Eur. J. Immunol. 1993. 23: 3003-3010 A soluble form of human CD58 3003

Jorg C. Hoffmann, Thomas J. Dengler, Percy A. Knolle, Marion Albert-Wolf, Matthias Roux, Reinhard Wallich and Stefan C. Meuer

Applied Immunology, German Cancer Research Center

A soluble form of the adhesion receptor CD58 (LFA-3) is present in human body fluids*

The human adhesion receptor CD58 (LFA-3) is expressed on most human cell types. Here we report on a soluble form of CD58 (sCD58) in human serum, human urine, and culture supernatants of several cell lines. sCD58 partially purified from human serum, from supernatant of the Hodgkin cell line L428, and purified sCD58 from human urine were found to have a molecular mass of 40-70 kDa under denaturating conditions (sodium dodecyl sulfate-polyacrylamide gel electrophoresis and Western blotting). However, gel filtration of sCD58 purified from human urine gave a molecular mass of 118-166 kDa, suggesting a noncovalent homotrimer conformation or its association with other molecules. Using an enzyme-linked immunosorbent assay specific for CD58 we found that sera from patients suffering from different forms of hepatitis contained elevated sCD58 levels (n = 108). Accordingly, there was a fivefold increase of supernatant sCD58 when the hepatocellular carcinoma cell line Hep G2 was incubated with 25 ng/ml recombinant tumor necrosis factor-a in vitro. In contrast, sCD58 serum levels of 337 additional patients suffering from various other immunological disorders were not found to be raised. At high concentrations sCD58 binds to CDZpositive cells and inhibits rosette formation of human T cells to human erythrocytes. Thus, local release of large quantities of naturally occurring sCD58 may interfere with intercellular adhesion in vivo.

1 Introduction

Soluble forms of membrane proteins such as cytokine receptors or cellular adhesion molecules (CD14, TNF receptor, CD25, IL-6 receptor, IFN-y-receptor, and CD54) have recently been detected in human body fluids [l-111. As some of them bind to their natural ligands, i.e. cytokines or cell surface molecules, they may have important func- tions in immune regulation by blocking receptorlligand interactions [3-5,8,11]. In addition, some of these soluble membrane proteins seem to be of clinical relevance since elevated serum levels have been observed in certain populations of patients [5, 6, 10, 111.

The cellular adhesion molecule CD58 (formerly known as LFA-3), a heavily glycosylated protein, is expressed by most human tissues including epithelial cells, endothelial cells and virtually all hematopoetic cells [12]. Both CD58 and its natural ligand CD2 consist of two domains and both molecules belong to the immunoglobulin superfamily [13-181. Using a panel of CD58 mAb we have recently identified four discrete epitopes on the N-terminal domain (domain 1) and two overlapping epitopes on the mem-

brane-proximal domain (domain 2) [ 17].The N terminus of the CD58 molecule represents the interaction site with CD2. In contrast, domain 2 links CD58 to the membrane anchor and does not participate in CD2 binding [17]. The interaction between CD58 and CD2 is important for antigen-specific T cell activation as well as activation processes in NK cells [14, 19-24]. In addition, the CD2/CD58 interaction is involved in adhesion between T lymphocytes and other CD58-positive cells. Thus, CD58 antibodies, purified soluble CD58, or recombinant soluble CD58 have been shown to block adhesion of T cells to thymic epithelium [25], cytotoxic T celvtarget cell interac- tion [12,13,16,23,26-301.T cell/B cell interaction [31], the mixed lymphocyte reaction, and rosette formation of human T lymphocytes to human or sheep red blood cells [22-351. As patients suffering from various diseases such as Hodgkin's disease or sarcoidosis were reported to have unknown serum factors that inhibit E rosette formation [36-381 we hypothesized that a circulating soluble form of CD58 (sCD58) could be responsible for this phenomenon.

Here we have purified and molecularly characterized a naturally occurring soluble form of CD58. Moreover, sera of normal controls and patients suffering from various immunological disorders were analyzed for sCD58 concen- trations in a newly established ELISA.

[I 120051 2 Materials and methods * Supported by a grant from Tumorzentrum Heidelberg-Mann-

heim 2.1 mAb

mAb directed at different epitopes of human CD58 [17] were AICD58.1 (epitope 4 of domain 1, IgGZ,), AICD58.5 (epitope 5 of domain 2), AICD58.6 (epitope 1 of

Correspondence: Stefan C. Meuer, Applied Immunology, German Cancer Research Center, Im Neuenheimer Feld 280, D-69120 Heidelberg, FRG (Fax: 49/62 21/42 25 61)

Key words: Lymphocyte activation / Intercellular adhesion / Immune disorder / CD2 / CD58

domain 1, IgGh), AICD58.16 (epitope 6 of domain 2, IgGI) and TS2/9 (epitope 1 of domain 1, IgGI) which was

0 VCH Verlagsgesellschaft mbH, D-69451 Weinheim, 1993 0014-2980/93/1111-3003$10.00 + .25/0

3004

obtained from the ATCC.The mAb GT2 (IgGI) directed at the CD2 molecule was a generous gift from D. Cantrell, London, GB. The mAb K7/65 (IgGz,) directed at H-2Kb was kindly provided by G. Hammerling, German Cancer Research Center [39].The anti-TllIB-mAb 3PT 2H9 (IgG1) was a gift from E. Reinherz, Boston, MA.

J. C. Hoffmann, T. J. Dengler, €? A. Knolle et al. Eur. J. Immunol. 1993. 23: 3003-3010

2.2 Lymphokines and mitogens

rIL-1P was purchased from Genzyme Laboratories (Bos- ton, MA) and was used at a final concentration of 100 U/ml. TNF-a (a generous gift from Knoll AG, Ludwigsha- fedFRG) was diluted to a final concentration of 25 ng/ml. A stock solution of PMA (Sigma, Munich, FRG) in 99% ethanol was diluted giving an end concentration of lop8 M. DMSO (Merck, Darmstadt/FRG) was used at 1.25% VN, and hemin (Sigma, Munich, FRG) at a concentration of 10 pM.

2.3 Cells and cell lines

The cell lines L428 and L540 were a generous gift from V. Diehl, University of Cologne, FRG, the cell lines HO and CO were kindly provided by H. Stein, Free University of Berlin, FRG. Hep G2 and A549 were obtained from the Tumor Bank at the German Cancer Research Center. The cell lines Hep G2, A549, and the cell lines K562, HL60, U937, the Jurkat variant JMP 1.4 [23], LAZ 509 and Daudi were grown in RPMI supplemented with 10% FCS, 2 mM glutamine and 1 YO penicillin/streptomycin (all from Gibco, Karlsruhe, FRG). The fibroblast cell line MS104 esta- blished form oral mucosa was kindly provided by E.-M. de Villiers, German Cancer Research Center, and was grown in DMEM supplemented with 15% FCS, 2mM glutamine and 1% penicillin/streptomycin at 37"C, 5% COz and 100% humidity. All cell lines were mycoplasma nega- tive by DNA staining with benzimidazole (Sigma, Munich, FRG) .

2.4 Urine and serum samples

Urine samples were taken from seven healthy human males, the pH was adjusted to pH 7.0 using 10 M NaOH, and the samples were centrifuged at 200 x g for 35 min. Sera from 36 healthy control persons were taken from employees of the German Cancer Research Center, sera from HIV-positive patients were obtained form D. A. Cooper, (Sydney, Australia) sera from patients with differ- ent forms of hepatitis were a generous gift form G. Hess (University of Mainz, FRG), sera of patients with Hodg- kin's disease, non-Hodgkin lymphoma, sarcoidosis, rheu- matoid arthritis, sero-negative spondylarthropathies and vasculitis were kindly provided by B. Heilig (University of Heidelberg, FRG). All samples were stored at -20°C.

2.5 ELISA for sCD58 in human serum and urine

Ninety-six well flat-bottom plates (Maxisorp, Nunc, Den- mark) were coated with 100 pl of mAb AICD58.16 ( 5 pg/ml) in PBS overnight at 4°C. All following steps were performed in the cold room unless otherwise stated. After

two washes with PBS/O.l YO Tween (washing buffer), the plates were blocked with PBS/1% BSA for 1 h, washed once and stored at -20°C for up to a fortnight.

After thawing the plates were washed once and samples were added in duplicates to two different trays at 50 pVwel1. As standards serial dilutions of rCD58 in PBUO.1 YO BSA were used giving concentrations of 200, 100, 50, 25, 2.5 ng/ml and PBS/O.l% BSA as blank. After a 2 h incubation, and five washes, 100 pl AICD mAb 58.6 (5 pg/ml) in PBS were added and incubated for 1 h. An irrelevant mouse mAb (K7/65, 5 pg/ml) in PBS was employed to exclude false positive values due to HAMA (human anti-mouse antibody) in patient sera.Three washes were folowed by The addition of 100 pl of a mouse Ig2a-specific goat antibody coupled to alkaline phospha- tase (1:600 v/v, Dianova, Hamburg, FRG) in PBS. This incubation was stopped after 1 h by washing three times, and the assay was developed at room temperature using 100 pl p-nitrophenyl-phosphate (Sigma) per well. The reaction was stopped after 15 min by the addition of 50 pl 0.6 M NaOH, and the absorbance was determined at 405 nm by a SLT ELISA reader (Salzburg, Austria). The concentration of sCD58 was expressed in ng/ml by refer- ence to the standard curve that was constructed from the above-mentioned standards using an SLT software pro- gram.

2.6 ELISA for sCD58 from cell culture supernatants

The protocol was essentially as described above with two modifications. Instead of PBS/O. 1% BSA complete medium (RPMT with 10% FCS) was used as the standard matrix at rCD58 concentrations of 25, 12.5, 2.5 and 0.25 ng/ml. The final enzyme reaction was stopped at 30 min and the absorbance was measured as above.

2.7 Purification of sCD58 from L428 supernatant

L428 cells were grown in RPMI/10% FCS, washed in serum-free CG-Medium (Camon, WiesbadedFRG), seeded at 3 x 105/ml serum-free CG medium supple- mented with 2 mM glutamine and 1% penicillidstreptomy- cin, and grown for 48 h.The viability was found to be above 97% by Trypan blue staining. Supernatant (1.8 1) was centrifuged at 2000 x g for 1 h, filtered (0,45 pm), and 1 pg/ml Aprotinin (Sigma) was added. Subsequently a sequential NH? SO? precipitation (70% .+ 78%) was per- formed resulting in a threefold enrichment of sCD58. This precipitate was dialyzed against PBS at 4°C and subjected to affinity chromatography using two columns in series at a flow rate of 20 ml/h. The first column consisted of 2 ml protein G-Sepharose (Fastflow, Pharmacia, Uppsala, Sweden), followed by a second column of 4ml CNBr- Sepharose (Pharmacia) coupled to the anti CD58 mAb TS2/9 (4mg/ml). The second column was washed with 30column volumes of PBS at a flow rate of 60mVh, followed by elution with 6 column volumes of 50 mM glycine HCl, pH 3, 0.25 M NaCl at a flow rate of 20 mVh. The collected fractions were immediately neutralized by the addition of 0.1 volume of 1 M Tris-HC1 pH 7.5, pooled and dialyzed against PBS.

Eur. J. Immunol. 1093. 23: 3003-3010 A soluble form of human CDS8 3005

2.8 Partial purification of sCD58 from human serum

Two liters of blood were obtained from 15 healthy controls and centrifuged at 2000 x g for 30 min. A sequential NH4SO4 precipitation (50% + 80%) was followed by affinity chromatography as above except for the preclearing first column. Instead of the protein G-Sepharose mouse IgG coupled to CNBr-Sepharose was used.

2.9 Purification of sCD58 from human urine

Human urine (15 I ) was prepared as above, filtered (0.45 pm), and passed over a series of three different columns in three separate runs. The first column consisted of 100ml of G-25 (Pharmacia) followed by the second column with 5 ml of mouse IgG coupled to CNBr Sepha- rose ( 5 mg/ml). The final column was a CD58 affinity column as for supernatant (see above). The third column was extensively washed with PBS and bound sCD58 was eluted as above. The eluate was dialyzed against 20 mM Tris, pH 8, h M PMSF and subjected to anion exchange HPLC. Briefly a Mono Q 5 X 50 mm column (Pharmacia) was loaded at a flow rate of 0.5 ml/min and sCD58 was eluted at a flow rate of 1 mumin in two peaks at 150 and 230 mM NaCl using a gradient from 65 to 500 mM NaCl. Both peaks were separately dialyzed against 15 mM NHdC03 and lyophilized.

2.10 SDS-polyacrylamide gel electrophoresis, electroblotting and deglycosylation

Cells (3 X lo6) were lysed for 1 h at 4°C with lysis buffer (20mM Tris, pH7.5, 10mM NaF, 150mM NaCl, 1 mM PMSF and 1 YO Triton X-100) and cell debris was removed by centrifugation at 15 OOO x g and 4°C for 15 min. Cell lysate or proteins were resolved by SDS-PAGE (12% acrylamide) under reducing or nonreducing conditions as indicated. Proteins were visualized either by silver staining [40], or electroblotted onto nitrocellulose sheets using a semidry blotting chamber (CTI, Idstein, FRG) according to the manufacturer’s instructions. Western blotting was per- formed as previously described using AICD58.6 (30 pg/ml) as first antibody and an alkaline phosphatase-conjugated isotype-specific goat anti-mouse antiserum (Dianova) as second antibody [41].

N-glycanase was purchased from Boehringer Mannheim. Lyophilizated purified sCD58 (1.25 pg) was deglycosilated according to the manufacturer’s instructions. In parallel 0.5 pg of sCD58 was mock treated without N-glycanase. 0.5 pg of the deglycosylated and 0.5 pg of the mock-treated sCD58 were subjected to SDS-PAGE (15 % acrylamide) under reducing conditions and silver stained. O n the same polyacrylamide gel 0.75 pg of deglycosylated sCD58 was separated and visualized by Western blotting using the AICD58.6 antibody.

2.11 Molecular weight determination by gel filtration

Analytical molecular weight determination of sCD58 was performed on a HiLoad Superdex 200 column (Pharma- cia). Briefly 200 p1 of purified sCD58 was loaded onto the

column at a flow rate of 0.5 ml/min. sCD58 was eluted with 50 mM Tris, 0.5 M NaCl, pH 7.6 at a flow rate of 1 ml/min and 1 ml fractions were collected. All fractions were analyzed for sCD58 by ELISA as above.

2.12 CD2 binding assay

JMP 1.4 cells or Daudi cells (4 x lo5) were incubated simultaneously with serial dilutions of purified sCD58 or recombinant CD58, and with saturating concentrations of the mAB AICD58.5 (50 yglml) in PBS for 1 h at 4°C. Subsequently cells were washed twice with PBS and incubated with FITC-labeled mouse IgG1-specific goat Ab (Dianova) at a 1:60 dilution in PBS. After two final washes labeled cells were resuspended in 200 pI PBS containing 1 % formalin, and fluorescence was determined using a Coulter Epics Profile Cytofluorograph.

2.13 E rosette formation

JMP 1.4 cells were washed in RPMI/2% FCS and subse- quently incubated with 100 pCi of Nas1Cr04 (Amersham, BraunschweigFRG) per lo7 cells for 1 h at room temper- ature. After two washes radiolabeled cells were resus- pended in 500 pl/107 cells and 50 p1 were aliquoted in 4-ml polystyrene tubes containing 1 ml of washing medium anti-CD2 hybridoma supernatant, rCD58 or sCD58 in washing medium (each sample in duplicate). Following an incubation over 1 h at room temperature 20 pl of 5% human erythrocyte suspension was added, gently mixed and centrifuged at 600 rpm. After an incubation of 45 min at room temperature, the pellet was gently resuspended and underlayered with 2 ml Ficoll (Pharmacia). Centrifugation at 1000 rpm for 20 min and 15 min at 2000 rpm (decelera- tion without brake) resulted in a gradient with erythrocytes and E rosettes as a pellet and nonadherent Jurkat cells at the interphase. Percentage of E-rosette formation was calculated by measuring the radioactivity of the interphase and the pellet after substraction of background values.

2.14 Statistical analysis

Statistical analysis was performed using Wilcoxan rank sum test for independent samples, i .e. Mann Whitney U-Test.

3 Results

3.1 Detection of sCD58 in cell culture supernatants, human serum, and urine

The mAb AICD58.16 directed against the CD58-domain 2 and the mAb AICD58.6 reactive with domain 1 [17] were employed for the detection of CD58 in a sandwich ELISA. sCD58 was detected in human serum, human urine and supernatants of certain cell lines (see below). To avoid measuring membrane-bound CD58 from cell debris rather than from soluble CD58, serum, urine, and supernatant samples were compared before and after centrifugation at 100000 x g for 1 h at 4°C. sCD58 levels were unaltered after ultracentrifugation (data not shown). For quantita- tion of the ELISA, recombinant human CD58 was

3006 J. C. Hoffmann, T. J. Dengler, F! A . Knolle et al. Eur. J. Immunol. 1993. 23: 3003-3010

employed [23]. The detection limit for sCD58 was calcu- lated to be 0.2 - 0.8 ng/ml for serum and urine and 0.05 - 0.1 ng/ml for supernatant (background f SD).

3.2 Serum sCD58 levels in patients and healthy controls

Thirty-two healthy controls had a mean serum sCD58 level of 20 k 5.4 ng/ml (mean ISD). While patients with rheu- matoid arthritis (n = 78, mean 15.3 f 3.6), vasculitis (n = 8, mean = 15.1 f 2.9), connective tissue diseases (n = 55, mean = 16.2 k 4.8), multiple myeloma (n = 26, mean = 15.7 f 5.8), Hodgkin's disease (n = 22, mean = 15.4 f 4.5) and HIV (n = 60, mean = 16.5 f 6.9) had slightly decreased sCD58 levels, patients with spondylar- thropathies (n = 16, mean = 18.6 f 6.6) and low malig- nant non-Hodgkin lymphomas (n = 72, mean = 18.9 f 6.1) had normal levels of sCD58 (Fig. 1). In contrast, some patients with different forms of hepatitis (n = 108) had significantly increased levels of sCD58 (23.3 f 8.6 ng/ml) when compared to normal controls (p = 0.0185 by Wilcoxan rank sum test). However, it was not possible to correlate increased sCD58 levels with any particular etiological type of hepatitis.

3.3 Urine s C D 5 8 levels of healthy controls

sCD58 was also detected in urine from seven healthy males with a mean of 6.8 k 1 ng/ml giving a normal range of 4.8-8.8 ng/ml.

3.4 sCD58 levels in supernatants of different cell lines

As CD58 is a widely expressed surface molecule, superna- tants of different cell lines grown in vitro were analyzed as potential sources of sCD58 (Table 1). Listed are sCD58 amounts per lo6 cells as well as cell surface expression of CD58 by the respective cells analyzed by flow cytometry. sCD58 was not detectable in the supernatant of the fibroblast cell line MS104 with and without rIL-lP. While CD58 surface expression by the premyloid cell line HL-60 was strongly induced after differentiation with both PMA (monocytic) and DMSO (granulocytic) sCD58-levels were below the detection limit (0.05 ng/106 cells) before and after stimulation. A similar result was found for the T cell

leukemia cell line MOLT4. In contrast, the erythroleu- kemic cell line K562 was not only strongly CD58+ but small amounts of sCD58 were also found in the supernatant. In addition, megakaryocytic differentiation induction by PMA markedly increased the sCD58 level more than tenfold. Induction of erythroid differentiation by hemin also resulted in a higher sCD58 level while surface expres- sion remained unchanged. The B lymphoblastoid cell line LAZ509 as well as the Hodgkin cell line L428 (B cell like) showed both strong surface expression as well as sCD58 levels of more than 1 ng/106 cells. Interestingly, Hodgkin cell lines L540 and CO (T cell like) were found to contain sCD58 in their supernatants.This was, however, not true for the Hodgkin cell line HO which is also thought to be of T cell origin. Furthermore, sCD58 in the supernatant of the adenocarcinoma cell line A549 was only detectable follow- ing stimulation with rIL-lP.

As significantly elevated serum sCD58 levels were found in different forms of hepatitis we investigated whether differ- ent cytokines or PMA could induce the hepatoceullular carcinoma cell line Hep G2 to release sCD58. Indeed both PMA and TNF-a were able to increase the sCD58 amount per lo6 Hep G2 cells (threefold and fivefold, respectively), while the sCD58 level remained unchanged after stimula- tion with rIL-1P.

3.5 Characterization of sCD58 with monoclonal antibodies

Since the membrane-bound form of CD58 consists of two separate extracellular domains [18] it was of interest to investigate whether sCD58 in human serum exists as the complete molecule, or whether truncated forms of CD58 might be circulating. To this end, sera of 15 normal controls were analyzed by two different ELISA specific for domain 1 or domains 1 and 2, respectively. mAbTS2/9 and AICD58.1 were employed for detection of domain 1 only, and the mAb AICD58.16 (domain 2) and AICD58.6 (domain 1) were used to detect the whole extracellular part of the CD58 molecule. Since similar concentrations were found in both systems (correlation coefficient = 0.92) it is clear that sCD58 expresses epitopes existing on domains 1 and 2, respectively, and, thus, likely comprises the majority of the extracellular portion of the CD58 molecule.

J Nc RA VuculiUr CTD Smddonin HD MM HIV SA LMNHL H.p.aan

Figure I. Serum cCD58 concentra- tions in patients with various immuno- logic disorders. cCD58 was measured in the CD58 ELISA for the indicated patient populations: normal controls (NC), RA (rheumatoid arthritis), con- nective tissue disease (CTD), Hodg- kin's disease (HD), multiple myeloma (MM), spondylarthropathies (SP), low malignant non-Hodgkin lympho- ma (LMNHL).

Eur. J. Immunol. 1993. 23: 3003-3010 A soluble form of human CD58 3007

Table 1. sCD58 in supernatants of different human cell lines")

Cell type Cell line Stimulus sCD58/106 Mean cells fluorescence

Fibroblast

Adenocarcinoma

Hcpatocellular carcinoma

Premyloid leukemia Granulocytic Monocytic Erythroleukemia Megacaryocytic Erythroid B lymphoblastoid Hodgkin

T cell leukemia

MS 104

A 549

Hep G2

HL60 HL60 HL60 W 6 2 W 6 2 K562

LAZ509 LA28 L540 co HO

Molt 4

- IL-1

IL-1

IL-1 TNF PMA

DMSO PMA

PMA Hemin

-

-

-

-

-

- PMA

< Min < Min < Min 0.50 0.35 0.20 1.53 1.28

< Min < Min < Min 0.19 2.40 0.50 1 .oo 1.10 2.30 0.35

< Min < Min < Min

3.6 Purification and biochemical characterization of sCD58 from cell culture supernatant, human serum and human urine

sCD58 was isolated from L428 supernatant employing a sequential NHdSOd precipitation and affinity chromatogra- phy using the mAb TS2/9. Western blot analysis of purified sCD58 yielded a broad band at 40-60 kDa as compared to 45-70 kDa for total cell lysate (Fig. 2A and B, respective- ly). The molecular mass of sCD58 partially purified from human serum was similar (Fig. 2C). For purification of larger quantities of sCD58,15 1 of human urine was passed over the above-mentioned affinity column, and further purified by HPLC/anion exchange chromatography.

7.60 8.80

16.50 20.70 10.10 13.50 17.40 13.00 9.20

24.90 16.00 68.60 97.60 72.60 53.50 46.50 18.10 20.70 15.60 10.60 26.60

a) Indicated cell lines were seeded at a cell density of 4 x lo5 cells/ml and cultured for 72 h in complete medium (-) or with 100 Ulml rIL-p, 25 nglml TNF-a, 1 O P X M PMA, 1.25% V N DMSO or lo-* M hemin. Cells were harvested, counted by Trypan staining and 4 X lo5 cells were used for indirect immunofluorescence employing mAb TS2/9 (20 &ml) For cell lines having a higher than 95% viability sCD58 con- centrations were measured by ELISA and sCD58/106 cells were determined by using the respective cell counts.

Fig. 2D shows samples of each purification step on SDS- PAGE under nonreducing conditions. sCD58 eluted from the anion exchange column in two separate peaks at 150 and 230 mM NaCl, respectively, both giving a similar signal of 40-70 kDa in a Western blot (Fig. 2D and E). The molecu- lar weight remained unchanged under reducing conditions and silver staining (Fig. 2F).

Subsequently the molecular mass of sCD58 as determined on SDS-PAGE under denaturating conditions was com- pared to the molecule obtained by gel filtration on a HiLoad Superdex200 column. In the latter system the molecular mass of sCD58 was found to be in the range 118-166 kDa with a peak at 139 kDa (data not shown).

Figure2. Different forms of CD58. Partially purified cCD58 from L428 cell culture supernatant (A) was ana- lyzed in comparison to CD58 from LA28 cell lysate (B) by SDS-PAGE under nonreducing conditions fol- lowed by Western blotting employing the mAb AICD58.6. (c) rCD58 (2) and partially purified cCD58 from human serum (1) are shown after SDS-PAGE and Western blotting as above. (D) Different purification steps

of cCD58 from human urine are analyzed by SDS-PAGE under nonreducing conditions and silver staining: (1) total urinary protein, (2) cCD58 afterTS2/9 affinity purification, (3) cCD58 from the first peak at 150 mM NaCl and (4) the second peak at 230 mM NaCl on a Mono Q anion-exchange column. (E) Western blotting of cCD58 as in D, 3, and D, 4, respectively. (F) cCD58 as in D, 4, after SDS-PAGE under reducing conditions and silver staining.

Eur. J. Immunol. 1993. 23: 3003-3010 3008 J. C. Hoffmann, T. J. Dengler, P. A. Knolle et al.

A B an mAb which is directed at domain 2 of the molecule and does not inhibit the CD2/CD58 interaction. Both, oligo- meric rCD58 as well as the sCD58 bound to CD2 positive cells in a dose-dependent fashion (Fig. 4). In contrast, the CD2 cell line Daudi did not show CD58 binding. Further- more, binding of sCD58 to CD2 on JMP 1.4 cells was blocked by preincubation with a CD2 mAb which is directed at a binding site for CD58 (data not shown).

3.8 Blocking of E rosette formation by sCD58 compared to rCD58

To investigate whether sCD58 might interfere with the CD2/CD58 interaction sCD58 was tested at various con- centrations for its ability to inhibit E rosette formation between the T cell leukemia line JMP 1.4 and human erythrocytes. As shown in Fig. 5 , sCD58 significantly

tions above 400 ng’ml (23 ’ 5 % (SEM).

Figure 3. Protein backbonc of cCDS8 after deglycosylation.

compared to N-glycanase alone (A, 1) by SDS-PAGE under reducing conditions and silver staining. (B) cCD58 was deglycosy- lated and Western blotting was performed from the same gel as in A using the mAb AICD58.6.

cCD58 was digested by N-glycanase (A, 2) or untreated (A, 3) and inhibits rosette formation, however, Only at concentra-

% Inhibition

Purified sCD58 (1.75 pg) was deglycosylated employing N-glycanase. Two bands at 25.5 and 27 kDa were detected after SDS-PAGE under reducing conditions (15% acryl-

Western blotting with the mAb AICD58.6 (Fig. 3A and B). Mock-treated sCD58 and N-glycanase are shown for com-

*O-

amide) by silver staining, and confirmed to be CD58 by 60-

parison. 40 -

3.7 Binding of sCD58 to CD2

To determine whether sCD58 binds to CD2, the Jurkat variant JMP 1.4 was incubated with purified sCD58 or * 1 10 100 1000

recombinant CD58. Bound CD58 was detected employing

%positive

CD58 (ng/ml)

Figure 4. sCD58 binds to CD2+ cells. Different concentrations of rCDS8 (0) or sCD58 ( W ) were mixed with a non-E-rosette blocking antLCD58 mAb (AICD58.S) and incubated for 1 h with either thc Jurkat variant JMP 1.4 (0 and W) or the CD2- cell line Daudi (0). After incubation with a FITC-labeled second antibody the percentage of CDSW cells was determined by flow cytometry. Shown is one of three rcprcscntativc cxpcrimcnls.

CD58 (nglml)

Figure 5. Dose-dependent inhibition of E-rosette formation by purified sCD58 from human urine. Radiolabeled JMP 1.4 cells were incubated with increasing concentrations of rCD58 (0) or sCD58 (0). After addition of human erythrocytes, rosetting Jurkat cells were separated from non-rosetting cells by Ficoll separation. Inhibition of E-rosette formation was calculatcd from values obtained from rosetting versus non-rosetting cells after subtraction of the background. Indicated are means of duplicates with the SEM .

4 Discussion

We have identified a soluble form of CD58 in human urine, serum and cell culture supernatants using a sensitive sandwich ELISA. In addition, sCD58 was purified from human urine and partially purified from supernatant of the Hodgkin-derived cell line L428 as well as from human serum. The apparent molecular mass of sCD58 from these sources was found to be 40-70 kDa on SDS-PAGE and Western blotting. Thus, the molecular mass of sCD58 is similar to CD58 from total cell lysate of L428 cells (45-70 kDa) or the reported molecular mass of CDS8 from erythrocytes (45-66 kDa) [42] or the B lymphoblastoid cell

Eur. J. Immunol. 1993. 23: 3003-3010 A soluble form of human CDS8 3009

line JY (50-70 kDa) [43]. Western blotting of sCD58 using the mAb AICDS8.6 directed against the N-terminal domain 1 suggests that sCD58 contains most of the extra- cellular portion of the membrane-anchored CD58. It is unlikely that a truncated domain 1 was lost during the purification procedure because sCDS8 was purified over a TS2/9 affinity column which also binds domain 1. No additional peak was detected in anion exchange chromato- graphy. Furthermore, comparison of sera from 15 healthy controls in two different sandwich ELISA employing mAb directed against domain 1 only or against domain 1 plus domain 2 gave similar sCD58 concentrations (correlation coefficient of 0.92).

Two distinct cell surface forms of CDS8 have previously been reported with a similar extracellular portion but differing in their membrane attachment. One has a trans- membrane and a short cytoplasmic domain whereas the other one is attached to a phosphatidylinositol (PI)-linker (PI-linked) [18, 42-44]. These two forms are generated from the same chromosomal DNA by alternative RNA splicing. To understand the mechanisms which might under- lie the generation of a circulating form of CD58, several possibilities have to be considered. sCDS8 might result from cell debris of CD58+ cells. However, since sCD58 was not detected in culture supernatants of several CD58+ cell lines, but clearly detectable in B cell-derived lymphoma cell lines, some T cell-derived Hodgkin cell lines, the erythro- leukemic cell line K562 and in the hepatocellular carcinoma cell line Hep G2, release of sCD58 appears to be restricted to certain cell types rather than being a general feature of CD58+ cell types. Southern blotting of different cell lines and tissues, as well as cloning of the genomic CD58 DNA have clearly demonstrated that only one gene encodes CDS8 ([18,43] and Roux et al., manuscript in preparation). cDNA for the PI-linked and the transmembrane form have been isolated from the JY cell line and PBL, respectively [18,43]. It cannot be excluded, however, that a third secreted form is generated by differential RNA-splicing as reported for circulating MHC class I molecules [45]. Final- ly, sCDS8 might be released from the cell surface by shedding or enzymatic cleavage. Preliminary data exist in this laboratory which support the latter possibility. Accord- ingly, differential amounts of enzymes in the cell lines investigated here would result in release of differential quantities of sCDS8 from the cell surface. If shedding or enzymatic cleavage were solely responsible for the exis- tence of sCDS8, the predicted size of the protein backbone of a shed transmembrane form or a cleaved PI-linked form would be 25.5 kDa.

Shedding of surface molecules is usually accompanied by down-modulation of its surface expression. Since we were unable to find down-modulation of CD58 cell surface expression accompanying release of sCD58 in different cell lines, shedding appears to be an unlikely mechanism. However, phospholipase D which is present in human serum might well cleave the PI-linked form off the cell surface and generate the 25.5 kDa form which was detected after deglycosylation. In addition, deglycosylation of sCDS8 gave a second band at 27 kDa by SDS-PAGE and Western blotting.The origin of this 27 kDa band remains a matter of speculation until C-terminal amino acid sequences are available.

Previous studies with recombinant CDS8, purified CDS8 from PI-specific phospholipase C-treated JY cells, or purified CD58 from cell lysates have shown that its functional potency is directly linked to its valency [23, 32, 331. While sCD58 has a relative molecular mass of 40-70 kDa under denaturating conditions, its molecular size was found to be 118-166 kDa, with a peak at 139 kDa, by gel filtration. This suggests that sCDS8 exists as a noncovalently associated homotrimer or, alternatively, its association with other molecules. The reduced capacity of sCD58 to block E rosette formation as compared to rCD.58 which exists as a decamer [23] could be explained on the basis of its quaternary composition. Nevertheless, sCDS8 binds cells and blocks the CD2/CDS8 interaction at con- centrations above 400 ng/ml. These findings are in accord- ance with data reported by Pepinsky et al. about chemically cross-linked recombinant CD58 where rCDS8 trimers were shown to block E rosette formation [33]. However, such rCD58 trimers had a very low comitogenic activity towards T lymphocytes when used in combination with submito- genic concentrations of CD2 antibodies.

Elevated serum levels of sCD58 were detected in patients suffering from various forms of hepatitis. This may simply reflect a quantitative aspect since one would expect much more soluble CD58 to be released by an inflamed liver rather than by any other organ.Whether cytokines might be involved in the generation and local release of sCDS8 in vivo was investigated by treating the hepatocellular carcinoma cell line Hap G2 withTNF-a which resulted in a fivefold increase of sCD58 in vitro. In addition, hepatocytes of patients suffering from chronic active hepatitis stain strongly positive for CD58, while hepatocytes of normal controls show no CD58 expression [46, 471. Importantly, hepatocyte staining was not restricted to the membrane, but was also found in the cytoplasm. Therefore, it seems likely that cytokines induce secretion of sCDS8 by hepato- cytes in the course of an inflammatory reaction. The serum levels would simply reflect the local production of high quantities of sCD58 in the liver.

While binding of T lymphocytes to various cell types via adhesion receptors is well documented, little is known about mechanisms of deadhesion. For the receptor/ligand pair LFA-KDS4 (ICAM-1) the conformational modula- tion and avidity change of the LFA-1 molecule is held responsible for both adhesion strengthening and deadhe- sion [48]. While strengthening of adhesion via the CD2/CDS8 interaction appears to be regulated on a surface expression level, a mechanism of deadhesion for the CD2/CD58 adhesion pathway is still lacking. One could, however, speculate that binding of T cells to CD58+ cells results in T cell activation via CD2 followed by up- regulation of CD2 surface expression and release of lymphokines. The cellular counterpart (e.g. monocytes) becomes activated through CDS8, produces cytokines (e .g . TNF-a and IL-lp), and up-regulates membrane expression of CDS8 and the release of sCDS8 [49]. Release of sCD58 which blocks intercellular adhesion might support de- adhesion.

Taken together, under physiological conditions sCDS8 might be involved in de-adhesion of T lymphocytes from CD58+ cells. Under pathological conditions, local release of high amounts of sCDS8, thus altering the balance

3010

between membrane versus soluble CD58 molecules, may block intercellular adhesion and T lymphocyte activation. Such a mechanism could be involved in modulating the course of an inflammatory reaction and - depending on inter-individual variations in the regulation of sCD58 induction - contribute to chronicity.

J. C. Hoffmann, T. J. Dengler, P. A. Knolle et al. Eur. J. Immunol. 1993. 23: 3003-3010

We thank B. Heilig, D. A . Cooper and G. Hess for generously providing patient sera. We gratefully acknowledge the help of H. Kirchgeflner for assistance with chromatography procedures. Finally we wish to thank M . W Hoffmann for critical reading of the manuscript and L. Edler for support in statistical analyses.

Received June 22,1993: in revised form August 16,1993; accepted August 18, 1993.

5 References

1 Brazil,V., Horejsi,V., Baudys, M., Kristofova, H. , Strominger, J. L. and Hilgert, I., Eur. J. Immunol. 1986. 16: 1583.

2 Brazil,V., Baudys, M., Hilgert, I., Stefanova, I., Low, M. G., Zbrozek, J. and Horejsi,V., Mol . Immunol. 1989. 26: 657.

3 Engelmann, H., Aderka, D., Rubinstein, M., Rotman, D. and Wallach, D., J. Biol. Chem. 1989. 264: 11974.

4 Engelmann, H., Novick. D. and Wallach, D., J. Biol. Chem. 1990. 265: 1531.

5 Symons. J. A.,Wood, N. C., Di Giovine, F. S. and Duff, G . W., J. Immunol. 1988. 141: 2612.

6 Marcon, L., Fritz, M. E., Kurman, C. C., Jensen, J. C. and Nelson, D. L., Clin. Exp. Immunol. 1988. 73: 29.

7 Hintzen, R. Q., de Jong, R., Hack, C. E., Chamuleau, M., de Vries, E. F. R., ten Berge, I. J. M., Borst, J. and van Lier, R. A. W., J. Immunol. 1991. 147: 29.

8 Novick, D., Engelmann, H.,Wallach, D. and Rubinstein, M., J. Exp. Med. 1989. 170: 1409.

9 Seth, R., Raymond, E D. and Makgoba, M. W., Lancer. 1991. 338: 83.

10 Tsujisaki, M., Imai, K., Hirata, H., Hanzawa,Y., Masuya, J., Nakano,T. and Sugiyama,T., Clin. Exp. Immunol. 1991.85: 3 .

11 Rothlein, R., Mainolfi, E. A., Czajkowski, M. and Martin, S. D., J. Immunol. 1991. 147: 3788.

12 Krensky, A. M., Sanchcz-Madrid, F., Robbins, E.,Nagy, J. A, , Springer,T. A. and Burakoff, S. J., J. lmmunol. 1983.131: 611.

13 Takai,Y., Reed, M. L., Burakoff, S. J. and Herrmann, S. H. , Proc. Natl. Acad. Sci. USA 1987. 4: 6864.

14 Hiinig, T., Tiefenthaler, G., Meyer zum Biischenfelde, K.-H. and Meuer, S. C., Nature 1987. 326: 298.

15 Selvaraj, P , Plunkett, M. L., Dustin, M., Sanders, M. E., Shaw, S. and Springer, T. A., Nature 1987. 326: 400.

16 Dustin, M., Sanders, M. E . , Shaw, S. and Springer, T. A., J. Exp. Med. 1987. 165: 677.

17 Dengler, T. J., Hoffmann, J. C., Knolle, P., Albert-Wolf, M., Roux, M.,Wallich, R. and Meuer, S. C., Eur. J. Immunol. 1992. 22: 2809.

18 Seed, B., Nature 1987. 329: 840. 19 Bierer, B. E., Peterson, A., Gorga, J. C.. Herrmann, S. H. and

Burakoff, S. J., J. Exp. Med. 1988. 168: 1145.

20 Koyasu, S., Lawton, T., Novick, D., Recny, M. A., Siliciano, R. F. ,Wallner, B. P. and Reinherz, E. L., Proc. Natl. Acad. Sci. USA 1990. 87: 2603.

21 Siliciano, R. F., Pratt, J. C., Schmidt, R. E., Ritz, J. and Reinherz, E. L., Nature 198.5. 317: 428.

22 Tiefenthaler, G., Hiinig,T., Dustin, M. L., Springer,T. A. and Meuer, S. C., Eur. J. Immunol. 1987. 17: 1847.

23 Albert-Wolf, M., Wallich, R. and Meuer, S. C., Int. Immunol. 1991. 3: 1335.

24 Moingeon, P., Chang, H.-C.,Wallner, B. P., Stebbins, S., Frey, A. Z. and Reinherz, E. L., Nature 1989. 339: 312.

25 Denning, S. M., Tuck, D. T., Vollger, L. W., Springer, T. A., Singer, K. H., Haynes, B. F., J. Immunol. 1987. 139: 2573.

26 Sanchez-Madrid, F., Krensky, A. M.,Ware, C. E, Robbins, E., Strominger, J. L., Burakoff, S. J. and Springer, T. A., Proc. Natl. Acad. Sci. USA 1982. 79: 7489.

27 Krensky, A. M., Robbins, E., Springer, T. A. and Burakoff, S. J.. J. Immunol. 1984. 132: 2180.

28 Shaw, S., Luce, G. E. G., Quinones, R., Gress, R. E., Sprin- ger, T. A. and Sanders, M. E., Nature 1986. 323: 262.

29 Shaw, S. and Luce G. E. G., J. Immunol. 1987. 139: 1037. 30 Schirren, C. A. ,Volpel, H. and Meuer, S. C., Blood 1992. 79:

138. 31 Emilie, D.,Wallon, C., Galanaud, P., Fischer, A., Olive, D. and

32 Dustin, M. L., Olive, D. and Springer, T. A., J. Exp. Med.

33 Pepinsky, R. B., Chen, L. L., Meier,W. and Wallner, B. P , J.

34 Hiinig,T., J. Exp. Med. 1985. 162: 890. 35 Plunkett, M. L., Sanders, M. E., Selvaraj, P and Springer,

T. A., J. Exp. Med. 1987. 165: 664. 36 Fuks, Z., Strober, S. and Kaplan, H. S., N. Engl. J. Med. 1976.

295: 1273. 37 Roux, M., Schraven, B., Roux, A., Gamm, H., Mertelsmann,

R. and Meuer, S., Blood 1991. 78: 2365. 38 Davies, B. H.,Williams, J. D., Smith, M. D., Jones-Williams,

W. and Jones, K., Pathol. Res. Pract. 1982. 175: 97. 39 HSmmerling, G. J., Riisch, E.,Tada, N. and Hammerling, U.,

Proc. Natl. Acad. Sci. USA 1982. 79: 4737. 40 Morrissey, J. H., Anal. Biochem. 1981. 117: 307. 41 Schraven, B., Kirchgessner, H., Gaber, B., Samstag, Y and

Meuer, S., Eur. J. Immunol. 1991. 21: 2469. 42 Wallner, B. P., Frey, A. Z., Tizard, R., Mattaliano, R. J.,

Hession, C., Sanders, M. E., Dustin, M. L. and Springer, T. A, , J. Exp. Med. 1987. 166: 923.

43 Selvaraj, P., Dustin, M. L., Silber, R., Low, M. G. and Springer,T. A., J. Exp. Med. 1987. 166: 1011.

44 Dustin, M. L., Selvaraj, P., Mattaliano, R. J. and Springer, T. A., Nature 1987. 329: 846.

45 Cosman, D., Khoury, G. and Jay, G., Nature 1982. 295: 73. 46 Autschbach, F., Meuer, S. C., Moebius, U., Manns, M., Hess,

G., Meyer zum Biischefelde, K.-H., Thoenes, W. and Dienes, H . 2 , Hepatology 1991. 14: 223.

47 Volpes, R., van den Oord, J. J. and Desmet,V J., Hepatology 1991. 13: 826.

48 Figdor, C. G., van Kooyk, Y. and Keizer, G. D., lmmunol. Today 1990. 11: 277.

49 Webb, D. S., Shimizu, Y.,Van Seventer, G. A., Shaw, S. and Gerrard, T. L., Science 1990. 249: 1295.

Delfraissy, J.-F., J. Immunol. 1988. 141: 1912.

1989. 169: 503.

Biol. Chem. 1991. 266: 18244.