Eur. J. Immunol. 1990.20: 539-543

Jennifer A. Cairns and John Gordon Department of Immunology, The Medical School, Vincent Drive, Birmingham

Membrane CD23 is mitogenic

539

Intact, 45-kDa (membrane) form of CD23 is consistently mitogenic for normal and transformed B lymphoblasts* When isolated from lysates of the RPh4I 8866 B lymphoblastoid cell line, the CD23 antigen was found to be present be in two forms correspondingto the intact 45-kDa membrane species and to a 25-kDa product that is more usually found as a released fragment in the extracellular medium. By contrast with preparations of extracellular species of CD23 which were seen to be variable in their biological activity, cell-associated CD23 was consistently mitogenic both for autogenous transformed B lymphoblasts and for pre-activated normal B cells. Addition to the normal cocktail of protease inhibitors of Na-tosyl-L-lysine chloromethylketone (TLCK), which has selectivity for trypsin-like serine proteases, resulted in preparations of CD23 from RPMI 8866 cell lysates that were exclusively in the 45-kDa intact form; such material retained full and reliable activity in the biological assays. The implications of these observationsfor the autocrine control of B lymphocyte growth are discussed.

1 Introduction CD23 is a type I1 integral membrane protein (that is, its aminoterminus lies inside the cell) of molecular mass 45 kDa [l] that is readily detected on activated B lymphocytes [2] but can also be found on monocytes [3],Tcells [4], eosinophils and platelets [5], Langerhans cells [6] and follicular dendritic cells (FDC) [7]. A significant portion of its extracellulardomain bears a strikinghomology to C-type (Ca2+-dependent)lectins [ l , 71 although a carbohydratebinding function has yet to be described; furthermore, at least in man, the extracellularextremity contains an inverse “RGD’ sequence [7] that can be recognized by antibodies that define this functional recognition motif in adhesion molecules such as fibronectin [8]. While it is unequivocal that CD23 can bind IgE with low affinity [l], neither the lectin-like nor “RGD” regions appear to be involved [9]. This leaves open the possibility that these potentially functional structural sequences might contribute to diverse biological properties of the molecule that have been mooted but are presently more contentious [7, 91 than its IgE-binding function. Although absent or expressed at a very low level on resting B lymphocytes, CD23 can be induced or up-regulated by a number of agents which include IL 4 [lo], phorbol esters [ll] and, more recently described, I L 2 [12]. On the transformation of B cells by EBV, CD23 is super-induced and remains constitutivelyexpressed at high level [13];both transformed and appropriately activated normal B cells release soluble species of CD23 by proteolysis of membrane-bound protein over a matter of hours [14, 151. Processing is complex. Initially, transient species of around [I 80001

*

This work was supported with funding from the Medical Research Council (GB).

Correspondence: Jennifer Cairns, Department of Immunology, The Medical Schoo1,Vincent Drive, Birmingham B15 2TJ, GB Abbreviations: BCGF B cell growth factor FDC: Follicular dendritic cells TLCK: Na-Tosyl-L-lysine chloromethylketone 0 VCH Verlagsgesellschaft mbH, D-6940 Weinheim, 1990

35-kDa are produced [16] which degrade rapidly to a more stable 25-kDa form; eventually, a long-lived 14-kDa fragment, that has lost IgE-binding capacity put possibly retains the inverse “RGD” sequence, is generated [7,9]. Swendeman and Thorley-Lawson [17] were the first to demonstrate that affinity-purified soluble fragments of CD23 possessed growth-promotingactivity which appeared to be equivalent to a previously described B cell-derived B cell growth factor (B-BCGF) [28]. Several studies have since confirmed this observation [9,19,20]. However, a more recent report has now cast doubt on this concept; thus, it was shown that recombinant CD23 derived from constructs containing cDNA whch coded directly for the major 25-kDa soluble species was completely inactive in conventional BCGF assays [21]. Our own studies, while supporting the notion of a growthpromoting function for soluble CD23, have, nonetheless, indicated that the associated activity can be highly labile making its full characterization problematic [7, 19, 221. Partly in an effort to resolve this issue, but also to investigate any potential function in its own right, we attempted to purify the intact form of the molecule directly from cell lysates. The outcome of the study reported herein has important implications for understanding the control of B lymphocyte growth under conditions of high B cell density and during the formation of homotypic adhesions.

2 Materials and methods 2.1 Cells Resting (Go) tonsil B cells were prepared by negative selections and density separation on Percoll (Pharmacia, Uppsala, Sweden) as described elsewhere [2]. RPMI 8866 is a cell line of EBV-transformed lymphoblastoid cells. Cultures were initiated or maintained in RPMI 1640 culture medium supplemented with 10% FCS and antibiotics at 37 “C in a humidified 5% CO:! atmosphere. Pre-activation of normal B cells was carried out for 24 h with PMA at 1 ng/ml. 0014-2980/90/0303-0539$02.50/0

540

Eur. J. Immunol. 1990.20: 539-543

J. A. Cairns and J. Gordon

2.2 Purification of CD23

RPMI 8866 cells or culture SN were used as the source of CD23. Soluble CD23 was purified by affinity chromatography using the MHM6 antibody as described previously [191. In some experiments soluble CD23 was purified from spent culture SN of mouse L cells that had been transfected with the full-length cDNA encoding for intact (45-kDa) CD23 (a kind gift of Drs. T. Kishimoto and H. Kikutani, Tokyo, Japan). Preparations of soluble CD23 were assayed within a day of isolation. Cell-associated CD23 was purified from lo8 RPMI 8866 cells. Cells were washed two times in PBS, pH7.2, containing a cocktail of protease inhibitors (1mM PMSF, 10 m~ benzamidine hydrochloride, 50 mM E-aminocaproic acid, 20 m~ iodoacetamide); where indicated, 0.2 mM TLCK (Nu-tosyl-L-lysine chloromethylketone; Sigma, Poole, Dorset, GB) was also included.The cells were lysed in 0.5 ml of 0.5% NP40 containing the cocktail of inhibitors. Lysates were centrifuged at 10000 x g for 30 min at 4 "C, sterile-filtered and rotated overnight at 4 "C with -6-Sepharose gel. The gel was washed with 50 bed volumes of PBS containing 0.5% NP40 and 0.2 mM TLCK where indicated before eluting bound material with 1ml of 0.2 M glycine-HC1, pH 2.3 (plus TLCK where indicated) which was immediately neutralized by the addition of 1 M Tris base. The eluted material was assayed by ELISA for CD23 as previously described [15]. For bioassays, the eluted material was dialyzed against four successive changes of PBS at 4 "C and used immediately. To assess the purity of CD23 preparations, the material was dialyzed against several changes of distilled water and lyophilized. The lyophilized material was electrophoresed on a 7.5% SDS gel and visualized by silver staining [15]. 2.3 Measurement of B cell stimulation Cells cultured for 3 days as described in the text were pulsed with 0.5 pCi (18.5 kBq) of [3H]dThdin 50 p1 of medium for 16 h. The amount of radioactivity incorporated into triplicate cultures was determined; individual determinations never varied by > 10% and were usually within 5% of each other.

3 Results 3.1 Soluble fragments of recombinant CD23 can be stimulatory for B cells

Mouse L cells transfected with the full-length cDNA encoding for the membrane-bound (intact) form of CD23 were found to release into their culture SN soluble fragments of CD23 in amounts similar to those found in the culture medium of RPMI 8866 B lymphoblastoid cells (70-300 ng/106 cells/ml/24 h). Affinity-purified preparations of soluble (s) CD23 isolated from the transfected L cell SN were found to consist of a major band migrating on SDS gels at molecular mass equivalent to 25-kDa, a minor band of 14-kDa and a variable band at 35-kDa (not detailed); similar heterogeneity in the molecular species of sCD23 has been noted previously for B cell-derived fragments [7, 9, 161.

Table 1. Soluble fragment of rCD23 can be stimulatory

0 2 3 (aM)b)

rH]dThd inmxporation8)(cpm) with natural recombinant

cD23 5

1 0.2 0.04 0.008 0

29 683 329u) 23 055 17 187 8 383 4 265

30221 30893

25883 14230 5065

-

Tonsil B cells were activated for 24 h with PMA at 1 nglml before addition of CD23 for a further 2 days before receiving a pulse of t3H]dThd for the final 16 h; results expressed as mean cpm of triplicate determinations on 105 celldwell, individual determinations were always within 10% and usually within 5% of each other. sCD23 prepared from RPMI 8866 cells (natural) or mouse L cells transfected with full-length cDNA for CD23; material isolated from serum-free culture SN generated in parallel. Preparations from SN of control mouse L cells (kindly provided by €? Beverley, London) gave 3754 cpm in this experiment.

In some experiments, an example of which is given in Table 1, sCD23 isolated in parallel from the SN of transfected L cells or RPMI 8866 cells was found to be effective at amplifying the stimulation of B cells that had been pre-activated with PMA; the activity of the sCD23 was similar from the two sources. Control (non-transfected) L cells were never seen to produce such activity. However, as noted in previous studies on B cell-derived sCD23 [19,22], the "BCGF"-activity of sCD23 from both sources appeared to be extremely labile and, on some occasions, failed to be detected (for example, see Sect. 3.2).This problemencountered with soluble fragments of the molecule has also been the experience of others working in this area (D. ThorleyLawson and R. Armitage, personal communications). 3.2 Cell-derived CD23 is stimulatory for B cells Partly in an attempt to generate a more reliable source of bioactive CD23 but also to investigate a possible biological role in its own right, affinity-purified isolates of CD23 were prepared directly from lysates of RPMI 8866 cells. Initial preparations which were derived from cells lysed in the presence of a standard cocktail of protease inhibitors were comprised of material which migrated on SDS gels as two bands equivalent to 45-kDa and 25-kDa (Fig. 1); the two bands presumably reflect the membrane (intact) form and its major degradation product, respectively. CD23 isolated from RPMI 8866 cell lysates was found to be consistently active in B cell stimulation assays. InTmble 2, an example is shown of an experiment where CD23 isolated in parallel form the SN and lysate of RPMI 8866 cells was assayed for activity against both the autogenous cell line and against pre-activated normal B cells. On this particular occasion, sCD23 was relatively inactive while the cell-derived CD23 was stimulatory in both assays examined. For both autogenous lymphoblastoid cells and normal cells, a critical cell density was observed to be necessary in order to develop the stimulatory activity of the CD23 preparations (see also

Eur. J. Immunol. 1990. 20: 539-543

Membrane CD23 is mitogenic

541

Table 2. Cell-associated CD23 is stimulatory for B cells

cells

rH]dThd incorporationa) (cpm) with control ccD23b’ SCD23

per well

RPMI 8866 5 x lcy

2685

1 x 104 5 x 104

25869 109 195

21 481 67719 113 116

3035 13080 106300

1 x 104 1 x 1oT 5 x 10s

476 1460 56 954

1637 5 478

732 1758 50 575

Activated tonsil B 55 604

Cells plated at number/well indicated were pulsed with [3H]dThdfor 16 h at end of 3-day culture; results expressed as mean cpm of triplicate determinations which never varied by > 10% and were usually within 5% of each other. Cell-associated(c) or sCD23 were prepared in parallel from RPMI 8866 lysates or culture SN, respectively, and introduced into the stimulation assays at final concentrations of 2.5 nM.

Table 3. Intact. 45-kDa CD23 is stimulatory for activated B cclls

[3H]dThd incorporationa)(cpm) in Activated tonsil B

Exp

rnCD23(nhQb)

1

0 2.5

2 3

4

5

kDa

0 2.5 0 2.5 0.125 0.025 0 2.5 0.5 0.1 0.02 0 2.5

RPMI 8866 5x103 104 165 765 49564 87 462 191 533 427 280 3764 18032 19378 15525 4834

-

104

10s

267 10288 77419

6927 12 147

9091 22449 15 339 11 656 6729 40439 43677 16418 7550

-

iman

-

-

(a)

180 -+ 45-

25-

Figure 1. Characterization of cell-associated CD23. Material eluting from MHM6 an~ibody-coup~edgel isolated from 107 RPMI 8866 cells electrophoresed on 7.5% SDS-polyacrylamide and visualized by silver stain: (a) prepared in absence of TLCK; (b) prepared in presence of 0.2 mM TLCK.

4302

33 984 35 806 9513 4708 10145 63039

a) As for Table 2 in number of cells per well indicated. b) Membrane (m) form of CD23 prepared from RPMI 8866 cell lysates in presence of TLCK.

Table 3). At high cell number, the cells were presumably providing sufficient stimulatory activity of their own while at very low cell numbers, even the addition of exogenous material was not enough to provide a sustained stimulation over the culture period.

3.3 Intact, 45-kDa CD23 is mitogenic for B cells The results already described indicate that preparations of cell-associated CD23 are more reliable than soluble preparations at displaying “BCGF”-activity but do not resolve whether it is the 45-kDa or 25-kDa form that is responsible. The formation of the 25-kDa species of CD23 results from cleavage of an Arg-Met bond suggesting the possible involvement of a trypsin-like serine protease. For this reason. the selective protease inhibitor TLCK was introduced into the purification and resultant preparations from RPMI 8866 cell lysates were now found to consist of a single 45-kDa component (see Fig. 1). Thus, a reliable means of preparing the intact (membrane) form of CD23 has been found. On repeated occasions, homogeneous 45-kDa

542

J. A. Cairns and J. Gordon

CD23 prepared in this way was found to be stimulatory for autogenous cells when they were plated at a limiting cell density (Table 3). Where tested, it also displayed activity against normal pre-activated tonsillar B cells. Titration of the 45-kDa material revealed that stimulatory activity could still be detected at the 100 PM level (Table 3). It is worth noting that for both RPMI8866 cells and preactivated normal B cells neither E l (aor p form) nor IL 6, both being cytokinesthat have been implicated in autocrine B cell growth, were able to replace 45-kDa CD23 in the stimulation assays over a wide range of concentrations; thus, in this respect, stimulation appeared to be specificfor 0 2 3 (data not detailed).

4 Discussion The findings presented support the concept of CD23 being involved in B cell growth regulation and extend it by indicating a stimulatory function for the membrane-bound form of the molecule. The ability to isolate intact CD23 from cell lysates represents an important advance in the study of this glycoprotein whose complex processing has, until now, rendered full biological characterization problematic. Following the original observation that preparations of CD23 can be mitogenic for B cells [17] it has become clear that not all the extracellular fragments generated possess stimulatory activity. Thus, it has been reported that, in isolation, the major 25-kDa fragment whether in recombinant or highly purified natural form is inactive as a “BCGF’ [9,21] as are preparations that have degraded to comprize primarily the 14-kDa fragment [19]. Nevertheless, we found in the present study that mouse L cells were able to process intact human CD23 that had been introduced by gene transfer into extracellular species identical in size to the B cell-derived material and that isolates of such material were similarly of variable efficacy in the stimulation assays employed. It might seem reasonable to assume, therefore, that the highly labile BCGF activity of CD23 preparations is associated with the transient, initial cleavage products of around 35-kDa molecular mass that are variably present in the culture SN although the possibility that other fragments - or, indeed, combinations of species - are responsible must remain open. When recombinant forms of the different soluble species become available, it will be important to determine whether some proteolytic cleavage products are able to antagonize the biological actionsof others. Also, there is a need to generate more (perhaps polyclonal) antibodies to different CD23 species in order to identify those that might be neutralizingin these assays; to date, all those available to us react with the 25-kDa fragment and are stimulatory or inert in functional systems [7,11]. One distinct possibility is that the B cell stimulatory moiety of 45-kDa CD23 that is probably retained in the 35-kDa fragment is liberated as the molecule is processed further to the 25-kDa species; as all our currently available antibodies bind to the latter fragment, this would explain why the stimulatory species is so difficult to capture when using antibody affinity as the means of isolation. The consistent and reliable activity observed with intact 45-kDa CD23 could reflect a function for the molecule when membrane-anchored. This activity could be

Eur. J. Immunol. 1990.20: 539-543

expressed during the formation of homotypic adhesions allowing for the triggering of neighboring cells; such an activity had been indicated in earlier studies on B cell stimulation [20,23].Thus, CD23might belong to a family of membrane-anchored growth factors; for example, it has recently been shown for transforming growth factor a that cells expressing a mutant form of the factor where the proteolytic cleavage sites had been deleted were able to stimulate directly adjacent cells bearing appropriate receptors [24]. It is still unclear whether CD23 in either its intact membrane-bound form or as soluble fragments interacts with neighboring cells in a classical receptor-ligand fashion. An alternative explanation to account for the stimulatory capacity of 45-kDa CD23 is that when introduced into the cell assays it is converted by membrane-bound proteases to a form which can then exert a growth-promoting effect. Such a situation is seen with IL 16 where an inital35-kDa precursor needs to be cleaved to a mature 17-kDaspecies to be biologically active [25]. It should be noted that the stimulatory activity of CD23 has been observed only in situations where cells are already substantially activated (our own unpublished data) arguing in favor of a role for a membrane-associated protease. This requirement may explain, in part, the inability of one group to demonstrate BCGF activity of CD23 on B cells that had been minimally triggered by anti-Ig alone [21], a finding with which we would concur. It has been mooted that the processing of CD23 might even be autoproteolytic [9, 191; we are currently investigating whether TLCK, which binds to histadine residues [26], interacts directly with CD23. It is an attractive notion that, whether as a membraneanchored growth factor or as a transient cleavage fragment, CD23 should be active only in the immediate proximity of its production. We have recently speculated on two major roles for CD23 during the development of a B cell response to thymus-dependent antigen [7].The first of these reflecting CD23’s function as an autocrine factor would be relevant as antigen-primed B cells that had been activated by cognate interaction with Th cells on interdigitating cells moved into the follicles giving rise to areas of high B cell density during the subsequent follicular blast reaction [27]. This phase of rapid B cell proliferation feeding off CD23 produced autogenously would then give way to the development of centroblasts and centrocytes, the latter being selected for on the basis of its presumed somatically mutated antibody’s ability to bind antigen presented by follicular dendritic cells (FDC)in the germinal center light zone [7,27]. The FDC in this region display an intense expression of CD23 and we have argued that this could be involved in facilitating the survival of antigen-selected centrocytes in those locations [7]. It would, therefore, be clearly inappropriate for CD23 presented at these sites to influence nonselected cells that were destined to die thus providing a rationale both for the extremely labile nature of soluble CD23 as a “ B C G F and also for the activity of the membrane-bound precursor. In this model it is essential that the centrocytes themselves do not express CD23 and this, indeed, is what is observed [7]. It follows that the ectopic or overexpression of CD23 by some B cell subtypes might contribute to the dysregulated growth associated with malignancieswhere CD23 is expressed in an abnormal abundance [28, 291.

Eur. J. Immunol. 1990.20: 539-543 The excellent technical assistance of M. J. Millsum is acknowledged. We are indebted to Dr. C, J. Gray (Department of Chemistry, University of Birmingham) for suggesting the use of TLCK in the preparation of CD23.

Received October 9, 1989; in revised form November 30, 1989.

5 References 1 Kikutani, H., Inui, S., Sato, R., Barsumiam, E. L., Owaki, H., Yamasaki, K., Kaisho, T., Uchibayashi, N., Hardy, R. R., Hirano, T., Taunasawa, S., Saklyama, F., Suemura, M. and Kishimoto, T., Cell 1986. 47: 657. 2 Walker, L., Guy, G. R., Brown, G., Rowe, M., Milner, A. and Gordon, J., Immunology 1986. 58: 583. 3 Gonzalez-Molina, A. and Spiegelberg, H. L., J. Immunol. 1976. 117: 1838. 4 Armitage, R. J., Goff, L. K. and Beverley, I? C. L., Eur. J. Immunol. 1989. 19: 31. 5 Capron, A., Dessaint, J. I?, Capron, M., Joseph, M., Ameisen, J. C. and Tonel, A. B., Immunol. Today 1986. 7: 15. 6 Bruynzeel-Koomen, C.,Van der Donk, E. M. M., Bruynzeel, F! L. B., Capron, M., De Gast, G. C. and Mudde, G. C., Clin. Exp. lmmunol. 1988. 74: 137. 7 Gordon, J., Flores-Romo, L., Cairns, J. A., Millsum, M. J., Lane, F! J., Johnson, G. D. and MacLennan, I. C. M., Immunol. Today 1989.10: 153. 8 Grangette, C., Gruart,V., Ouaissi, M. A., Rivi, F., Delespesse, G., Capron, A. and Capron, M. J., J. Immunol., in press. 9 Delespesse, G., Sarfati, M. and Hofstetter, H., Zmmunol. Today 1989. 10: 159. 10 Defrance,T., Aubry, J. F!, Rousset, F.,Vaubervliet, B., Bonnefoy, J.Y., Arai, N.,Takebe,Y.,Yokota,T., Lee, F., Arai, K., De Vries, J. and Banchereau, J., J. Exp. Med. 1987. 165: 1459. 11 Gordon, J.,Webb, A., Walker, L., Guy, G. R. and Rowe, M., Eur. J. Immunol. 1986. 16: 1627.

Membrane CD23 is mitogenic

543

12 Hivroz, C., Valle, A., Brouet, J. C., Banchereau, J. and Grillot-Courvalin, C., Eur. J. Immunol. 1989. 19: 1025. 13 Thorley-Lawson, D. A. and Mann, K. P., J. Exp. Med. 1985. 162: 45. 14 Thorley-Lawson, D. A., Edson, C. M. and Swendeman, S. L., J. Immunol. 1986. 136: 1745. 15 Cairns, J. A., Flores-Romo, L., Millsum, M. J., Guy, G. R., Gillis, S., Ledbetter, J. and Gordon, J., Eur. J. Immunol. 1988. 18: 349. 16 Guy, G. R. and Gordon, J., Proc. Natl. Acad. Sci. USA 1987. 84: 6239. 17 Swendeman, S. L. and Thorley-Lawson, D. A., EMBO J. 1987. 6: 1637. 18 Gordon, J., Ley, S. C., Melamed, M. D., Aman, P. and Hughes-Jones, N. C., Nature 1984. 310: 145. 19 Gordon, J., Cairns, J. A., Millsum, M. J., Gillis, S. and Guy, G. R., Eur. J. Immunol. 1988. 18: 1561. 20 Armitage, R. J. and Goff, L. K., Eur. J. Immunol. 1988. 18: 1753. 21 Uchibayashi, N., Kikutani, H., Barsumian, E. L., Hauptmann, E J., Schneider, R., Schwendenwein, R., Somergruber, W., Spevak, W., Maurer-Fogy, I., Suemera, M. and Kishimoto, T., J. Imrnunol. 1989.142: 390. 22 Flores-Romo, L., Cairns, J. A., Millsum, M. and Gordon, J., Immunology 1989. 67: 547. 23 Gordon, J.,Walker, L. and Guy, G. R., Immunology 1985.56: 329. 24 Steele, R. E., Trends Biochem. Sci. 1989. 14: 201. 25 Black, R. A., Kronheim, S. R., Cantrell, M., Deeley, M. C., March, C. J., Prickett, K. S.,Wignall, J., Conlon, F! J., Cosman, D., Hopp,T. I? and Mochizuki, D.Y., J. Biol. Chem. 1988.263: 9437. 26 Schoellman, G. and Shaw, E., Biochemistry 1963. 2: 252. 27 MacLennan, I. C. M. and Gray, D., Immunol. Rev. 1986. 91: 61. 28 Sarfati,M.,Bron, D.,Larneaux,L.,Fonteyn, C.,Frost,H. and Delespesse, G., Blood 1988. 71: 94. 29 Gordon, J. and Cairns, J. A., Adv. Cancer Res., in press.

Intact, 45-kDa (membrane) form of CD23 is consistently mitogenic for normal and transformed B lymphoblasts.

When isolated from lysates of the RPMI 8866 B lymphoblastoid cell line, the CD23 antigen was found to be present be in two forms corresponding to the ...
556KB Sizes 0 Downloads 0 Views