the conversion of the human membrane-associated folate binding

THE JOURNAL OF BIOLOGICAL CHEMISTRY Vol. 266, No. 4, Issue of February 5, pp. 2346-2353,1991 Printed in U. S A .

The Conversion of the Human Membrane-associated Folate Binding Protein (Folate Receptor) to the Soluble Folate Binding Protein by a Membrane-associated Metalloprotease*

(Received for publication, March 22,1990)

Patrick C. ElwoodSS, J. C. Deutschq, and J. Fred Kolhousen From the $Medicine Branch, Division of Cancer Treatment, National Cancer Institute, Bethesda, Maryland 20892 and the 1Department of Medicine, Division of Hematology, University of Colorado Health Sciences Center, Denver, Colorado 80262

The membrane-associated (M-FBP) and soluble (S- FBP) forms of human folate binding proteins (FBP) have been well characterized. Although related in a precursor-product manner, the mechanism of conversion and the basis for differences between M-FBP and S-FBP are not known.

The conversion of M-FBP to S-FBP in crude human nasopharyngeal carcinoma (KB) cell preparations is demonstrated based on characteristic gel filtration elution profiles of M-FBP and S-FBP (Ve/Vo = 1.3 and 1.7, respectively) in Triton X-100. M-FBP is stoichio- metrically converted to S-FBP in a time- and temperature-dependent reaction by a metalloprotease which is: heat-labile; particulate; contained in human KB cell and placental membranes, and rat kidney homogenates; inhibited by EDTA, 1,lO-phenanthroline, and parahydroxymercuribenzoate; requires divalent cations; is maximally active at neutral pH; and is active in the presence or absence of detergent. The purified soluble FBP product appears to be identical to S-FBP. Conversion of purified endogenously [‘Hlleucine-labeled M-FBP yields a soluble FBP characterized by a 45% decrease in specific activity (moles of ‘H/mol folate bound) relative to M-FBP and a non-folate binding fragment which contains 45% of the [3H]leucine from M-FBP, requires detergent and/or urea to remain soluble, and migrates aberrantly on gel filtration in 1% (v/v) Triton X-100 and 8 M urea. Based on changes in the specific activity and the gel filtration elution profiles of purified labeled M-FBP associated with conversion to S-FBP, the endoproteolytic cleavage site is predicted between residues 226 and 229 of the cDNA predicted human FBP amino acid sequence. These results suggest that the cDNA predicted hydrophobic carboxyl terminus (residues 227-257) remains intact on the fully processed, membrane-anchored M-FBP, contains the Triton binding domain, and is involved in the formation of the membrane anchor of M-FBP.

AM07709-01 (to P. C. E.), March of Dimes Grant 1-805 (to J. F. K.), * This work was supported by National Research Service Award

National Institutes of Health Grant 26486 (to J. F. K.), and National Research Service Award GM122610-01 (to J. C. D.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “aduer- tisernent” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

dressed Medicine Branch, Div. of Cancer Treatment, NCI, Bldg. 10, 3 To whom correspondence and reprint requests should be ad-

Rm. 12N226,9000 Rockville Pike, Bethesda, MD 20892.

Folate binding proteins (FBP),’ characterized by their high affinities for folic acid as well as reduced physiologic folates and antifolates, have been isolated and characterized from a variety of tissues (1-16). Based on differences in physico- chemical properties of homogeneous human proteins, human FBPs exist in two forms: a membrane-associated (hydrophobic) species referred to as M-FBP (folate receptor), and a soluble (hydrophilic) species referred to as S-FBP. In cultured human nasopharyngeal carcinoma (KB), T47 mammary carcinoma, fibroblasts, and CaCo colon carcinoma cells (8, 10, 17, 18) and human reticulocytes (6), M-FBP is a necessary component of folate and classic antifolate (methotrexate) (19) cellular transport since: (a) M-FBP binds significant quantities of total cell-associated and internalized ligand (17), ( b ) polyclonal monospecific antihuman M-FBP antiserum inhibits transport of folates (17) and methotrexate (19), ( c ) polyclonal anti-human folate receptor monospecific antiserum induces megaloblastic morphologic changes associated with parallel reductions in intracellular folates in cultured human bone marrow cells (ZO), ( d ) the N-hydroxysuccinamide ester of methotrexate specifically and covalently labels M-FBP and blocks binding and transport of methotrexate (19); and ( e ) cloned transport-mediated methotrexate-resistant human nasopharyngeal epidermoid carcinoma (KB) cells express significantly (1-30% of control) lower quantities of M-FBP (21).

The functional role of S-FBP is less well defined but the presence of S-FBP in human sera (1) and milk (4) suggest a biologic function. Several lines of evidence suggest a role of the S-FBP in transport and in folate hepatic uptake. S-FBP increases in human serum with folate deficiency or other disease states associated with folate deficiency (2) and the expression of S-FBP is directly related to intra- and extracellular folate concentrations in vitro (8). While several conflict- ing reports describe enhanced cellular transport (22, 23) or inhibition of cellular transport (24) in the presence of S-FBP in vitro, covalently labeled purified human S-FBP adminis- tered to rats rapidly appeared in the liver, biliary tract, and proximal gastrointestinal tract (25). The presence of FBPs (and presumably bound folate) in human milk may effectively increase milk folate levels or facilitate folate delivery to the neonate.

Characterization of M-FBP and S-FBP from human sources (4-9) has identified several distinctive characteristics specific for each form of FBP despite the following common features: ligand binding characteristics (4-7); immunologic cross-reactivity (8); identical determined amino-terminal

The abbreviations used are: FBP, folate binding protein; M-FBP, membrane-associated FBP; S-FBP, soluble FBP; GPI, glycosylphos- phoinositol; SDS-PAGE, sodium dodecyl sulfate-polyacrylamide gel electrophoresis.

2346

Conversion of Membrane-associated Folate Binding Protein 2347

amino acid sequence (26) and internal amino acid sequences (27); single-chained glycoproteins (4, 7); and a similar apparent M, (approximately 40,000) determined by SDS-PAGE analysis (26, 27). Based on Northern blot analysis probed with radiolabeled human FBP cDNA, both forms of FBPs appear to be encoded for by an 1,100-base pair species of mRNA (27). Distinct features of the M-FBP are: ( a ) localization to crude membrane fraction (5); ( b ) particulate nature (4, 7, 28); (c) hydrophobicity and detergent requirement for solubilization (4, 5, 7); and ( d ) Triton X-100 binding such that M-FBP aberrantly migrates with an apparent M , of 160,000 on gel filtration in the presence of Triton X-100 (4, 28). In contrast, features characteristic of the S-FBP are: ( a ) extracellular localization (7, 28); ( b ) solubility in the absence of detergents (4, 5, 7, 28); and (c) lack of detectable Triton X-100 binding (4, 5, 28).

To further understand the basis of these characteristics for both forms of the human FBPs, the cDNA for the human FBP has recently been cloned from human KB cell and placental cDNA libraries by our laboratory (27) as well as other laboratories (29-31). Lacey et al. (29) and Sadisavan et al. (30) reported similar cDNA clones from human CaCo-2 carcinoma cell and KB cell cDNA libraries, respectively, whereas Ratnam et al. (31) have recently reported a human placental cDNA encoding a homologous but distinct FBP. The full length KB cell FBP cDNA nucleotide sequence encodes a 257-residue protein which, when compared to determined amino (26, 32) and carboxyl-terminal sequences of the human soluble FBP (32), contains a characteristic signal peptide (25 residues long) and a hydrophobic 31-amino acid residue carboxyl terminus which are absent from the processed soluble form (S-FBP). Based on the hydrophobicity of the constituent amino acids and the predicted a-helical secondary structure of the predicted carboxyl terminus (27), and based on computer analysis using a database derived from the known structures of other integral membrane proteins (33), we suggested that the carboxyl-terminal domain (including any co- or post-translational modifications) unique to M-FBP could serve a membrane anchoring function by spanning the membrane and that this segment would also contain the Triton binding domain. In contrast, others (29, 34) have suggested that the M-FBP is anchored to the membrane and distinguished from the soluble FBP by a glycosylphosphati- dylinositol amide (GPI) linkage based on release of FBP from membranes by phosphatidylinositol-specific phospholipase C.

To study further the nature of the differences between the two forms of human FBPs and to characterize the factor(s) involved in the conversion of the M-FBP to the S-FBP, we have established conditions under which M-FBP is converted to S-FBP in vitro by crude KB cells or KB cell membrane fractions. The results of these studies indicate that the conversion of M-FBP to S-FBP involves proteolysis of the primary amino acid structure of M-FBP rather than simple removal of a fatty acid, GPI linkage, or other covalent modification.

MATERIALS AND METHOD$

RESULTS

Rate of Conuersion-The elution profiles, determined by Sephacryl S-200 gel filtration in the presence of 1% (v/v) Triton X-100, of the folate binding proteins contained in the

The “Materials and Methods” are presented in miniprint at the end of this paper. Miniprint is easily read with the aid of a standard magnifying glass. Full size photocopies are included in the microfilm edition of the Journal that is available from Waverlv Press.

crude solubilized particulate pellet of human nasopharyngeal epidermoid carcinoma (KB) cells following incubation at 37 “C for 0, 1,2, and 22 h are shown in Fig. 1, A, B, C, and D, respectively. KB cells solubilized at 4 “C contain a single peak of bound folic acid which represents the particulate membrane-associated folate binding protein (M-FBP) (4, 5, 7), or folate receptor, which is characterized by a VJV0 = 1.3 (Fig. lA) on Sephacryl S-200 gel chromatography in 1% (v/v) Triton X-100 and which corresponds to an apparent M , of 160,000. We have previously described (4, 7, 28) a soluble FBP (S-FBP), which does not bind Triton X-100, and which has a characteristic VJV0 = 1.7 on Sephacryl S-200 gel chromatography corresponding to an apparent M, of 40,000, that is released by proliferating KB cells into their conditioned media (data not shown) but is not detectable within KB cells solubilized at 4 “C (see Fig. L4). As shown in Fig. 1, B, C, and D, the hydrophobic M-FBP ( Ve/Vo = 1.3) contained in solubilized KB cell particulate membrane fraction is pro- gressively converted to a soluble hydrophilic FBP (Ve/Vo = 1.7) in a time and temperature-dependent manner, such that 38, 58, and >97% of the KB cell M-FBP is converted to a soluble FBP at 1, 2, and 22 h, respectively (for quantitation, see “Materials and Methods”). The change in VJV0 (1.3 to

VE ’ vo FIG. 1. Sephacryl S-200 gel filtration analysis of the con-

version of human KB cell M-FBP to S-FBP. Following incubation, the folate binding proteins from each sample of crude solubilized KB cell membrane fraction were saturated with radiolabeled folic acid and analyzed by Sephacryl S-200 gel filtration in 1% (v/v) Triton X-100 (see “Materials and Methods”). The elution profiles A , B, C , and D represent the results when samples were incubated at -20 ‘C, at 37 “C for 1 h, at 37 “C for 2 h, and at 37 “C for 22 h, respectively. ’

2348 Conversion of Membrane-associated Folate Binding Protein

1.7) of folate binding indicates loss of Triton binding to the FBP (4, 28). This conversion is stoichiometric (moles/mol of bound folate) and the resultant soluble species of FBP is stable under these conditions since there is no detectable loss (less than 2%) in total folate binding capacity during incubation as determined by folate binding assays. Soluble FBP was also released from unsolubilized KB cell particulate membranes and intact cells at a rate similar to the release of S- FBP by KB cells in tissue culture. The factor(s) involved in this conversion is heat-labile since boiling crude KB cells or their particulate membranes inhibits conversion of M-FBP to S-FBP and is nondialyzable using dialysis tubing with a M, cutoff of 12,000 (data not shown).

Localization-The percent conversion of M-FBP by the KB cell particulate pellet, which contains approximately 75% of the total folate binding capacity as determined by folate binding assays, was equal to or greater than that of solubilized whole cells (which contained an equivalent quantity of crude membranes) for each point studied (see Fig. 2). In other experiments, solubilized KB cell particulate pellet converted 70% of added purified M-FBP compared to 23% conversion of added purified M-FBP by the soluble KB cell fraction when analyzed by gel filtration following a 4-h incubation at 37 "C (data not shown). These results indicate that the majority of the factor(s) involved in converting M-FBP to a soluble FBP is contained within the crude membrane fraction of KB cells.

Protease Inhibitors-Whereas serine protease inhibitors (diisopropyl phosphoflouridate, and phenylmethylsulfonyl flouride) and alkylating inhibitors (iodoacetic acid, iodoacetamide, and N-ethylmaleimide) a t concentrations of 10 mM exhibit no detectable inhibition on the conversion of M-FBP to S-FBP, 1 p~ of EDTA and 1 mM of 1,lO-phenanthroline ?xhibit marked inhibition (>95 and 77%, respectively) of conversion of M-FBP to S-FBP, and 10 mM of 1,lO-phenanthroline or parahydroxymercuribenzoate result in >95% inhibition of this conversion. These results are similar to those recently reported (35) for the human myeloid CD13/amino- peptidase N, a zinc-requiring metalloprotease, and suggest that conversion of M-FBP to S-FBP involves a divalent cation requiring proteolytic enzyme.

Effect of Divalent Cations-Since EDTA exhibited marked inhibition of conversion of M-FBP to S-FBP, the role of various divalent cations in this process was investigated by adding the specified cation to a KB cell particulate pellet which had been pre-incubated with 1 mM EDTA to chelate endogenous divalent cations. In the absence of added divalent cations, no detectable conversion occurred. Calcium and magnesium chloride are not able to significantly reconstitute

I

0 2 4 22 Time (Hr)

FIG. 2. Rate of conversion of M-FBP to S-FBP. Solubilized KB cells (0) or solubilized KB cell membrane fractions (0) were incubated for the time specified at 37 "C. The endogenous folate binding proteins from an aliquot of each sample were saturated with radiolabeled folic acid and analyzed by gel filtration (see "Materials and Methods").

converting activity while the addition of manganese and zinc chloride (1-10 mM) are able to overcome the EDTA induced inhibition of conversion (Fig. 3). In the absence of EDTA, similar results were obtained by addition of the divalent cations directly to the KB cell particulate pellet indicating that these differences are not solely due to differences in the affinity of these cations for EDTA (36). Analysis of a purified form of the factor(s) involved will be necessary to determine which cation is complexed to the factor in uivo. Similarly, the gel filtration profile of rat kidney solubilized in the presence of EDTA contained no detectable binding of Iz5I-labeled folic acid at 1.7 VJVo, whereas greater than 90% of the folate binding occurred at 1.7 VJV, when the kidney was solubilized without EDTA or when the kidney was solubilized in EDTA that was neutralized with excess zinc chloride.

Determination of pH Optimum-The percent maximum conversion of M-FBP to the soluble species of FBP by KB cells in potassium phosphate, pH 4.5, 6.0,7.0,7.5,8.0, and 9.0 was 10, 50, 100, 100, 78, and 66%, respectively. Since the pH optimum is approximately 7-7.5 under these conditions, it is unlikely that the conversion of M-FBP to the soluble FBP occurs within lysosomes where the pH is 4-5.

Characterization of the Soluble FBP-Based on SDS-PAGE and gel filtration analysis, the soluble FBP product from conversion of M-FBP was purified to apparent homogeneity by affinity chromatography. The apparent M, (42,000) on SDS-PAGE, Ve/Vo (1.7) on gel filtration in 1% (v/v) Triton X-100, and the amino acid composition of this soluble FBP are indistinguishable from the purified S-FBP (7) released by KB cells into their conditioned media (data not shown). The amino acid composition of M-FBP contains 1.5-2.0 mol % more leucine residues than the soluble FBPs consistent with the previously reported excess of leucine and hydrophobic amino acid residues in M-FBP from human KB cells (7) and human milk (4) relative to S-FBP. Since the apparent M, determined by SDS-PAGE of purified M-FBP or S-FBP purified from either media or KB cells have virtually identical apparent M, values (7, 26), differences in primary structure or covalent modifications thereof are not detectable by SDS- PAGE analysis.

Conversion of Endogenously Labeled M-FBP-TO study changes in the primary structure of M-FBP occurring during conversion to the soluble FBP, to study the potential rele- vance of the carboxyl-terminal 31-amino acid residues of the

40- o Ca++

.- 5

s !!? al 5 c

E 20- 2 0

a al

0' ' 1 1 0.001 0.01 0.1 1 10

Concentration (mM)

FIG. 3. Determination of divalent cation requirements. Crude solubilized KB cell membrane fractions were preincubated with 1 mM EDTA a t 4 "C for 15 min, the respective cation was added at the concentration specified, and the samples were incubated for 2 h a t 37 "C. Endogenous folate binding proteins from an aliquot of each sample were saturated with radiolabeled folic acid and analyzed by gel filtration (see "Materials and Methods"). The legend inset contains the designation for each divalent cation studied.


predicted amino acid sequence of M-FBP, and to study these changes in exogenously added labeled substrate (M-FBP), M- FBP was purified with affinity chromatography from KB cells which had been pulsed with 1-[4,5-'HH]leucine. SDS-PAGE analysis (Fig. 4) of purified endogenously labeled M-FBP stained with Coomassie Blue (Fig. 4A) and autoradiographed (Fig. 423) shows a single, homogeneous, ['Hlleucine-labeled protein band. While internal controls ([I4C]stearic and ["C] palmitic acids) were easily extracted into the organic phase during fatty acid analysis (26) of the purified endogenously labeled M-FBP, less than 1% of the incorporated "H from purified M-FBP could be extracted following fatty acid hydrolysis (26). Furthermore, amino acid composition analysis of purified endogenously labeled M-FBP revealed that greater than 98% of the "H co-eluted with [14C]leucine standard while the remainder co-eluted with ['4C]isoleucine standard. These data indicate that virtually all of the 'H contained in the endogenously labeled M-FBP is leucine (<2% isoleucine) and was not converted from leucine to fatty acids with subsequent attachment to M-FBP.

Gel filtration analysis of purified endogenously labeled M- FBP incubated in the conversion assay for 6 h at 37 "C with buffer or with boiled KB cells is shown in Fig. 5A. The peak of folate binding (open circles, dashed lines) co-elutes with the peak of ['H]leucine (solid circles, solid lines) a t VJVo = 1.3 providing further evidence of purity of the M-FBP and indicating that the factor(s) involved in the conversion of M-FBP is heat-labile. Following incubation (37 "C for 6 h) of the purified endogenously labeled M-FBP in the conversion assay, the peak of folate binding elutes a t V,/Vo = 1.7 (Fig. 5B) indicating that greater than 95% of the M-FBP had been converted to the soluble FBP with loss of the Triton binding region and the membrane anchor of M-FBP. Although the major peak of ["Hlleucine co-elutes with the peak of folate binding, the specific activity (picomoles of ['H]leucine/pmol folic acid binding) was decreased to approximately 55% of its original value indicating that approximately one-half of the ["Hlleucine was removed during conversion. The remainder of the ['Hlleucine eluted independent of the folic acid binding capacity as a broad plateau (which contained approximately 35% of the total 'H eluted) preceding the soluble FBP and as a discrete peak (V,/Vn = 3.4) which contains 10% of the total "H eluted and co-elutes with standard ['Hlleucine under these conditions.

The above data indicate that [3H]leucine is released during conversion of M-FBP to S-FBP either as a result of exo-

A. B.

0-

z X 66.2- E .- 46- s w

1 . 1 i L

S 31- u 0)

P 21.5- :

FIG. 4. SDS-PAGE analysis of purified endogenously ["HI leucine-labeled M-FBP. Purified endogenously labeled M-FBP (7 pg containing 55,000 cpm) was resolved on a 5-15% gradient SDS- PAGE, stained with Coomassie Blue (lane A ) , and autoradiographed (lane R ) .

O ~ 1.0 2.0 3.0 4.0

vE I vO

FIG. 5. Gel filtration analysis of the conversion of purified endogenously ['Hlleucine-labeled M-FBP. Purified endogenously labeled M-FBP (7-10 pg containing 55,000-72,000 cpm) was incubated with boiled crude KB cell membrane fractions ( A ), crude KB cell membrane fractions ( R ) , and crude KB cell membrane fractions in the presence of 10 mM each of diisopropyl phosphoflouridate and iodoacetamide (C) for 22 h a t 37 "C. Each sample was analyzed by Sephacryl S-200 gel filtration in the presence of 1% (v/ v) Triton X-100 and 8 M urea, and the "H (0) and folate binding capacity (0) eluted into each fraction was determined. Fig. 5 0 contains the elution profile of "H from 7 p g of purified endogenously labeled M-FBP following incubation with a crude placental membrane fraction.

proteolytic cleavage of M-FBP with the release of the S-FBP and amino acids or small peptides, or as a result of specific endoproteolytic cleavage of M-FBP releasing S-FBP and another fragment which subsequently undergoes secondary "nonspecific" proteolysis. To determine the possible contri- bution of such nonspecific proteolysis of a 'H-labeled fragment which lacks folate binding properties to the elution profile shown in Fig. 5B, nonspecific protease inhibitors (diisopropyl phosphofluoridate and iodoacetamide, 10 mM each) which do not inhibit conversion were added to the conversion assay. As shown in Fig. 5C, the addition of protease inhibitors did not affect the conversion of endogenously labeled M-FBP to soluble FBP as determined by the shift in elution of the folate binding activity or of 'H from VJVo of 1.3 to 1.7. However, in contrast to the results shown in Fig. 5B, the 'H eluting independent of the folate binding capacity elutes in a single, sharp peak near the void volume (V,/Vn = 1.1). It is important to note that no measurable folate binding elutes with the 'H a t VJVO = 1.1, and that no "H was detected a t V,/Vn = 3.4 in the presence of "nonspecific protease" inhibitors. Further attempts to study this peak of "H eluting a t Vp/


Vo = 1.1 by SDS-PAGE or gel filtration after the removal of urea (by dialysis) and/or Triton X-100 (with ether or ethanol) resulted in an insoluble precipitate even in the presence of excess SDS (10%). In preliminary experiments: a synthetic peptide composed of an amino-terminal lysine (a polar amino acid added to facilitate immobilization of the peptide) followed by the 14 carboxyl-terminal residues of the cDNA predicted FBP amino acid sequence are insoluble in phosphate-buffered saline (10 mM potassium phosphate, pH 7.5, containing 0.15 M NaCl) containing 10% (v/v) SDS, 1% (v/v) Triton X-100, and 35% (v/v) dimethyl sulfoxide after boiling for 5 min. These results are compatible with a [3H]leucine-labeled peptide fragment being released during the conversion of M-FBP to S-FBP which: (a ) migrates aberrantly on gel filtration at an apparent M, ( Ve/Vo = 1.1) greater than M-FBP ( VJV0 = 1.3) presumably as a result of Triton binding which accounts for >75% of the apparent M, of M-FBP in the presence of Triton X-100 (4, 28); ( b ) requires urea and/or Triton X-100 to remain soluble; (c) does not contain the folate binding domain; ( d ) is further degraded by nonspecific proteolysis to leucine or low molecular weight peptides that co-elute with the leucine standard; and ( e ) contains approximately 45% of the total leucine residues of mature M-FBP. Since the labeled S-FBP eluting at 1.7 Ve/Vo contains the folate binding domain (see above) and is soluble in the absence of Triton X-100, the fragment characterized by a VJV0 = 1.1 lacks the folate binding domain but appears to contain the Triton binding domain and the membrane anchoring domain which are features of the M-FBP. It is important to note that no detectable loss (less than 5%) in total folate binding capacity was observed in the studies shown in Fig. 5, and that the recovery of applied radioactivity from the columns was greater than 90% for these experiments.

Since Antony et al. (5, 37) have previously postulated that the placental folate receptor ( VJV0 = 1.7 on gel filtration) contained in crude solubilized human placenta results from the proteolytic cleavage of a higher M, particulate placental FBP and have recently demonstrated the conversion of the hydrophobic form of the placental receptor to the hydrophilic form by a magnesium-dependent factor (37), and since the conversion of M-FBP to S-FBP as described in this study may be unique to KB cells, we studied the ability of a placental particulate pellet to convert the endogenously labeled KB cell M-FBP. As shown in Fig. 50, two distinct peaks of 3H eluted at Ve/Vo = 1.1 and VJV0 = 1.7 comprising an elution profile identical to that shown in Fig. 5C. Folate binding was not measured in this experiment since crude placental tissue would contribute the majority of the folate binding capacity and since greater than 95% (5) of the placental FBP elutes at Ve/Vo = 1.7. Identical results were seen in experiments in which rat kidney served as the source of protease. In addition, when EDTA was added during the preparation of rat kidney homogenates to inhibit proteolytic cleavage of M-FBP, the major form of FBP contained in crude solubilized rat kidney was M-FBP in contrast to previous reports (15) where, in the absence of appropriate protease inhibitors, the S-FBP was the predominant form of FBP in this tissue. These experiments demonstrate that crude human placenta and rat kidney contain a similar if not identical protease to that contained in human KB cells, may explain the observation (5) that >95% of the placental FBPs exists as the S-FBP, and demonstrate that the protease involved in the conversion of M- FBP to S-FBP is not unique to human KB cells. These observations may also explain the predominance of the S- FBP described in crude rat kidney tissue (15) as well as the

P. C. Elwood, unpublished observation.

1 .o 2.0

FIG. 6. Gel filtration analysis of the conversion of covalently labeled purified M-FBP. Purified M-FBP (10 pg) was iodinated by the chloramine-T method and incubated with boiled crude KB cell membrane fractions (0) or with crude KB cell membrane fractions (0) at 37 "C for 24 h and analyzed by Sephacryl S- 200 gel filtration in the presence of 1% (v/v) Triton X-100.

VENO

appearance of lower M, FBPs in KB cells (7, 26), L1210 cells (38, 39), human granulocytes (11, 40), and human milk (4).

Fig. 6 contains the gel filtration elution profile of covalently labeled (iodinated by the chloramine-?' method) M-FBP following incubation with crude KB cells (open circles) compared to the elution profile of labeled M-FBP incubated with boiled KB cells (solid circles). Purified M-FBP iodinated by this method elutes as a sharp peak of radioactivity on gel filtration at VJV0 = 1.3. Following partial (-40%) conversion, sharp peaks are seen at VJV0 = 1.1 and 1.3 with a broad plateau of radioactivity between VJV0 = 1.5 to 2.0, an elution profile that is qualitatiaely similar to that shown in Fig. 5C. Since iodination of proteins by the chloramine-T method labels tyrosine residues, these results demonstrate that M-FBP is endoproteolytically cleaved such that both fragments (at Ve/ Vo = 1.1 and 1.5-2.0) contain labeled tyrosine residue(s). The decreased rate of conversion was not further studied but may have resulted from the covalent modification (iodination) of M-FBP interfering (e.g. steric) with the protease involved in this process.

DISCUSSION

Conversion of the human nasopharyngeal epidermoid carcinoma (KB) cell M-FBP involves a time- and temperature- dependent reaction mediated by a heat-labile, particulate, membrane-bound factor(s) contained in KB cells, rat kidney, and possibly human placenta. The presence of the converting activity in other tissues, e.g. human placenta and rat kidney, demonstrates that the factor(s) is not unique to KB cells and most likely explains the presence of variable quantities of the soluble and membrane forms of the FBP reported in these (5, 15) and other tissues (7, 11, 26, 38-40). Inhibition of the conversion reaction by EDTA, 1,lO-phenanthroline, or parahydroxymercuribenzoate suggested that the factor was a metalloprotease but did not exclude the possibility that a lipase mediated conversion.

The conversion of hydrophobic placental FBPs to hydrophilic placental FBPs was recently reported (37) to be inhibited by EDTA and to involve a divalent cation requiring factor. The placental system differs from that observed in KB cells in that much higher concentrations of EDTA (20 mM) are required to inhibit conversion (37), magnesium is the preferred cation rather than zinc or manganese (37), the hydrophilic placental folate receptor requires Triton X-100 for solubilization (5) in contrast to the human KB cell or milk S-FBP, and placental hydrophobic membrane-associated


FBPs are heterogeneous (31). Although these observations suggest that KB cell and placental converting factors are different, further comparisons will require purification and characterization of the involved factors involved in conversion.

Another major observation in these studies is that conversion of purified [3H]leucine-labeled M-FBP results in a 45% decrease in the specific activity (moles of 3H/mol folate binding) of the soluble FBP and in the release of a radiolabeled non-folate binding fragment. The decrease in specific activity strongly suggests that a proteolytic event is involved in the conversion process and correlates with the previously reported difference in leucine content between M-FBP and S-FBP purified from human milk (4) and KB cells (7) where the M- FBP from each of these sources contained approximately 2- fold more leucine/mol protein. To determine the site of proteolytic cleavage of M-FBP by this factor(s), it was useful to compare the cDNA-predicted amino acid sequence of human FBP (27) to the determined amino-terminal amino acid sequence of KB cell M-FBP and S-FBP (26) and the determined carboxyl terminus of human milk S-FBP (32) as shown in Fig. 7. Although the predicted amino acid sequence of the human M-FBP (27) contains a total of 22 leucine residues, the amino terminus (residues 1-25) contains 5 leucines and the extreme carboxyl terminus (residues 244-257) contains 8 leucines. The proteolytic loss of the amino terminus from the purified M-FBP would decrease the specific activity by only 23% (5/22 leucines). Sadasivan et al. (26) and Svendsen et at. (32) have demonstrated that the membrane and soluble forms of human FBPs share a common and identical amino-terminal sequence beginning at IleZ6 (see Fig. 7) indicating that the amino terminus is a signal peptide (27,29). For these reasons, it is not possible to explain the changes in specific activity by cleavage of the amino terminus from M-FBP during conversion to S-FBP. Nine of the 17 remaining leucine residues are randomly distributed between amino acids 26 and 226 which constitutes the determined amino acid sequence of human milk S-FBP (32). At least 8 of these 9 leucines would have to be released during the conversion to account for the observed reduction in specific activity. Since the apparent M, of the product (S-FBP) was unchanged relative to native S-FBP (M, = 40,000) and since conversion resulted in less than 2% loss of folate binding capacity, it is very unlikely that the observed loss of [3H]leucine involved these randomly dis- persed leucines. In contrast, 8 or 47% (8/17) of the remaining 17 leucines are concentrated at the carboxyl terminus between residues 244 and 257. The observations that 1) conversion of purified labeled M-FBP results in two radiolabeled peaks, 2) the conversion of purified labeled M-FBP to S-FBP results in a 45% decrease in the specific activity of the converted M- FBP, 3) TyrPg is at least partially iodinated since this residue is the only tyrosine contained in the carboxyl terminus, 4)

NH211te2’1 COOH (Ats”’t 50 100 150 m 250 257

&LLLd.L-L-M-L--L-L I I & M ~ T ~ L L . ~ M L L U !

t FIG. 7. The 267-residue amino acid sequence of the human

FBP predicted from the cDNA. The numbers indicate amino acid residues 1-257 and are based on the amino acid sequence of the human FBP predicted from the cDNA (27). The determined amino termini (IleZ6) of purified human KB cell M-FBP and S-FBP (26) are indicated by -NH2(llez6) and the determined carboxyl terminus (Alazz6) of the human soluble FBP (29) is depicted by -COOH(AlaZz6). The signal peptide is indicated by the box (0); and the leucine, methionine, and tyrosine 229 are designated by L, M , and T, respectively. The arrow (4) points to the predicted site of endoproteolytic cleavage of M-FBP by the metalloprotease.

the determined carboxyl terminus of the human milk soluble S-FBP corresponds to residue 226 of the predicted mature M- FBP sequence, and that 5) the predicted 31-amino acid hydrophobic carboxyl terminus potentially unique to the mature M-FBP contains 47% of the leucine residues in M-FBP, taken together, indicate that the conversion of M-FBP to S-FBP involves endoproteolytic cleavage of the carboxyl terminus of the mature M-FBP most likely between residues 226 and 229. Thus, the products of conversion appear to be the soluble form of FBP (S-FBP) and a leucine-rich carboxyl-terminal fragment. Since M-FBP is distinguished from S-FBP by virtue of its membrane attachment and Triton X-100 binding properties, the hydrophobic carboxyl terminus appears to contain the Triton X-100 binding domain and a membrane anchor such as a GPI tail (see below) and/or a transmembrane domain (27, 33).

Data in support of proteolysis have been published in previous studies (28) of the precursor-product relationship of M-FBP and S-FBP where the proteins were endogenously labeled with [36S]methionine and the specific activity (moles of 35S/mol bound folate) of purified M-FBP and purified S- FBP was determined. The maximum specific activity of [35S] methionine-labeled S-FBP was 60% of the [35S]methionine- labeled M-FBP. Since the M-FBP contains 5 methionines (see Fig. 7), 2 of which are contained in the carboxyl-terminal 31 amino acids, the observed reduction (40%) in specific activity of S-FBP relative to that of M-FBP is compatible with the loss of 2 of the 5 methionine residues from M-FBP during conversion to S-FBP.

We considered the possibility that purified labeled M-FBP might be heterogeneous relative to processed amino and/or carboxyl terminus. Since at least 80% (34) of KB cell M-FBP is localized to the plasma membrane and therefore fully processed, the quantity of potentially unprocessed M-FBP represents only 20%, a t maximum, of the t o t a l cellular M- FBP. The observed reduction (45%) of specific activity of [3H]leucine-labeled M-FBP cannot be accounted for by processing of the amino-terminal signal sequence (see above). Furthermore, since this would involve processing of only 20% (or less) of the M-FBP, removal of the signal peptide could only account for an approximate 5% (20% of the total M- FBP would have a 23% decrease in specific activity) reduction in overall specific activity. Likewise, if the carboxyl-terminal amino acid sequence is cleaved (e.g. during attachment of a GPI anchor) from 80% of the mature M-FBP prior to conversion to s-FBP, then a maximum of 20% of KB cell M- FBP would contain an unprocessed carboxyl terminus, and we would expect the specific activity to decrease by approximately 10% (20 x 47%). The observations that approximately 45% of the t o t a l leucine contained in purified fully processed M-FBP is released during the conversion to S-FBP and that the difference in amino acid sequence between M-FBP and S-FBP resides solely within the carboxyl terminus, indicate that purified [3H]leucine-labeled M-FBP must be nearly homogeneous relative to the carboxyl terminus in order to account for the changes in the specific activity observed during conversion,

Although the membrane localization of M-FBP has been well characterized (5,17,41,42), the means by which M-FBP is anchored to the membrane is controversial (27). Recent studies demonstrating the release of a FBP from KB (34) and CaCo (29) cells suggest that M-FBP is covalently attached to the plasma membrane via a GPI amide linkage and that the S-FBP was derived from the mature M-FBP by hydrolysis of this GPI linkage (34). The anchoring of all described GPI- anchored membrane proteins (43-45) occurs within 2 min of


translation in the rough endoplasmic reticulum (44, 45) and involves cleavage of a signal carboxyl-terminal peptide with simultaneous formation of an amide bond between the remaining carboxyl-terminal amino acid and the ethanolamine of the GPI moiety (43-45). Although M-FBP contains a hydrophobic carboxyl terminus similar to other unprocessed GPI-anchored proteins, the current studies suggest that the carboxyl terminus of M-FBP remains intact in the mature processed membrane form (see above). Furthermore, in contrast to other GPI-linked proteins, M-FBP is readily solubilized in 1% (v/v) Triton X-100 (4 ,5 ,7) ; the conversion of the M-FBP to the S-FBP occurs under conditions which inhibit eukaryotic phosphatidylinositol-specific phospholipase C (43, 44) and is inhibited by EDTA and 1,lO-phenanthroline; and the release of the S-FBP involves endoproteolytic cleavage of the M-FBP. Although available evidence (29, 34) supports attachment of a GPI anchor to M-FBP, the above observations strongly suggest that M-FBP is not anchored in the manner described for all other well characterized GPI-linked proteins (43-45) in that the GPI-linkage would be to an unprocessed carboxyl terminus of M-FBP.

The biologic function of the metalloprotease involved in the conversion of KB cell M-FBP to S-FBP is unclear. How- ever, other membrane transport proteins or receptors (e.g. human transferrin receptor (46), platelet glycoprotein Ib (47), secretory component of immunoglobin (48), and the interleu- kin-2 receptor (49)) also release a soluble form of the receptor as a result of proteolytic cleavage of their respective membrane form. These observations suggest that proteolysis of a membrane protein to a soluble form may represent a more generalized biologic phenomena. Although it is possible that release of the S-FBP represents nonspecific proteolysis, the stability of S-FBP such that there was no detectable change in the apparent M , and less than 2% loss in folate binding capacity following incubation for up to 24 h at 37 "C, the stoichiometry of conversion, the release of S-FBP by proliferating cultured cells into conditioned media (7, 28), and, more importantly, the presence of S-FBP in human serum (2) and human milk (4) strongly argue against this possibility. Since both the M-FBP and the protease involved in the release of the S-FBP are contained in the plasma membrane, it is possible that cells may regulate surface expression of membrane receptors by either alteration of synthesis of M- FBP or the metalloprotease, or by altering the activity of the metalloprotease. The S-FBP contained in human serum may function to facilitate the hepatic uptake of folate (25) while the S-FBP in human milk may serve to concentrate folates in milk or facilitate intestinal absorption of folate in the neonate (22, 23).

REFERENCES

1. Waxman, S. (1975) Br. J. Haematol. 2 9 , 23-29 2. Waxman, S., and Schreiber, C. (1973) Blood 42,291-301 3. Suleiman, S. A,, and Spector, R. (1981) Arch. Biochem. Biophys. 208,87-

94

4. Antonv. A. C.. Utlev. C. S.. Marcell. P. D.. and Kolhouse. J. F. 11982) J. Bioi: Chem. '257,~10081-10089 '

Biol. Chem. 256,9684-9692

Clin. Invest. 80, 711-723

Biol. Chem. 2 6 1 . 15416-15423

5. Antony, A. C., Utley, C., Van Horne, K. C.,

6. Antony, A. C., Kincade, R. S., Verma, R. S.,

7. Elwood, P. C., Kane, M. A., Portillo, R. M.,

and

and

and

Kolhouse, J. F.

Krishnan, S. R.

Kolhouse, J. F.

. .

(1981) J .

(1987) J.

(1986) J.

8. Kane, M. A., Elwood, P. C., Portillo, R. M., Najfeld, V., Finley, A,, Waxman,

9. Kamen, B. A,, and Capdevila, A. (1986) Proc. Natl. Acad. Sci. U. S. A. 8 3 ,

10. Kamen, B. A,, Wand, M. T., Streckfuss, A. J., Peryea, X., and Anderson,

11. Fischer, C. D., da Costa, M., and Rothenberg, S. P. (1975) Blood 46,855-

S., and Kolhouse, J. F. (1988) J. Clin. Inuest. 8 1 , 1398-1406

5983-5987

R. G. W. (1988) J . Bioi. Chem. 263,13602-13609

I l K 7

12. 13.

14.

15. 16. 17.

18.

19.

20.

21.

dayosta, M., and Fischer, C. (1981) J. Lab. Clin. Med. 98,956-964 da Costa, M., and Rothenberg, S. P. (1988) Biochim. Biophys. Acta 9 3 9 ,

Svendsen, 1. B., Hansen, S. I., Holm, J., and Lyngbye, J. (1984) Carlsberg

Selhub, J., and Franklin, W. A. (1984) J. Biol. Chem. 259,6601-6606 Leslie, G. I., and Rowe, R. B. (1972) Biochemistry 11,1696-1703 Antony, A. C., Kane, M. A., Portillo, R. M.. Elwood, P. C., and Kolhouse,

Kane, M. A,, Portillo, R. M., Elwood, P. C., Antony, A. C., and Kolhouse,

Deutsch, J. C., Elwood, P. C., Portillo, R. M., Macey, M. G., and Kolhouse,

Antony, A. C., Bruno, E., Briddell, R. A,, Brandt, J. E., Verma, R. S., and

Knight, C. B., Elwood, P. C., and Chabner, B. A. (1988) Blood 7 2 , 274

533-541

Res. Commun. 49,123-131

J. F. (1985) J. Biol. Chem. 260,14911-14917

J. F. (1986) J. Biol. Chem. 261,44-49

J. F. (1989) Arch. Biochem. Biophys. 274,327-337

Hoffman, R. (1987) J . Clin. Invest. 80, 1618-1623

22. Salter, D. N., and Blakeborough, P. (1988) Br. J. Nutr. 59 , 497-507 23. Mason, J. B., and Selhub, J. (1988) Am. J. Clin. Nutr. 48,620-625 24. Waxman, S., and Schreiber, C. (1975) Biochemistry 14,5422-5428 25. Deutsch, J. C., and Kolhouse, J. F. (1989) Blood 74 , 675 (abstr.) 26. Luhrs, C. A,, Pitiranggon, P., da Costa, M., Rothenberg, S. P., Slomiany,

B. L., Brink, L., Tous, G. I., and Stein, S. (1987) Proc. Natl. Acad. Sci. U. S. A. 84,6546-6549

(abstr.)

27. Elwood, P. C. (1989) J. Biol. Chem. 264,14893-14901 28. Kane, M. A,, Elwood, P. C., Portillo, R. M., Antony, A. C., and Kolhouse,

29. Lacey, S. W., Sanders, J. M., Rothberg, K. G., Anderson, R. G. W., and J. F. (1986) J . Biol. Chem. 2 6 1 , 15625-15631

Kamen, B. A. (1989) J . Clin. Invest. 8 4 , 715-720 30. Sadasivan, E., and Rothenberg, S. P. (1989) J. B~ol. Chem. 264,5806-5811 31. Ratnam, M., Marquardt, H., Duhring, J. L., and Freisheim, J. H. (1989)

32. Svendsen, I. B., Hansen, S. I., Holm, J., and Lyngbye, J. (1984) Carlsberg

33. Viswanadhan, V. N., Weinstein, J. N., and Elwood, P. C. (1990) J. Biomol.

34. Luhrs, C. A., and Slomiany, B. L. (1989) J. Biol. Chem. 264 , 21446-21449 35. Ashmun, R. A., and Look, A. T. (1990) Blood 75,462-469 36. Peters, D. G., Hayes, J. M., and Hieftje, G. M. (1974) in Chemical Separa-

tions and Measurements: Theory and Practice o AnalytLcal Chemistry, 3rd Ed., pp. 172-173, W. B. Saunders Co., Philadelphia, PA

37. Antony, A. C., Verma, R. S., Unune, A. R., and LaRosa, J. A. (1989) J .

38. Freisheim, J. H., Price, E. M., and Rutnam, M. (1989) Adu. Enzyme Regul. Biol. Chem. 264,1911-1914

39. Price, E. M., and Freisheim, J. H. (1987) Biochemistry 26,4757-4763 29,13-26

40. Sadasivan, E., da Costa, M., Rothenberg, S. P., and Brink, L. (1987)

Biochemistry 28,8249-8254

Res. Commun. 49,123-131

Struct. Dyn. 7,985-1001

41. McHugh, M., and Cheng, Y.-C. (1979) J. Biol. Chem. 2 5 4 , 11312-11318 42. Rothbere. K. G.. Yine. Y.. Kolhouse. J. F.. Kamen. B. A.. and Anderson.

Biochim. Biophys. Acta 925,36-47

R. G. E. (1990) J. &il bioi. 110,637-649

Natl. Acad. Sci. U. S. A . 86,22-26

9 ~ ~ 9 9 n

43. Bailey, C. A., Gerber, L., Howard, A. D., and Udenfriend, S. (1989) Proc.

44. Ferguson, M. A. J., and Williams A. F. (1988) Annu. Reu. Biochem. 57,

45. Doering, T. L., Masterson, W. J., Hart, G. W., and Englund, P. T. (1990)

46. Chitambar, C. R., and Zivkovic, Z. (1989) Blood 74,602-608 47. Lopez, J. A., Chung, D. W., Fujikawa, K., Hagen,. F. S., Papayannopoulou,

T., and Roth, G. J. (1987) Proc. Natl. Acad. Set. U. S. A. 84,5615-5619 48. Mostov, K. E., Kraehenbuhl, J., and Blobel, G. (1980) Proc. Natl. Acad.

Sci. U. S. A . 77,7257-7261 49. Waldman, T. A. (1989) Annu. Reu. Biochem. 58,875-911 50. Lowry, 0. H., Rosebrough, N. J., Farr, A. L., and Randall, R. J. (1951) J.

"" ""

J. Biol. Chem. 265,611-614

Biol. Chem. 193,265-275

Conuersion of Membrane-associated Folate Binding Protein 2353

the conversion of the human membrane-associated folate binding

Documents