2622 A. Iglesias, A. Nichogiannopoulou, G . S. Williams et al. Eur. J. Immunol. 1993. 23: 2622-2630

Antonio Iglesias., Aliki NichogiannopoulouA,

Early B cell development requires p signaling G. Stuart Wlliamso, Heinrich Flaswinkel and Georges Kohler

M a x Planck-Institut fur Immnnbiologie, Freiburg

In vitro studies with Abelson murine leukemia virus (AMuLV)-transformed murine pre-B cell lines demonstrated that wild-type p but not mutant p chains lacking the first constant domain (pA1) can efficiently induce Ig light (L) chain gene rearrangement. Using antibodies against the cytoplasmic tail of the immunoglobulin co-receptor p (Igp) chain we find p, but not pA1 chains associated with Igp. Since a heterodimer of surface-labeled proteins was co-precipitated with p we conclude that only wild-type p is associated with the Iga/Igp co-receptor on the surface of pre-B cell lines. Mutant pA1 chains achieve their surface expression by utilizing a glycophospholipid anchor. In vivo analysis of transgenic mcie expressing either p or pA1 transgenes revealed the expected “normal” B cell development in the case of wild-type p transgenic lymphocytes, but a block in differentiation of pA1 transgenic lymphocytes. The maturation block occurs at the developmental transition of pre-B lymphocytes from the CD43/S7+, CD45R/B2201°W stage to the CD43/S7-, B220’0w/high stage in which the majority of L chain gene rearrangements occur. These results, together with the observed inability of the pAl chains to signal activation of L chain gene joining and to associate Iga/Igp in pre-B cell lines suggest that signals mediated by the protein complex composed of dIga/Ig@ are crucial during differentiation of pre-B lymphocytes.

1 Introduction

In B lymphocytes the antigen receptor is a multichain complex composed of immunoglobulin (Ig) heavy (H) and light (L) chains and a heterodimer of two trans-membrane proteins, named Iga and Igp, which are necessary both for surface transport of, and signal transduction through Ig ([1-4], for a review see [5]). Binding to a Lchain and association with the Iga/Igp heterodimer were shown to be sufficient prerequisites to allow the membrane form of the p protein (pm) to be displayed on the surface of even non-B cell lines [l, 61. Surface expression of p in the absence of conventional L chains has been demonstrated in murine and human pre-B cell lines [7-101 and in human pre-B lymphocytes [ l l ] . Two pre-B cell-specific proteins called w and 1, encoded by the genes h5 and VpreB have been identified that have homology to the constant and variable regions of h L chains, respectively [12, 131. These two proteins were shown to form a so-called pseudo L chain protein complex (WL) and to be bound to p proteins before L chains are expressed in pre-B cells [ 141. In some Abelson murine leukemia virus (AMuLV)-transformed pre-B cells rearrangement and expression of L chain genes can be brought about upon expression of endogenous or exoge- nous p genes [15,16].These results provided evidence for a regulated model of B cell differentiation, where p activates joining of L chain gene segments [15, 171. The importance

[I 111771 MPI fur Psychiatrie, Am Klopferspitz 18a, D-82152 Martins- ried, FRG

A Center for Blood Research, Harvard University, Boston, USA CNRS, 11 rue Humann, 67085 Strasbourg, France

Correspondence: Georges Kohler, MPI fur Immunbiologie, Stii- beweg 51, D-79108 Freiburg, FRG

Key words: Pre-Bcell receptor / Transgenic mice / Mutant immunoglobulin

of pm and q L for B cell development is demonstrated in recent experiments, where abolition of expression of pm or 15 via gene targeting led to a blocking of B cell differen- tiation and B cell deficiency in homozygously mutated mice [18, 191.

We have previously shown that in pre-B cells, p induces x gene rearrangement with high efficiency only if asso- ciated with a heterodimer of proteins of 33 kDd36 kDa M, [16]. Mutant p proteins with deletions within the constant region or having the trans-membrane and cytoplasmic domains substituted by the corresponding regions of MHC class I proteins do not associate to the 33 kDd36 kDa heterodimer and are unable to induce x gene rearrange- ment [ l , 16, 201. Here we show that the 33 kDd36 kDa proteins represent an immature glycosylation form of the Iga/Igp heterodimer.

Mutant pA1 chains are transported to and displayed on the surface of pre-B cell lines independently of the Iga/Igp membrane proteins. We show that GPI linkage is the mechanism of surface deposition of pA1 chains. The absence of associated, signal-transducing, Igahgp proteins and the inability of pA1 chains to induce L chain gene rearrangement suggest an important role for the Iga/Igp heterodimer in B cell differentiation. Indeed, transgenic mice expressing the mutant pA1 chains are characterized by an arrest in development of early B lymphocytes at a stage prior to L chain gene rearrangement and by peripher- al B cell depletion.

2 Materials and methods

2.1 Cell lines

The pre-B cell lines used here, p-1, p-4 and p-6 are obtained by transfection of the AMuLV-transformed pre-B cell line 33.1.1- with the pgene from the hybridoma Sp6 [21]. These lines and the other cell lines used here, 33.1.1+ and

0014-2980/93/1010-2622$10.00+ ,2510 0 VCH Verlagsgesellschaft mbH, D-6945 1 Weinheim, 1993

Eur. J. Immunol. 1993. 23: 2622-2630 Early B cell development requires p signaling 2623

WEHI-231, have been described in detail elsewhere [16, 221. The cell line p-6xE12 was generated by supertransfec- tion of the cell line p-6 with the x gene described in [21] but including the second x enhancer located 3' of the x constant region.The 1c gene construct was a kind gift of Dr. H. Eibel. The cell line pTM-1 was generated by transfection of the cellline33.1.1-withthechimericy genedescribedin [l],in which the trans-membrane region of p has been substituted by the corresponding portion of the H-2Kb gene.

goat anti-mouse IgM antibodies and analyzed. Nucleated spleen or bone marrow cells were stained with fluorescei- nated or biotinylated monoclonal antibodies, the latter followed by incubation with streptavidin-phycoerythrin (Southern Biotechnology).The antibodies used were RS3.1 (anti-IgMa, [24]), MB86 (anti-IgMb, [25]), 14.8 (anti- CD45R-B220, [26]), 20-5 (anti-Sp6 idiotype, €! A. Cazen- ave, unpublished), 1M41 (anti-mouse Cfl , [27]) and anti-CD4367 (Pharmingen).

2.2 Immunoprecipitations 2.5 ELISA assays

Metabolic labeling was performed by incubating 5 x lo7 cells in methionine-depleted medium for 0.5 h followed by incubation for 3 4 h in medium containing 0.25 mCi [35S]methionine. After washing, the cells were lysed with digitonin buffer (1% digitonin, 20 mM triethanolamine, 10mM Tris pH7.5, 150mM NaCI, 1 mM EDTA, 1 mM PMSF) on ice for 30 min, followed by centrifugation for 20 min at 24000 x g. For immunoprecipitation, cell lysates were first pre-cleared with 5-10 pg of normal rabbit serum for 1 h at room temperature, followed by incubation for a further 1 h with protein A-sepharose and centrifugation. The supernatant was then incubated with rabbit anti-mouse IgM or rabbit anti-Igp antibodies (H. Flaswinkel and M. Reth, unpublished) for 1 h at room temperature, followed by a 1 h incubation with protein A-Sepharose and centri- fugation. Where indicated, the pellets were resuspended in endoglycosidase H or N-glycanase buffer for treatment with these enzymes as indicated by the supplier (Boehringer Mannheim, FRG). Surface iodinations were performed using the water-soluble Bolton-Hunter reagent (Pierce, Illinois) to surface-label lo8 cells with 0.5 mCi Na125 I (Amersham Buchler) as described [23]. After labeling, cells were washed extensively, lysed and immunoprecipi- tated as described above. Pellets were resuspended in sample buffer. One- and two-dimensional SDS-PAGE were performed as described [ 161. Molecular weight markers were purchased from Sigma.

2.3 Western blot analysis

Digitonin lysates of 5 x lo7 cells were immunoprecipitated and protein samples were electrophoresed in 12% polya- crylamide gels as above. Proteins were then transferred electrophoretically to nitrocellulose paper (Fluorotrans, Pall, GB) and incubated either with the monoclonal anti-Igp antibody AB29-04 (mouse IgG1, x , H. Flaswinkel and M. Reth, unpublished) or with rabbit anti-mouse IgM antibodies, followed by incubation with horseradish perox- idase-labeled goat anti-mouse IgGl (Southern Biotechno- logy) or donkey anti-rabbit Ig (Amersham). Detection was performed with the enhanced luminescence ECL detection system (Amersham).

Supernatants of hybridomas were screened for transgenic or endogenous Ig proteins by an ELISA assay using either goat anti-mouse IgM antibodies (Southern Biotechnology) or purified 20-5 monoclonal antibodies as coating antibod- ies. After incubation of the hybridoma supernatants in the coated plates, bound Ig proteins were detected with alkaline phosphatase-labeled RS3.1 or MB86 monoclonal antibodies (to reveal transgenic or endogenous p mole- cules, respectively) or with alkaline phosphatase-labeled 1M41 monoclonal antibodies to detect all IgM proteins (after coating with goat anti-mouse IgM antibodies) or transgenic p and pA1 proteins (after coating with the monoclonal 20-5 antibody).

3 Results

3.1 Igp is part of the pre-B cell receptor

In AMuLV-transformed pre-B cells expressing transfected p genes, we have detected a heterodimer of proteins, of 33 kDa and 36 kDa M,, noncovalently associated with p in immunoprecipitates made with IgM-specific rabbit anti- bodies ([16], Fig. l).The same complex of p and associated proteins is co-immunoprecipitated from digitonin lysates with rabbit antibodies directed against the cytoplasmic tail of Igp. The 33 kDa/36 kDa proteins are also precipitated with the rabbit anti-Igp-tail antibodies independently of p and oh in the p-cell line 33.1.1-. This shows that the 33 kDa/36 kDa proteins are either directly associated with Igp or that at least one of them is Igp itself. Endoglycosy- dase H-treatment reduces the molecular mass of both biosynthetically labeled proteins to one band of approxi- mately 24 kDa (Fig. 1, lower panel). This band has the expected size of the deglycosylated p chain [2]. As two intracellular forms of Igp of 33 and 36 kDa are visualized in Western blots (see Sect. 3.3), we conclude that both bands contain Igp. From the results shown below we infer, however, that Iga most likely co-migrates with the 33 kDa band and is involved in surface p receptor assembly.

3.2 Glycosylation-extent of co-receptor chains differs in clonally related cells

2.4 Flow-cytometric analysis In the IgM-expressing pre-B cell line 33.1.1+, p and x are also precipitated with the same anti-Igp antibodies,

One- or two-color analysis were performed using a FAC- although the bands corresponding to the associated pro- Scan instrument (Becton Dickinson).Transfected cells were teins are weaker and more diffuse (Fig. 1, upper panel). incubated with or without 1 U of phosphatidylinositol- The 33.1.1f line has been characterized as being in the late specific phospholipase C (Boehringer Mannheim) for pre-B or early B cell differentiation stage [16]. Using 30 min at 37 "C. Then cells were stained with FITC-labeled surface iodination, the associated IgaAgp protein heterod-

2624 A. Iglesias, A. Nichogiannopoulou, G. S. Williams et al. Eur. J. Immunol. 1993. 23: 2622-2630

Non-Red.

33.1.1+

Figure2. Two-dimensional SDS gel analysis of the surface p- associated IgdIgP heterodimer. 1251-labeled surface proteins of the indicated cell lines were immunoprecipitated with rabbit anti- mouse IgM antibodies and the protein samples were electropho- retically separated, first under non-reducing (Non-Red.) and then under reducing (Red.) conditions; the position of the IgdIgP heterodimer is indicated by arrows.

Figure 1. Immunoprecipitation of p-associated proteins. The indi- cated cell lines were biosynthetically labeled, lysed with digitonin and the protein preparations were immunoprecipitated with either anti-lgP (A-B29) or anti-IgM (A-p) rabbit antibodies (upper panel) and run in SDS-PAGE under reducing conditions. Anti-IgM precipitates of the k-6 transfectant (a line similar to p-4) were treated with endoglycosidase H as indicated (lower panel). The position of p, x , and the associated proteins 33 kDd36 kDa and O/L

is indicated.

imer is easily visualized as co-precipitated with surface IgM in the cell line 33.1.1+ (Fig. 2). In the cell line p-6 (a similar transfectant as p-4), surface p is also associated with a heterodimer of proteins (Fig. 2).The sizes of these proteins (33 kDa/38 kDa) are, however, smaller than the ones of 33.1.1+ (34 kDd40 kDa). This is better visualized in one- dimensional SDS gel electrophoreses (Fig. 3). To raise the

Figure 3. Deglycosylation analysis of Ig-associated surface pro- teins. lZsI surface-labeled proteins from digitonin lysates of the indicated cell lines were immunoprecipitated with rabbit anti- mouse IgM antibodies and the samples were analyzed in SDS- PAGE under reducing conditions in 12% polyacrylamide gels after overnight incubation at 37°C in the absence (Mock) or presence of 1 U N-glycosidase F (PNGase F, Boehringer). The positions of Ig and associated proteins as well as molecular weight markers (lane M) arc indicated by bars or arrows.

Eur. J. Immunol. 1993. 23: 2622-2630

low levels of surface receptors, the p-6 line was supertrans- fected with a x gene. Upon FACS and immunoprecipitation analysis of a series of such x transfectants,we found that the increase in surface p was dependent on the amount of n expressed (not shown) and that enhanced surface expres- sion is not accompanied by changes in the size of the associated Igahgp proteins (compare lanes labeled p-6 and p-6xE12 in Fig. 3). These differences in size are due to different glycosylation, as treatment of the immunopreci- pitated surface proteins with N-glycanase F (Boehringer) converts them to polypeptides of the same size (around 24 kDa, Fig. 3).The band around 56 kDa (or 37 kDa when deglycosylated) is probably not associated with p, because it is also precipitated from 33.1.1- lysates, where no p is expressed (not shown).

Early B cell development requires p signaling 2625

A

B

The line 33.1.1- and its p transfectants represent a less mature p r e - ~ cell stage than 33.1.1+. pool size differences of particular glycosylated forms of the Iga/Igp proteins may7 the predominance and sharpness Of the 33 kDa/36 kDa bands Seen after biosynthetic labeling as compared to the more diffuse Pattern of 33.1.1' (Fig. 1).

We conclude that different glycosylations of Iga/Igp can exist in clonally related cells and that these differences are not imposed by L chain expression.

Figure 4. Western blot analysis of and associated proteins. Digitonin lysates of the indicated cell lines were immunoprecipi- tated with rabbit anti-mouse IgM antibodies. After mock (-) or Endo H treatment (+) the samples were electrophoresed as in Fig. 1 and the proteins were transferred to nitrocellulose mem- brane. The filter was first developed with the monoclonal anti- I&-antibody AB29-04 (A) and, after stripping, with rabbit anti- mouse IgM antibodies (B). Detection with second antibodies and the ECL system was as described in Sect. 2.4. The position of surface (b) Or intracehlar (pi) Ig8 proteins is indicated by arrows and that of molecular weight markers and of p proteins is indicated by bars.

3.3 IgulIgp are associated with surface p but not with surface pAl

In contrast to wild-type p chains, mutant pA1 chains lacking the CH1 domain were previously shown to lack the capacity to induce L chain gene rearrangement upon expression in pre-B cells. Anti-p precipitations of biosyn- thetically labeled lysates revealed a heterodimer of 33 kDa/36 kDa which was associated with p but not with pAl, although both types of proteins are equally well displayed on the cell surface [16]. The Iga/Igp co-receptor is needed for surface expression of IgM [l]. To confirm that pA1 was not associated wit these proteins also on the surface of the pre-B cell lines, we performed surface iodination experiments. Again no association was seen with pAl, whereas p in the lines p-6, 33.1.1+ and p-6xE12, a x-expressing supertransfectant of p-6, did show the expected co-receptor Iga/Igp association (Fig. 3). Further- more, using a monoclonal antibody specific for the cyto- plasmic tail of Igp (AB29-04) in Western blot experiments, both the intracellular and surface forms of Igp can be found associated with wild-type p but not with pA1. In Fig. 4, immunoprecipitates from digitonin lysates of the indicated cell lines using rabbit anti-p antibodies are shown after SDS-PAGE, transfer to a nitrocellulose filter and incuba- tion with AB29-04 (Fig. 4A) or goat anti-mouse IgM antibodies (Fig. 4B). As indicated, the antibody AB29-04 recognizes the highly glycosylated surface form (ps), as identified in surface iodination experiments (Fig. 3 and not shown), and two different glycosylation forms of intracel- lular Igp (pi), similar in size to the 33 kDa/36 kDa biosyn- thetically labeled proteins seen in Fig. 1. The size of ps in the lymphoma line WEHI 231, used here for comparison, is somewhat heterogeneous and slightly larger (40-42 kDa) than its counterpart in the cell line 33.1.1+ (40 kDa), whereas the size of pi is identical for the cell lines shown. In

the cell lines WEHI 231 and 33.1.1+ the surface Igp pro- teins associated with IgM are resistant to Endo H, while the intracellular forms of the same protein are converted by Endo H treatment into a polypeptide of the expected size of 24 kDa (Fig. 4A). In the cell line p-1, no Endo H- resistant Igp proteins are detected with the antibodyAB29- 04, probably due to the low amounts of surface p proteins in this line (not shown, [16]). However, as co-precipitated with anti-p antibodies, Igp is detected intracellularly in this cell line in higher amounts than in the cell lines WEHI 231 and 33.1.1+ (Fig. 4A). This result indicates that, in the absence of L chains, most p and associated Iga/Igp do not reach the cell surface. Thus, the easier visualization of the latter proteins in biosynthetic labeling versus surface iodination experiments in p-expressing pre-B cells (see also Fig. 1) is probably due to their intracellular accumulation. As can be seen from Fig. 4A, no Igp proteins can be demonstrated in the pAl-10 immunoprecipitate, although similar amounts of p proteins are detectable in this cell line as compared with p-1 or WEHI 231 (Fig. 4B).

3.4 Surface pAl is GPI-anchored

As glycosyl-phosphatidylinositol (GPI) linkage has been demonstrated as an alternative mechanism for cell surface expression of IgD or mutant IgM [28, 291, we tested the possibility of GPI-facilitated surface expression of pA1 proteins. As shown in Fig. 5, no reduction of the levels of surface p is observed after treatment with phosphatidyl- inositol-specific phospholipase C (PI-PLC) in the cell lines 33.1.1+, p-6, and pTM-1, whereas an eight- to nine-fold reduction is found in the pA1-10 line. Therefore, pA1 chains can be displayed on the cell surface via GPI linkage, thus explaining the efficient surface expression of

2626 A. Iglesias, A. Nichogiannopoulou, G. S. Williams et al Eur. J. Immunol. 1993. 23: 2622-2630

IgM

Figure 5. Flow-cytometric analysis of pre-B cell lines.The staining profiles of the indicated cell lines are shown before and after (dark curves) treatment with phosphatidylinositol-specific phospholi- pase C and staining with FITC-labeled goat anti-mouse IgM antibodies.The profile obtained with the parental cell line 33.1.1-, used here as negative control, is shown superimposed on the profiles of the line pA1-10 (C). The line pTM-1 expresses a chimeric p chain (see Sect. 2.1) whose surface expression is independent of the I g d g p co-receptor.

this protein in the absence of associated Iga/IgP proteins. From the results shown in Figs. 1-5 we conclude that (i) in the pre-B cells studied here, the difference in M, observed

between the IgdIgP proteins associated with surface p in our pre-B cell lines and the described IgM-associated proteins in mature B cells, is mainly due to different glycosylation extents, and (ii) in contrast to wild-type p or IgM, surface expression of pA1 does not require associa- tion with the surface co-receptor complex of Iga/IgP, but can be brought about by GPI linkage.

3.5 Expression of pAl in transgenic mice

The data shown in Figs. 2-4, together with previous results [16] suggest that signaling through sp in association with I g d g P is required for efficient induction of x gene rear- rangements. Thus, we asked next whether the observed in vitro inefficiency of pA1 proteins in inducing x gene rear- rangement in pre-B cell lines [16], also interferes in vivo with B cell development. To this end we generated trans- genic mice with the same wild-type p and pAl-encoding DNA constructs used for expression in the pre-B cell lines [16]. Two mouse lines expressing the wild-type p construct (designated Bp: 50 copies and Ep: 10 copies), and three mouse lines carrying the pAl construct (named NpA1: 10 copies; OpAl: 20 copies and CpAl: 2 copies) were analyzed in this study. IgM molecules bearing the trans- genic idiotype (as detected in ELISA assays using the monoclonal anti-idiotypic antibody 20-5) are detected in the serum of pAl transgenic mice at much lower concen- trations (0.5 pg/ml in mice of the OpAl and Np A1 lines and 1 pg/ml in mice of the CpAl line) than in the serum of mice from both wild-type Bp and Ep transgenic lines (about 100 pg/ml). Two days after birth the spleen of OpA1- and NpA1-transgenic mice show a strong reduction in the number of B lymphocytes, not observed in p-transgenic mice, that increases with age to reach between 20 and 50% of the values of control littermates at adult stages (Table 1). T lymphocytes are not affected (not shown). In the CpA1

Table 1. Number of B lymphocytes in spleen and bone marrow of transgenic micea)

Spleen Bone marrow

'Itansgenic Age No. of mice B220+, IgM+B220+, IgM+B220+, IgM- Ratio line (TGLM) (A) (B) (A : B)

BP 2 d 7 w 8 w

EP 2 d 6 w 6 w

OpA1 2 d 4 w 8 w 9 w 9 w

NpAl 2 d 7 w 7 w

11 w CpAl 2 d

3 w 10 w

0.730

0.750 0.330 0.550 0.830 0.085 0.130 0.500 0.350 0.300 0.017 o.200 0.500 0.200 4)

0.020 0.100

0.710

-

0.870 0.600

0.370 0.730

0.190 0.500 0.230 0.250

0.400 0.100 0.300

0.090 0.080

1.200 0.725 0.930 0.645

0.310 1.120 0.790 0.920

0.980 0.190 2.600 0.190 0.720 0.310 0.700 0.350

1.700 0.230 0.520 0.190 2. 100 0.140

0.040 0.160

The number of B lymphocytes in spleen and bone marrow of transgenic mice at different ages (days: d; weeks: w) is given as fraction of the corre- sponding littermate values. Spleen and bone marrow cells were analyzed in FACS after staining with antibodies specific for the indicated cell surface markers as described in detail in Sect. 2.4. The number of transgenic (TG) and littermate (LM) mice used in each experiment is given in parenthe- sis. The number of B lymphocytes 2 days after birth is underlined. Only background values were found.

Eur. J. Immunol. 1993. 23: 2622-2630 Early B cell development requires p signaling 2627

transgenic line the depletion of B lymphocytes is particu- larly strong. This line shows an almost complete absence of splenic B lymphocytes at day 2 after birth and the levels of peripheral B cells in adult mice make only up to 10% of the littermate values (Table 1). In the bone marrow of the Bp- and Ep-transgenic mice the overall number of lymphocytes of the B cell lineage is reduced to different extents, as compared with non-transgenic littermates but the ratio of B220+/IgM+ to B220+/IgM- lymphocytes is comparable to the ratio found in normal mice (Table 1). However, in the bone marrow of OpAl- and NpA1-transgenic mice the relative amount of the B220+/IgM- pre-B lymphocytes is increased, apparently at the expense of the more mature B220+/IgMf lymphocyte subset (Table 1). In the bone marrow of CpAl-transgenic mice a profound reduction of lymphocytes is found that affects both subsets of pre- B lymphocytes (Fig. 6 and Table 1). The results suggest that, in contrast to wild-type p chains of the same specific- ity, the expression of pA1 leads to a block in the differen- tiation of transgenic B lymphocytes at a stage prior to expression of surface immunoglobulin. In order to further disclose the defect, we performed double staining experi- ments on bone marrow cells of transgenic and normal mice with antibodies specific for CD45RD3220 and for CD43/S7, as these markers have been shown to define earlier stages of B cell differentiation [30]. As described [30], the double- positive population represents only 2-5% of total bone marrow cells in normal littermate mice (Fig. 6, left panels). Although the number of lymphocytes is often reduced in the bone marrow of the wild-type p transgenic mice (Table 1), the percentage of CD43/S7+ B220+ early precur- sors is unchanged in both Ep (not shown) Bp (Fig. 6, compare percentages in R1 and R2 to littermate control values). In contrast, about 50% of B220-positive bone marrow pre-B lymphocytes in OpAl- and NpA 1-transgenic mice are positive for the CD43/S7 surface antigen and show characteristically lower levels of B220 as compared with normal pre-B lymphocytes. In the bone marrow of CpAl- transgenic mice the subset of CD43/S7+ pre-B lympho- cytes, although not absolutely increased, constitutes the majority of B220+ lymphocytes, as more mature stages are strongly reduced in these mice (Fig. 6).We conclude that, in contrast to transgenic mice expressing wild-type p chains, normal transition from very early stages of B cell differen- tiation into more mature stages is disturbed in the bone marrow of transgenic mice expressing pA1 proteins.

LM I 9 1

b I

I = CD43

Figure 6. Flow cytometric analysis of bone marrow pre-B lympho- cytes.Tota1 nucleated bone marrow cells of the indicated transgenic (right panels) and their littermate mice (left panels, LM) were double-stained with phycoerythrin-labeled anti-CD45RB220 and FITC-labeled anti-CD43K7 antibodies and analyzed in FACS. Mature Band CD43- pre-B lymphocytes are included in the region defined by the small oval (Rl) and CD43+ early pre-B lymphocytes are the region of largeroval (R2).The percentage of cells present in each region is indicated.

Table 2. Analysis of B cell hybridomas derived from transgenic and littermate spleensa)

Ig produced Rearrangement status

Mouse L Ht + L Ht + H e + L H e + L GL DJ VDJ 'Ibtal Line

BP EP Total @

OpA1 NpAl CpAl 'Ibtal pAl mtal of littermates

1 17 8 12 9 29 0 0 0 0 2 1 2 1 6 -

1 2 3

11 19 12 42 -

0 4 13 0 17 4 2 7 1 10 4 6 2 0 1 27 9 N D N D N D ND 9 0 9 9 18

11 3 11 18 32 29 3 20 27 50

40 0 12 32 44

a) Fusion of LPS-stimulated splenocytes was done as described [32]. The pre- sence of L chains and transgenic (Ht) or endogenous (He) Ig Hchains in hybridoma supernatants was deter- mined in an ELISA assay as described in Sect. 2.5. The analysis of the rear- rangement status of the endogenous Ig H genes was done essentially as de- scribed [32]. The number of unrear- ranged germ-line (GL), partially (DJ) or completely (VDJ) rearranged al- leles is given. ND: not determined.

2628 A. Iglesias, A. Nichogiannopoulou, G . S. Williams et al. Eur. J. Immunol. 1993.23: 2622-2630

3.6 Analysis of transgenic B cell hybridomas

Similar to earlier studies with other H chain-transgenic mice [21,31,32], the majority of B cell hybridomas derived from Bp- and Ep-transgenic mice express exclusively transgenic p H chains (29 of 36 H chains-expressing hybri- domas), reflecting the fact that a large fraction (26 of 27) of analyzed endogenous alleles is not or is only incompletely rearranged (Table 2). In contrast, B cell hybridomas de- rived from pA1 transgenicmice express transgenic H chains almost exclusively in association with endogenous H chains and only in 43 of 72 hybridomas expressing Ig H chains (Table 2). Nonetheless, an accumulation of unrearranged or partially rearranged endogenous alleles (23 of 50) is detected in these hybridomas, as compared with B cell hybridomas derived from normal littermate mice (12 of 44).

4 Discussion

Using rabbit antibodies raised against the trans-membrane tail of Ig& we have identified Igp to be part of the formerly described 33/36 kDa protein heterodimer associated with p in AMuLV-transformed pre-B cell lines [16]. From this result, together with surface iodination data (Figs. 2 and 3) we conclude that the pw i pre-B cell receptor is also associated with the IgdIgp heterodimer that is normally found associated with IgM in B cells [ l-S].The chains of the surface heterodimer in p-expressing pre-B cell transfec- tants have slightly different sizes than in the IgM-bearing pre-B cell line 33.1.1+ and in the B lymphoma WEHI 231. These differences can be ascribed to variable glycosylation, since deglycosylation converts all surface forms in the cell lines analyzed, as well as the intracellular forms, into polypeptides of about 24 kDa (Figs. 1, 3 and 4). Thus, murine Iga and Igp have a very similar core protein size, as is also the case for their human counterparts [33]. Previous reports suggesting size differences for the deglycosylated forms of Iga versus Igp proteins [2] probably resulted from incomplete trimming of sugar residues.

Membrane deposition of IgM needs the association of the IgcdIgP co-receptor. This association depends on the pre- sence of the membrane proximal Cp4 domain and of the trans-membrane domain of the p chain [1, 41. Deletion of the cytoplasmic tail of human pchains resulted in GPI- anchored surface pexpression [29]. Since none of these domains are altered in pA1, it is not clear why pA1 pro- teins, in contrast to intact p chains, are not associated with IgdIgp and are instead displayed on the cell surface via GPI linkage (Fig. 5 ) . An unintentionally introduced C-terminal mutation in our pA1 construct is unlikely since an indepen- dently derived pA1 chain also reached the membrane without Iga/Igp association (S. Pillai, personal communi- cation). It is conceivable that deletion of the first constant p domain,which is the binding site for BiP, a chaperonin with a retention signal for the endoplasmic reticulum [34], results in an altered pathway of transit through the ER which then allows GPI addition. In any case, the altered pA1 receptor lacks the ability to induce Lchain gene joining in pre-B cell lines, even when expressed at high density [16]. As the IgdIgp co-receptor is required for signal transduction, we explain the observed deficiency of

pA1 proteins in inducing L chain gene rearrangement by their lack of signal transmission potential.

In vivo, the pAl transgene caused a block in B cell development and B cell depletion, that is not observed in transgenic mice expressing the control wild-type p chain (Table 1 and Fig. 6). During ontogeny of B lymphocytes in the bone marrow, different stages of differentiation have been identified as defined by the coordinate expression of surface antigens other than Ig, such as CD45RB220, CD43/S7 (leukosyalin), CD24/HSA (heat-stable antigen) and BP-1 (30). CD43/S7+ early pre-B lymphocytes consti- tute only 2 to 5% of total bone marrow cells and are shown to be the immediate precursors of the majority of immature CD43-B220'Ow/IgM- pre-B lymphocytes subset [30]. CD43/S7+ pre-B lymphocytes characteristically express high levels of the CD24/HSA and the lowest levels of the CD45R/220 antigen among the pre-B cell precursors [30]. Pre-B cells of this phenotype constitute the predominant subset of lymphocytes in the bone marrow of pAl- transgenic mice (Fig. 6). Progression into the CD43- stage is apparently blocked in these mice while normal in wild-type p-transgenic mice. The developmental block observed is, therefore, specific for the pA1 transgene and is more pronounced in the CpAl line, which expresses higher serum and surface levels of transgenic Ig molecules (not shown) than the lines OpA1 and NpA1, albeit still very low when compared with p-transgenic mice. A strong depletion of mature B lymphocytes and an increase of the CD43+ pre-B lymphocyte subset has also been described in mice carrying a targeted mutation of the membrane form of p [18] or of hS [19].The CD43+ pre-B lymphocytes are shown to carry a high proportion of rearranged H gene loci and only few rearranged L genes, whereas most L gene rearran- gements are found in the later, CD43- stages [30, 351. Hence we interpret the differentiation block at the CD43+ stage observed in pA1-transgenic mice as reflecting the inability of this transgenic Ig molecule to promote L gene rearrangement. However, mature B lymphocytes are still generated (Table 1 and Fig. 6), indicating that the arrest at the CD43+ stage can be overcome in these mice. Interest- ingly, also in AS mutant mice a small amount of B cells are generated in spite of the general differentiation block of B lymphocytes [19]. In a recent report [35], it was found that Ig L gene rearrangement can occur in the complete absence of p H chains in mice carrying a deletion of the entire JH cluster. In these as well as in the mice mutant for the pm and the h5 genes, xgene rearrangements are detected in a fraction of the CD43+ pre-B lymphocytes subset. Based on data obtained in separated subsets of pre-B lymphocytes from the bone marrow of mutant and of normal mice, the authors calculated that L gene rearran- gements can occur independently of p proteins in about 5% of B cells in normal mice (pathway 1). In 95% of pre B lymphocytes of the mouse, expression of the pre-B cell receptor composed of In/w/i could precede and strongly up-regulate rearrangement of the L gene loci (pathway 2) [3S]. Obviously, pathway 1 would not be affected in pA1-transgenic mice and could be responsible for the observed B cells accumulating with time in transgenic pAl mice (Table 1).

As a result of allelic exclusion, most B cell hybridomas of wild-type p-transgenic mice express transgenic IgM only, while the majority of B cell hybridomas of pA1 transgenic

Eur. J. Immunol. 1993. 23: 2622-2630 Early B cell development requires signaling 2629

signaling through IgdIgP is essential during early B cell development.

mice express either endogenous IgM alone or together with transgenic pA1 chains (Table 2). Apparently, B cell matu- ration in these mice cannot be driven by the transgenic pA1 alone and requires the expression of endogenous (intact) p proteins. However, the strong reduction of CD43- pre- B lymphocytes and early IgM+ mature B lymphocytes in the bone marrow of these mice indicates that the pA1 transgene is not an unnoticed passenger molecule. Indeed, the genomic analysis of B cell hybridomas derived from these mice revealed an increased number of germ-line or incompletely rearranged DJ loci among the silent endoge- nous Ig H alleles, as compared with hybridomas from non-transgenic littermate mice (Table 2). Thus, the p A l transgene inhibits rearrangement of endogenous Ig H genes. Current models postulate that negative feedback signaling through p w i shuts down rearrangement of H genes at the CD43+ stage and, at the same time, triggers the transition into the next, CD43- stage, while up- regulating L gene rearrangement. The newly generated IgM receptor then signals general rearrangement shut-off [30,35,36].Within the frame of this model it seems difficult to reconcile the absence of signaling capacity of the pA1 chain for one function (induction of x gene rearrange- ment) with the retention of signaling potential for another function, namely allelic exclusion of endogenous Ig H genes through inhibition of their rearrangement. In an alternative model onset of H gene rearrangements is signaled via the pseudo L (QL) complex in association with the pseudo ' H chain protein X [37]. Protein X is then replaced by newly generated p H chains, which, in associa- tion with the WL chains oh, will now signal opening of the L locus for rearrangement. In this model, different com- plexes (X/QL versus p/QL) would serve to signal the activation of different genomic (H and L) regions. Pre- B cells in p and pA1 chain transgenic mice would express the H/QL surface complex earlier and bypass the stage(s) of development which require the X/QL complex for opening and rearrangement of the (endogenous) H chain locus, the consequence being an accumulation of unrearranged or partially rearranged endogenous.Ig H alleles (Table 2). On the other hand, and as a consequence of their lack of associated IgdIgP co-receptor proteins, pA1 chains are unable to activate L gene rearrangement both in v i m [16] and in vivo, leading to the observed block in differentiation of pre-B lymphocytes.

Cross-linking of surface pchains lacking V and pA1 domains (and associated cpL chains) of a pre-B cell line resulted in increased L gene rearrangements [38]. Similar treatment of p h l transfectants of this study was, however, ineffective (not shown). As the former mutant chains were associated with IgdIgB, but p A l is not, we conclude that signaling through IgdIgP affects L gene rearrangement.

In summary,we have shown that, in contrast to wild-type p, the pA1 chains are not associated with the co-receptor proteins Iga/IgP in pre-B cells but are displayed on the cell surface via GPI linkage. The lack of associated, signal transducing IgcdIgP co-receptor apparently results in the incapacity to signal activation of L chain gene rearrange- ment. When expressed in transgenic mice, the mutant p A l protein, but not the wild-type version of the same u chain, leads to a block in B cell differentiation at a stage

The excellent technical help of C. Strohmeier, l? Renard and E. Stratling is acknowledged. We also thank Dr. M . Reth for the generous gift of the p T M clone and Drs. l? Nielsen and N. Nohen for their very helpful comments on the manuscript.

Received October 28,1992; in revised form June 8, 1993; accepted June 28, 1993.

5 References

1 Hombach, J.,Tsubata,T., Leclercq, L., Stappert, H. and Reth,

2 Campbell, K. S. and Cambier, J. C., EMBO J. 1990. 9: 441. 3 Chen, J., Stall, A. M., Herzenberg, L. A. and Herzenberg, L.

4 Venkitaraman, A. R., Williams, G. T., Dariavach, P. and

5 Reth, M., Annu. Rev. lmmunol. 1992. 10: 97. 6 Matsuuchi, L., Gold, M. R., Travis, A., Grosschedl, R.,

DeFranco, A. L. andKelly, R. B., Proc. Natl. Acad. Sci. USA 1992. 89: 3404.

M., Nature 1990. 343: 760.

A., EMBO J. 1990. 9: 2117.

Neuberger, H. S., Nature 1991. 352: 777.

7 Pillai, S. and Baltimore, D., Nature 1987. 329: 172. 8 Kerr, W. G., Cooper, M. D., Feng, L., Burrows, P. D. and

Hendershot, L. M., lnt. lmmunol. 1989. 4: 355. 9 Karasuyama, H., Kudo, A. and Melchers, F., J. Exp. Med.

1990. 172: 969. 10 Tsubata,T. ,Tsubata, R. and Reth, M., J. Exp. Med. 1990.172:

973. 11 Nishimoto, N., Kubagawa, H., Ohno, T., Garland, G. L.,

Stankovic, A. K. and Cooper, M. D., Proc. Natl. Acad. Sci. USA 1991. 88: 6284.

12 Sakaguchi, N. and Melchers, F., Nature 1986. 324: 579. 13 Kudo, A. and Melchers, F., EMBO J. 1987. 6: 2267. 14 Miesener,V., Jongstra-Bilen, J.,Young, A. J., Atkinson, M. J.,

Wu, G. E. and Jongstra, J., J. lmmunol. 1990. 145: 905. 15 Reth, M., Petrac, E.,Wiese, P., Lobel, L. and Alt, F.W., EMBO

J. 1987. 6: 3299. 16 Iglesias, A., Kopf, M.,Williams, G. S., Buhler, B. and Kohler,

G., EMBO J. 1991. 10: 2147. 17 Alt, F. W., Blackwell, T. K. and Yancopoulos, G. D., Science

1987. 238: 1079. 18 Kitamura, D., Roes, J., Kuhn, R. and Rajewsky. K., Nature

1991. 350: 423. 19 Kitamura, D., Kudo, A., Schaal, S., Miiller,W., Melchers, F.

and Rajewsky, K., Cell 1992. 69: 823. 20 Nichogiannopoulou, A., DiplomaThesis 1992, Albert-Ludwig-

Universitat, Freiburg, FRG. 21 Rusconi, S. and Kohler, G., Nature 1985. 314: 330. 22 Ralph, R., lmmunol. Rev. 1979. 48: 107. 23 Thompson, J. A., Lau, A. L. and Cunningham, D. D.,

24 Schiippel, R.,Wilke, E. and Weiler, E., Eur. J. Immunol. 1987.

25 Nishikawa, S. I., Sasaki,Y., Kina,T., Amagai,T. and Katasura,

26 Coffman, R. L., lmmunol. Rev. 1983. 69: 5. 27 Leptin, M., Potash, M. J., Grutzmann, R., Heusser, C.,

Shulman, M. J., Kohler, G. and Melchers, F., Eur. J. lmmunol. 1984. 14: 534.

Biochemistry 1987. 26: 743.

17: 739.

Y., lmmunogenetics 1986. 23: 137.

28 Wienands, J. and Reth, M., Nature 1992. 356: 246. 29 Shaw, A. C., Mitchell, R. N.,Weaver,Y. K., Campos-Torres, J.,

Abbas, A. K. and Leder, I?, Cell 1990. 63: 381. 30 Hardy, R. R., Carmack, C. E., Shinton, S. A., Kemp, J. D. and

prior to L chain gene rearrangement, thus indicating thit Hakayama, K., J. Exp. Med. 1991. 173: 1213. .

2630

31 Storb, U., Pinkert, C., Arp, B., Engler, P., Gollahon, K., Manz, J., Brady, W. and Brinster, R., J. Exp. Med. 1986. 164: 627.

32 Iglesias, A,, Lamers, M. and Kohler, G., Nature 1987. 330: 482.

33 Van Noesel, C. J. M., van Lier, R. A.W., Cordell, J. L.,Tse, A. G. D.,van Schijdel, G. M.W., deVries, E. F. R., Mason, D.Y. and Borst, J., J. Immunol. 1991. 146: 3881.

A. Iglesias, A. Nichogiannopoulou, G. S. Williams et al. Eur. J. Immunol. 1993. 23: 2622-2630

35 Ehlich, A., Schaal, S., Gu, H., Kitamura, D., Miiller, W. and Rajewsky, K., Cell 1993. 72: 695.

36 Hayashi, S. I., Kunisada,T., Ogawa, M., Sudo,T., Kodama, H., Suda,T., Nishikawa, S. and Nishikawa, S. I., J. Exp. Med. 1990. 171: 1683.

37 Rolink, A. and Melchers, F., Cell 1991. 66: 1081. 38 Tsubata,T.,Tsubata, R. and Reth, M., Int. Immunol. 1992. 4:

34 Munro, S. and Pelham, H. R. B., Cell 1987. 48: 899. 637.