Assembly and secretion of recombinant chains of human inter-α-trypsin inhibitor in COS-7 cells

Authors

  • Nathalie Martin-Vandelet,

    1. Laboratoire de Physiopathologie et Génétique Rénale et Pulmonaire, Institut National de la Santé et de la Recherche Médicale, INSERM Unité 295, Faculté de Médecine de Rouen, and IFR 61: physicochimie et biologie des systèmes intégrés.
    Search for more papers by this author
  • Sébastien Paris,

    1. Laboratoire de Physiopathologie et Génétique Rénale et Pulmonaire, Institut National de la Santé et de la Recherche Médicale, INSERM Unité 295, Faculté de Médecine de Rouen, and IFR 61: physicochimie et biologie des systèmes intégrés.
    Search for more papers by this author
  • Jeannette Bourguignon,

    1. Laboratoire de Physiopathologie et Génétique Rénale et Pulmonaire, Institut National de la Santé et de la Recherche Médicale, INSERM Unité 295, Faculté de Médecine de Rouen, and IFR 61: physicochimie et biologie des systèmes intégrés.
    Search for more papers by this author
  • Richard Sesboüé,

    1. Laboratoire de Physiopathologie et Génétique Rénale et Pulmonaire, Institut National de la Santé et de la Recherche Médicale, INSERM Unité 295, Faculté de Médecine de Rouen, and IFR 61: physicochimie et biologie des systèmes intégrés.
    Search for more papers by this author
  • Jean-Pierre Martin,

    1. Laboratoire de Physiopathologie et Génétique Rénale et Pulmonaire, Institut National de la Santé et de la Recherche Médicale, INSERM Unité 295, Faculté de Médecine de Rouen, and IFR 61: physicochimie et biologie des systèmes intégrés.
    Search for more papers by this author
  • Maryam Diarra-Mehrpour

    1. Laboratoire de Physiopathologie et Génétique Rénale et Pulmonaire, Institut National de la Santé et de la Recherche Médicale, INSERM Unité 295, Faculté de Médecine de Rouen, and IFR 61: physicochimie et biologie des systèmes intégrés.
    Search for more papers by this author

M. Diarra-Mehrpour, INSERM, Unité 295, Faculté de Médecine de Rouen, 22 Boulevard Gambetta, F-76183 Rouen Cedex, France. Tel. + 332 35 14 82 80; Fax: + 332 35 14 82 37;
E-mail: maryam.diarra-mehrpour@univ-rouen.fr

Abstract

The inter-α-trypsin inhibitor (ITI) family is a group of structurally related plasma serine protease inhibitors. The ITI family members consist of combinations of mature heavy chains named HC1, HC2, HC3 linked to bikunin (a Kunitz-type protease inhibitor) by a covalent interchain protein–glycosaminoglycan–protein cross-link. The biosynthesis of the ITI family members takes place in the liver. In this report we examine the biosynthesis of these proteins using transient transfected COS-7 cells expressing one or more combinations of human ITI chains. The processing and secretion of α1-microglobulin and bikunin does not require the ITI heavy chains. A small proportion of the H3 chain seems to be processed into the HC3 form in the absence of the other ITI chains. In contrast, the processing of H2 into HC2 needs the presence of the L chain. The COS-7 cells are able to link the HC2 and HC3 heavy chains with bikunin by means of a chondroitin sulfate bridge, and thus to generate 260-kDa ITI-like proteins as well as pre-α-trypsin inhibitor (PαI). However, the maturation of the Hl chain into HC1 and the assembly of HC1 inside multichain proteins may take place according to a mechanism which differs from that of the H2 and H3 chains. These results indicate that the assembly of the constituent chains of the ITI-like proteins and PαI is not dependent on the liver machinery.

Abbreviations
ITI

inter-α-trypsin inhibitor

PI

pre-α-trypsin inhibitor


L

light chain

H

heavy chain

PGP

protein-glycosaminoglycan protein

DMEM

Dulbeccos modified Eagle's medium

MEM

minimum Eagle's medium

PhMeSO2F

phenylmethanesulfonylfluoride


SV40

Simian virus 40.

The inter-α-trypsin inhibitors form a family of structurally related plasma serine protease inhibitors capable of stabilizing specific components of the extracellular matrix [1,2]. These glycoproteins are also involved in pathological conditions such as tumor invasion, metastasis, and arthritis, which reinforces the interest in understanding their biosynthesis [3].

This family is composed of multiple proteins made up of given combinations of polypeptide chains after complex post-translational maturation [4–7]. The mature heavy chains named HC1, HC2, HC3 are covalently linked to bikunin (a Kunitz-type protease inhibitor) by an unusual protein–glycosaminoglycan–protein (PGP) cross-link [8–11]. The subunits are assembled by a chondroitin-4-sulfate chain that originates from bikunin. Three combinations of bikunin and heavy chains have been identified in human serum: (a) inter-α-trypsin inhibitor (ITI) itself, a 220-kDa glycoprotein, composed of HC1, HC2 and bikunin (b) pre-α-trypsin inhibitor (PαI, 125 kDa) composed of HC3 and bikunin and (c) HC2/bikunin (140 kDa) composed of HC2 and bikunin [for reviews, see ref. 12]. The ITI-related proteins are cleaved by chondroitinase into HC chains and bikunin [13]. The structure of the different polypeptides of the ITI family has been determined by sequencing of their corresponding cDNAs [14–17]. The four human ITI chains are encoded in the liver by four distinct genes: one codes for a hybrid polypeptide chain called L, which is cleaved into α1-microglobulin and bikunin; and the three others code for the homologous heavy-chain polypeptides H1, H2, H3 [for reviews, see ref. 18].

The investigations concerning the biosynthesis of the ITI-related proteins in human hepatoma HepG2 cell line showed that these cells synthesized the heavy chain precursors H2, H3, the light chain L and subsequently secreted α1-microglobulin, bikunin, mature H2 and an ITI-like protein composed of two mature HC2 heavy chains linked to two bikunin chains by chondroitin sulfate bridges [6,7]. However, these cells were unable to synthesize H1 and therefore to make HC1 and an ITI similar to the protein identified in human serum. These cells were also unable to secrete PαI, whereas COS-1 cells cotransfected with cDNAs encoding rat H3 and bikunin did secrete a rat PαI-like protein [19]. α1-microglobulin, PαI and ITI were found both in the supernatants of primary cultures of rat [20] and human [7] hepatocytes and in human serum. Free bikunin, however, whilst found in these supernatants, was not found at significant levels in normal plasma. Given the structural complexity of these proteins and the limits of previously used models, we employed an original transfection system involving the expression of recombinant H1, H2, H3 and L chains in the COS-7 cells which do not synthesize these proteins constitutively. This model made it possible to study the biosynthesis of each separate chain and their combinations.

Materials and methods

Materials

The human hepatoma HepG2 cell line (No. 85011430) was purchased from the European Collection of Animal Cell Cultures. African green monkey kidney (COS-7) cells were kindly provided by Stany Chrétien (INSERM U363, Paris). Restriction enzymes, RPMI 1640, Dulbecco's modified Eagle's medium (DMEM), methionine-free minimum essential medium (methionine-free MEM) and NaCl/Pi (154 mm NaCl, 1.5 mm KH2PO4, 3 mm Na2HPO4, pH 7.2) were obtained from Life Technologies. The ProFection Mammalian Transfection System-DEAE-Dextran, pGL2-Basic and pGL2-Promoter were purchased from Promega. pcDNA3 was from Invitrogen. Fetal calf serum, chondroitinase ABC and BSA were from Boehringer. Methionine was purchased from Merck. The labeled products [α-32P]dCTP (3000 Ci·mmol–1) and [35S]l-methionine (1000 Ci· mmol–1) were from Amersham. Rabbit polyclonal antisera directed against human ITI and α1-microglobulin were purchased from Dako and Accurate Chemical & Scientific Corporation, respectively. Immunoprecipitations were performed using Protein A-Sepharose CL-4B from Pharmacia. All chemicals used were of analytical grade.

Cell culture

The cells were propagated at 37 °C under an atmosphere of 5% CO2 in a medium containing 2 mm glutamine and 10% (by vol.) fetal calf serum. The HepG2 cells were cultured in RPMI 1640 and COS-7 cells in DMEM.

Construction of the recombinant expression vector
containing the full length human ITI cDNAs

The human ITI cDNAs were inserted into the expression vector pcDNA3 which contains a multiple cloning site polylinker located downstream from the cytomegalovirus promoter/enhancer and upstream from the bovine growth hormone polyadenylation signal. The constructs are schematized in Fig. 1. The ITI L cDNA (1.2 kb long) extends from 64 bp upstream from the translation initiation codon to 44 bp downstream from the stop codon [14] and encodes both the N-terminal of α1-microglobulin and the C-terminal of bikunin. This fragment was inserted into EcoRI and XhoI restriction sites of pcDNA3 (p-L). The ITI H1 cDNA (2.9 kb long) extends from 6 bp upstream from the translation initiation codon to 58 bp downstream from the stop codon [15]. This fragment was inserted into HindIII blunt and BamHI of pcDNA3 (p-H1). The ITI H2 cDNA (3 kb long) extends from 55 bp upstream from the translation initiation codon to 99 bp downstream from the stop codon [16]. This fragment was inserted into HindIII and EcoRI of pcDNA3 (p-H2). The ITI H3 cDNA encompasses 3 kb from 19 bp upstream from the translation codon to 99 bp downstream from the stop codon [17]. This fragment was blunt ended and inserted into the EcoRV site of pcDNA3 (p-H3). After amplification in the Escherichia coli host strain JM109, all constructs were sequenced using the dideoxynucleotide chain-termination method [21] and found to be identical to previously published sequences.

Figure 1.

Recombinant expression vectors. (A) Cloning of the full-length ITI cDNAs. The L, H1, H2 and H3 cDNAs were inserted under control of the cytomegalovirus (CMV) promoter of the pcDNA3 vector. The poly A tract is provided by the bovine growth hormone polyadenylation signal (bGH poly A). The open boxes represent protein coding regions and the stippled boxes represent untranslated domains. The cloning sites are shown. (B) Schematic representation of the ITI L, H1, H2 and H3 proteins. The black boxes represent signal peptides, gray boxes N-terminal cleavage polypeptides, hatched boxes mature chains and open boxes C-terminal cleavage polypeptides.

COS-7 cell transfection

Ten micrograms of ITI construct or control (vector alone) plasmid DNA were introduced for transient expression in COS-7 cells by the ProFection Mammalian Transfection System-DEAE-Dextran according to Promega's instructions. Briefly, 1 day before transfection the COS-7 cells were seeded at a density of 106 cells in 10-cm dishes. These cells were washed twice with NaCl/Pi. A DNA (18 ng·µL–1) and DEAE-Dextran (28 µL) mixture was made in 540 µL of NaCl/Pi and dispersed over the cells. After 30 min at room temperature, the mixture was removed and the cells were washed twice with DMEM supplemented with 10% (by vol.) fetal calf serum. The cells were incubated for 4 h with a culture medium containing 100 µm chloroquine [22]. The cells were then washed twice with DMEM supplemented with fetal calf serum and incubated for 48 h at 37 °C. The efficiency of the transfection was determined by luciferase activity using pGL2-Basic and pGL2-Promoter vectors according to Promega's instructions.

RNA extraction and Northern blot analysis

The total cellular RNA was extracted 48 h after transfection by the guanidinium isothiocyanate method [23]. Northern blot analysis was carried out as previously described [24].

Metabolic labeling

The cells were incubated 48 h after transfection in methionine-free MEM without fetal calf serum for 1 h at 37 °C. The cells were washed twice with methionine-free, serum-free MEM and incubated in the same medium with 100 µCi·mL–1[35S]l-methionine for 10 min (pulse). The cells were then washed twice and chased for various times with 3.35 mm unlabeled methionine in MEM. The culture medium containing the secreted labeled proteins was recovered at various intervals. In order to prepare cell lysates, transfected COS-7 cells were washed twice with ice-cold NaCl/Pi. The cells were scraped using a rubber policeman into 2 mL of lysis buffer containing 100 mm Tris/HCl pH 7.5, 1% (by vol.) Triton X100, 0.5% (mass/vol.) sodium deoxycholate, 2 mm phenylmethylsulfonylfluoride (PhMeSO2F) and 2 mm benzamidine. The culture media and cell lysates were clarified by centrifugation at 10 000 g and immediately frozen at −70 °C.

Chondroitinase ABC digestion

Prior to treatment with chondroitinase ABC, the sample (500 µL) was incubated overnight at 4 °C in the presence of 20 µL of non-immune rabbit serum and 50 µL Protein A-Sepharose. The Protein A-Sepharose was removed by pelleting. The sample was dialyzed 2 h against 40 mm Tris/HCl, pH 8.0, 40 mm sodium acetate and treated at 37 °C for 2 h with 75 mU·mL–1 chondroitinase ABC in 0.01% (mass/vol.) BSA. To check for non-specific degradation, a parallel sample was incubated in the absence of enzymes.

Immunoprecipitation and SDS/PAGE analysis

The specific antisera were obtained from rabbits. The anti-bikunin serum was obtained by immunization with the human urinary ITI derivative as described [25]. Specific anti-(H1, H2, or H3) sera were obtained by immunization with corresponding peptides produced in a bacterial expression system as described [26]. The specificity of the antisera was investigated by: (a) immunoprecipitation of the translation products of human liver mRNAs; (b) analysis by Western blotting of serum proteins submitted to chondroitin AC lyase digestion according to [13] except that 4 mU of enzyme was used. Cell lysates and culture media were immunoprecipitated by various antisera as in [6]. Briefly, 20 µL of non-immune rabbit serum and Protein A-Sepharose (50 µL) were added to 500 µL of lysates or medium and the mixture was incubated overnight at 4 °C. After collection of the liquid phase by centrifugation, 5 µL of antiserum and new Protein A-Sepharose (20 µL) were added. After overnight incubation at 4 °C, the immunoprecipitates were washed 10 times with a buffer containing 150 mm Tris/HCl pH 7.5, 1 m NaCl, 1% (mass/vol.) sodium deoxycholate, 1% (by vol.) Triton X100 and then twice in 150 mm Tris/HCl pH 7.5. The immunoprecipitates were released from the Protein A-Sepharose by boiling in a sample buffer (0.1 m dithiothreitol, 4% (mass/vol.) SDS, 80 mm Tris/HCl pH 6.9 and 10% (by vol.) glycerol) and analyzed by SDS/PAGE (stacking gel 5%, resolving gel 14%) [27], followed by fluorography. The rainbow-colored protein molecular-mass markers from Amersham (myosin, 200 kDa; phosphorylase b, 92.5 kDa; BSA, 66 kDa; ovalbumin, 46 kDa; carbonic anhydrase, 30 kDa; soybean trypsin inhibitor, 21.5 kDa and lyzozyme 14.3 kDa) were routinely included in the run.

Results

We used a transient transfection system in COS-7 cells to explore a large series of ITI chain combinations and study the assembly and secretion of ITI-related proteins.

In order to assess the transcription of each transfected ITI cDNA in COS-7 cells, comparative Northern blot analyses were performed at high stringency utilizing ITI probes. A single ITI mRNA species (3.4 kb for the heavy chains and 1.5 kb for the light chain) was detected for each cDNA, indicating that the transfected ITI cDNAs were correctly transcribed in COS-7 cells (results not shown).

The biosynthesis of ITI-related proteins was analyzed by immunoprecipitation of radiolabeled cell lysates and culture media, using antisera specific for the individual components. The presence of the PGP cross-link was probed by dissociation with chondroitinase ABC. This property was employed to probe for the PGP-mediated chain assembly of the ITI-related proteins.

Addition of a glycosaminoglycan chain to bikunin in a system expressing the L chain alone

COS-7 cells were transfected with p-L alone to attain transiently high levels of expression. Cells expressing the L chain were labeled with [35S]methionine for 10 min and chased for various times with an excess of unlabeled methionine. At the beginning of the biosynthesis, four main bands were immunoprecipitated by anti-bikunin serum in lysates (Fig. 2, lane 1): the first one (49.5 kDa) exhibited a molecular weight similar to that of the hybrid L chain found in HepG2 lysates (Fig. 2, HepG2) and three faster migrating components (40, 30 and 28 kDa). A smear stretching between 45 and 97 kDa appeared in lysates after 30 min of chase (Fig. 2, lane 2) representing the L chain with GAG. This has also been observed in the cell lysates of primary rat and human hepatocytes [7,20]. The 28-kDa form, the L chain and a heterogeneous polypeptide species ranging from 56 to 70 kDa (Fig. 2, lane 6) were secreted in this system. The proteins detected in the medium with anti-α1-microglobulin serum are a free form of α1-microglobulin (32 kDa) and the L chain (49.5 kDa) (Fig. 2, lanes 8–10). Free α1-microglobulin appeared in the medium, indicating that the processing of the L chain occurs correctly. However, it should be noted that no other α1-microglobulin related products were detected in the medium, suggesting that the diffuse band (57–70 kDa) seems to represent bikunin with glycosaminoglycan (Fig. 2, lane 6). The treatment with chondroitinase ABC caused the disappearance of this diffuse band and the appearance of bikunin without glycosaminoglycan (28 kDa) and a minor band of 22 kDa (Fig. 2, lane 7). The 22-kDa form seems to be generated from the 28-kDa form detected before the treatment. This 22 kDa form also detected in primary rat hepatocytes [20] and in pulse-chase experiments in HepG2 performed with tunicamycin [6] could represent bikunin lacking an N-linked carbohydrate. A chondroitinase-resistant form reacting with anti-bikunin serum as well as with anti-α1-microglobulin serum was observed, confirming the presence of the L chain without the glycosaminoglycan chain (Fig. 2, lanes 7 and 10) in the medium. These results corroborate the evidence that the diffuse band (57–70 kDa) detected in the medium is only constituted of bikunin with glycosaminoglycan. Our results show that ITI heavy chains are required for neither the processing of the L chain nor the secretion of α1-microglobulin and bikunin with glycosaminoglycan. Bikunin with glycosaminoglycan is not a degradation product released during the biosynthesis of ITI-related proteins in the system expressing heavy chains.

Figure 2.

Expression and secretion of
α1-microglobulin-bikunin-related proteins in transfected COS-7 cells.
COS-7 cells transfected with the p-L construct were labeled with [35S]methionine for 10 min then chased for the times indicated. Cell lysates (lanes 1–3) or media (lanes 4–10) immunoprecipitated by anti-bikunin or anti-α1-microglobulin (α1-m) sera were subjected to 14% SDS/PAGE according to Laemmli. The [35S]methionine-labeled media obtained after 4 h of chase were digested with chondroitinase ABC (CABC, lanes 7 and 10) and immunoprecipitated by anti-bikunin or anti-α1-microglobulin sera. Control was performed using HepG2
hepatoma cells pulsed in the presence of [35S]methionine for 10 min and lysates were immunoprecipitated with anti-bikunin serum. The protein molecular-mass standards are indicated on the left. Bikunin and glycosaminoglycan are represented as bik and GAG, respectively.

Processing of H2 in HC2 required the L chain in COS-7 cells

When the H2 chain was expressed alone, the primary translation product was a polypeptide chain with a molecular mass of 100 kDa (Fig. 3A, lane 1). During the chase, the molecular mass of intracellular H2 increases to about 120 kDa (Fig. 3A, lanes 2–4), and then these forms are secreted in the medium (Fig. 3A, lanes 7–8). The secreted forms (120–125 kDa) were not affected by the treatment with chondroitinase ABC (Fig. 3A, lane 9) and were considered to be H2. These data show that HC2 could not be generated in this case.

Figure 3.

Synthesis of the H2 chain and its coexpression with the L chain in transfected COS-7 cells. (A) COS-7 cells transfected with the p-H2 construct alone (B) or with p-H2 and p-L were labeled with [35S]methionine for 10 min then chased for the times indicated. Cell lysates (lanes 1–4) or media (lanes 5–9) were immunoprecipitated with anti-ITI serum and subjected to 14% SDS/PAGE. The [35S]methionine labeled media obtained after 4 h of chase were digested with chondroitinase ABC and immunoprecipitated by anti-ITI serum (CABC, lanes 9). Bikunin and glycosaminoglycan are represented as bik and GAG, respectively.

For the continuation of our study, it should be noted that when the heavy chains were expressed alone, no species could be detected in samples immunoprecipitated with anti-bikunin serum (data not shown).

To study the effects of the coexpression of the H2 and L chains, COS-7 cells were cotransfected with equal amounts of p-H2 and p-L plasmids. The lysates showed the H2 precursor (100 kDa), the L chain (49.5 kDa) and the 28 kDa form immunoprecipitated with anti-ITI serum (Fig. 3B, lane 1). A high-molecular-weight complex (> 200 kDa) with anti-ITI antigenicity was secreted along with H2 and the L chain (Fig. 3B, lanes 7 and 8). The heterogeneous proteins (> 200 kDa) presented immunologic reactivity with anti-bikunin as well as with anti-H2 (results not shown).

The treatment with chondroitinase ABC caused the disappearance of the high-molecular-weight complex and the appearance of HC2 (80 kDa) and bikunin without glycosaminoglycan (28 kDa) (Fig. 3B, lane 9). These observations suggest that: (a) the processing of H2 in HC2 required the presence of the L chain (b) the coexpression of H2 and bikunin does not produce HC2/bikunin but a heterogeneous protein with a high molecular weight. These data indicate that the proteolytic cleavage of the H2 and the carbohydrate attachment for the formation of the PGP cross-link between HC2 and bikunin are carried out simultaneously.

COS-7 cells coexpressing H3 and L chains produce PαI and heterogeneous proteins

When the H3 chain was expressed alone, a major band of 116 kDa was observed in cell lysates at the beginning of the biosynthesis (Fig. 4A, lane 1). During the chase, this band increased in intensity whereas a faint faster migrating band (90 kDa) appeared in lysates (Fig. 4A, lanes 2 and 3). After 4 h of chase, the H3 recombinant chains were secreted in the medium (Fig. 4A, lane 7). The 90-kDa species secreted most likely represents the HC3 form. In the absence of bikunin, only a small amount of the H3 chain seems to be processed into the HC3 form.

Figure 4.

Synthesis of the H3 chain and its co-expression with the L chain in transfected COS-7 cells. (A) COS-7 cells expressing the H3 chain alone or (B and C) coexpressing the H3 and L chains were labeled with [35S]methionine for 10 min and chased for the indicated times. Cell lysates (lanes 1–4) or media (lanes 5–9) were immunoprecipitated with anti-H3 serum (A and B) or with an anti-bikunin serum (C) and analyzed by 14% SDS/PAGE. Digestion with chondroitinase ABC before immunoprecipitation was performed on the media obtained after 4 h (CABC, lanes 9) or after 24 h (CABC, lane 10) of chase. Bikunin and glycosaminoglycan are represented as bik and GAG, respectively.

When the H3 and L chains were coexpressed, the same bands observed in the system expressing the H3 chain alone were detected with anti-H3 serum except for the intensity decrease of the 116 kDa band at 4 h (Fig. 4B, lane 4). After 2 h of chase, three species displaying H3 specificity were secreted in the medium: a heterogeneous protein with a high molecular weight (> 200 kDa), H3 and HC3 (Fig. 4B, lanes 7 and 8). When samples were subjected to immunoprecipitation with anti-bikunin serum, the L chain and a 28-kDa form were observed in lysates (Fig. 4C, lane 1). After 30 min of chase, a trailing smear stretching from the top of the gel to 46 kDa was formed in lysates (Fig. 4C, lanes 2–4). The band (116 kDa) reacting with anti-bikunin serum appeared more distinctly (Fig. 4C, lanes 2–4). In media, the pattern observed after 2 h of chase (Fig. 4C, lanes 7 and 8) with anti-bikunin serum was equivalent to that obtained in lysates after 30 min of chase (Fig. 4C, lane 2). The emergence of heterogeneous proteins (> 200 kDa) displaying reactivity with bikunin and the HC3 chain attests the assembly of the HC3 chain with bikunin. The heterogeneous proteins disappeared after treatment with chondroitinase ABC while HC3 increased (Fig. 4B, lane 9) and bikunin without glycosaminoglycan appeared (Fig. 4C, lane 9). This digestion did not affect the pattern of the H3 chain in COS-7 cells expressing H3 alone (Fig. 4A, lane 10). The 116-kDa form may represent PαI because: (a) no species could be immunoprecipitated with anti-bikunin serum in a system expressing the H3 chain alone (results not shown) (b) it displayed a double immunoreactivity towards H3 and bikunin. The heterogeneous proteins (> 200 kDa) are composed of HC3 and bikunin. The cross-linking of HC3 and bikunin is most likely mediated by the PGP cross-link as it can be dissociated by chondroitinase ABC digestion.

Expression of H1 chain in COS-7 cells

Shortly after the start of the labeling experiment in which the H1 chain was expressed alone, a major band of 100 kDa was observed in cell lysates (Fig. 5A, lane 1). After 4 h of chase, the H1 chain was secreted as a protein of 100 kDa (Fig. 5A, lanes 7 and 8). A labeled polypeptide with a high molecular mass of 240 kDa was also detected (Fig. 5A, lanes 7 and 8) that is an artifact related to non-specific binding to Protein A-Sepharose as already described [6]. These proteins were not affected by the treatment with chondroitinase ABC (Fig. 5A, lane 9).

Figure 5.

Synthesis of the H1 chain and its coexpression with the L chain in transfected COS-7 cells. Transfected COS-7 cells were labeled with [35S]methionine for 10 min and chased for the indicated times. (A) The cells transfected with the p-H1 construct alone: cell lysates (lanes 1–4) or media (lanes 5–9) were immunoprecipitated with anti-ITI serum. (B) the cells co-transfected with the p-H1 and p-L constructs: cell lysates (lanes 1–4) or media (lanes 5–8) were immunoprecipitated with anti-ITI serum. The media obtained after 24 h of chase were digested with chondroitinase ABC (CABC, lanes 9) before immunoprecipitation with anti-ITI (A) or anti-H1 (B). The products were analyzed by 14% SDS/PAGE. Bikunin and glycosaminoglycan are represented as bik and GAG, respectively.

In a system coexpressing H1 and L chains, the lysates contained the H1 precursor (100 kDa) and the L chain (49.5 kDa) as revealed by immunoprecipitation with anti-ITI serum (Fig. 5B, lane 1). In the media, the pattern obtained after 4 h of chase, as revealed with anti-ITI serum, showed no difference from systems in which the H1 chain and bikunin were expressed separately, that is, no additional form was detected (Fig. 5B, lanes 7 and 8). The 240-kDa and 100-kDa proteins reacting with anti-H1 serum were resistant to the treatment with chondroitinase ABC and HC1 was not produced (Fig. 5B, lane 9). In summary, in a system coexpressing H1 and L, H1 was not processed into HC1 despite the expected syntheses of both mRNA and protein. In consequence, the assembly of HC1 and bikunin could not occur.

Expression of the H1 chain is essential to the structural conformation of the HC2 and L complex

We wanted to determine whether the system coexpressing H1, H2 and L chains could generate a protein similar to ITI 220 kDa observed in serum. For the sake of clarity, only the immunoprecipitations of cell lysates performed with anti-ITI serum are shown. In the system coexpressing the H1, H2 and L chains, the correct synthesis of intracellular ITI chains was illustrated by the presence of the heavy chains (100 kDa) and the L chain (Fig. 6, lanes 1 and 2). The immunoprecipitations of cell lysates performed with anti-H1 or anti-H2 serum confirmed the identity of the 100 kDa forms and indicated that the addition of each chain did not modify the synthesis of the others (results not shown). After 4 h of chase, besides the forms obtained in the system expressing each chain alone, a high-molecular-weight complex (260 kDa) appeared in the medium (Fig. 6, lanes 7 and 8). This complex (260 kDa) was detected with anti-H2 serum (Fig. 6, lanes 16 and 17), whereas it was not detected with anti-H1 serum (Fig. 6, lanes 11 and 12). These results suggest that the complex of 260 kDa does not contain H1. The treatment with chondroitinase caused the disappearance of the complex (260 kDa) and the appearance of HC2 (Fig. 6, lane 18). The 240-kDa protein detected with anti-H1 (Fig. 6, lanes 11 and 12) and anti-H2 (Fig. 6, lanes 16 and 17) was resistant to the chondroitinase ABC treatment (Fig. 6, lanes 13 and 18).

Figure 6.

Co-expression and maturation of the
H1, H2 and L chains in transfected COS-7 cells.
Radiolabeled cell lysates (lanes 1–4) and media (lanes 5–18) were immunoprecipitated with anti-ITI, anti-H1 and anti-H2 sera and analyzed by 14% SDS/PAGE. The media obtained after 24 h of chase were digested with chondroitinase
ABC (CABC, lanes 13 and 18) before
immunoprecipitation.

These data suggest that the complex of 260 kDa resulted from the assembly of the HC2 chain and bikunin via a glycosaminoglycan cross-link. Gel electrophoresis revealed that a defined band was obtained from coexpression of H1, H2 and L than from coexpression of H2 and L. This observation led us to consider the synthesis of the H1 chain as essential to the spatial arrangement of the HC2 and bikunin chains in the multichain protein.

The absence of formation of ITI (HC1, HC2 and bikunin) suggested that COS-7 cells would be unable to assemble the H1 chain into a multichain protein for two reasons: (a) a species incompatibility occurs when monkey kidney cells are transfected with human cDNA encoding these proteins; (b) one or more functions specific to liver cells are needed. In order to investigate these possibilities, we transfected the human hepatoma HepG2 cells with the p-H1 construct or the pcDNA3 vector. The biosynthesis of ITI-related proteins was analyzed using specific anti-(H1, H2, H3 or ITI total) sera. The presence of the PGP cross-link was probed by treatment with chondroitinase ABC. Our results showed that synthesis of the recombinant H1 chain in HepG2 cells does not affect the synthesis of the H2, H3 and L chains or of the ITI-like protein. Transfection with the pcDNA3 vector had no effect. A 240-kDa protein was observed in the supernatants of transfected HepG2 cells with each plasmid. This 240-kDa form, observed regardless of the antiserum used, turned out to be an artifact related to non-specific binding to Protein-A-Sepharose [4]. The pattern of synthesis and secretion of the 100 kDa component in the HepG2 cells transfected with the p-H1 construct is similar to that of the H1 recombinant chain expressed in the COS-7 cells. Hence we were unable to detect cleavage of H1 or formation of a multi-chain protein with H1 antigens in either of the cell lines used (results not shown). Hence neither species incompatibility nor cell specific requirements can be invoked to explain our inability to detect cleavage of H1 or formation of the PGP link between the HC1 and bikunin chains in the cell lines used.

COS-7 cells coexpressing human H2, H3 and L chains produce an ITI-like protein composed of HC2, HC3 and bikunin

An ITI-like protein composed of HC2, HC3 and bikunin was found in bovine serum. This protein was as yet unknown in humans. This model could be studied in our system to determine if the assembly of the three human recombinant chains could occur. Only the immunoprecipitations of cell lysates performed with anti-ITI serum are shown. In the system coexpressing the H2, H3 and L chains, the correct synthesis of the intracellular ITI chains was obtained, indicating that the addition of each chain did not modify the synthesis of the others. After 4 h of chase, besides the forms obtained in the system expressing each chain alone, the high-molecular-weight complexes (260 kDa) appeared in the medium (Fig. 7A, lanes 7 and 8). These complexes (260 kDa) were detected with anti-H2 (Fig. 7B, lanes 3 and 4) and anti-H3 sera (Fig. 7B, lanes 8 and 9). These results suggest that the complexes of 260 kDa contain the H2 and H3 chains. The chondroitinase ABC digestion caused the disappearance of high-molecular-weight material and the appearance of bikunin without glycosaminoglycan and the bands of 22 kDa (Fig. 7A, lane 9), HC2 (Fig. 7B, lane 5) and HC3 (Fig. 7B, lane 10). The 240-kDa form described above as reacting with anti-H2 and anti-H3 serum was not affected by the treatment with chondroitinase ABC; therefore it is reasonable to consider that this 240-kDa form is an artifact. These data suggest that the form of 260-kDa results from the assembly of the HC2 chain or of the HC3 chain and bikunin via a glycosaminoglycan cross-link.

Figure 7.

Co-expression of the H2, H3 and L chains in transfected COS-7 cells. Radiolabeled cell lysates (lanes 1–4) or media (lanes 5–9) were immunoprecipitated with anti-ITI serum (A). The media were also immunoprecipitated with anti-H2 serum (lanes 1–5) or with anti-H3 serum (lanes 6–10) (B). The samples were analyzed by 14% SDS/PAGE. The media obtained after 24 h of chase were digested with chondroitinase ABC (A, CABC, lane 9, B, CABC, lanes 5 and 10) before immunoprecipitation.

Expression of the H1 chain is essential in the structural conformation of the HC3 and L complex

As yet, no protein composed of H1, H3 and bikunin has been described. We wanted to investigate whether associations between H1, H3 and bikunin could be accomplished in our system.

In the system coexpressing the H1, H3 and L chains, the synthesis of each intracellular ITI chain was correctly performed. After 4 h of chase, besides the forms obtained in the system expressing each chain alone, a high-molecular-weight complex (260 kDa) appeared in the medium (Fig. 8, lanes 7 and 8). This complex (260 kDa) was detected with anti-H3 serum (Fig. 8, lanes 17 and 18), whereas it was not detected with anti-H1 serum (Fig. 8, lanes 12 and 13). These results suggest that the 260-kDa complex does not contain H1. The treatment with chondroitinase caused the disappearance of the complex (260 kDa) and the appearance of bikunin without glycosaminoglycan, of the 22-kDa band (Fig. 8, lane 9) and of HC3 (Fig. 8, lane 19). It is noticeable that the complex has a better defined appearance than when the H3 and L chains are expressed without the H1 chain. These data suggest that the 260-kDa form results from the assembly of the HC3 chain with bikunin via a glycosaminoglycan cross-link. The spatial arrangement of this complex is well-defined, suggesting that the presence of the H1 chain is essential in the structural conformation of the HC3 chain and bikunin complex.

Figure 8.

Co-expression of the H1, H3 and L
chains in transfected COS-7 cells.
Radiolabeled cell lysates (lanes 1–4) or media (lanes 5–19) were immunoprecipitated with anti-ITI serum (lanes 1–9), with anti-H1 serum (lanes 10–14) or with anti-H3 serum (lanes 15–19). The samples were analyzed by 14% SDS/PAGE. The media obtained after 24 h of chase were digested with chondroitinase ABC (CABC, lanes 9, 14 and 19) before immunoprecipitation. Bikunin and glycosaminoglycan are represented as bik and GAG, respectively.

Co-expression of the four recombinant ITI chains in the
COS-7 cells

When the four chains were co-expressed (results not shown), in addition to the forms obtained in the system expressing each chain alone, a complex of 260-kDa molecular weight with the same antigenic reactivity as those observed in the systems coexpressing two heavy chains (H1 and H2; H1 and H3; H2 and H3) and the L chain is detected. In these conditions, we do not detect a protein composed of HC1, HC2 and bikunin.

These data indicate that COS-7 cells expressing the ITI chains are able to assemble the ITI related complex via a glycosaminoglycan cross-link.

Discussion

ITI-related proteins present a multi-chain structure in which polypeptide chains are covalently joined through a chondroitin sulfate bridge. To investigate the synthesis and maturation of these proteins, we examined the individual or combined expression of the human recombinant H1, H2, H3 and L chains and their subsequent assemblies in COS-7 cells.

We show that COS-7 cells transfected with human L cDNA synthesized α1-microglobulin and bikunin as in human hepatocyte cells [7]. Our data indicate that the ITI heavy chains are not required for the processing and secretion of α1-microglobulin and bikunin. The free recombinant bikunin observed in the medium of the COS-7 cells suggests that the presence of bikunin in the medium of human hepatocyte cells is not the result of degradation during the biosynthesis of ITI-related proteins [7,20]. This runs counter to the fact that free bikunin is not found in normal serum at significant levels. This absence may be due to rapid transfer of this protein from the blood to the tissues as described for plasma clearance of rat bikunin [28]. Our data support the idea that bikunin is secreted into the blood by hepatocytes and cleared by renal glomerular filtration; in which case, bikunin may be the source of urinary trypsin inhibitor immunologically related to ITI [29].

Two α1-microglobulin-bikunin precursors (L chains, 40 and 42 kDa) containing one and two N-linked oligosaccharides were observed in rat hepatocyte cells [20,30] or in the COS-1 cells which express the rat α1-microglobulin-bikunin precursor cDNA [31] whereas only one L chain was observed in our system. Our result is similar to that obtained in primary human hepatocyte cells [7]. This difference of N-glycosylation of in the L chain observed in the human and the rat is reproduced in the recombinant COS cell models. This indicates that the recombinant COS cell models are able to reproduce the N-glycosylation specific to each species.

Despite their homologies [17], the H1, H2 and H3 heavy chains do not exhibit comparable behaviors during their maturation. The processing of H2 into HC2 requires the presence of the L chain in COS-7 cells and is thus in agreement with the suggestion that the release of the C-terminal propeptide is mediated by the chondroitin sulfate chain of bikunin [9]. Moreover, the absence of free HC2 in a system without bikunin suggests that the proteolytic cleavage of the H2 and the glycosaminoglycan attachment for the formation of the PGP cross-link between HC2 and bikunin are carried out simultaneously. On the other hand, the observation of a small quantity of HC3 in the absence of bikunin, a phenomenon never observed for H2, even during long labeling times (results not shown), led us to consider the possible existence of an assembly mechanism of the HC3 and bikunin chains that differed from that of HC2 and bikunin. The two-chain (HC/bikunin) form is only observed in the system which coexpresses the H3 and L chains. The absence of production of the HC2/bikunin form in all models of biosynthesis investigated to date [6,7] suggests that HC2/bikunin in plasma might correspond to a degradation product. Except for H2, our results indicate that the synthesis and the maturation of each chain are not interdependent. The different behavior of each of the heavy chains supports the assumption of a function specific to each chain.

The systems coexpressing the L chain and only one H2 or H3 chain produce heterogenous proteins (> 200 kDa). However, the systems coexpressing the L chain and at least two H2 or H3 chains give rise to a protein of 260 kDa resulting from the assembly of two heavy chains linked to bikunin via a glycosaminoglycan cross-link. These results indicate that the presence of at least two different heavy chains is indispensable to the ITI-like protein structural conformation.

The absence of ITI (HC1, HC2 and bikunin) formation in the COS-7 and HepG2 cells suggests that the maturation of H1 into HC1 and the assembly of HC1 in a multichain protein may involve a mechanism which differs from that governing the assembly of the H2 and H3 chains.

Our results lead us to conclude that: (a) ITI heavy chains are not required for the processing and secretion of α1-microglobulin and bikunin. However, the processing of H2 into HC2 does require the presence of the L chain (b) COS-7 cells are able to assemble the H2 or H3 heavy chains with bikunin by means of a chondroitin sulfate bridge and thus to generate 260-kDa ITI-like proteins and PαI. This assembly is not dependent on the liver machinery.

Acknowledgments

We thank S. Chrétien for the COS-7 cells and for a helpful discussion. We are indebted to V. Norris for careful reading of the manuscript. We thank F. Bonnet for technical assistance and N. Porchet for secretarial assistance. N.M-V. is the recipient of a doctoral fellowship from the Ministère de la Recherche et de la Technologie (MRT) and from the Association de Recherche contre le Cancer (ARC).

Footnotes

  1. Enzymes: chondroitinase ABC (EC 4.2.2.4)

Ancillary