 O. J. Cordero et al. (Eur. J. Biochem. 268) 3. Turner, B.M. (1979) Purification and characterization of a-lfucosidase from human placenta. pH-dependent changes in molecular size. Biochim. Biophys. Acta 578, 325±336. 4. Kretz, K.A., Cripe, D., Carson, G.S., Fukushima, H. & O'Brien, J.S. (1992) Structure and sequence of the human a-l-fucosidase gene and pseudogene. Genomics 12, 276±280. 5. Seo, H.C., Willems, P.J., Kretz, K.A., Martin, B.M. & O'Brien, J.S. (1993) Fucosidosis: four new mutations and a new polymorphism. Hum. Mol. Genet. 2, 423±430. 6. Takeshita, H., Yasuda, T., Nadano, D., Iida, R., Nakanaga, M., Tenjo, E., Sawazaki, K. & Kishi, K. (1994) Genetically polymorphic a-l-fucosidase (FUCA1) isozymes detected in blood plasma. Hum. Genet. 94, 224±230. 7. Willems, P.J., Gatti, R., Darby, J.K., Romeo, G., Durand, P., Dumon, J.E. & O'Brien, J.S. (1991) Fucosidosis revisited: a review of 77 patients. Am. J. Med. Genet. 38, 111±131. 8. Barker, C., Dell, A., Rogers, M., Alhadeff, J.A. & Winchester, B. (1988) Canine a-l-fucosidase in relation to the enzymic defect and storage products in canine fucosidosis. Biochem. J. 254, 861±868. 9. Marquardt, T., Brune, T., Luhn, K., Zimmer, K.P., Korner, C., Fabritz, L., van der Werft, N., Vormoor, J., Freeze, H.H., Louwen, F., Biermann, B., Harms, E., von Figura, K., Vestweber, D. & Koch, H.G. (1999) Leukocyte adhesion deficiency II syndrome, a generalized defect in fucose metabolism. J. Pediatr. 134, 681±688. 10. Scanlin, T.F., Matacic, S.S., Pace, M., Santer, U.V. & Glick, M.C. (1977) Abnormal distribution of a-Lfucosidase in cystic fibrosis: increased activity in skin fibroblasts. Biochem. Bioph. Res. Co. 79, 869±875. 11. Scanlin, T.F., Smit-Matacic, S. & Glick, M.C. (1979) Abnormal distribution of a-l-fucosidase in cystic fibrosis: decreased activity in serum. Clin. Chim. Acta 91, 197±202. 12. Ayude, D., Fernandez-Rodriguez, J., Rodriguez-Berrocal, F.J., Martinez-Zorzano, V.S., de Carlos, A., Gil, E. & Paez de La Cadena, M. (2000) Value of the serum a-l-fucosidase activity in the diagnosis of colorectal cancer. Oncology 59, 310±316. 13. Paez de la Cadena, M., Fernandez, J., de Carlos, A., MartinezZorzano, V., Gil-Martin, E. & Rodriguez-Berrocal, F.J. (1996) Low levels of a-l-fucosidase activity in colorectal cancer are due to decreased amounts of the enzymatic protein and are related with Dukes' stage. Int. J. Oncol. 9, 747±754. 14. Hopfer, R.L., Johnson, S.W., Masserini, M., Giuliani, A. & Alhadeff, J.A. (1990) Hydrolysis of fucosyl-GM1 ganglioside by purified pellet-associated human brain and human liver a-lfucosidases without activator proteins or detergents. Biochem. J. 266, 491±496. 15. Giardina, M.G., Matarazzo, M., Morante, R., Lucariello, A., Varriale, A., Guardasole, V. & De Marco, G. (1998) Serum a-lfucosidase activity and early detection of hepatocellular carcinoma: a prospective study of patients with cirrhosis. Cancer 83, 2468±2474. 16. Alhadeff, J.A. (1978) Lysosomal hydrolases in cystic fibrosis livers. Clin. Chim. Acta 89, 469±473. 17. DiCioccio, R.A., Barlow, J.J. & Matta, K.L. (1982) Substrate specificity and other properties of a-l-fucosidase from human serum. J. Biol. Chem. 257, 714±718. 18. Aviles, M., Martinez-Menarguez, J.A., Castells, M.T., Skalaban, S.R., Ballesta, J. & Alhadeff, J.A. (1996) Immunocytochemical localization and biochemical characterization of a novel plasma membrane-associated, neutral pH optimun a-l-fucosidase from rat testis and epididymal spermatozoa. Biochem. J. 318, 821±831. 19. Abascal, I., Skalaban, S.R., Grimm, K.M., Aviles, M., MartinezMenarguez, Castells, M.T., Ballesta, J. & Alhadeff, J.A. (1998) Alteration of isoform composition of plasma-membraneassociated rat sperm a-l-fucosidase during late epididymal maturation: comparative characterization of the acidic and neutral isoforms. Biochem. J. 333, 201±207.

q FEBS 2001 20. Cordero, O.J., Salgado, F.J., VinÄuela, J.E. & Nogueira, M. (1998) Interleukin 12 -dependent activation of human lymphocyte subsets. Immunol. Lett. 61, 7±13. 21. Salgado, F.J., Vela, E., MartõÂn, M., Franco, R., Nogueira, M. & Cordero, O.J. (2000) Mechanisms of CD26/dipeptidyl peptidase IV cytokine-dependent regulation on human activated lymphocytes. Cytokine 12, 1136±1141. 22. Maeda, T., Balakrishnan, K. & Mehdi, S.Q. (1983) A simple and rapid method for the reparation of plasma membranes. Biochim. Biophys. Acta 731, 115±120. 23. Fernandez, J., Rodriguez-Berrocal, F.J., de Carlos, A., de Castro, G. & Paez de la Cadena, M. (1996) Nonradiactive immunoquantification of a-l-fucosidase protein in human colon tissues. J. Biochem. Biophys. Methods 31, 39±47. 24. Cordero, O.J., Sarandeses, C. & Nogueira, M. (1994) Prothymosin a receptors on peripheral blood mononuclear cells. FEBS Lett. 341, 23±27. 25. DiCioccio, R.A. & Miller, A.L. (1991) Biosynthesis, processing and secretion of a-l-fucosidase in lymphoid cells from patients with I-cell disease and pseudo-Hurler polydystrophy. Glycobiology 1, 595±604. 26. Chuang, N.N., Yeh, C.C. & Lin, K.S. (1991) The basic isoelectric form of a-l-fucosidase from the hepatopancreas of the shrimp Penaeus monodon (Crustacea: Decapoda). Comp. Biochem. Phys. 99B, 395±398. 27. Tusnady, G.E. & Simon, I. (1998) Principles governing amino acid composition of integral membrane proteins: application to topology prediction. J. Mol. Biol. 283, 489±506. 28. Hofmann, K. & Stoffel, W. (1993) TMbase ± a database of membrane spanning proteins segments. Biol. Chem. Hoppe-Seyler 347, 166. 29. Von Heijne, G. (1992) Membrane protein structure prediction. Hydrophobicity analysis and the positive-inside rule. J. Mol. Biol. 225, 487±494. 30. Carlsen, R.B. & Pierce, J.G. (1972) Purification and properties of an a-l-fucosidase from rat epididymis. J. Biol. Chem. 247, 23±32. 31. Goi, G., Bairati, C., Massaccesi, L., Lovagnini, A., Lombardo, A. & Tettamanti, G. (2000) Membrane anchoring and surface distribution of glycohydrolases of human erythrocyte membranes. FEBS Lett. 473, 89±94. 32. Fleischer, B. (1994) CD26: a surface protease involved in T-cell activation. Immunol. Today 15, 180±183. 33. Martin, S.J., Reutelingsperger, C.P., McGahon, A.J., Rader, J.A., van Schie, R.C., LaFace, D.M. & Green, D.R. (1995) Early redistribution of plasma membrane phosphatidylserine is a general feature of apoptosis regardless of the initiating stimulus: inhibition by overexpression of Bcl-2 and Abl. J. Exp. Med. 182, 1545±1556. 34. DiCioccio, R.A. & Miller, A.L. (1993) Phosphorylation and subcellular location of alpha-l-fucosidase in lymphoid cells from patients with I-cell disease and pseudo-Hurler polydystrophy. Glycobiology 3, 489±495. 35. Hennebicq Reig, S., Kim, I., Janin, A., Grard, G., Hemon, B., Moreau, O., Porchet, N., Aubert, J.P., Degand, P. & Huet, G. (1996) Regulation of cathepsin D dependent on the phenotype of colon carcinoma cells. Int. J. Cancer 68, 479±484. 36. Helseth, D.L. Jr & Veis, A. (1984) Cathepsin d-mediated processing of procollagen: lysosomal enzyme involvement in secretory processing of procollagen. Proc. Natl Acad. Sci. USA 81, 3302±3306. 37. Wilson, J.M., Whitney, J.A. & Neutra, M.R. (1991) Biogenesis of the apical endosome±lysosome complex during differentiation of absorptive epithelial cells in rat ileum. J. Cell. Sci. 100, 133±143. 38. Braun, M., Waheed, A. & von Figura, K. (1989) Lysosomal acid phosphatase is transported to lysosomes via the cell surface. EMBO J. 8, 3633±3640. 39. Cabezas, J.A., Reglero, A. & Calvo, P. (1983) Glycosidases. Ffucosidases, galactosidases, glucosidases, hexosaminidases and

q FEBS 2001 glucuronidase from some mollusc vertebrates, and neuraminidase from virus. Int. J. Biochem. 15, 243±259. 40. Kreisel, W., Hanski, C., Tran-Thi, T.A., Katz, N., Decker, K., Reutter, W. & Gerok, W. (1988) Remodeling of a rat hepatocyte plasma membrane glycoprotein. De- and reglycosylation of dipeptidyl peptidase IV. J. Biol. Chem. 263, 11736±11742. 41. Tauber, R., Park, C.S. & Reutter, W. (1983) Intramolecular heterogeneity of degradation in plasma membrane glycoproteins: evidence for a general characteristic. Proc. Natl Acad. Sci. USA 80, 4026±4029.

Cell surface human a-l-fucosidase (Eur. J. Biochem. 268) 3331 42. Horstkorte, R., Fan, H. & Reutter, W. (1996) Rapid isolation of endosomes from BHK cells: Identification of DPP IV (CD26) in endosomes. Exp. Cell. Res. 226, 398±401. 43. Drickamer, K. & Taylor, M.E. (1998) Evolving views of protein glycosylation. Trends. Biochem. Sci. 23, 321±324. 44. Smith, R.E., Talhouk, J.W., Brown, E.E. & Edgar, S.E. (1998) The significance of hypersialylation of dipeptidyl peptidase IV (CD26) in the inhibition of its activity by Tat and other cationic peptides. CD26: a subverted adhesion molecule for HIV peptide binding. AIDS Res. Hum. Retroviruses 14, 851±868.