Microglobulin by a Rat Thymic Epithelial Cell Line That Expresses MHC Class I Heavy Chain

Overproduction and secretion of /^-microglobulin by a rat thymic epithelial cell line that expresses MHC class I heavy chain

CATHERINE DARGEMONT1*, DOMINIQUE DUNONH, JEAN SALAMERO2, MARIE-ANGE DEUGNIER1, JEAN DAVOUST2 and JEAN PAUL THIERY1

1Laboratoire de Physiopathologie du Developpement, CNRS URA 1337 and Ecole Normale Supeneure, 46 rue d'Ulm, 75230 Paris Cedex 05, France 2Centre d'Immunologie de Marseille-Luminy, Case 906, Marseille Cedex 09, France • Author for correspondence t Present address: Basel Institute for Immunology, 4058 Basel, Switzerland

Summary epithelial cell line), all three of these cell lines Major histocompatibility complex (MHC) class I expressed the same amount of membrane and antigens are constituted of dimers consisting of a intracellular MHC class I heavy chain. These data peripheral light chain, ftj-microglobulin(/J2m ) and a are thus consistent with a constitutive secretion of transmembrane heavy chain whose cell surface /32m dependent upon an overexpression of MHC expression depends on its assembly with /?2m. In class I light chain as compared to the heavy chain. contrast, soluble p2m can be secreted in the absence The amount of /J2m mRNA and the ratio of /32m of heavy chain expression. The presence of /J2m in versus MHC class I heavy chain transcripts were medium conditioned by a rat thymic epithelial cell higher in IT45 than in IT26 cells, indicating that line, IT45-R1 (IT45) prompted us to investigate overexpression of /J2m in IT45 cells could be due to an whether /32m could be secreted by cells that express enhanced level of /J2m mRNA. These results demon- MHC class I antigens. IT45 cells produce three to five strate that /52m can be oversynthezdsed and conse- times more /J2m in the culture supernatant than quently secreted by cells that express MHC class I another rat thymic epithelial cell line, IT26-R21 heavy chains. It may thus suggest that p2m could be (IT26). The IT45 cell line exported p2m through a secreted differentialy at specific body sites, which constitutive pathway of secretion, as indicated by the remain to be determined. kinetics of production and localization of intracellular /32m. Although cells from the IT45 cell line expressed a much higher amount of /J2m as compared to IT26 and NBT II cells (a rat bladder Key words: /32m, secretion, MHC class I antigens.

Introduction Furthermore, the small intestine of these animals fails to bind immunoglobulin G (Zijlstra et al. 1990). ^2-microglobulin (/32m) is the invariant light chain Several unexpected biological activities of free /32m 3 (12xlO Mr) of the major histocompatibility complex have been described. /32m produced by synovial fibroblasts (MHC) class I antigens present on virtually all cells (Grey stimulated with phorbol esters can induce collagenase et al. 1973). Association with /32m is generally required for synthesis in these cells (Brinckerhoff et al. 1989). /32m can the transport of class I heavy chain from the endoplasmic also increase the number of insulin-like growth factor I reticulum to the cell surface (Ploegh et al. 1979; Arce- transcripts and polypeptides as well as their receptors in Gomez et al. 1979) and /32m additionally assists in bone cell cultures (Centrella et al. 1989). These data stabilizing the heavy chain conformation recognized by suggest that /32m could be involved in the modulation of alloantisera (Lancet et al. 1979). /S2m has also been connective tissue breakdown and bone remodeling. On the reported to be associated with other proteins of the other hand, we advanced the hypothesis that /32m might immunoglobulin superfamily including an Fc receptor on be one of the thymic chemotactic factors involved in the intestinal epithelial cells of neonatal rats that is involved migration of hemopoietic precursors from their site of in the transcytosis of immunoglobulin G (Simister and emergence, the bone marrow, to their site of differen- Mostov, 1989) and CD1, a cortical thymocyte marker tiation, the thymus. This suggestion was based on the (Amiot et al. 1986). It has been shown recently that mice observation that /32m originally produced by the rat homozygous for a /32m gene disruption do not express class thymic epithelial cell line IT45-R1 (IT45) (Itoh et al. 1982) I molecules (except a small amount of H-2D) and is able to induce the oriented migration of immature consequently lack CD4~CD8+ cytotoxic T cells (Zijlstra et lymphoid rat bone marrow cells in vitro (Imhof et al. 1988; al. 1989; Zijlstra et al. 1990; Roller et al. 1990). Deugnier et al. 1989; Dargemont et al. 1989). Journal of Cell Science 98, 559-566 (1991) Printed in Great Britain © The Company of Biologists Limited 1991 559 In adults, /32m is present at high concentration in body pulse-chase experiments, subconfluent cells in 26 cm2 tissue fluids (Berggard and Beam, 1968) and elevated serum culture dishes were incubated for 1 h in MEM lacking methionine levels of /32m are found in pathological conditions such as and cysteine, 0 % serum and 5 mM Hepes, then labeled for 1 h at 37°C with 200/(Ci of PSJmethionine (>1000Cimmol~1; Amer- neoplastic and inflammatory states (Child and Kushwaha, 1 sham) and 200 //Ci of PSJcysteine (>1000 Ci mmol" ; Amersham) 1984). It has been suggested that /52m could be released in the same medium. After washing with DMEM, cells were during cell death (Child and Kushwaha, 1984) or through incubated in DMEM, 1 mM cysteine, 1 mM methionine, 0% serum membrane turnover (Creswell et al. 1974) but a constitut- and 5 mM Hepes for 0, 1, 2 or 5 h. The cells were washed in PBS, ive secretion of /32m has never been demonstrated, harvested with a rubber policeman and extracted for 30min at although /32m production has been described for lymphoid 4°C in lml 1% Triton X-100, 2mM EDTA, 150 mM NaCl and cells (Bernier and Fanger, 1972; Poulik and Bloom, 1973; 30 mM Tris-HCl (pH8.6). Insoluble material was removed by Hiitteroth et al. 1973). The secretion of/32m at specific body centrifugation at 15 000 gfor 15minat4°C. Cell lysates or culture supernatants were precleared by the addition of rabbit anti- sites, which remain to be defined, could explain its mouse IgG antibodies (Fc fragment specific; Jackson) for 1 h and presence in body fluids. It is clear that human /32m then a protein A-Sepharose (Pharmacia) suspension (1.5 g per translated in Xenopus laevis oocytes, in the absence of 10 ml) containing 2% Triton X-100, 150 mM NaCl, 2mM EDTA, MHC class I heavy chains, is secreted into the medium 30 mM Tris-HCl (pH 8.6). The precleared lysates were incubated (Severinsson and Peterson, 1984). These data prompted us overnight at 4°C in the presence of anti-rat, -mouse and -human to investigate whether this protein can be secreted by cells MHC class I antibodies and rabbit anti-human /32m antibodies that expressed MHC class I heavy chain, since MHC class I (Serotec, UK), which are known to crossreact with rat /32m (Simister and Mostov, 1989). The lysates were subsequently antigens are present on virtually all cells. reacted with rabbit anti-mouse IgG antibodies for 1 h at 4°C. The In this study, we used two thymic epithelial cell lines precleared supernatante were incubated overnight at 4°C with IT45 and IT26-R21 (IT26; Itoh, 1979) to investigate the rabbit polyclonal anti-human /32m antibodies. Immune complexes production mechanisms of /32m. IT45 cells exported to the were precipitated using protein A-Sepharose (Pharmacia) for 1 h culture supernatant three times more /32m than IT26 cells. at 4°C. After centrifugation (3min, 15 000g) the pellets were The accumulation of /32m in the IT45 conditioned medium extensively washed and the immune complexes were dissociated was due to an active secretion of this protein that occurred by heating with sample buffer under reducing conditions. SDS-PAGE analysis of the samples was carried out on 10 % to in a constitutive manner. IT45 cells overexpressed /32m as 17.5% gradient gels that were processed for fluorography with compared to IT26 cells, whereas the two cell lines Amplify (Amersham) prior to drying and autoradiography. expressed the same amount of both membrane and intracellular MHC class I heavy chain. The overexpression of /32m in IT45 cells correlated with an elevated Preparation of fS2m from conditioned media amount of /32m mRNA. A high level of /32m mRNA was Conditioned media from IT45 and IT26 cells were prepared by also found in stromal cells from rat thymus but not in incubating cells with serum-free DMEM for 48 h at 37 °C. Then, thymocytes. the supernatants were fractionated on reverse-phase HPLC (TSK ODS-120T, 5 mm, lml min"1, LKB). Proteins were eluted from the column with a gradient of 30% to 50% acetonitrile, 0.1% trifluoroacetic acid for 30 min. Fractions (1 ml) were immediately Materials and methods lyophilized, reconstituted, analyzed on a 15% SDS-PAGE gel (reducing conditions) and stained using a silver nitrate method Reagents (Dargemont et al. 1989). Rabbit polyclonal anti-human /92m and mouse monoclonal anti- rat MHC class I antibodies (Ox-18 and MN4-91-6) were obtained Surface labeling of MHC class I antigens from Serotec (Oxford, UK). Rabbit polyclonal antibodies to After fast trypsinization with a solution containing 0.05 % (w/v) human HLA and mouse monoclonal antibodies to class I H2: 20 trypsin and 0.02 % (w/v) EDTA, cells were incubated for 1 h at 4°C -8.4S (H-2Kd specific; Ozato and Sachs, 1981), T1.4C (H-2Dd d with 50 ml of a mouse monoclonal antibody against rat MHC class specific; Figueroa et al. 1981) and 28-14.8S (H-2L specific; Ozato I (Ox-18 used at 1:100). Cells were then labeled with an FITC- et al. 1980) were kindly provided by Dr F. Lemonnier. FrTC- coupled goat anti-mouse Ig antibody (1:100, 30 min, 4°C). conjugated goat anti-mouse IgG was purchased from Nordic Controls were done with second step reagent only. Cells were then (Tilburg, The Netherlands) and rabbit polyclonal anti-mouse IgG, Fc fragment specific, from Jackson Laboratory (West Grove, fixed overnight in a 4 % formaldehyde solution in PBS at 4°C and USA). Rat /32m cDNA (Mauxion and Kress, 1987) and rat class I analyzed by flow fluorometry with a FACScan (Becton Dickinson, MHC heavy chain cDNA (RT16 clone; Mauxion et al. 1989) were CA, USA). Fluorescence intensity was measured with logarithmic generous gifts from Dr M. Kress. amplification. Lysis software was used to construct fluorescence histograms with data collected from 3xlO3 cells.

Cell culture Immunostaining of f$2m and confocal laser scanning IT45 and IT26 cell lines were generously provided by Dr T. Itoh microscopy (Department of Anatomy, Tohoku University School of Medecine, Sendai, Japan; Itoh et al. 1982; Itoh, 1979). The NBT H cell line, Cells were seeded onto coverslips placed in 24-well plates and originally established by Toyoshima et al. (1971) was obtained incubated overnight in DMEM, 10% fetal calf serum (Flow from Dr M. Mareel (Laboratory of Experimental Cancerology, Laboratories). After washing with 0.2 % gelatin-PBS and PBS, University Hospital, Ghent, Belgium). These cells were grown at cells were fixed for 15 min in 2 % formaldehyde, 0.1 % glutaraldehyde in PBS (Munro and Pelham, 1987). The cells were washed 37 °C in 5 % CO2 and in standard medium (DMEM supplemented with glutamine, antibiotics and 10% heat-inactivated fetal calf three times in PBS and then incubated twice for 10 min with 0.1 M serum). glycine. The cells were extracted for 6 min with 1 % Triton in PBS, and then treated for 10 min with 1 mgml"1 sodium borohydrate. Primary antibody was applied for 20 min, followed by three Radiolabeling and immunoprecipitation washes in PBS, a 20 min incubation with FITC-conjugated goat For long-term labeling studies subconfluent cells were incubated anti-rabbit IgG (1/100 in 0.2% gelatin-PBS) and three final for 12 h at 37 °C in 75 cm2 tissue culture dishes with 200/tCi of washes with PBS. The coverslips were mounted in Moviol [^Slmethionine (209mCimmor1; Amersham) and 200/jCi of (Hoechst GmbH, Frankfurt, FRG) and viewed with a confocal [^SJcysteine (144mCimmol~1; Amersham) in MEM lacking laser scanning microscope. Confocal laser scanning microscopy methionine and cysteine, 0% serum and 5mM Hepes. For (Shotton and White, 1989; White et al. 1987; Wijnaendts van

560 C. Dargemont et al. Resandt et al. 1985) was performed using a Wild Leitz CLSM B instrument based on a Leitz orthoplan microscope interfaced with an argon ion laser (model 2020; Spectra-Physics) adjusted at a b 488 nm. We have used a 40xNPL, fluotar objective with a numerical aperture of 1.3 to detect the fluorescence emission from the FITC-coupled antibodies (A em.>530nm). Electronic zooming was performed by reducing the amplitude of two angular displacements of the scanning mirror. Each image was averaged by a scan lasting 0.5 s. A focal series of 16 horizontal sections % vertically spaced every 0.5-2/an were collected from each specimen. There was little appreciable bleaching of the specimens (less than 10% of the total fluorescence). The 8 bit encoded I 512x256 pixel images were represented using a home-made look- up table that was translated into a rainbow color scale. Images together with the linear pseudo-color scales were photographed using a Freeze Frame system from Polaroid.

Northern blots Total RNA from organs or cell lines was prepared by a modified -•12.4 guanidium/CsCl centrifugation method (MacDonald et al. 1987). B2m. Cells were homogenized in 10 volumes of 4 M guanidium thiocyanate, 0.05 M sodium acetate, 2 HIM EDTA and 1 M 2-mercaptoethanol. Solid CsCl was added to reach a concentration -74 of 4.5 M. Homogenates were layered onto a cushion of 6.7 M CsCl (density 1.82 gml"1), 0.05M sodium acetate, lmM EDTA. After an overnight centrifugation at 40 000 revs min"1 at 20 °C in a 70.1 Ti rotor (Beckman), the RNA band in the CsCl gradient was recovered with a syringe and ethanol-precipited. About 5mg of Fig. 1. Production of /32m by IT45 or IT26 cells. (A) Serum-free total RNA from the different organs was heated at 60 °C in 50 % media conditioned by IT45 or IT26 cells were purified by formamide, 3-(Ar-morpholine)propane sulfonic acid buffer (pH7; reverse-phase HPLC. Fractions eluted with 30 % to 50 % 0.2 M 3-(iV-morpholine)propane sulfonic acid, 50mu sodium acetonitrile were collected, lyophilized, reconstituted in acetate and 10 mM EDTA) and then electrophoresed in a 1 % distilled water and analyzed on 15 % SDS-PAGE (under agarose gel containing 6 % formaldehyde in the running buffer. reducing conditions) stained with silver nitrate. Using this RNAs were blotted onto nylon membrane (Hybond N, Amer- elution gradient, /32m was found in only one fraction. The sham). The probes used to hybridize these blots were rat /32m microsequence of the /32m band eliminated the possibility of cDNA (Mauxion and Kress, 1987), rat class IMHC antigen cDNA contamination by other proteins (Dargemont et al. 1989). Lane (RT16 clone; Mauxion et al. 1989) and rat glyceraldehyde a: fraction eluted from the IT45 medium. Lane b: fraction phosphate dehydrogenase cDNA (Fort et al. 1985). These probes eluted from the IT26 medium. (B) Subconfluent monolayers of were labeled by random priming (random priming kit, Boehr- IT45 (lane a) and IT26 (lane b) cells were radiolabeled for 12 h. inger) with [32P]dCTP. Filters were hybridized with the probe at Equivalent amounts of labeled conditioned media were 42°C in 50% formamide, 5xstandard saline/phosphate/EDTA immunoprecipitated with /S2m polyclonal antibodies and (0.18 M NaCl, 10 mM sodium phosphate, pH7.7, lmin EDTA), analyzed by SDS-PAGE on 10 % to 17.5 % gradient gels. Just 2xDenhardt's solution (0.2% polyvinylpyrrolidone, 0.2% Ficoll, one part of the gel is shown here, since no protein except /32m 0.2 % BSA), 5 % dextran sulfate and 20 mg of salmon sperm DNA could be observed. per ml. Blots were washed in O.lxSSC (0.15M NaCl, 15mM sodium citrate, pH7.0), 0.1% SDS at 55°C and were finally exposed to Kodak X-OMAT film for 1-3 days. The production of p2m by IT45 cells resulted from active secretion and not from release during cell death Either secretion or release during cell death could explain Results /32m production in culture supernatants. To distinguish between these two possibilities, the kinetics of /32m /32m is overproduced by the IT45 cell line production in the IT45 and IT26 cell culture supernatants IT45 and IT26 cell lines were established from the thymus were analyzed. Culture supernatants were collected 1, 2 or of the adult rat by Itoh (Itoh et al. 1982; Itoh, 1979). Both 5 h after a 1 h metabolic labeling of IT46 and IT26 cells cell lines displayed typical epithelial characteristics such with [^SJcysteine and [35S]methionine. /32m was then as the presence of desmosomes and keratin filaments. /32m immunoprecipitated from the conditioned media with was purified from both serum-free IT45 and IT26 48 h rabbit polyclonal anti-human /32m antibodies. As shown in conditioned media using reverse-phase HPLC. The Fig. 2 (A, lanes a-d, and B), the amount of labeled /32m amounts of /32m present in these conditioned media were produced in the culture supernatant of IT45 cells was determined using SDS-PAGE stained by the silver nitrate maximal after a 2h chase. The level of /32m did not method (Fig. 1A). These experiments show that /32m was increase after a 5h chase, indicating: (1) that /S2m present in a higher amount in the cell supernatant of IT46 synthezised during a lh period was already exported 3 compared to that of IT26. Indeed, the 11 x 10 MT band was within 2 h; and (2) that cell death was not involved in the constituted of pure rat /32m, as indicated by its microse- production of /32m by IT45 cells. IT26 cells produced /32m quence analysis (Dargemont et al. 1989). This difference at a lower rate than IT45 cells; after a 2 h chase, the /32m could be quantified by immunoprecipitation of /32m in concentration was 4.5 times lower in IT26 than in IT45 serum-free media conditioned by either IT45 or IT26 cells conditioned medium (data not shown). These results are after a 12 h radiolabeling. Scanning of fluorograms showed consistent with a secretion of /32m by IT45 cells and, that under these conditions IT45 cells produced at least moreover, the kinetics suggest that /32m is secreted via a three times more /S2m than IT26 cells (Fig. IB). constitutive pathway.

Overexpression and secretion of fy-microglobulin 561 istic of that found for antigens localized throughout the abed secretory pathway with intense labeling in the endoplasmic reticulum. Intensely stained dots could also be observed in lamellipodia of some IT45 cells but the significance of such dots remains to be determined. This experiment did not provide any evidence for the presence of secretory granules containing /32m. The diffuse intracellular localization of /32m supports the hypothesis of a constitutive pathway for the secretion of this protein in the IT45 cell line. B2m Secretion correlates with an overexpression of p2m in IT45 cells /S2m is the common small subunit of the MHC class I molecule (Grey et al. 1973) and is required for the transport of MHC class I heavy chain from the endoplas- B mic reticulum to the cell surface (Ploegh et al. 1979; Arce- Gomez et al. 1979). In contrast, soluble /32m can be secreted in the absence of the heavy chain expression (Severinsson and Peterson, 1984). It follows that the most likely hypothesis to explain the secretion of /32m by cells that express MHC class I heavy chain proposes overexpression of /32m as compared to the amount of heavy chain. To test such an hypothesis, an immunofiuorescence study was used to compare the cell surface expression of MHC class I in three epithelial cell lines: two thymic epithelial cell lines, IT45 and IT26, and NBTII, a rat bladder carcinoma- derived cell line. Flow fluorometry analysis indicated that the intensity of staining with a monoclonal antibody Fig. 2. Kinetics of the /32m production by IT46 cells. Culture against rat MHC class I, Ox-18, was identical for IT45 and supernatants were collected from 0-5 h after 1 h metabolic IT26 cells and slightly lower in NBTII cells (Fig. 4A, lanes labeling of subconfluent IT45 cells and immunoprecipitated a, b, c). Although the cell surface expression of MHC class I with anti-|62m polyclonal antibodies. (A) Immunoprecipitates was identical in IT45 and IT26 cells, it was also necessary from supernatants of IT46 cells chased for 0 (lane a), 1 (lane to measure the total amounts of /32m and MHC class I b), 2 (lane c) and 5 h (lane d) were analyzed by SDS-PAGE on heavy chain in whole cells. After labeling these three 10% to 17.5% gradient gels. (B) Scanning of gel fluorograms 35 was used to determine peak areas of immunoprecipitated /i2m epithelial cell lines with [ S]cysteine and [^SJmethionine from IT45 supernatants as a function of time. for 12 h, /32m and MHC class I heavy chain were immunoprecipitated with rabbit polyclonal antibodies against human /32m and human HLA, a monoclonal antibody against rat MHC class I (MN4-91-6) and The intracellular distribution of p2m indicates a monoclonal antibodies against mouse H-2Kd (20-8.4S; constitutive secretion pathway for this protein Ozato and Sachs, 1981), H-2Dd (T1-4C; Figueroa et al. The intracellular localization of/32m in IT45 and IT26 cell 1981) and H-2Ld (28-14.8S; Ozato et al. 1980). The lines was investigated to determine whether IT45 cells monoclonal antibody MN4-91-6 is the only available contained specialized structures for /32m storage, such as antibody that is able to immunoprecipitate rat MHC class secretory granules. Cells were fixed with 2% formal- I antigens. Since it is directed only at the products of the dehyde, 0.1% glutaraldehyde, a procedure described as RTI-A locus, it was necessary to use such a mixture of preserving the structure of intracellular compartments antibodies to immunoprecipitate the largest amount of (Munro and Pelham, 1987), and then stained with MHC class I molecules. As shown in Fig. 4B the amount of polyclonal antibodies that recognize both free and bound MHC class I heavy chain in whole cell extracts of the three /32m. The stained cells were visualized with a confocal cell lines was in good agreement with their cell surface laser scanning microscope to provide good spatial resol- expression: peak areas determined by fluorogram scan- ution of the intracytoplasmic immunoreactivity. Single- ning were 7.38, 7.5 and 4.19 mm2 for IT45, IT26 and NBTII point laser illumination combined with single-point cells, respectively. In contrast with the case of MHC class I detection ensures a very strong rejection of the out of focus heavy chain expression, the IT45 cell line expressed five fluorescent light and allows direct measurement of the times more /32m than the two other cell lines. Indeed, the fluorescence emitted from 0.5—2 jsm thick optical sections ratios of labeled /32m/class I heavy chain measured by (Shotton and White, 1989; White et al. 1987; Wijnaendts fluorogram scanning were 0.53, 0.1 and 0.16 for IT45, IT26 van Resandt et al. 1986). This principle was used here to and NBTH cells, respectively. These different amounts of obtain serial sections from IT45 cells. These cells uni- immunoprecipitated /52m correlated with the relative formly exhibited a bright ring of immunoreactivity around intensity of intracellular /32m staining in IT45 and IT26 the nucleus and a diffuse one throughout the cytoplasm cells, confirming that the excess of/32m immunoreactivity (Fig. 3). A similar distribution of /32m was obtained with observed in IT45 cells was not associated with MHC class I IT26 cells but /32m immunoreactivity was about five times heavy chain. Thus, secretion of /32m by IT45 cells that lower in these cells than in IT45 cells (data not shown). express MHC class I antigens is related to an overexpres- The observed pattern of /32m distribution was character- sion of intracellular /32m.

562 C. Dargemont et al. Fig. 3. /32m staining of IT45 cells viewed with the Confocal Scanning Laser Microscope. IT45 cells were fixed in 2 % formaldehyde, 0.1 % glutaraldehyde and processed for immunofluorescence using a polyclonal antibody against /32m as the primary antibody and FITC-conjugated anti-mouse IgG as the secondary antibody. Two optical sections collected from the same IT45 cell with a Confocal Scanning Laser Microscope are represented here. The fluorescence signal is encoded in blue, green yellow and red. Red is the color corresponding to the most intense signal. Note the bright ring of fluorescence around the nucleus and the diffuse emission from the cytoplasm: this staining pattern corresponds to the immunoreactivity found for the endoplasmic reticulum network. Scale: 30 (imx 30 /an per image.

"h B a b c

69 • 3 •-class I 46 • heavy chain

30 • Relative fluorescence

Fig. 4. Expression of MHC class I heavy chain and /32m in three epithelial cell lines: IT45, IT26 and NBT II. (A) Surface expression of MHC class I antigens. MHC class I expression was determined by indirect immunofluorescence staining with the Ox-18 21 • monoclonal antibody. Controls, shown as white histograms, were stained with the secondary FITC-conjugated antibody only. Fluorescence intensity distributions were obtained by flow cytometry. (a) IT45; (b) IT26; (c) NBTH. (B) subconfluent monolayers of IT45, IT26 and NBT II cells were radiolabeled and an equivalent amount of 14 • labeled whole cell extract was immunoprecipitated with a mixture of antibodies against human, mouse and rat MHC class I and a /J2m polyclonal antibody. Immunoprecipitates from IT45 (lane a), IT26 (lane b) and NBT II Gane c) were analyzed by SDS-PAGE on 10% to 17.5% gradient gels.

The high level expression of /52m in the IT45 cell line is correlated with a high level of fi2.m mRNA The high level of expression of /92m in the IT45 cell line could be due to a more efficient translation, a more active 12 3 4 transcription and/or a better stability of the /32m 82m transcripts or polypeptides. To discriminate between these 2.10 • hypotheses, Northern blot experiments were performed to analyze and compare the quantity of /32m mRNA in IT45 and IT26 cells. A rat /32m cDNA probe was used to hybridize mRNAs and the measured concentrations of /32m mRNA were normalized by probing RNAs with a rat 1.05, glyceraldehyde phosphate dehydrogenase cDNA. As shown in Fig. 5 (lanes 1 and 2) the level of/32m mRNA was 0.75, three times higher in IT45 than in IT26 cells, suggesting a higher level of transcription of the /32m gene or higher stability of/S2m mRNA in IT45 cells. Moreover, this result indicated a good correlation between polypeptide and transcript levels in IT45 and IT26 cells. However, it Class I appeared from the hybridization profile obtained with a 1.90*. rat MHC class I cDNA that the quantity of MHC class I heavy chain mRNA was also lower in IT26 cells even though the expression of the corresponding protein was similar in the two cell lines (Fig. 5, lanes 1 and 2). Northern blot experiments were also performed on the GAPDH mRNAs isolated from crude preparations of thymocytes and thymic stroma after elimination of thymocytes from a 3-week-old rat. The level of /32m transcripts was 4.8 times higher in the stroma preparation than in thymocytes (Fig. 5, lanes 3 and 4). Furthermore, the ratios of /?2m Fig. 5. Northern blot analysis of the /32m and MHC class I mRNA/MHC class I heavy chain mRNA obtained from heavy chain in IT45 and IT26 cells, and in thymic stroma or in thymocytes. RNA from cell lines or organs were prepared, the IT45 cell line and thymic stroma were comparable and electrophoresed in a 1 % agarose gel, blotted onto nylon were 1.5-2 times higher than those obtained from the membrane and hybridized with rat /J2m cDNA (Mauxion and IT26 cell line or thymocytes. This experiment suggested Kress, 1987) or rat MHC class I cDNA (RT16 clone; Mauxion et that different mechanisms control the metabolism of /32m ai. 1989). They were also hybridized with rat glyceraldehyde and class I heavy chain mRNA in IT45 cells and thymic phosphate dehydrogenase cDNA to normalize the mRNA stroma. concentrations. Lane 1, IT45 cells; lane 2, IT26 cells; lane 3, thymic stroma; lane 4, thymocytes.

Overexpression and secretion of fe-microglobulin 563 Discussion present in IT46 cells (Morello et al. 1985; Kimura et al. 1986). Our aim in these experiments was to test whether /32m This report demonstrates that /32m can be secreted in could be secreted by cells that express MHC class I vitro by thymic epithelial cells. Other observations antigens. Indeed, we have described (Dargemont et al. suggest that thymic stromal cells could secrete this 1989) the presence of ^2m in the medium conditioned by protein. Indeed, Northern blot analysis indicated that the the rat thymic epithelial cell line, IT45. Similarly, /32m amount of /32m transcripts and the ratio of /32m/heavy production has already been described for lymphoid cells chain mRNA were higher in rat thymic stroma than in (Bernier and Fanger, 1972; Poulik and Bloom, 1973; thjrmocytes. These observations suggest that regulation of Hutteroth et al. 1973) and phorbol ester-stimulated the /32m mRNA amounts might be similar in thymic fibroblasts (Brinckerhoff et al. 1989). Such a /32m accumu- epithelium and IT45 cells. On the other hand, during lation in cell culture supernatants has generally been chicken embryogenesis, peaks of /32m transcripts and free considered to result either from a /32m release during cell /32m synthesis, unrelated to the amounts of MHC class I death (Child and Kushwaha, 1984) or through membrane heavy chain mRNAs, can only be detected in the thymus turnover (Cresswell et al. 1974), but the hypothesis of an concomitantly with the second wave of thymus coloniz- active /32m secretion by cells that express MHC class I ation by T cell precursors (Dunon et al. 1990). These data heavy chains has never been demonstrated. In contrast, in suggest that chicken thymus could secrete /32m and, the absence of MHC class I heavy chain, /32m is secreted in consequently, that this differential production of /32m the culture supernatant (Severinsson and Peterson, 1984). during embryogenesis would influence chemotactic mi- In our system, the kinetics of /32m production by the gration of T cell precursors. thymic epithelial cell line, IT46, showed evidence of an active secretion rather than a release during cell death. This work was supported by the CNRS, the Fondation pour la Moreover, they suggested a constitutive pathway for the Recherche Medicale and the Association de Recherche sur la /32m secretion, an hypothesis further supported by the Polyarthrite. CD. is a fellow of the Association pour la Recherche contre le Cancer. We thank Drs B. Imhof, T. Rabilloud, B. Boyer, intracellular localization of /32m: /32m could not be G. Levi and B. Sather very much for helpful discussions and detected in any specialized structures involved in regulat- critically reading the manuscript, and M. Denoyelle for technical ory secretion pathway. assistance. Although the amount of /32m immunoprecipitated in whole cell extracts was five times higher in IT45 than in References IT26 and NBTH cells, cell surface and intracellular expression of MHC class I heavy chain was identical in AMIOT, M., BERNARD, A., RAYNAL, B , KNAPP, W., DESCHILDRE, C. AND IT45 and in IT26 cells and only slightly lower in NBTII BOUMSELL, L. (1986). Heterogeneity of the first cluster of differentiation: characterization and epitopic mapping of three CD1 cells. Thus, the steady level of MHC class I heavy chain molecules on normal human thymus cells. J. Immun. 138, 1752-1768. does not explain the difference between IT46 and the other ABCE-GOMBZ, B., JONES, E. A., BAHSTABLE, C. J., SALOMON, E. AND cell types with respect to the expression and secretion of BODMER, W. (1979). The genetic control of HLA-A and B antigens in /32m. On the other hand, it has been shown that fibroblasts somatic cell hybrids: requirement for /32-microglobulin. Tiss. Antigens or B lymphocytes do not internalize their MHC class I 11, 96-112. BBRGGARD, I. AND BEARN, A. G. (1968). Isolation and properties of a low molecules in the absence of antibodies against MHC class I molecular weight ^2-microglobulin occunng in human biological antigens (Grey et al. 1982; Tse and Pernis, 1984). fluids. J. bwl. Chem. 243, 4096-4103. Furthermore, in T lymphocytes, which do spontaneously BIBNIER, G. M AND FANGKR, M. W. (1972). Synthesis of p2- internalize their MHC class I antigens, /S2m seems to be microglobulin by stimulated lymphocytes J. Immun. 109, 407-409. BRINCKERHOTF, C. E., MITCHELL, T. I., KARMILOWICZ, M. J , KLUVK- internalized at the same rate as the heavy chain (Tse and BECKERMAN, B. AND BENSON, M. D. (1989). Autocrine induction of Pernis, 1984; Vega and Strominger, 1989). Consequently, collagenase by serum amyloid A-like and /i2-microglobulin-like /52m found in the IT45 supernatant probably did not result proteins. Science 243, 655-667. CENTRELLA, M., MCCARTHY, T. L. AND CANAUS, E. (1989). /32- from an active turn-over of MHC class I. Our data are thus microglobuhn enhances insulin-like growth factor I receptor levels consistent with a constitutive secretion of /32m not and synthesis in bone cell cultures. J. biol. Chem. 264, 18268-18271. dependent upon the MHC class I expression or metabolism CHILD, J. A. AND KUSHWAHA, M. R. (1984). Serum ^2-microglobulin in but related to an overexpression of /32m. The most likely lymphoproliferative and myeloproliferative diseases. Hemat. Oncol. 2, 391-401. hypothesis that could account for this /32m secretion is that CRESWELL, P., SPRINGER, T., STROMINOBR, J. L., TURNER, M. J., GEEY, H. the excess of /32m synthesized by IT45 cells can escape M. AND KUBO, R. T. (1974). Immunological identity of the small passively from these cells. This leakage of/32m could occur subunit of HLA antigens and /ffi-microglobulin and its turnover on in the same way as in the absence of MHC class I heavy the cell membrane. Proc. natn. Acad. Sci. U.S.A. 71, 2123-2127. DARGEMONT, C, DUNON, D., DEUGNIER, M. A., DENOYELLE, M., GDIAULT, chain (Severinsson and Peterson, 1984). J. M., LEDERER, F., LE, K. D. H., GODEAU, F., THIERY, J. P. AND The overexpression of /32m by IT45 cells is probably due IMHOF, B. A (1989) Thymotaxin, a chemotactic protein is identical to to an enhanced level of transcription or mENA stability. ^2-microglobulin Science 246, 803-806. Indeed, the excess of/32m in IT45 cells was correlated with DEUGNIER, M. A., iMHor, B. A., BAUVOIS, B., DUNON, D., DENOYELLE, M. AND THIERY, J P. (1989). Characterization of rat T cell precursors a high level of /32m mRNA and likewise, the smaller sorted by chemotactic migration toward thymotaxin. Cell 56, amount of /32m in IT26 cells was correlated with a lower 1073-1083. level of /32m transcripts, suggesting a difference in the DUNON, D., KAUFMAN, J., SALOMONSEN, J., SKJOEDT, K., VAINIO, 0., regulation of /S2m gene transcription or in /32m mRNA THIERY, J. P. AND IMHOF, B. A. (1990). T-cell precursor migration toward /?2-microglobulin is involved in thymus colonization of chicken stability in these two cell lines. This difference in the embryos. EMBO J. 9, 3315-3322. amounts of /32m transcripts is not due to a general FIGUEROA, F., DAVIS, W. C. AND KLEIN, J. (1981). Ten new monoclonal antibodies detecting antigenic determinants on class I H-2 molecules. regulation of MHC class I antigens mRNA, since the ratio ;; /32m/class I heavy chain mRNAs is 1.5-fold higher in IT45 ImmunogenetiCB 14, 177-180. FORT, P., PIECHACZYK, M., EL SABHOUTY, S., DANI, C, JEANTEUR, P. AND than in IT26 cells. It is intriguing to consider the BLANCHARD, J. M. (1986). Various rat adult tissues express only one possibility that a /32m gene-specific transcription factor majeur mRNA species from the glyceraldehyde-3-phosphate not shared by MHC class I heavy chain genes may be dehydrogenase multigemc family Nucl. Acids Res. 13, 1431-1442. 564 C. Dargemont et al. GREY, H. M., COLON, S. M AND CHESNUT, R. W. (1982) Requirements OZATO, K. AND SACHS, D. H. (1981). Monoclonal antibodies to mouse for the processsing of antigen by antigen-presenting B cells. J. MHC antigens. HI. Hybrydoma antibodies reacting to antigens of the Immun. 129, 2389-2395. H-2b haplotype reveal genetic control of isotype expression. J. Immun. GREY, H. M., KUBO, R. T., COLON, S. M., POULJK, M. D., CRESSWELL, P., 126, 317-321. SPRINGER, T., TURNER, M. AND STROMINOBR, J. L. (1973). The small PLOEGH, H. L , CANNON, L. E. AND STROMINGEE, J. L (1979). Cell-free subunit of HLA-A antigens is /J2-microglobulin. J. exp. Med. 138, translation of the mRNAs for the heavy and light chains of HLA-A 1608-1612. and HLA-B antigens. Proc. natn. Acad. Sci. U.SA. 76, 2273-2277. HUTTKROTH, T. H., CLEVE, H , LITWIN, S. D. AND POULJK, M. D. (1973). POULJK, M. D. AND BLOOM, A. D. (1973). /32-microglobulin production The relationship between /ffi-microglobulin and immunoglobulin in and secretion by lymphocytes in culture. J. Immun 110, 1430-1433. cultured human lymphoid cell lines. J. exp. Med. 137, 838-843. SEVEJUNSSON, L. AND PETERSON, P A. (1984). /ffi-microglobulin induces iMHOr, B. A., DEUGNIER, M. A., GIRAULT, J. M., CHAMPION, S., DAMAIS, intracellular transport of human class I transplantation antigen C, ITOH, T. AND THIERY, J. P. (1988). Thymotaxin. a thymic epithelial heavy chain in Xenopus laevis oocytes. J. Cell Bwl. 99, 226-232. peptide chemotactic for T cell precursors. Proc natn. Acad. Sci. U.S.A. SHOTTON, D. AND WHITE, N. (1989). Confocal scanning microscopy: three 85, 7699-7703. dimensional biological imaging. Trends biochem. Sci. 14, 435-439. ITOH, T. (1979). Establishment of an epithelial cell line from rat thymus. SIMISTER, N. E. AND MOSTOV, K. E. (1989). An Fc receptor structurally Am. J. Anat. 156, 99-104. related to MHC class I antigens. Nature 337, 184-187 ITOH, T., AIZU, S., KASAHARA, S. AND MORI, T. (1982). Establishment of TOYOSHIMA, K., ITO, N., HIASA, Y., KAMAMOYO, Y. AND MAKIURA, S. a functioning epithelial cell line from the rat thymus. A cell line that (1971). Tissue culture of urinary bladder tumor induced in a rat by N- induces the differentiation of rat bone marrow cells into T cell butyl-N-(-4-hydroxybutyl) mtrosamine: establisment of cell line, Nara lineages. Bwmed. Res. 2, 11-19. Bladder Tumor II. J. natn. Cancer Inst. 47, 979-985. KIMURA, A., ISRAEL, A., LE BAIL, 0. AND KOURILSKY, P. (1986). Detailed TSE, D. B. AND PERNIS, B. (1984). Spontaneous internalization of class I analysis of the mouse H-2Hb promoten enhancer-like sequences and major histocompatibility complex molecules in T lymphoid cells. J. their role in the regulation of class I gene expression. Cell 44, exp. Med. 159, 193-207. 261-272. VEGA, M. A. AND STROMINGER, J L (1989). Constitutive endocytoBis of ROLLER, B. H., MARRACK, P., KAPPLER, J. W. AND SMITHIES, 0. (1990). HLA class I antigens requires a specific portion of the Normal development of mice deficient in /92-microglobulin, MHC clase intracytoplasmic tail that shares structural features with other I proteins, and CD8+ T cells. Science 248, 1227-1230. endocytosed molecules. Proc. natn. Acad. Sci. U.SA. 86, 2688-2692. LANCET, D., PARHAM, P. AND STROMINOER, J. L. (1979). Heavy chain of WHITE, J. G., AMOS, W. B. AND FORDHAM, M. (1987). An evaluation of HLA-A and HLA-B antigens iB conformationally labile: a possible role confocal versus conventionnal imaging of biological structures by for ^2-microglobulin Proc. natn. Acad. Sci. U.S-A. 76, 3844-3848. fluorescence microscopy. J. Cell Bwl. 105, 41-48. MACDONALD, R. J., SWIFT, G. H., PRZYBYLA, A. E. AND CHIRGWIN, J. M. WlJNAENDTS VAN RjSANDT, R. W., MARSMAN, H. J. B., KAPLAN, R., (1987). Isolation of RNA using guanidium salts Meth. Enzym 152, DAVOUST, J., STELZBR, E. H. AND STRICKER, R. (1985). Optical 219-227. fluorescence microscopy in three dimensions: microtomoscopy. J. MAUXION, F. AND KRESS, M. (1987). Nucleotide sequence of rat p2- Microsc. 138, 29-34 microglobulin cDNA. Nucl. Acids Res. 15, 7638. ZULSTRA, M., Brx, M., SIMISTER, N. E., LORINO, J. M., RAULET, D H. MAUXION, F., SOBCZAK, J. AND KRESS, M. (1989). Characterization of five AND JAENISH, R. (1990). /J2-microglobulin deficient mice lack distinct cDNA clones encoding for class I RT1 antigens. CD4"CD8+ cytolytic T cells. Nature 344, 742-746. Immune-genetics 29, 397-401. ZIJLSTRA, M., La, E., SAJJADI, F., SUBRAMANI, S. AND JABNISH, R. (1989). MORgLLO, D., DUPREY, P., ISRAEL, A. AND BABINET, C. (1985). Germ-line transmission of a ^2-microglobulin gene produced by Asynchronous regulation of mouse H-2D and y92-microglobulin RNA homologous recombination in embryonic stem cells. Nature 342, transcripts. Immunogenetics 22, 441-462. 435-438. MUNRO, S. AND PELHAM, H. R. B. (1987). A C-terminal signal prevents secretion of luminal ER proteins Cell 48, 899-907. OZATO, K., MBYBR, N. AND SACHS, D. H. (1980). Hybrydoma cell lines secreting monoclonal antibodies to mouse H-2 and la antigens. J. Immun 124, 533-540. (Received 16 November 1990 - Accepted 17 December 1990)

Overexpression and secretion of fo-microglobulin 565