EXPERIMENTAL

CELL

RESEARCH

193,

59-71

(1991)

A Novel 43-kDa Glycoprotein Is Detected in the Nucleus of Mammalian Cells by Autoantibodies from Dogs with Autoimmune Disorders MICHEL~OULARD,* JEAN-PHILIPPE BARQuE,t V~RONIQUEDELLAVALLE,* DANI~LEHERNANDEZ-VERDUN,~: CLAUDEMASSON,SFRANCOISEDANON,§ANDCHRISTIAN-JACQUESLARSEN*" ‘INSERM

U-301, Institut de Ge’nCtique Moleculaire, 27 rue Juliette Dodu, 75010 Paris, France; tDiuision d’Immunocytologie applique’e (DICA) et URA 523 CNRS, Unioersitd de Technologie de Compkgne, Rue R. de Couttolenc, BP 649, 60206 Compiegne, France; $Institut Jacques Monod-CNRS, 2 place Jussieu, 75251 Paris Ceden 05, France; and §Laboratoire d’lmmunologie, INSERM U-108, Hopital St. Louis, 1 avenue C. Vellefeaux, 75010 Paris, France

We have characterized a new antibody specificity in a panel of sera from dogs developing systemic lupus erythematosus (SLE) or clinically related autoimmune disorders. This antibody stains in a speckled fashion the nucleus of cells of different mammalian origins. The target antigen is a basic (PI 9.2) nuclear polypeptide with an apparent molecular weight of 43 kDa (~43) which is detected in various mammalian cell nuclei. ~43, as studied in HeLa cells, appears to be cell cycle-independent. It is released from nuclei by salts (0.5 M NaCl or 0.25 M ammonium sulfate). Upon subfractionation of nuclear components, p43 is found in the fraction containing HnRNPs and is recovered in immunoprecipitates obtained with 4F4 monoclonal antibody to HnRNP C proteins. Immunoelectron microscopy revealed that p43 is concentrated over the dense chromatin periphery and interchromatin granule clusters. Another important feature of p43 is its ability to specifically bind wheat germ agglutinin lectin but not concanavalin A nor Ulex europaeus I, supporting the notion that p43 is a glycoprotein bearing an N-acetylglucosamine moiety. Consistent with this result, a radioactive p43 band is specifically immunoprecipitated by canine anti-p43 autoantibodies from HeLa cells metabolically labeled with [‘4C]glucosamine. Finally, antip43 antibodies do not immunoprecipitate SnRNA, indicating that p43 has no apparent association with cc Issl Academic press, IDA. SnRNPs.

INTRODUCTION Autoantibodies to nuclear antigens (ANA) are present in sera of patients with systemic lupus erythematosus (SLE) and various multisystemic autoimmune disorders (for a review, see [l]). They recognize a large variety of nuclear autoantigens including DNA, RNA, ’ To whom dressed.

correspondence

and

reprint

requests

should

be ad-

histone and nonhistone proteins, and ribonucleoproteins. As new ANA specificities were characterized, the notion has progressively emerged that particular autoimmune diseases are associated with characteristic profiles of ANA of different specificities. In this respect, ANA have become an important clinical biology tool for the diagnosis and the classification of autoimmune diseases. Besides their use as diagnostic markers, ANA have now become particularly efficient tools for identifying new nuclear molecules and elucidating their functions. In the beginning of the 198Os, Lerner and Steitz inaugurated a fruitful series of studies when they discovered that nuclear Sm and RNP antigens are associated with a category of small nuclear RNA (the so-called URNAs) in ribonucleoprotein complexes that participate in splicing of mRNA [2]. Since then, many other autoantigens have been identified and their functions investigated [ 1, 31. The decisive importance of ANA in molecular biology is currently illustrated by works in which ANA were used for screening cDNA expression libraries in order to isolate cDNAs encoding autoantigens [4-71. Autoimmune disorders with clinical features very similar to those occurring in man have been described in dogs [B]. According to Monier et al., anti-ds-DNA antibodies are rare in canine SLE [9] but in some cases they share the same idiotypic cross-reactivity as murine anti-DNA lupus antibodies [lo]. Inversely, anti-histone antibodies are frequently observed in lupic canine sera, but they recognize a very different pattern of histones from those in human SLE [ll]. Antibodies to soluble nuclear antigens have also been reported [12J. In addition to the common anti-Sm and anti-RNP activities, two other specificities designated anti-type 1 and antitype 2 have been described whose corresponding antigens are resistant to trypsin and RNAse and correspond to molecular complexes between 50,000 and 150,000 Da [ 121. Finally, canine antibodies to nucleolar antigens have been described [ 13,141. In our laboratory, we have

59

0014.4827/91

$3.00

Copyright 0 1991 by Academic Press, Inc. All rights of reproduction in any form reserved.

60

SOULARD

recently characterized a family of canine sera that identify three antigenically related llO-, 95, and 45kDa polypeptides located in nucleoli and nucleoplasm of mammalian cell nuclei [15]. In addition, we demonstrated the llO-kDa antigen not to be Nucleolin (or C23 protein, the llO-kDa nucleolar polypeptide involved in ribosomal gene transcription complex [16-181). In this work, we describe a new variety of ANA present in dogs developing SLE and clinically related syndromes. These antibodies generate a speckled staining of the nucleus excluding nucleoli, and specifically react with a 43-kDa (~43) apparent molecular weight strongly basic polypeptide. The 43-kDa antigen has no relationship with the nuclear soluble antigens commonly detected by SLE autoantibodies (such as Sm, RNP, and SS-A and SS-B polypeptides). It is recovered from HnRNP-containing nuclear fractions by 4F4, a monoclonal antibody specific to HnRNP C proteins [19]. This antigen also has the important capacity to bind the wheat germ agglutinin (WGA) lectin and incorporates thus rejoining the cohort of [ 14C]glucosamine, previously characterized nuclear glycoproteins [20]. Subfractionation and immunoelectron microscopy studies indicate that p43 is not associated with nuclear membranes, chromatin or nuclear matrix, but is preferentially located at the chromatin periphery and in regions containing interchromatin granules. A preliminary report has been published elsewhere [21]. MATERIALS

AND

METHODS

Sera All canine sera tested in this work were collected in the immunology department of the “Ecole Veterinaire d’Alfort” or were kindly provided by private veterinary practitioners. The routine procedure for screening sera from dogs presenting overt symptoms of autoimmunity has been reported elsewhere [15]. Those sera displaying a positive nuclear fluorescence on rat hepatocytes at dilutions higher than 1:lOO were further analyzed on HEp-2 cells. Absence of anti-DNA antibodies was tested by a modified Farr assay [22] and by the indirect immunofluorescence assay on the kinetoplast of Crithidia luciliae [23]. Antihistone specificity was performed by ELISA test and immunoblotting as previously described 115, 241. The test was repeatedly negative. As negative control sera, a pool of 10 canine sera was constituted, which gave no immunofluorescence at a 1:30 dilution. Cells HeLa S3 cells were grown in agitated suspension in MEM supplemented with 10% heat-inactivated FCS, 5 mM glutamine, and antibiotics [25]. They were labeled with [?S]methionine for 18 hr (25 &i/ ml, New England Nuclear), with [i4C]6-glucosamine for 48 h (11 &Y ml, CEA Saclay, France) as described elsewhere [26]. Other cells including HEp-2 human tumor cells, KE-37 (a leukemic human Tcell line), HL-60 (a promyelocytic human cell line), NIH/3T3 mouse cells, ED equine cells, NRK rat cells, BHKPl hamster cells, CCL64 mink cells, and dog fetal thymocytes were employed for immunofluorescence and immunoblot assays. Human tumor TG cells used for immunoelectron microscopy were grown as monolayers in MEM supplemented with 10% FCS and were

ET AL. harvested fixation. Preparation

by EDTA

+ trypsin

treatment

of Cell Nuclear Extracts

and washed with PBS before

and Subnuclear

Fractions

HeLa cells were suspended for 5-7 min in a low Low-salt nuclei. ionic strength buffer (1 mM Tris-HCl, pH 8.0) and quickly transferred to buffer A (10 mM TrissHCl, pH 7.4, 10 mM NaCl, 1.5 mM MgCl,) containing 1 mM PMSF and aprotinin (100 pg/ml). The cells were then disrupted in a Dounce homogenizer and the resulting homogenates were centrifuged through a 30% sucrose cushion in the same buffer for 10 min at 2500 rpm. Detergent-purified nuclei. Cells were suspended in buffer A and disrupted by addition of NP-40 at a 0.25% (v/v) final concentration. The nuclear pellets were then sonicated as previously reported [27] and homogenized in Laemmli immunoblot buffer [28]. Fractionation of nuclear extracts according to Pederson et al. [29]. Low-salt nuclei were resuspended in buffer A (pH 7.0), washed three times with this buffer, and sonicated as previously reported [29]. The homogenate was layered onto a 30% sucrose cushion in the same buffer and centrifuged for 15 min at 5000 rpm in a Beckman SW-27 rotor. All steps were carried out at 4°C. Preparation of nuclear matrix. The procedure described by Fey et al. was employed, except that vanadyl adenosine was omitted [30]. Saline extraction of HeLa cell nuclei. Low-salt HeLa cell nuclei were suspended in increasing NaCl concentrations and stirred at 4°C for 2 h. The suspensions were then layered onto the top of a lo-40% sucrose gradient in the same buffer formed over a 72% cushion. The tubes were centrifuged in a Beckman SW-27 rotor at 24,000 rpm for 4 h at 4°C. Samples from each fraction were analyzed by gel electrophoresis and immunoblotting. Immunoblotting The procedure was as previously described [31]. Dogimmunoglobulins complexed with antigens were revealed by peroxidase-coupled anti-dog IgG immunoglobulin (l/40) or protein A coupled to peroxidase (l/2000). 4F4 monoclonal antibody initially prepared in the laboratory of Dr. G. Dreyfuss [19] was generously provided to us by Dr. Brunel. We are indebted to Dr. Dreyfuss who authorized the use of this reagent. For glycoconjugate assays, blots were immersed for 2 h in a blocking solution (2% PVP in PBS, pH 7.4) and incubated for 2 h with horse radish peroxidase-conjugated lectins (5 @g/ml in PBS containing 2% PVP) with or without 0.6 M N-acetylglucosamine as a competitor. Blots were then washed five times (for 10 min) in PBS, pH 7.4, containing 0.5% Tween 20. Immunoprecipitation Nuclear extracts from [?S]methionineor [i4C]glucosaminelabeled HeLa cells were suspended in SDN buffer (150 mM NaCl, 100 mM Tris, pH 8.5,5 mM EDTA, 0.5% SDS, 0.5% DOC, 0.5% NP-40,50 pg/ml aprotinin, 1 mM PMSF) and sonicated. Nuclear supernatants (12,OOOg for 15 min) corresponding to 8 X lo6 cells were adjusted to a 500-11 vol with SDN buffer and incubated with 7 ~1 of canine nonimmune serum for 2 h at 4°C under soft agitation. One hundred microliters of a 10% protein A-ultrogel suspension (IBF, France) were added and incubation was continued for 30 min at 4°C. Upon centrifugation at 12,OOOgfor 90 s, 7 ~1 of autoimmune canine serum was added to the supernatants and incubation was continued for 2 h at 4°C. The immune complexes were recovered in the protein A-ultrogel pellets that were washed four times in SDN buffer. Antigens were released by addition of Laemmli buffer to the pellets and boiling for 2 min. Supernatants were directly loaded onto 15% polyacrylamide gels. For characterization of RNA in immune complexes, HeLa cells were labeled for 18 h with inorganic 32P (20 &i/ml, CEA Saclay).

CANINE

AUTOANTIBODiES

DETECT

Upon incubation of nuclear extracts (6 X 10s nuclei) with appropriate sera, RNA were isolated from nuclear extracts suspended in NET2 buffer (50 mM Tris-HCl, pH 7.4, 150 mM NaCl, 0.05% NP-40, v/v) and extracted by two phenol treatments [2]. Samples were analyzed in 8.5% polyacrylamide-urea gels, as previously described (32). Gel ~~etrophoresis Conditions for one-dimensional and two-dimensional gel electrophoresis in 15% polyacrylamide gels were as previously reported [27]. For 2-D gel analysis of immunoprecipitates, immune complexes recovered from [?S]methionine HeLa cell nuclear extracts were suspended in 25 ~1 of a solution containing 9.5 M urea, 5% P-mercaptoethanol, 0.2% ampholines (3.5-10, LKB), and 0.2% SDS. After boiling for 2 min, the mixture was centrifuged for 2 min at 12,OOOgand was diluted with an equal volume of the same solution containing 4% NP-40 instead of SDS. For pi determinations, the O’Farrell technique modified by Hames et al. was used in order to obtain a pH range from 2.5 to 9.5 [33]. Immunofluorescenw

Studies

Tests were initially performed on HEp-2 cells fixed with acetone for 3 min at 4OC and HeLa cells fixed with a l/l mixture of methanol/ acetone for 5 min. Because of the possibility of a poor cell fixation by acetone, HeLa cells were also fixed with 3% formaldehyde in PBS, pH 7.4, for 5 min at room temperature, washed in PBS, and treated with 0.5% Triton X-100 (5 min at 4°C) or a methanol/acetone mixture (1 vol/l ~01, 5 min at 4°C). For RNase A, and DNase I digestions, HEp-2 cells grown on microscope slides were fixed and incubated for 30 min at 37°C with PBS containing 5 mM M&l, + 2 mM PMSF and variable amounts (10 to 100 pg/ml) of RNase A and DNase I (Sigma). Controls of DNase I and RNase activity were performed in parallel. Trypsin digestion was performed at 10 @g/ml in PBS. Antitrypsin inhibitor (6 @g/ml) was added before incubating the slides with canine antibodies. Finally, the slides were washed three times in PBS and incubated for 60 min at, 2O’C with a 1:200 dilution of fluorescein-conjugated rabbit anti-dog IgG (Biosys, Compibgne, France). Immunodiffusion

Tests

Tests were performed as already described using either a calf thymus extract (Pel-Freeze) at 20 mg/ml in PBS, pH 7.4, or a HeLa cell nuclear extract in PBS adjusted to a protein concentration of 10 mg/ ml [27]. To analyze the protein material contained in the precipitation lines, agar bands were cut. Upon addition of 3~ Laemmli buffer, the mixtures were boiled for 3 min and loaded to the top of SDSPAGE gels. Protein transfer was carried out as described above. Afinity

Purification

of Autoantibodies

The procedure described by Olmsted et al. 134) was modified as previously reported 1151. Briefly, the antibodies were eluted from nitrocellulose sheets with 3.0 M KSCN, dialyzed against PBS, and concentrated by lyophilization. Immunoelectron

Microscopy

TG cells were fixed for 1 h at room temperature with 0.1 A4 phosphate-buffered 4% paraformaldehyde. Free aldehyde groups were then quenched by incubation in 0.5 M ammonium chloride for 2 h at room temperature and the cell pellet was washed in buffer before being embedded in Lowicryl K4M 1353. Ultrathin sections were mounted on Formvar-coated gold grids. For the immunogold labeling procedure, nonspecific sites of adsorption were blocked overnight at 4°C by a solution of 5% BSA in PBS. The sections were then incubated for 2 h with various dilutions of autoimmune serum in PBS with 5% BSA (dilutions tested up to

A NUCLEAR

61

GLYCOPROTE~N

1:lOO). They were washed for 25 min with PBS, floated on 1% BSA/ PBS for 5 min, and incubated with lo-nm protein A-gold particles (Janssen) diluted 1:lO in 1% BSA:PBS for 1 h. After washing with PBS for 25 min, the grids were rinsed with distilled water and stained for 15 min with 4% aqueous uranyl acetate. TWO types of control assay were carried out: either the immune serum was replaced by a nonimmune dog serum or the incubation was done with the protein A-gold complex alone. The final preparations were examined in a Philips EM 410 microscope at 80 kV. Distribution of colloidal particles in cells was examined on micrographs of randomly chosen fields at a final magnification of X20,000. Cell Synchronization The procedure where [15, 241.

of double thymidine

block has been reported

else-

RESULTS

(I) Characterization Autoantibodies

of a New Family

of Canine

Thirty-two sera from dogs with SLE or lupus-like syndromes reacting positively with rat hepatocyte nuclei at a 1:lOO dilution were further examined onto human cells (HeLa and HEp-2 cells, Fig. 1). While 17 of these sera displayed homogeneous nucleoplasmic or nucleolar fluorescence patterns, 15 displayed a nuclear speckled staining excluding nucleoli. No cytoplasmic staining was observed, whatever the conditions of cell fixation (acetone, methanol + acetone, Triton X-100 + 3% paraformaldehyde). The reactivity of these sera was supported by IgG antibodies as judged from a positive fluorescence reaction with FITC-conjugated dog anti-IgGs and a negative one with FITC dog anti-IgM. A pool of normal dog sera did not stain the cell preparations at a l:50 dilution (data not shown). Seven sera tested on different mammalian cells (Fig. 1) generated a very similar staining pattern. Importantly, a homogeneously speckled pattern of nuclear staining was observed in normal cells (including fetal cells) as well as in malignant cells (see discussion on this point). Treatment of Triton X-lOO-permeabilized HeLa cells with RNase A (lo-30 @g/ml) or DNase I (100 pg/ml) did not appear to change the fluorescence patterns (data not shown). In contrast, the staining was completely abolished by trypsin (10 yglml) or Pronase (500 pg/ml). Saline conditions were also found to modify this pattern as the fluorescence was totally abolished when the sodium chloride concentration was raised up to 0.5 M. To further characterize these dog sera, immunodiffusion tests were carried out with saline extracts of rabbit thymus (Pel-Freeze). All 15 sera giving a speckled staining pattern also generated a common precipitation line which crossed the lines from reference human sera to Sm (Fig. 21, RNP, and SS-B/La (data not shown), and also the line generated by recently reported antinucleolar dog sera [ 151. RNase and DNase treatments did not change this pattern but incubation with trypsin re-

62

SOlJLARD

ET AL

FIG. 1. Indirect immunofluorescence patterns on different mammalian cells. Cells were fixed using a methanol/acetone mixture and stained with one canine autoimmune serum (1:40(l) followed by Huorescein-conjugated mouse anti-dog IgG. Depending on the cell origin, a more or less speckled pattern was observed in all nuclei. Nucleoli were never stained. The 15 sera gave an identical pattern. (A) NRK cells. (B) Dog fetal thymocytes. (C) NIH/3T3 mouse cells. (D1 HEp-2 human tumor celis. (E) HEp-2 cells in phase contrast.

sulted, as expected, in the disappearance of the precipitation line. Taken together, these data indicate that these 15 sera detected a new antigenic specificity which is brought by polypeptides. (II)

Biochemical

Characterization

of

the Antigen

Immunoblot analysis. Immunoblot analysis of nuclear HeLa cell extracts was performed in order to identify the antigen(s) reacting with these dog sera (Fig. 3). A unique and intense band (Fig. 3A, lane 3) was revealed by peroxidase coupled to protein A or to dog anti-IgG at the position of a 43-kDa protein (for this reason, the material of this band was referred to as ~43). Normal dog sera were negative (Fig. 3A, lane 2). In order to

ascertain that antibodies to p43 were responsible for the immunofluorescence staining, affinity-purified antibodies were prepared by elution from t.he 43-kDa band of an immunoblot (Fig. 3B, lane 1). These purified antibodies reacted only with a 43-kDa mat.erial on the immunoblot (Fig. 3B, lane 2) and generated exactly the same fluorescence pattern on HeLa cells as that obtained with total serum (Fig. 3C). In addition, the immunoprecipitation line generated by a double immunodiffusion test on a HeLa cell nuclear extract (see Fig. 2) was analyzed by SDS-PAGE. The resulting immunoblot incubated with the same autoimmune dog serum revealed t.he presence of a faint but reproducible band running at 43 kDa [data not shown]. An identical p43 band was found in nuclear extracts of different mammalian cells including dog fetal t,hymo-

CANINE

AUTOANTIBODIES

DETECT

A NIJCLEAR

63

GLYCOPROTEIN

In order to detect possible associations between p43 and SnRNA 32P-labeled HeLa cell nuclear extracts were processed and analyzed as reported (see Ref. [3]). Results presented in Fig. 6 clearly show that canine anti-p43 antibodies did not precipitate any SnRNA (lane 2) when experimental conditions allowing Ul to U6 SnRNA to be immunoprecipitated by anti-Sm (lane 3) or anti-RNP antibodies (lane 4) were used. (III) FIG. 2. Comparative immunodiffusion of different autoimmune canine sera with a reference anti-Sm human serum. Immunodiffusion was carried out with a rabbit thymus extract (Pel-Freeze, central well). Four different canine sera exhibiting the speckled staining depicted in Fig. 1 were loaded in wells 2,3,5,6. A human anti-Sm serum was loaded in wells 1 and 4. The common precipitation line generated by the canine sera clearly crosses the line issued from anti Sm-serum. In total, 15 canine sera displaying the same staining characteristics were found to generate a common precipitation line that crosses those generated by anti-Sm, anti-RNP, and anti-%-B/la sera (data not shown).

Distribution

of p43

Because recovery of p43 in immunoprecipitation tests required nonionic detergents and especially DOC + alkaline pH, we first examined the influence of these reagents on the subcellular distribution of ~43. These studies showed that p43 is strictly nuclear when hypotonic buffer is employed for preparing nuclei (Fig. 7A, lane 4, see Materials and Methods). In contrast, the

KD

cytes, man (KE37 and HL-60 leukemic cells, HEp2 and HeLa tumor cells), mouse (liver and NIH3T3 cells grown in culture), rat (liver and normal rat kidney cells grown in culture), Syrian hamster (BHK21 tumor cells), horse (ED cells), and mink (CCL64 fibroblasts). These data supported the notion of an evolutionary conservation of the p43 antigen through mammals. Although the fluorescence patterns were not consistent with the notion that p43 was actin, an actin extract (Sigma) was subjected to electrophoresis and blotted onto nitrocellulose. No signal was detected by dog sera. In addition, we observed no loss of activity in sera that were incubated with actin solutions before immunofluorescence study. Immunoprecipitation studies. Preliminary tests using amino acid-labeled HeLa cell extracts indicated that DOC (0.5%) and pH 8.5 were required for solubilization of the antigen. Under these conditions, a specific p43 band (large arrow) was observed (Fig. 4, lanes 3 and 5). Upon longer autoradiographic exposure, a much fainter 28- to 30-kDa band was sometimes observed with several sera (Fig. 4, small arrow). This molecular species was not detected by affinity-purified anti-p40 antibodies. A 43-kDa material was detected by nonimmune serum (Fig. 4, lanes 2 and 4). To distinguish this material from ~43, immune precipitates were analyzed by nonequilibrium 2-D gel electrophoresis (NEpHGE, Fig. 5). A wide spot was detected in the basic area of the p43-containing gel (Fig. 5A). Its isoelectric point was independently estimated to 9.2 by IEF. In contrast, the 43-kDa material recognized by nonimmune serum in one-dimensional gels (Fig. 4, lane 4) was resolved in 2-D gels as a spot running in the acidic area (Fig. 5B, arrow).

Subcellular

97

M .~L

/,

/ A

123

B

1

2C

FIG. 3. The autoimmune canine antibodies responsible for the fluorescence pattern also recognize a polypeptide with a 43.kDa apparent molecular weight (~43) antigen. (A) Immunoblot of a HeLa cell nuclear extract. A nuclear extract representing approximately 5 X lo6 cells/lane was analyzed by SDSPAGE. One part of the gel was stained with Coomassie blue (lane 1). The other part was transferred onto nitrocellulose. The resulting blot was revealed with a nonimmune dog serum (lane 2, dilution 1:50) or with one of the canine sera displaying the speckled staining in Fig. 1 (lane 3, dilution 1:500). Detection of antibodies was performed with peroxidase coupled to protein A (or goat anti-dog IgG). (B) The canine IgG detecting p43 display a speckled fluorescence pattern. Monospecific autoantibodies were eluted by KSCN from the HeLa cell p43 band of a nitrocellulose sheet (the position of the band was verified by immunoblot of a parallel lane in the same blot, see lane 1 in B). Upon dialysis and concentration, these affinity-purified antibodies recognized p43 on an immunoblot of HeLa cell nuclear proteins (B, lane 2) and generated a speckled pattern of HeLa cell nuclei (C, notice that the more diffuse staining of cells compared to Fig. 1 resulted from the use of acetone instead of acetone/methanol for cell fixation).

64

SOULARD

12345 FIG. 4. Immunoprecipitation ol’HeI,a cell nuclear extracts using canine anti-p43 autoantibodies. A nuclear extract was prepared from HeLa cells labeled with [%]methionine and precleared with a nonimmune dog serum and protein A before incubation with autoimmune serum. Immune complexes were recovered with protein A-ultrogel and run on 15% SDS-PAGE gels before autoradiography. Lane 1, total nuclear extract; lanes 2 and 4, complex recovered with nonimmune serum exposed for 2 h (lane 2) or 48 h (lane 4); lanes 3 and 5, autoimmune complex exposed for 2 h (lane 3) and 48 h (lane 5). The large arrow points to ~43. The small arrow points to an unrelated 30.kDa polypeptide which is detected by some of the pathologic sera. Equal amounts of radioactive material were submitted to immunoprecipitation.

ET AL.

attempted. In a first approach, HeLa cell nuclei prepared by low-salt buffer containing 0.5% Triton X-100 were first extracted with 0.25 M ammonium sulfate, then successively digested with DNAse I (100 pg/ml in the presence of 0.25 M ammonium sulfate) and RNAse A (25 pg/ml). The resulting supernatants from each treatment and the final pellet containing nuclear matrix were examined for p43 content. As seen in Fig. 8, a part of p43 was liberated by ammonium sulfate treatment (lane 2) and also by DNAse I (lane 3). In fact, the DNAse I treatment appeared only to complete the action of ammonium sulfate, since incubation of nuclei with the enzyme in the presence of 110 mM NaCl did not release any p43 material (data not shown). RNAse had no apparent effect (lane 4). The pellet contained only residual p43 (lane 5, notice that the protein amounts in lanes 4 and 5 were twice those of the other lanes). From these data and those provided by sodium chloride treatments, we concluded that p43 is not an intrinsic component of the nuclear matrix. In a second approach, HeLa cell nuclei prepared in hypotonic buffer without detergent were processed according to the protocol described by Pederson et al., i.e., centrifugation of sonicated extracts through a 30% su-

NEPHGE

presence of 0.25% NP-40 in the extraction buffer releases a 43-kDa material in the cytoplasmic fraction (lane 7). DOC (0.5%) gave the same eff’ect (data not shown). This cytoplasmic material has the same 2-D electrophoretic behavior as ~43. In addition, antibodies eluted from this cytoplasmic 43-kDa band stain only the nucleus of HeLa cells and react with the nuclear p43 band in immunoblotting assay (data not shown). These data confirmed that p43 is a nuclear antigen which is redistributed into the cytoplasm in the presence of nonionic detergents or lipophilic reagents. To investigate the importance of saline conditions on the localization of ~43, HeLa cell “low-salt” nuclei were incubated with increasing sodium chloride molarities (from 0.2 to 2.0 M) and submitted to sucrose gradient centrifugation (Fig. 7B). p43 was found exclusively in the nuclear pellet up to 0.4 M NaCl, but it was shifted to the top of the gradient when the NaCl molarity was equal to or exceeded 0.5 M. (IV) Nuclear Distribution of the 43-kDa Protein In order to determine precisely the nature of the nuclear structures containing ~43, several analyses were

FIG. 5. Two-dimensional gel autoradiograms ofcomplexes recovered by immunoprecipitation of HeLa cell nuclear extracts with nonimmune serum (H) or an autoimmune canine serum (A). HeLa cells were labeled with [“‘Slmethionine.

CANINE

AUTOANTIBODIES

DETECT

A NIJCLEAR

65

GLYCOPROTEIN

buffer at 37°C [37]. Again p43 turned out to be almost exclusively found in the supernatant fraction containing HnRNP (data not shown). (V) p43 Has a Glycosylated Status The fact that p43 required DOC to be solubilized suggested that the antigen might bear glycosylated residues. To get insight on this point, a 2-D NEPHGE gel of a nuclear HeLa cell extract was transferred onto nitrocellulose and revealed with WGA coupled to peroxidase. A spot migrating at the position of p43 was detected (data not shown). This spot disappeared when an excess of N-acetylglucosamine was added. A more direct approach consisted of immunoprecipitating a nuclear ex-

KD

M

FIG. 6. Anti-p43 antibodies do not coprecipitate SnRNA. Nuclei were isolated from HeLa cells labeled for 18 h with 20 pCi/ml of 32P. Upon incubation of equal amounts of radioactive material with appropriate sera, SnRNAs were recovered from immune precipitates and subjected to electrophoresis in 8.5 % polyacrylamideeurea gels. Lane 1, canine nonimmune control serum; lane 2, canine anti-p43 serum; lane 3, human anti-Sm serum; lane 4, human anti-RNP serum; lane 5, total nuclear extract.

crose cushion [29]. Under such conditions, HnRNPs are concentrated at the top of the tube (fraction I). The bulk of chromatin preparations that penetrates the sucrose layer constitutes the fraction II. The pellet (fraction III) contains nucleoli and nuclear matrix. The presence of HnRNP restricted to fraction I was ascertained by submitting blots of each fraction to 4F4 monoclonal antibody which specifically reacts with HnRNP proteins Cl and C2 (Fig. 9B, lane 6). Probing the immunoblotted fractions with a dog serum demonstrated that p43 was mostly concentrated in fraction I (Fig. 9B, lane 3). To confirm this point, the following experiment was designed. HnRNP samples were immunoprecipitated with anti-p43 canine antibodies under conditions which preserve the integrity of HnRNP [36]. Upon migration and transfer, the resulting blot was revealed by anti-p43 antibodies (Fig. 9A, lane 2) and 4F4 antibody (lane 1). As expected, p43 was detected in the complex, and also the Cl/C2 doublet (notice that Cl is the major band). The inverse experiment consisting of immunoprecipitating HnRNP with 4F4 antibody and revealing the blot with 4F4 (lane 3) and anti-p43 antibody (lane 4) confirmed that p43 is immunoprecipitated by both antibodies. Taken together, these results support the notion that p43 is an HnRNP protein or a protein associated with HnRNPs. An identical result was obtained if HnRNPs were prepared according to the procedure of Samarina et al. which involves incubation of nuclei in a pH 8.0

A

12

BOTTOM

34

56

78

TOP

0,4M 0,5M B FIG. 7. Importance of the detergent and saline conditions on ~43 recovery. (A) Nuclear and cytoplasmic HeLa cell fractions were prepared under low-saline conditions in the absence (lanes l-4) and the presence (lanes 5-8) of NP-40 (0.25%) and analyzed on 15% SDS-polyacrylamide gels. Gels were stained with Coomassie blue (lanes 1, 2, 5, 6) or transferred onto nitrocellulose sheets which were revealed using an autoimmune canine serum (lanes 3, 4, 7, 8). Lanes 1 and 3, low-salt cytoplasmic extract; lanes 2 and 4, low-salt nuclear extract; lanes 5 and 7, low-salt + NP-40 cytoplasmic extract; lanes 6 and 8, low-salt + NP-40 nuclear extract. (B) Low-salt nuclear extracts were incubated with 0.4 or 0.5 M NaCl for 2 h at 4°C and submitted to centrifugation through a 10-40s sucrose gradient (in buffer A) during 4 h at 24,000 rpm. Centrifugation was performed in a Beckman SW-27 rotor at 4°C. Twenty-five fractions were recovered and each fraction was analyzed for its p43 content by immunoblot. This experiment indicates that p43 is released from structures sedimenting to the bottom of the tube by 0.5 M NaCl.

66

SOLJLARD

ET AL

To further assess this point, immunoprecipitation was carried out on a nuclear extract from HeLa cells that had been labeled for 48 h with [ 14C]glucosamine (11 pCi/m1/48 h). A band of material running at the position of p43 was precipitated by the autoimmune canine antibodies from two sera with anti-p43 specificity (Fig. lOB, lane 1). Under the same conditions, a nonimmune dog serum did not precipitate any radioactive material (Fig. lOB, lane 2). (VI) Nuclear Microscopy

E 2345

FIG. 8. p43 is not an intrinsic component of nuclear matrix. A HeLa cell nuclear extract was fractionated according to the procedure described by Fey et al. [28]. Fractions from the different steps were analyzed by Coomassie blue staining (A) and by immunoblotting with autoimmune canine serum (B). Lane 1: Cytoplasmic supernatant recovered after centrifugation of nuclei purified in the presence of 0.5% Triton X-100 (in CSK buffer consisting of 100 mM NaCl, 300 mM sucrose, 10 mM Pipes, pH 6.8, 3 mM M&l,, 1 mM EGTA, 1.2 mM PMSF (281). Lane 2: 2OOOgsupernatant of a nuclear homogenate extracted by 0.25 M ammonium sulfate (5 min at 4°C). Lane 3: The pellet of the previous step was homogenized in CSK I buffer rontaining only 50 mM NaCl [28] and treated with DNAse I (100 pg/ml). At the end of the incubation, ammonium sulfate was added at a 0.25 M final concentration. The mixture was centrifuged and a supernatant sample was analyzed. Lane 4: The pellet from the previous step was homogenized in modified CSK buffer and incubated with RNAse A (25 pg/ml). The supernatant was recovered by centrifugation and analyzed. Lane 5: The final pellet containing nuclear matrix [28] was homogenized. Notice that the samples analyzed in lanes 1 :j were obtained from 4 X lo6 cells, whereas the samples in lanes 4 and 5 resulted from treatment of 8 X 10” cells.

Localization

of ~43

by Immunoelectron

Immunoelectron microscope examination was performed onto human TG cells that were actively growing at the time of the cell fixation. One autoimmune dog serum was selected which exhibited a high fluorescence titer onto HEp-2 cells and was verified to detect only ~43. On an ultrathin Lowicryl section, this serum exhibited a positive reaction at 1:50 and 1:lOO dilutions, but better result,s were obtained with the 1:lOO dilution as the levels of nonspecific immunostaining were consistently low. As illustrated in Fig. 11A, gold labeling was mainly distributed in the nucleus, whereas it was negligible or low in the cytoplasm. In addition, it was more concen-

kd 97 66

31

tract with anti-p43 dog antibodies. Upon electrophoresis in a 15% polyacrylamide gel and transfer onto nitrocellulose, the p43 band was revealed by WGA-peroxidase and was completely abolished by N-acetylglucosamine but not by a-D-glucopyrannoside (Fig. 10). Importantly, the intensity of the signal was higher than that of a total nuclear extract, reflecting enrichment of the WGA-reacting material through immunoprecipitation (compare lanes 7 and 9). In contrast, the unspecific material recovered in the precipitate with nonimmune serum was negative (lane 8). Negative results were also observed when the immunoprecipitation protocol was repeated using concanavalin A coupled to peroxidase or Ulex europaeus agglutinin I (UEA I) which react with glycoconjugates bearing respectively a-D-mannosides and cu-L-fucosides (data not shown). These data established that p43 contains a carbohydrate moiety with Nacetylglycosamine residues but no a-D-mannosyl or cyL-fucosyl residues.

14 A

1234Bl234567

8

910

FIG. 9. Nuclear distribution of ~40. A nuclear extract of HeLa cells was prepared according to the procedure of Pederson et al. [29] and fractionated by centrifugation at 5000 rpm in a SW-27 rotor for 15 min. Samples from the fractions I (top) and 11 and III (bottom) which contain, respectively, HnRNPs, chromatin, and nuclear matrix + nucleoli [29] were analyzed by immunoblot for ~43 content. (A) Lanes 1 and 2, HnRNP sample immunoprecipitated with anti-p43 autoimmune antibody, hlotted onto nitrocellulose, and revealed with 4F4 monoclonal antihody (lane 1) or anti-p43 autoimmune antibodies (lane 2); lanes 3 and 4, HnRNP sample immunoprecipitated with 4F4 antibody and revealed with 4F4 (lane 3) and anti-p43 antibody (lane 4). (B) Lanes I, 4, and 9, fraction III (bottom of the gradient) revealed by canine serum (lane l), 4F4 (lane 4), and Coomassie blue stained; lanes 2, 5, and 8, fraction II (same order); lanes 3, 6, and 7, fraction I (top of the gradient, same order); lane 10, molecular weight markers.

CANINE

AIJTOANTIRODIES

DETECT M

-66

-43

-31

123

4 56

7891011

-12

34

FIG. 10. p43 is a glycoprotein hearing GlcNAc residues. (A) A nuclear HeLa cell extract was successively treated with nonimmune canine antibodies and an anti-p413 serum (as presented in Fig. 4). Both complexes were recovered hy protein A-ultrogel and run on polyacrylamide gels. One part of’ the gel was blotted onto nitrocellulose and revealed with immune serum or WGA-peroxidase. The other part was stained by silver. Silver staining of’ the gel. Idane 1, nuclear extract; lane %, nonimmune complex; lane :s, anti-p43 immune complex; lane 4, nuclear extract blotted and revealed hy anti-p43 antihodies; lane 7. nuclear extract hlotted and revealed by WGA-peroxidase; lanes 5 and 8, immunohlot of’ the nonimmune complex revealed by immune serum (lane 5) and WGA-peroxidase (lane 8); lane 6, immun&lot of the immune complex revealed by canine autoimmune serum; lanes 9. 10, and 11, immunohlot of’ the immune complex revealed by WGA-peroxidase (lane 91, WGA-peroxidase + 0.6 M GlcNac (lane l(J), WCA-peroxidase + 0.6 M