Endothelin-3-converting Enzyme Activity of the KEL1 and KEL6 Phenotypes of the Kell Blood Group System*

The Kell blood group protein is a metalloendopeptidase that preferentially cleaves a Trp21-Ile22 bond of big endothelin-3 producing bioactive endothelin-3. Kell is a polymorphic protein, and 25 different phenotypes, because of point mutations resulting in single amino acid substitutions, have been described. It was recently reported that a recombinant form of KEL1 (K, K1) phenotype, expressed in K562 and HEK293 cells, had no endothelin-3-converting activity, in contrast to the common KEL2 (k, K2) phenotype. We demonstrate that KEL1 red blood cells and also a soluble recombinant form of KEL1 protein (s-Kell KEL1) have similar enzymatic activity as the common Kell phenotype. In addition we show that KEL6 red blood cells, which are more prevalent in persons of African heritage than in Caucasians also have endothelin-3-converting enzyme activity and that the recombinant soluble form of KEL6 protein (s-Kell KEL6) has similar Km values as the wild-type.

The Kell blood group system is highly polymorphic, expressing over 25 antigens that have been classified into five antithetical sets of high and low prevalence antigens with the others being independently expressed or having unknown antithetical partners. The different phenotypes are due to single nucleotide mutations that result in an amino acid substitution (1,2). Two antithetical allelic sets of antigens, KEL1 (K, K1)/KEL2 (k, K2) and KEL6 (Js a )/KEL7(Js b ) are of particular interest because they are more prevalent in certain racial groups. For example, KEL1 is present in 9% of Caucasians and in 2% of persons of African heritage, whereas KEL6 is present in 20% of persons of African heritage and in less than 1% of Caucasians. KEL1 is due to a 698C3 T mutation that results in a Thr 193 3 Met substitution at an N-glycosylation consensus sequence (3). KEL6 is due to a 1910T3 C mutation that results in a Leu 597 3 Pro substitution, which is located between two closely placed non-conserved cysteine residues (596CPAC599), and the antigen is more sensitive to reducing agents (4).
The Kell blood group protein, a 93-kDa, type II membrane glycoprotein, is a member of the M13 family of zinc endopeptidases (5). Within this large group, Kell has significant similarity with six other type II membrane glycoproteins: neutral endopeptidase 24.11 (6), neprilysin 2 (NL1/SEP) (7-9), two different endothelin-converting enzymes (ECE-1 and ECE-2) (10, 11), PEX (PHEX) (12), and X-converting enzyme (endothelin-converting enzyme-like 1/damage-induced neutral endopeptidase; ECEL1/DINE) (13). As a group, the M13 family of zinc endopeptidases either degrades bioactive peptides or cleaves intermediate inactive peptides to produce smaller bioactive peptides (14). However, in some cases, for example X-converting enzyme, the substrates are unknown. Unlike endothelin-converting enzyme-1 and -2 that preferentially activate endothelin-1 (ET 2 -1) (10, 11), Kell has strong preference for big endothelin-3 (big ET-3) and has much less activity with big endothelin-1 (big ET-1) and big endothelin-2 (big ET-2) (15). Within the M13 family, Kell is unique in that it is linked, through a single disulfide bond (16,17), with another protein, XK, a 50.9-kDa putative membrane transporter with 10 possible transmembrane domains (18). However, the extracellular domain of Kell expressed as a soluble recombinant form retains its endothelin-3-converting enzyme activity. A three-dimensional model of Kell protein, constructed on the basis of the crystal structure of neutral endopeptidase 24.11, showed that Kell has two globular extracellular domains (19). The domain closest to the cell membrane contains the enzyme-active site. The surface of the other domain, away from the enzyme active site contains the amino acid residues whose substitutions lead to the different phenotypes.
Because of the unequal distribution of KEL1 and KEL6 in different racial groups we studied their catalytic activity. In contrast to the recent report by Clapéron et al. (20), we demonstrate that KEL1 red blood cells and a recombinant form of KEL1 phenotype have a similar big ET-3cleaving activity to the common KEL2 phenotype.

Construction of Recombinant Soluble KEL1 Protein (s-Kell KEL1)-
Two overlapping PCR products were produced using each one of the primers containing the KEL1 mutation. The template for the PCR was the common Kell cDNA. The two overlapping PCR products were joined by another PCR to yield the complete cDNA. The primer sequences of the two PCRs are as follows: PCR1, (EX1F) 5Ј-CAGTCC-TCCGAATCAGCTCCTAGA-3Ј and (K1R) 5Ј-ACTGACTCATCAG-AAGTCTCAGCATTCGG-3Ј; PCR2, (NK1F) 5Ј-GGACTTCCTTAA-ACTTTAACCGAATGCTG-3Ј and (2316R) 5Ј-GCTGTGGCATCT-TTGGTACC-3Ј. The underlined bold letter in the primer sequence is the incorporated KEL1 mutation.
The 230-bp cDNA segment with the KEL1 mutation, obtained by cutting the PCR product with EcoRI and AvrII, replaced a DNA segment between EcoRI and AvrII sites of a wild-type s-Kell construct with an N-terminal six-histidine tag, which was in a pAcGp67A vector (BD Biosciences) (15,20). The sequence of the construct was verified by DNA sequencing. The preparation of common s-Kell and s-Kell KEL6 expression constructs has been described previously (15). manufacturer's protocol. The absence of cross-contamination in s-Kell KEL1 with common s-Kell was determined by isolation of the baculovirus DNA from the expression culture and sequencing the KEL1 region that was amplified by PCR.
Isolation of Recombinant s-Kell Proteins-The secreted s-Kell proteins were harvested by affinity chromatography on nickel columns basically following the procedure described previously (15). s-Kell in the cell medium (3 ϫ 600 l) was applied to a nickel-nitrilotriacetic acid resin column (Ni-NTA spin kit, Qiagen, Valencia, CA) that had been equilibrated with serum-free cell culture medium (Ex-cell 400 or SF-900 II from Invitrogen). The column was washed once with 50 mmol/liter HEPES buffer, pH 7.4, containing 0.3 mol/liter NaCl and 20 mmol/liter imidazole and once with the same buffer but with imidazole increased to 50 mmol/liter. Bound protein was eluted with 50 mmol/liter HEPES, pH 7.4, containing 0.3 mol/liter NaCl, and 250 mmol/liter imidazole. The above solutions were made in the serum-free media instead of H 2 O. The serum-free medium contains non-ionic detergent, which is necessary to maintain the Kell enzyme activity. If the solutions are not made with the serum-free medium, 0.1% Tween 20 is required to retain enzyme activity.
Endothelin-converting Enzyme Analysis of KEL1 and KEL6 and Common Kell Phenotype-The recombinant Kell proteins and the human red cells of KEL1, KEL6, Kell null, and common Kell phenotype were assayed for their abilities to process big ET-1, big ET-2, and big ET-3 using an endothelin enzyme immunoassay kit (Cayman Chemical Co., Ann Arbor, MI) that was previously described (15). Big ET-1, big ET-2, or big ET-3 (0.1 mol/liter) were incubated at 37°C for 10 min with assay buffer (50 mmol/liter HEPES, pH 6.0, 50 mol/liter ZnCl 2 , and 150 mmol/liter NaCl containing 0.25% (v/v) of supernatant of boiled bovine serum albumin (1 mg/ml)) and various amounts of s-Kell. The final volume was 200 l. At the end of the incubation time, an equal volume of 5 mmol/liter EDTA was added to terminate the reaction, and the samples were kept in ice until enzyme immunoassay was performed. To determine whether the KEL1 protein is inactivated by the conditions used by Clapéron et al. (20), the purified s-Kell and s-Kell KEL1 were stored at 4°C for 16 h in the presence of 2% Brij-58 (Sigma-Aldrich) and 1% Triton X-100 (Sigma) prior to the enzyme assay. When K m values were determined for big ET-3, the substrate concentrations ranged from 0.1 to 0.6 mol/liter, and the incubation time was 10 min.

KEL1 and KEL6 Phenotype Red Blood Cells Have Endothelin-3-converting Enzyme Activities Comparable to Common Kell-type Red Blood
Cells-The endothelin-converting enzyme activities of red blood cells with KEL1, Kell null and common phenotype were compared. As previously shown, common Kell-type red blood cells preferentially cleave big ET-3 releasing ET-3 and have much less activity with big ET-1 and  MARCH 17, 2006 • VOLUME 281 • NUMBER 11

JOURNAL OF BIOLOGICAL CHEMISTRY 7181
big ET-2 as substrates. Kell null red blood cells have minimal background activity (15,20). The KEL1 red blood cells were comparable to the common Kell red blood cells and also preferentially activated ET-3 (Fig. 1A). The KEL6 (Js a ) red blood cells, like the common Kell-type red blood cells, were also capable of cleaving big ET-3 and producing ET-3 (Fig. 1B). KEL6 red blood cells also preferentially cleaved big ET-3 (data not shown). Soluble

DISCUSSION
In contrast to a recent publication by Clapéron et al. (20), which reported that recombinant KEL1 protein expressed on the surface of HEK293 and K562 cells did not have endothelin-3-converting enzyme activity, whereas the common KEL2 did, our results clearly demonstrated that both KEL1 red blood cells and a soluble recombinant form of KEL1 have similar enzyme activity as the wild-type forms. The reasons for the discrepancy are not clear. The possibility of loss of enzyme activity of KEL1 protein during purification of the recombinant native Kell KEL1 (K1) was checked by storing the purified wild-type s-Kell and s-Kell KEL1 in the same conditions as described by Clapéron et al. (20). The soluble recombinant proteins were kept in the presence of 2% Brij-58 and 1% Triton X-100 at 4°C for 16 h prior to the enzyme assay. There was no reduction in s-Kell KEL1 enzyme activity.
The KEL1 phenotype arises from a Thr 193 3 Met substitution that occurs in an N-glycosylation site, and KEL1 is expected to have four instead of five N-linked sugar moieties. It is possible that this mutation may affect the processing and insertion of Kell protein on the cell surface in a recombinant system. This is unlikely, however, because the KEL1 recombinant protein was detected by antibodies on the cell surface.
In our study both intact KEL1 and KEL6 red blood cells, which mimic the in vivo conditions, and soluble recombinant proteins that express the extracellular domains show very similar activity to their wild-type counterparts. This is perhaps not surprising because a three-dimensional model of the extracellular sequences of Kell protein indicates the presence of two separate globular domains. The mutation that leads to the KEL1 amino acid substitution is present in the outer globular domain, quite distinct from the globular domain that is closer to the plasma membrane and contains the enzyme active site. Although the Leu 597 3 Pro mutation that leads to the KEL6 amino acid substitution appears in a linear model to be present near the zinc-binding motif, a three-dimensional model places it on the surface, away from the enzyme active site (19).
The presence of endothelin-3-converting enzyme activity in KEL1 red blood cells is not due to an unrelated protease because Kell null (Ko) red blood cells have no endothelin-converting activity (15,21). The specificity of the assay for the endothelin-3-converting enzyme activity is also shown by the fact that a recombinant s-Kell (E582G), in which the enzyme-active site was mutated and was prepared in the same way as s-Kell KEL1, also had no endothelin-3-converting enzyme activity (15).
The KEL6 phenotype occurs in ϳ20% of persons with African heritage and in Ͻ1% in other populations. This has led to the suggestion that the KEL6 phenotype may offer a selective advantage to Africans. However, our study indicates that if this is the case, it is not due to a difference in endothelin-converting activity, because KEL6 red blood cells have similar endothelin-3-converting enzyme activity as the wild-type red blood cells and also showed similar substrate preference for big ET-3 over big ET-1 and big ET-2.