Molecular cloning of endo-beta -galactosidase C and its application in removing alpha -galactosyl xenoantigen from blood vessels in the pig kidney.

Galalpha1-3Gal is the major xenoantigenic epitope responsible for hyperacute rejection upon pig to human xenotransplantation. Endo-beta-galactosidase C from Clostridium perfringens destroys the antigenic epitope by cleaving the beta-galactosidic linkage in the Galalpha1-3Galbeta1-4GlcNAc structure. Based on partial peptide sequences of the enzyme, we molecularly cloned the enzyme gene, which encodes a protein with a predicted molecular mass of about 93 kDa. The deduced protein sequence of the enzyme has limited homology in the C-terminal half with endo-beta-galactosidase from Flavobacterium keratolyticus and beta-1,3-glucanases. The enzyme expressed in Escherichia coli removed the alpha-galactosyl epitope nearly completely from pig erythrocytes and from pig aortic endothelial cells. The enzyme-treated endothelial cells in culture were greatly reduced in cell surface antigens, which were recognized by IgM, IgG, or IgA in human sera, and became much less susceptible to complement-mediated cytotoxicity caused by human sera. When the pig kidney was perfused with the enzyme, the vascular endothelial cells became virtually devoid of the alpha-galactosyl epitope, with concomitant decrease in binding to IgM in human plasma. These results demonstrated that the recombinant endo-beta-galactosidase C is a valuable aid in xenotransplantation.

Xenotransplantation, especially from pig to human, has been considered as means to overcome the limitations in the number of donor organs available for transplantation surgery (1). However, the occurrence of hyperacute rejection whereby vascular endothelial cells are destroyed by an antibody-and complementdependent mechanism, remains an obstacle to xenotransplantation (1,2). Gal␣1-3Gal antigen has been identified as the major xenoantigen responsible for hyperacute rejection (2)(3)(4). Non-primate mammals express Gal␣1-3Gal, while old world monkeys and humans do not, and their sera have antibodies against this carbohydrate antigen (5,6).
Several approaches have been proposed to eliminate the Gal␣1-3Gal antigenic epitope from pigs or mice (2). Removal of the ␣-1,3-galactosyltransferase gene by gene targeting is a straightforward method but has so far been successful only in mice (7,8). Introduction and overexpression of genes for glycosyltransferases (i.e. ␣-1,2-fucosyltransferase (9 -12), N-acetylglucosaminyltransferase III (13), and ␣-2,3-sialyltransferase (14)) results in decreased expression of Gal␣1-3Gal antigen by competition over the acceptor structure and other mechanisms. Furthermore, transgenic pigs with ␣-1,2-fucosyltransferase have been produced and have been reported to be reduced in the ␣-galactosyl antigen expression (9,10,12). Expression of the dominant negative form of ␣-1,3-galactosyltransferase has also been tried (2). However, by none of the methods has production of pig organs virtually devoid of the ␣-galactosyl antigen been achieved so far (2). As an alternative approach, ␣-galactosidase from the coffee bean was used to remove the ␣-galactosyl antigen from cultured pig endothelial cells. The enzyme-treated cells became resistant to complement-dependent lysis by human sera (15,16). Although very recently the ␣-galactosidase has been applied to remove the ␣-galactosyl antigen from isolated femoral veins from the pig (17), this enzyme has not been applied to treat pig organs, probably because the acidic optimum pH of the enzyme makes it necessary to use enormous amounts of the enzyme.
Endo-␤-galactosidase C, an endo-␤-galactosidase found in the culture fluid of Clostridium perfringens, cleaves the Gal␤1-4GlcNAc linkage in the Gal␣1-3Gal␤1-4GlcNAc structure, releasing Gal␣1-3Gal disaccharide (18). Due to its neutral optimum pH (18), the enzyme is expected to be valuable as an aid for xenotransplantation. As reported here, we cloned the gene for this enzyme, produced the recombinant enzyme, and applied it to digest out the ␣-galactosyl antigen from pig blood vessels in the kidney.

EXPERIMENTAL PROCEDURES
Partial Peptide Sequence of Endo-␤-galactosidase C-Endo-␤-galactosidase C was purified from culture fluid of C. perfringens as described previously (18). The enzyme preparation obtained from 18 liters of culture medium was subjected to SDS-polyacrylamide gel electrophoresis using an 8% gel according to Laemmli (19). After light staining with Coomassie Brilliant Blue, bands were excised, and proteins in the gel were digested with trypsin as described by Rosenfeld et al. (20). The resulting peptides were separated on a reverse phase column (Monitor C 18, 1 ϫ 150 mm) with a linear gradient of acetonitrile and 0.1% trifluoroacetic acid at a flow rate of 50 l/min using a MAGIC 2002 (Michrom Bio Resources, Inc.). Amino acid sequences of the resulting peptides were determined with a PE Applied Biosystems protein sequencer model 494A.
Cloning of the Endo-␤-galactosidase C Gene-PCR 1 was performed using the sense primer 5Ј-GAYGARAAYGARTAYGT-3Ј (based on the amino acid sequence of peptide I) and antisense primer 5Ј-TCYTCRTT-NARRTCNCCRTC-3Ј (based on the amino acid sequence of peptide IV) and genomic DNA of the bacteria as the template, after denaturation at 94°C for 3 min, with 35 cycles of 94°C for 1 min, 45°C for 1 min, and 72°C for 2 min. The PCR product, which corresponded to a partial genomic sequence of the enzyme gene, was subcloned into the pGEM-T Easy Vector (Promega). The 5Ј-and 3Ј-ends of the DNA were extended by HindIII cassette (Takara), using 5Ј-TGATACTTGTGGTACAGTA-GAATGCCCACT-3Ј, 5Ј-ATGTTTACTATTAGTAGTAGGTATTAAGCC-3Ј, 5Ј-AGATGGTATGCATACTAGACCTATGTTCCC-3Ј, and 5Ј-AAGT-TGGAGATGGTTGGGTTGGTGATGTTG-3Ј as primers. Cassette PCR was performed with 30 cycles of 94°C for 30 s, 55°C for 2 min, and 72°C for 1 min. Full-length DNA was obtained by PCR using two primers, 5Ј-GCAGCCGAATTCATGAAATTTTTCTCGAAATTCAAAAA-GGTA-3Ј and 5Ј-GCGTCGCTCGAGTTATTTATTTTGTAAAAAAGTTA-CTTTAGC-3Ј, which were designed according to the information of the 5Ј-and 3Ј-DNA sequences obtained by cassette PCR with genomic DNA as a template. PCR was performed after denaturation at 94°C for 1 min and 30 cycles of 94°C for 30 s, 55°C for 2 min, and 72°C for 1 min.
Assay of Endo-␤-galactosidase C-The enzymatic reaction was performed in 10 l of 50 mM phosphate buffer, pH 7.2, containing 20 g of Gal␣1-3Gal␤1-4GlcNAc␤1-3Gal␤1-4Glc (Calbiochem). After 30 min at 37°C, the reaction was stopped by adding 20 l of ethanol, and after desalting with Dowex 50W ϫ 2 H ϩ form and 1 ϫ 2 acetate form, one-tenth of the hydrolysate was spotted to a Silica Gel 60 TLC plate (Merck). The plate was developed in n-propyl alcohol/ethyl acetate/H 2 O (7:2:1), and sugar spots were revealed by spraying with orcinol-H 2 SO 4 reagent and heating (22). The amounts of released products were determined using a densitometer, Gel Doc 1000 (Bio-Rad). Under these assay conditions, enzymatic activity was proportional to the amount of the enzyme and reaction time until 40% of the substrate was hydrolyzed. One unit of the enzymatic activity was defined as the amount of the enzyme required to hydrolyze 1 mol of the substrate per min.
Mass Spectrometry-Mass spectrometry was performed using a triple quadrupole mass spectrometer, API 300 (Perkin-Elmer) equipped with an electrospray ion source. The sample was directly infused into the electrospray ion source at a flow rate of 0.3 ml/h. The spray was obtained by a potential difference of 4.8 kV.
Other Biochemical Analyses-The amount of protein in the purified enzyme was determined using a Micro BCA protein assay reagent kit (Pierce), with bovine serum albumin as a standard. Possible contamination of other glycosidases was determined by incubating respective p-nitrophenyl glycosides in 0.2 ml of reaction mixture with 29.4 g of the purified enzyme at 37°C for 1 h at pH 7.0 (50 mM phosphate buffer) and also at pH 6.0 (50 mM citrate phosphate buffer). Possible contamination of proteases was determined as described previously (23) also at pH 7.0 and 6.0 under the assay conditions described above.
Production of Recombinant Endo-␤-galactosidase C-The coding sequence of the enzyme was ligated into the expression vector pFLAG-Shift 12 (Sigma), which produces a gene product that can be secreted into the periplasm. The recombinant gene was transformed into Escherichia coli BL-21 cells, which were cultured at 37°C in 2 liters of LB medium containing 50 g/ml of ampicillin and 0.4% of glucose and were induced to express the enzyme by adding 0.1 mM isopropyl ␤-D-thiogalactopyranoside to the culture medium. After 30 min, the bacterial cells were collected by centrifugation at 5000 ϫ g for 10 min and washed with 300 ml of Tris-HCl, pH 8.0, twice and then 300 ml of 0.5 M sucrose, 30 mM Tris-HCl, pH 8.0, and 1 mM EDTA at room temperature. The cells collected by centrifugation were immediately suspended in 250 ml of ice-cold distilled water to release the enzyme from the periplasm. The supernatant containing the released enzyme was collected by centrifu-gation at 3500 ϫ g for 10 min at 4°C and purified by chromatography on columns of Sephadex G-200 (2.7 ϫ 100 cm) and DEAE-Sephadex A-25 (1.0 ϫ 10 cm) as described previously (18). The purified fraction contained about 700 g of protein. When large amounts of the enzyme were required, this purification procedure was repeated.
Removal of ␣-Galactosyl Antigen from Living Cells-Pig erythrocytes were collected by centrifugation (1000 ϫ g) from citric acid-treated blood and washed twice with PBS(Ϫ) containing 0.2% bovine serum albumin (PBS/BSA). Then they were incubated with 3 mg/ml dimethyl suberimidate dihydrochloride in 0.1 M Na 2 CO 3 containing 0.15 M NaCl and 0.1 mM EDTA at 37°C for 20 min to prevent agglutination. Erythrocytes were washed with PBS/BSA twice and suspended in PBS/BSA to make a 1% suspension. Aliquots of 50 l of the erythrocyte suspension thus prepared were centrifuged at 1000 ϫ g, and the precipitated erythrocytes were suspended in 50 l of PBS/BSA to which 20 milliunits of recombinant endo-␤-galactosidase C in 4 l of PBS was added, and the reaction was continued at 37°C for 30 min.
Pig aortic endothelial cells were isolated from a pig as described previously (16). The cells were harvested with trypsin-EDTA (Life Technologies, Inc.) and washed twice with PBS/BSA and resuspended in PBS/BSA. Recombinant endo-␤-galactosidase C (70 milliunits in 7 l of PBS) was added to the aortic endothelial cells (2ϫ10 5 cells/50 l), and the reaction was continued at 37°C for 1 h.
After the enzymatic reaction, erythrocytes or endothelial cells were washed twice with PBS/BSA and reacted with either FITC-labeled Griffonia simplicifolia isolectin-IB 4 (GS-IB 4 ) or human sera. In the former case, cells were suspended in 50 l of PBS/BSA containing 1 g of FITC-GS-IB 4 (Sigma) and kept on ice for 30 min. In the latter case, the washed cells were incubated with 100 l of normal human sera (1:1 diluted by PBS/BSA) pooled from 50 healthy volunteers. After a 30-min incubation on ice, the cells were washed twice with PBS/BSA and then incubated with 100 l of FITC-conjugated rabbit anti-human IgG, IgM, or IgA (Dako), each diluted 1:20 with PBS/BSA, on ice for 30 min. FITC-conjugated mouse monoclonal anti-human IgG 1 (Zymed Laboratories Inc.), IgG 2 , IgG 3 , or IgG 4 (Sigma) was also used. Then the stained erythrocytes or endothelial cells were washed twice with PBS/BSA, suspended in 0.5 ml of PBS/BSA, and analyzed using a FACS (FACS Calibur, Becton Dickinson).
Assay of Complement-mediated Cytotoxicity-Pig aortic endothelial cells were plated at 1 ϫ 10 4 cells/well in a 96-well culture plate 1 day prior to the assay. After washing with PBS(Ϫ), the cells were incubated with 50 l of Dulbecco's modified Eagle medium containing 70 milliunits of recombinant endo-␤-galactosidase C at 37°C for 2 h. The control cells were incubated without the enzyme. The cells were washed twice and incubated with 50 l of 25 or 50% normal human sera diluted with PBS(Ϫ) at 37°C for 1 h. The human sera were pooled from 10 healthy volunteers. After incubation, the viability of the cells was determined by an assay using 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (Sigma) as described by Nagasaka et al. (24). The absorbance of each well was measured by an automatic plate reader (EAR340; SLT-Labinstruments, Austria) at 540 nm, and the values were referred to as "cell viability in the presence of complement." In each assay, "cell viability in the absence of complement" was determined using the sera treated at 56°C for 30 min to inactivate complement. Complement-mediated cytotoxicity was calculated from the following formula: percentage cytotoxicity ϭ ((cell viability in the absence of complement Ϫ cell viability in the presence of complement)/cell viability in the absence of complement) ϫ 100. Experiments were performed in triplicates.
Endo fusion with or without endo-␤-galactosidase C. Then the kidneys were perfused twice with 100 ml of freshly frozen human plasma obtained from three healthy volunteers. For histological studies, biopsied samples were fixed with formalin and stained with hematoxylin-eosin and periodic acid-Schiff. The samples were also immediately frozen in liquid nitrogen and stained with FITC-GS-IB 4 . The perfused human plasma was collected, and remaining immunoglobulin levels were analyzed by FACS using pig erythrocytes as described above.

Cloning and Structural Analysis of Endo-␤-galactosidase
C-SDS-polyacrylamide gel electrophoresis of an endo-␤-galactosidase C preparation revealed several protein bands, among which a 50-kDa major band and a 95-kDa band (Fig. 1A) co-migrated with the enzyme activity on DEAE-Sephadex A-25 column chromatography. Sequence analysis of tryptic peptides indicated that the 50-kDa band was C. perfringolysin O protein (26), while the 95-kDa band was a novel protein.
Based on the sequences of four tryptic peptides derived from the 95-kDa band (Fig. 2), PCR primers were designed, and PCR using genomic DNA of C. perfringens as a template yielded a DNA of 1946 base pairs. Cassette PCR was performed to obtain the 5Ј and 3Ј sequences. The composite DNA encoded an open reading frame of 2535 bp corresponding to a protein of 845 amino acids (Fig. 2). Upstream from the putative translational initiation site, there was a typical Shine-Dalgarno sequence, AGGAGG (nucleotides 231-236) (27). The deduced protein sequence contained all of the peptide sequence identified in the 95-kDa protein (Fig. 2). In all but one case, amino acids next to the N terminus of the identified peptides were either arginine or lysine, which was consistent with the fact that the peptides were derived by trypsin digestion. In one case (peptide I, Fig.  2), the proximal amino acid was alanine. We confirmed the correctness of the sequence by 5Ј-rapid amplification of cDNA ends. Therefore, we inferred that peptide I of the 95-kDa protein was derived from the N-terminal portion of the 95-kDa protein, which was expected to be generated by proteolytic processing of the nascent protein. This was confirmed by Nterminal sequence analysis; the purified recombinant enzyme described in the following section had the N-terminal sequence DENEYVL, which is the N-terminal sequence of peptide I. Furthermore, the upstream peptide (amino acids 1-34, Fig. 2) had a cluster of hydrophobic amino acids typically found in a signal sequence (28). This putative signal sequence might be involved in secretion of the enzyme. Endo-␤-galactosidase from Flavobacterium keratolyticus is known to have a signal sequence (29). The molecular mass of the protein devoid of the putative signal sequence was 93 kDa and was in good agreement with the observed molecular mass of the protein, 95 kDa.
Expression of Recombinant Endo-␤-galactosidase C-The enzyme was expressed in E. coli as a periplasmic enzyme under the control of the ompA gene. The recombinant enzyme was purified by Sephadex G-200 and DEAE-Sephadex A-25 column chromatography. A major band with apparent molecular mass of 95 kDa and a closely related minor band of 97 kDa were found in the purified preparation (Fig. 1B). As mentioned above, the major band corresponded to the protein devoid of the putative signal sequence. From the molecular mass difference, the minor band appeared to correspond to a nascent protein with a signal sequence.
The recombinant enzyme hydrolyzed a pentasaccharide, Gal␣1-3Gal␤1-4GlcNAc␤1-3Gal␤1-4Glc, releasing two oligosaccharides, which could be separated by TLC (Fig. 4). The nature of the oligosaccharides was identified by electrospray mass spectrometry. The pentasaccharide before digestion gave m/z 892.2, which corresponded to the calculated mass of the oligosaccharide plus 23, presumably due to the addition of a Na ϩ ion (Fig. 5A). After enzymatic digestion, two prominent peaks appeared at m/z 568.1 and m/z 365.0 (Fig. 5B). The former corresponded to the mass of GlcNAc␤1-3Gal␤1-4Glc plus Na ϩ , and the latter to that of Gal␣1-3Gal plus Na ϩ . These observations established that the cloned enzyme was endo-␤-galactosidase C, which hydrolyzes Gal␤1-4GlcNAc linkage and releases Gal␣1-3Gal disaccharide. The specific activity of the enzyme was 10.5 units/mg protein, indicating the potent activity of the enzyme. The recombinant enzyme was free from proteases, ␣-galactosidase, ␤-galactosidase, and ␤-N-acetylglucosaminidase. The recombinant enzyme had a pH optimum of 7.0 -8.0 as reported for the enzyme from culture fluid of C.  7. Effects of endo-␤-galactosidase C treatment on susceptibility to complement-mediated cytotoxicity by human sera. Pig aortic endothelial cells with or without endo-␤-galactosidase C treatment were incubated with 25 or 50% human sera in the presence or absence of complement. Cell viability was determined by the 3-(4,5dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide assay, and complement-mediated cytotoxicity was calculated as described under "Experimental Procedures." Gray bars, without enzymatic treatment; black bars, with enzymatic treatment. Average values from triplicate assays are shown with S.E. values. perfringens (18). Furthermore, at 4°C the enzyme exhibited 40% of the activity observed at 37°C.
Action of Recombinant Endo-␤-galactosidase C on Pig Erythrocytes and Pig Endothelial Cells-After treatment with recombinant endo-␤-galactosidase C, more than 99% of the ␣-galactosyl epitope, which was recognized by FITC-GS-IB 4 , was removed from pig erythrocytes (Fig. 6A). Similarly, 98% of the ␣-galactosyl epitope could be enzymatically removed from pig aortic endothelial cells (Fig. 6B). Binding of IgM to the treated erythrocytes was reduced to 25% of that in untreated cells (Fig. 6A); in endothelial cells, the reduction was to less than 23% (Fig. 6B). Binding of IgA and IgG in human sera was also reduced to 21 and 33%, respectively, in the case of enzymatically treated endothelial cells (Fig. 6B), while less reduction was observed in treated erythrocytes for IgG (Fig. 6A). Furthermore, using antibodies against human IgG subclasses, we found that binding to IgG 1 , IgG 2 , or IgG 3 was markedly reduced in the treated cells (Fig. 6, C and D), while no changes in binding to IgG 4 were detected. The cells treated with endo-␤-galactosidase C became much less susceptible to complement-mediated cytotoxicity caused by human sera (Fig. 7).
Enzymatic Removal of the Xenoantigen from the Luminal Surface of Vascular Endothelial Cells in the Pig Kidney-We then examined the effects of the enzymatic digestion of the ␣-galactosyl epitope on intact blood vessels of the pig kidney. The pig kidney was perfused with the enzyme solution, and the reaction was performed at 4°C, since lower temperature is preferable for preservation of the tissue, and endo-␤-galactosidase C effectively worked even at low temperature. Staining with FITC-GS-IB 4 verified that the ␣-galactosyl epitope facing the lumen of the blood vessels was virtually removed by the enzymatic treatment at 4°C for 4 h (Fig. 8). No histological change was observed in the enzymatically treated kidneys.
With this powerful method of removing ␣-galactosyl epitopes in situ, we investigated whether binding of immunoglobulins in human plasma was decreased in enzymatically treated blood vessels as in the case of aortic endothelial cells or erythrocytes. Although quantitative analysis is possible for dissociated cells using FACS, direct measurement of bound immunoglobulins is not easy in intact blood vessels. Therefore, we measured the decrease in binding activity of immunoglobulins to pig erythrocytes after perfusion into the kidney. To avoid the effects of variation between pigs, we used two kidneys from the same animal; one kidney was treated with endo-␤-galactosidase C, and the other was treated with the medium without the enzyme. In three independent experiments, enzymatically treated kidneys bound significantly lower amounts of IgM than the kidney treated only with the medium. On the average, 31% of binding activity of IgM remained in human plasma after perfusion into the pig kidney without the enzyme treatment (Fig. 9). However, 64% of the binding remained after perfusion into the kidney with the enzyme treatment (Fig. 9). This difference was statistically significant. The result demonstrated that large amounts of IgM-binding antigenic epitope were eliminated after endo-␤-galactosidase C digestion. However, the ability to absorb IgG and IgA antibodies was less reduced by enzymatic treatment, and the differences were not statistically significant (Fig. 9). DISCUSSION Although the principal aim of the present investigation was to reduce ␣-galactosyl xenoantigenic determinants in blood vessels in situ using an enzyme, the results presented here raised other interesting points. First, the protein sequence of endo-␤-galactosidase C shares some amino acids in the more C-terminally located region with endo-␤-galactosidase from F. keratolyticus (29) and ␤-1,3-glucanases (30,31). Although the Flavobacterium endo-␤-galactosidase and ␤-1,3-glucanases act on polysaccharides to release oligosaccharides of various sizes, endo-␤-galactosidase C acts on carbohydrate moieties of glycoproteins and glycolipids and releases only a disaccharide (18). Nevertheless, the similarity in structure of these enzymes suggested that there are similarities in the mechanisms of action between endo-␤-galactosidase C and the other endoglycosidases mentioned above.
Second, endo-␤-galactosidase C, at the partially purified stage, has been proved useful for the structural analysis of glycoconjugates such as poly-N-acetyllactosamine in Ehrlich carcinoma cells (32) and teratocarcinoma glycopeptides (33). The newly available recombinant enzyme is of a high order of purity and lacks proteases and is expected to be broadly applicable in studies of various glycoconjugates because of the widespread occurrence of the Gal␣1-3Gal␤1-4GlcNAc structure.
The most interesting finding of this study was that after perfusion with recombinant endo-␤-galactosidase C, the pig vascular endothelial cells in the kidney became virtually devoid of the ␣-galactosyl antigen with concomitant loss of a significant portion of the IgM-binding xenoantigen, which is known to be primarily responsible for hyperacute rejection in pig to primate transplantation (34). The recombinant enzyme could act at physiological pH and even at a low temperature. Thus, the enzyme is expected to be a valuable aid in xenotransplantation. It will be especially helpful in removing residual ␣-galactosyl antigen in transgenic pigs, which express transgenes of appropriate glycosyltransferases and have reduced ␣-galactosyl antigen levels. The reappearance of the antigenic activity after the endoglycosidase digestion may be prevented if the enzyme is present in the bloodstream. Production of transgenic pigs with the endo-␤-galactosidase C gene should also be considered. Transfection of the enzyme gene mediated by liposome or virus vector might be also helpful. In these gene manipulation experiments, endo-␤-galactosidase C is more suitable than ␣-galactosidase, because levels of protein expression required will be less in case of the endoglycosidase than ␣-galactosidase. For nearly complete removal of the ␣-galactosyl antigen from isolated pig blood vessels, 20 units/ml (2 mg/ml) of ␣-galacto- FIG. 9. Decreased absorption of immunoglobulins in human plasma to blood vessels of pig kidney after treatment with endo-␤-galactosidase C. The right kidney from a pig was treated with endo-␤-galactosidase C, while the left kidney was treated only with the medium as described under "Experimental Procedures." Then, 100 ml of human plasma was perfused into the kidneys under gravity. The perfused plasma samples were reacted with pig erythrocytes, followed by reaction with FITC-conjugated antibodies to human immunoglobulins. Stained erythrocytes were analyzed by FACS. The results are expressed as percentages of the mean fluorescence intensity relative to that obtained using the plasma before perfusion. White bars, before perfusion; gray bars, after perfusion with the medium; black bars, after perfusion with the enzyme. Average values from three independent experiments are shown with S.E. values. Results were statistically analyzed by Student's t test (*, p Ͻ 0.05). sidase is required (17), while 0.45 units/ml (0.043 mg/ml) of endo-␤-galactosidase C was sufficient for the removal from blood vessels in the kidney. Finally, it might be appropriate to remind that xenografts initially protected from hyperacute rejection for a certain period might become resistant to rejection even in the presence of xenoantibodies, by a mechanism called accommodation (35).