Eur. J . Immunol. 1989.19: 1693-1699 Thrombocytopenia in mice tolerized to alloantigens 1693

Jesus Merino”, Hui-yu Qin, Stephane Schurmans, Denise Gretener, Georges E. Grau and Paul H. Lambert

WHO Immunology Research and Training Centre, Department of Pathology, Geneva

Thrombocytopenia associated with the induction of neonatal tolerance to alloantigens: immunopathogenic mechanisms*

BALB/c mice rendered tolerant to alloantigens by neonatal injection of semi- allogeneic (C57BL/6 x BALB/c)F1 spleen cells develop a thrombocytopenia in associ- ation with an autoimmune lupus-like syndrome. The possible mechanisms involved in the thrombocytopenia were investigated. The development of thrombocytopenia was first detected at 3 weeks of age coinciding with the start of the other autoimmune manifestations and was always related to a state of tolerance and B cell chimerism. There was a significant increase of megakaryocytes in bone marrow and spleens from thrombocytopenic tolerant mice and radiolabeled platelets from these mice were more rapidly eliminated from the bloodstream than normal platelets when injected into normal recipients. A significant correlation between the spleen weight and the decrease of the circulating platelets was observed, although some mice with severe thrombocytopenia had only a moderate spleen enlargement.

Thrombocytopenia significantly correlates with the levels of platelet-associated IgG (PAIgG) but not with anti-single-stranded DNA antibodies or circulating immune complexes. Platelets from mice with high levels of PAIgG had a shorter life-span when injected into normal mice than those from mice with low or normal PAIgG. The possibility that PAIgG are partially due to antibodies reacting specifically with platelet membrane components was analyzed. First, F(ab’)z Ig fragments from toler- ant mice were shown to bind to normal platelets, in contrast to F(ab’)z Ig fragments from normal mice. Second, some monoclonal antibodies produced by hybridomas derived from tolerant mice reacted in vitro with platelets and induced a transient thrombocytopenia after i.v. injection into normal mice.

These data suggest that the thrombocytopenia observed in tolerant mice is the result of a peripheral hyperdestruction of platelets associated with (a) hypersplenism, (b) nonspecific fixation of immunoglobulins, probably as immune complexes and (c) with autoantibodies reacting specifically with platelets. It may represent an interesting model for human chronic idiopathic thrombocytopenia.

1 Introduction

BALB/c mice neonatally injected with (C57BW6 X BALB/c) F1 semiallogeneic spleen cells became tolerant to the H-2b alloantigens [l], but also show a wide range of autoimmune manifestations similar to those seen in SLE [ 2 , 31. Indeed, these tolerized mice showed hypergammaglobulinemia, va- rious autoantibodies - including anti-single-stranded DNA

[I 74401

* This work has been supported by the Swiss National Foundation (grant no 3.803.0.86). Partially supported by a grant from Merck Sharp and Dohme, Spain, and by a grant from the “Fondo de Investigaciones Sanitarias de la Seguridad Social”, Spain. Current address: Laboratory of Immunology, Hospital Nacional “Marques de Valdecilla”, E-39008 Santander , Spain

Correspondence: Paul H. Lambert , WHO Immunology Research and Training Centre, Department of Pathology, CMU, 1 rue Michel Ser- vet, CH-1211 Geneva 4, Switzerland

(ss DNA), anti-Sm and anti-myosin antibodies - immune com- plexes and renal deposits of Ig, in association with thrombocy- topenia, splenomegaly and lymphadenopathy [3, 41.

In previous experiments, using allotypic markers, it was dem- onstrated that all autoantibodies appearing in these tolerant mice are produced by F1 donor B cells [5]. Further experi- ments, in which tolerized mice were treated with an anti-CD4 mAb, demonstrated that the depletion in CD4+ T cells com- pletely prevented the development of this autoimmune syn- drome [6]. Moreover, host CD4+ T cells are required for the activation of autoreactive FIB cells [7]. Conversely, donor T cells were not necessary for the development of the autoim- mune manifestations in this model, although they could play a modulatory role in the production of some autoantibodies**.

The present studies were initiated to investigate the patho- genic mechanisms involved in the thrombocytopenia observed in these mice, in particular relation to non-immunological and to immunological mechanisms including circulating immune complexes and anti-platelet antibodies.

Abbreviations: CIC: Circulating immune complexes CTLp: CTL precursors PAIgG: Platelet-associated IgG Lambert, P. H., submitted.

0 VCH Verlagsgesellschaft mbH, D-6940 Weinheim, 1989

* * Merino, J . , Schurmans, S . , Wen, L. , Brighouse, G. , Luzuy, S. and

0014-2980/89/0909-1693$02.50M)

1694 J. Merino, H.-Y. Qin, S . Schurmans et al.

2 Materials and methods

Eur. J. Irnmunol. 1989.19: 1693-1699

ng Indium). An average of 5 pCi (lo8 labeled platelets) in 0.2 ml were injected i.v. into each mouse. Blood samples were then obtained sequentially and counted in a Packard 5130 gamma-counter. The reading at 1 min was considered as 100% and the following amounts were expressed as a percentage of this determination. After the sequential bleedings, mice were killed and the radioactivity contained in spleen, lungs, liver, heart and kidneys was also measured.

2.1 Mice, induction of neonatal tolerance, serum samples and platelet counts

Mice of the inbred strains BALBic and C57BL/6 were pur- chased from Bomholtgard Laboratories (Ry, Denmark). (C57BL/6 x BALB/c)F1 hybrid mice were produced in our laboratory. Spleen cells ( lo8) from 2-3-month-old (C57BL/6 x BALB/c)FI hybrid mice, obtained as previously described [6], were injected i.p. into newborn BALB/c mice within 24 h after birth. Control mice were untreated littermates. Mice were bled by retro-orbital sinus puncture under ether anesthesia, and blood was kept for 90 min at room temperature to clot. Serum was then separated by centrifugation at 1500 x g for 10 min and stored at -70 "C. Blood (20 pl) was also obtained from retro-orbital plexus using siliconized microcapillaries and immediately diluted in a buffer containing ammonium oxalate (Unopette kit; Becton Dickinson, Basel, Switzerland). Platelets were counted in an improved Neubauer hematocy- tometer under phase contrast at 400x magnification. Mice with platelet counts under 600 OOO/mm3 were considered throm- bocytopenic.

2.2 Generation of CTL activities in mixed leukocyte culture (bulk CTL assay) and estimation of CTL precursor (CTLp) frequency

The CTL unresponsiveness to F, alloantigens in BALBlc mice neonatally injected with F1 spleen cells was assessed in a 5-day MLC, followed by a cytolytic 'lcr-release assay as previously described [8].

The frequencies of alloreactive CTLp in spleen cells were investigated in a micro-MLC under limiting dilution conditions as described [9]. CTLp frequencies were calculated by the maximum likelihood method [lo].

2.3 Preparation of anti-IgCH allotype antisera and determination of the Ig allotype

Anti-Ighb allotype antisera were obtained following the method described by Lieberman [ l l ] , in BALB/c mice immunized with C57BL16 IgG. The IgG fraction was purified with a protein A-Sepharose (Pharmacia, Uppsala, Sweden) column and conjugated to alkaline-phosphatase (APh; Sigma, St Louis, MO) by using glutaraldehyde (Serva Lab., New York, NY; [12]).

To evaluate the levels of Ig bearing the Ighb allotype in the serum of tolerant mice, a previously described [5] solid-phase ELISA was employed. Results were expressed in mg equiva- lent/ml of Ig-bearing Ighb allotype from normal C57BL/6 serum.

2.4 Platelet survival studies

The survival of "'Indium-labeled platelets was evaluated according to the previously described method [12]. Briefly, platelets isolated from BALB/c and C57BL/6 mice were washed three times in PBS containing 0.3% BSA (Sigma) and labeled with "'In-oxine (Amersham Int., Bucks, GB; 10 mCi/

2.5 Platelet-associated Ig (PAIg)

The content of the Unopette kit was harvested in polystyrene tubes and washed three times in PBS containing 1% BSA and 0.01 M NaN3 by centrifugation at 2200 x g for 10 min at room temperature. After the last washing the platelet pellet was solubilized in PBS containing 1% Tween-20 and 2% BSA. The platelet lysates (50 x lo6 platelets/ml) were kept at 4 "C until use. The amount of PAIgG and PAIgM was determined by an ELISA assay using affinity-purified goat anti-mouse anti- bodies specific for y or p chains (Cappel Lab., Cochranville, PA) following the standard methodology [14].

2.6 IgG binding to platelets

The serum levels of IgG binding to platelets were also deter- mined using an ELISA assay on mouse platelet monolayers as described [15]. A goat anti-mouse gamma chain (Cappel) was employed as APh conjugate. Results are expressed as absor- bance values after subtraction of the background values.

2.7 Solid-phase anti-DNA assay

To measure the anti-ssDNA IgG antibodies, an ELISA assay was employed as previously described [6]. Heat-denatured calf thymus DNA (Sigma) was used as ssDNA. Results are ex- pressed in titration units (TU) referring to the standard curve obtained by serial dilutions of a serum pool from 6- to 8- month-old MRL Ipr/lpr mice.

2.8 Circulating immune complexes (CIC)

The levels of CIC were determined by binding to conglutinin in solid phase, using an ELISA methodology as previously described [6]. Bovine conglutinin was obtained from normal adult bovine serum according to Maire et al. [16]. Results were expressed in pg equivalents referring to a standard curve obtained by serial dilutions of heat-aggregated mouse IgG pre- viously incubated for 45 min at 37 "C with fresh normal mouse serum.

2.9 Sucrose gradient ultracentrifugation and preparation of F(ab')* fragments

Serum pools from tolerant and control mice were subjected to ultracentrifugation at 28000 rpm for 18 h in a 5-20% sucrose gradient in veronal-buffered saline using a SW-60 rotor (Beck- man Instruments, Palo Alto, CA).

Affinity-purified Ig from a pool of sera was subjected to pepsin digestion as described by Lamoyi and Nisonoff [17]. The ratio

Eur. J . Immunol. 1989.19: 1693-1699 Thrombocytopenia in mice tolerized to alloantigens 1695

12001 T Ig/pepsin was 33/1. The efficiency of the digestion was tested by measuring the ability of Ig recovered to bind to staphy- lococcal protein A in an ELISA assay. Briefly, 50 yllwell of digested Ig, diluted at 30 yg/ml in carbonate buffer, was incu- bated in microtiter plates (Limbro, Flow) 3 h at 37 "C. After 3 washes, wells were incubated 1 h at room temperature with 1% BSA in PBS, and then 5 h at 4 "C with staphylococcal protein A (Pharmacia) conjugated to APh (10 yglml) in PBS containing 1% BSA. The results obtained with digested Ig were similar to the background.

2.10 Production of hybridomas secreting anti-platelet Mab

Spleen cells from severely thrombocytopenic 3-week-old toler- ant mice were mixed with myeloma cells (NS-2 line) at a ratio of 10: 1 and were fused using an 8% final concentration of polyethylene glycol (mol. wt. 1000; Merck AG, Zurich, Swit- zerland) following the method of Kohler and Milstein [18]. The fused cells were distributed in 96-well plates on a feeder layer of BALB/c peritoneal macrophages. IgG and IgM anti- platelet antibodies secreting hybrids were screened by ELISA (see above) and cloned out by limiting dilution as described [19]. Hybridoma cells were maintained in DMEM sup- plemented with 10% FCS, glutamine, sodium pyruvate, Hepes buffer and penicillin-streptomycin.

2.11 In vivo injection of anti-platelet mAb

Adult BALB/c, C57BL/6 or (C57BLl6 x BALBlc) F1 normal mice were injected i.v. with 200 yl of culture SN (Ig concentra- tion 50 to 100 yglml) from hybridomas producing anti-platelet antibodies. In order to avoid aggregated Ig, SN were ultracen- trifuged (150 000 g during 150 min) before injection. Mice injected with irrelevant mAb (IgGI anti-rheumatoid factor idiotype, IgGl anti-TNP or IgM anti-pneumococcal polysac- charide PS3) were used as control. The platelet count follow- up in mice injected with mAb carried out as previously described in Sect. 2.1.

2.12 Statistical analysis

Significance analysis on differences between results obtained from various groups of mice was performed by using the Wil- coxon rank test. To analyze the correlation between different parameters, Sperman's correlation test was employed.

3 Results

3.1 Thrombocytopenia after induction of neonatal tolerance

The number of circulating platelets in BALB/c parental mice, neonatally tolerized to alloantigens with (C57BL/6 x BALBlc) F1 semiallogeneic lymphoid cells, was investigated. Sequential analysis of the levels of circulating platelets showed a normal level in the first two weeks, then a drop at the third week of age was observed (Fig. 1).

In order to study the relation between the thrombocytopenia and the effective induction of tolerance, newborn BALB/c mice were injected with either lo8 or 25 x lo6 (C57BL/6 x BALB/c) F1 spleen cells. They were killed 4-5 weeks later and

01 . 1 2 3 4 5

age (weeks]

Figure 1. BALB/c mice, injected at birth with loR (C57BL/6 X BALBi c) F1 spleen cells (a), were bled once a week for platelet count from the first to the 5'h week of life. Noninjected BALB/c mice served as controls (A). Results represent the mean values obtained in groups of 6 to 12 mice. SEM are indicated.

0 0

4% w 0 0 0 0

BALB/r+ B A L B / c + BALM 1Wx lo6 2.5 lo6 control 5 cells F, cells

Figure 2. BALB/c mice, injected at birth with loR (C57BLi6 X BALBi c) F1 spleen cells, were bled at 4 weeks of age for platelet count and then killed. Noninjected BALB/c mice served as controls (0). Closed circles (0) represent tolerized injected mice (presence of Ig bearing the Ighb donor allotype in serum and low CTLp frequency against allogeneic H-2b target cells: l / lOOOOO). Open circles (0) represent non tolerized injected mice (no IgHb Ig in serum and CTLp frequency similar to noninjected controls).

studied for the existence of specific CTL alloreactivity against H-2b alloantigens, for the presence of the Ighb allotype as a marker of B cell chimerism and for platelet count. In mice receiving lo8 F, cells, tolerance, as demonstrated by CTLp lower than 1/100000 (CTLp in controls = 1/4000) and by the persistence of the Ighb allotype in serum, was observed in 17/19 mice. Thrombocytopenia, often severe, was observed in 15 of the 17 tolerant but not in the 2 non-tolerant mice (Fig. 2). In mice which received 25 X lo6 FI cells, 4 out of 9 were tolerant and all 4 were thrombocytopenic, whereas the platelet count was in the normal range in the remaining 5 mice.

3.2 Peripheral destruction as the major mechanism of thrombocytopenia

Histological studies were done in eight 1-month-old tolerant mice with severe thrombocytopenia. A marked increase (five- to tenfold) of the number of bone marrow and spleen

J. Merino, H.-Y. Qin, S. Schurmans et al. Eur. J. Immunol. 1989.19: 1693-1699

71bn-C57BL/6 platelets

. . 1 15 26 48

Hours after injection

Figure 3. BALB/c mice were injected at birth with 10' (C57BLI6 X

BALB/c) F, spleen cells. Five weeks later they were injected i.v. with los "'In-labeled platelets from normal BALBlc or from normal C57BL/6 mice (0). Twenty microliters of blood was obtained 1 min, 1 h, 15 h, 24 h and 48 h after the injection of platelets, and the radioactivity contained in these samples was measured. Non-tolerized BALB/c mice injected with radiolabeled platelets served as control (A). Results are expressed as % of the count obtained in the sample drawn 1 min after injection and represent the mean values obtained in groups of 6 to 10 mice. SD are indicated.

100 zoo 300 spleen weight lmgrl

Figure 4. Correlation analysis between the number of circulating platelets and spleen weight in 4-week-old BALB/c mice neonatally injected with lo8 (C57BL/6 x BALBlc) F, spleen cells (0). Nonin- jected BALB/c control mice (A) served as control. Discontinuous lines represented means k 2 SD of the values obtained in BALB/c control mice.

megakaryocytes was observed. In spleens there was also a significant increase in the size of the white and red pulps with prominent young lymphoid cells in the white pulp. No hemor- rhagic nor thrombotic lesions were found in the post mortem macroscopical or microscopical examination of the organs in these thrombocytopenic mice.

The survival of "'In-radiolabeled platelets was studied in tolerant mice. Tolerant and age-matched control BALBlc mice were injected i.v. with 10' "'In-platelets from BALB/c or

C57BW6 normal mice. In tolerant mice the platelets from either strain of mice were more rapidly eliminated from cir- culating blood than in control mice (Fig. 3). There was no significant difference in life-span between BALB/c and C57BL/6 platelets. The radioactivity levels were also mea- sured on day 3 in spleen, lungs, liver, heart and kidneys of these mice. High levels were detected only in spleen and liver, being insignificant in the other organs (data not shown). Therefore, an increased clearance of platelets in spleen and liver appears to take place.

Spleens of the tolerant mice were found to be 2 to 12 times enlarged as compared with the controls and there was a signifi- cant negative correlation between the platelet count and the spleen weight (p < 0.05; Fig. 4), although a severe thrombocy- topenia was observed in mice showing only a moderate spleen enlargement (lower than 150 mg; range of controls: 40 to 90 mg). There was also a correlation between the degree of early platelet clearance (decrease of radioactivity at 24 h) and the spleen weight (p < 0.01), but an accelerated clearance could also be observed in mice showing a moderate splenome- galy.

3.3 Involvement of immune mechanisms in the pathogenesis of thrombocytopenia in tolerant mice

The levels of PAIgG and PAIgM were measured at 3-5 weeks of age in 30 tolerant BALB/c mice, 5 neonatally injected but non-tolerant mice and 13 BALBlc control mice. The levels of PAIgG were markedly increased in tolerant mice as compared to the controls and to the non-tolerized mice (p < lo-'; Table 1). An increased binding of serum IgG to BALB/c or CBA normal platelets was also found in tolerant mice, and this could reflect either the presence of anti-platelet antibodies or of platelet-binding CIC. Indeed, CIC levels, by the conglutinin binding assay, were markedly increased in tolerant mice as compared to non-tolerant and control mice (Table 1).

In tolerant mice there was a significant negative correlation between the platelet counts and the levels of PAIgG

I .* 0

.. I

ad b

_-.

AA A .-

I I

I I 0

P I

0 5 10 12 platelet count I. 105/mrn3~

Figure 5 . Correlation analysis between the levels of PAlgG and the platelet count in 4-week-old BALBlc mice neonatally tolerized to H-2b alloantigens (0) by injection of 10' (C57BL/6 X BALB/c) F, spleen cells. Noninjected BALB/c mice (A) served as controls. Values in BALBic mice neonatally injected with F, cells but not tolerized to H-2b alloantigens (CTL alloreactivity and CTLp frequency similar to the controls) are also represented (U). Discontinuous lines represent means f 2 SD of the values obtained in BALB/c control mice.

Eur. J . Immunol. 1989.19: 1693-1699

Table 1. High levels of PAIgG, binding of serum IgG to platelets and CIC in mice tolerized to alloantigens by neonatal injection of F1 cells")

Thrombocytopenia in mice tolerized to alloantigens 1697

Groups B cell No. of Binding of serum IgG to CIC" chimerism mice PAIgMb' PAIgGb) BALBlc CBA

plateletsc' platelets"

12 0.2 k 0.2 0.1 f 0.1 0.05 rt 0.03 0.11 ? 0.05 < 1

BALB/c + F1 ell^^' - 5 N.D. 0.1 f 0.1 0.11 k 0.18 0.12 f 0.03 c1

BALB/c controls - BALB/c + F, cells + 30 0.6 f 0.7 4.3 f 4.8 0.64 f 0.18 0.54 f 0.13 12 rt 7

a) BALB/c mice were neonatally injected with 10' (C57BL/6 x BALBk) F, spleen cells. Platelets and serum samples, drawn from these mice, were tested at 4 weeks of age.

b) The levels of PAIgG are expressed in ng/106 platelets f 1 SD. c) The values of IgG bound to platelets are expressed as the absorbance at 405 nm k 1 SD. d) CIC levels are expressed in pg equivalents of aggregated mouse gammaglobulidml f 1 SD. e) BALBlc mice neonatally injected with F1 cells in which a normal CTL alloreactivity against H-2b target cells (no tolerance induction) was

observed.

(p < Fig. 5). Conversely, there was no significant corre- lation between the number of circulating platelets and the levels of anti-ssDNA IgG antibodies, CIC or serum IgG bind- ing to normal platelets.

3.4 Significance of platelet-bound IgG in tolerant mice

The life-span of radiolabeled platelets, either from tolerant mice with elevated PAIgG or from tolerant mice with normal PAIgG values, was measured. Three groups of normal BALB/ c mice were injected with "'In-platelets. The first one with platelets from normal BALB/c mice, the second with platelets from tolerant BALB/c mice with low levels of PAIgG (< 0.7 ng/106 platelets) and the third group with platelets from tolerant BALB/c mice with high levels of PAIgG (> 6 ng/106

Hours after injection

Figure 6. Adult normal BALB/c mice were injected i.v. with 10' "'In- labeled platelets from three groups of 4-week-old mice: (a) nonin- jected BALB/c controls (A); (b) tolerized BALB/c mice with low levels of PAIgG (0; < 0.5 ng/106 platelets) and (c) tolerized BALB/c mice with high levels of PAIgG (0; > than 8 ng/106 platelets). These mice were bled (20 p1) 1 min, 1 h, 15 h, 24 h and 40 h after injection of radiolabeled platelets. The radioactivity contained in the samples was counted. Results are expressed as % of the count obtained in the sample drawn 1 min after injection and represent the mean values obtained in groups of six to ten mice. SD are indicated.

platelets). The rate of elimination of platelets from mice with high levels of PAIgG was significantly greater than that observed with platelets from the two other groups (Fig. 6).

The nature of the platelet-binding serum Ig was analyzed by sucrose density-gradient ultracentifugation of sera from toler- ant mice. An increased binding of Ig to platelets was observed in all fractions from 7 S to 25 S with a maximum at 7-12 S corresponding with the IgG peak.

Since the binding of IgG to platelets may reflect the binding of CIC as well as that of anti-platelet antibodies, the specificity of this binding was also analyzed by using F(ab'):! fragments, obtained after pepsin digestion of a pool of serum IgG either from tolerant or normal mice. The F(ab'), fragments from tolerant mice showed a significantly higher binding to normal platelets than that observed with F(ab'):! fragments from nor- mal mice (Fig. 7). The platelet binding of F(ab'):! did not appear contaminated by undigested IgG since no binding was revealed when an anti-mouse Fc antibody was used as a sec-

I I

100 50 10 5 F q concentration @q/ml)

Figure 7. F (ab'), fragments were obtained from different pools of 10 sera from 4-week-old tolerant (0) and noninjected control (A) BALB/c mice. The binding to normal platelets of these different F (ab'), fragments is expressed as absorbance after subtraction of back- ground values (0.120).

1698

ond-step reagent in our ELISA assay, instead of the anti- F(ab')z antibody (data not shown).

J. Merino, H.-Y. Qin, S. Schurmans et al. Eur. J. Immunol. 1989.19: 1693-1699

3.5 Thrombocytopenia induced by mAb from tolerant mice

Hybridoma clones were derived from thrombocytopenic 3-week-old tolerant mice. Seven clones produced IgM anti- bodies and ten clones produced IgGl antibodies reacting with normal BALB/c platelets in ELISA. Ultracentrifuged SN of these hybridomas were injected i.v. (0.2 ml containing 10-15 pg Ig) in normal adult mice in order to study their abil- ity to induce thrombocytopenia. When injected in BALB/c or (C57BW6 x BALB/c) F1 mice, a decrease of more than 50% of the circulating platelets was observed within the first hour after the injection of SN of 2 IgM and 5 IgGl out these 17 mAb (Fig. 8). The SN of these seven hybridomas induced also a slight but significative decrease of the platelet counts when injected in C57BL/6 mice. Therefore, these mAb do not appear to be particularly directed against H-2b alloantigens.

Moreover, pristan-primed (C57BL/6 x BALB/c) F1 mice injected with anti-platelet IgM-secreting cells from one hybridoma also developed a significant thrombocytopenia as compared to mice injected with a hybridoma cell line produc- ing an irrelevant IgM mAb (data not shown).

15' lh 3h 24h

time after injection

Figure 8. Hybridoma cell lines, secreting mAb reacting with mouse platelets, were derived from thrombocytopenic tolerant BALB/c mice. This figure shows the thrombocytopenia induced after injection of 15 pg of an IgGl mAb reacting in vitro with normal BALBlc platelets (O), compared with the platelet count observed after injection of the same amount of an IgGl mAb of irrelevant specificity (A). Results represent the mean values [three (C57BL/6 X BALB/c) F, mice/group] of the sequential platelet counts after the injection of mAb. SD are indicated.

4 Discussion

These results suggest that immune mechanisms are at least partly involved in the pathogenesis of the thrombocytopenia which appears after induction of neonatal tolerance to alloan-

tigens. In this model it was already shown that donor B cells are stimulated in the presence of host CD4' T cells through an allogeneic-like interaction ( [ 5 , 61, Merino, J. et al., submit- ted). This activation of autoreactive B cells results in the pro- duction of various autoantibodies, such as anti-DNA and anti- SM and of CIC. The most characteristic organ involvement in this model is a membrano-proliferative glomerulonephritis mediated by the renal deposition of immune complexes [3]. Thrombocytopenia is also frequently observed in tolerant mice and the present results have shown that, like other manifesta- tions of this syndrome, it is closely related with the establish- ment of an effective state of tolerance.

Our results, showing an accelerated clearance of platelets from normal mice when they were injected in tolerant mice, are consistent with the idea that thrombocytopenia in tolerant mice is mainly produced by an excessive peripheral destruction of platelets. A central mechanism causing a reduced produc- tion of platelets seems unlikely considering the significant increase of megakaryocytes in bone marrow and spleens of thrombocytopenic tolerant mice, although some degree of dysthrombopoiesis cannot be excluded.

The peripheral hyperdestruction of platelets may be due to several mechanisms. Platelet hyperaggregation could lead to sequestration of platelets in the blood microcirculation as occurs in disseminated intravascular coagulation [20]. Our results do not support this possibility since no significant radioactivity was detected in lungs or kidneys of mice injected with radiolabeled platelets from tolerant mice and neither hemorrhagic nor thrombotic lesions were observed in tolerant mice. In contrast, high levels of radioactivity were detected in spleens and livers of these mice after injection of radiolabeled platelets. These data and the existence of a significant negative correlation between the number of circulating platelets and the spleen weight support the possible role of a hypersplenism in thrombocytopenia.

However, the finding that some mice showing only a very moderate spleen enlargement have a severe thrombocytopenia led us to consider that immunological mechanisms can be directly involved. Indeed, there was a strong negative correla- tion between the extent of thrombocytopenia and the levels of PAIgG. Moreover, the life-span of platelets from tolerant mice with high levels of PAIgG was decreased when they were injected in normal BALB/c mice.

The increase of PAIgG in tolerant mice can reflect two differ- ent events. On the one hand, a nonspecific binding can take place by attachment of immune complexes to the surface of platelets through Fc or complement receptors [21, 221, since the serum of tolerant mice contain high levels of immune com- plexes. On the other hand, some of the antibodies bound to the platelet membrane may represent specific anti-platelet autoantibodies. The demonstration that F(ab'h fragments from tolerant mice could bind to normal platelets suggests the existence of such anti-platelet antibodies. In addition, the pro- duction of hybridomas producing anti-platelet antibodies and the demonstration of their pathogenicity in vivo further sup- port the existence of these autoantibodies.

Anti-platelet antibodies in tolerant mice are likely to be induced by a similar mechanism to that proposed for other autoantibodies in this model [5]. In addition to allogeneic-like stimulations of B cells, self-antigens or cross-reacting antigens

Eur. J. Immunol. 1989.19: 1693-1699 Thrombocytopenia in mice tolerized to alloantigens 1699

6 Merino, J., Schurmans, S., Luzuy, S., Izui, S., Vassalli, P. and

7 Merino, J., Schurmans, S. , Duchosal, M., Izui, S. and Lambert, P.

8 Feng, H. M., Glasebrook, A. L., Engers, H. D. and Louis, J . A , ,

9 Ryser, J. E. , and MacDonald, H. R., J. Immunol. 1979.122: 1961. 10 Fazekas de St Groth, S . , J. Immunol. Methods 1982.49: Rll-R23. 11 Lieberman, R., Springer Semin. Immunopathol. 1978. I : 7. 12 Avrameas, S . , Immunochemistry 1969. 6: 43-52. 13 Grau, G. E., Morrow, D. M., Izui, S. and Lambert, P. H., J. Im-

14 Yoshida, H., Yoshida, M., Izui, S. and Lambert, P. H., J. Clin.

15 Grau, G. E., Piguet, P. F., Gretener, D., Vesin, C. and Lambert,

16 Maire, M. A,, Barnet, M. and Lambert, P. H., Mol. Immunol.

17 Lamoyi, E., and Nisonoff, A., J . Immunol. Methods 1983.56: 235. 18 Kohler, G., and Milstein, C., Eur. J. Immunol. 1976. 6: 511. 19 McKearn, T. J., in Kennett, R. H., McKearn, T. J. and Bechtol,

K. B. (Eds.), Monoclonal Antibodies. Hybridomas: A New Dimension in Biological Analyses, Plenum Press, New York 1980, p. 374.

20 Salmon, J., Lambert, P. H. and Hiernaux, M., in Linder, J . and Wilhelmi, E. (Eds.), Infective Inflamation S . Huber, Bern 1968, p. 429.

21 Endresen, G. K. M. and Fovre, O., Int. Arch. Allergy Appl. Immunol. 1982. 67: 33.

22 Taylor, R. P., Kujala, G., Wilson, K., Wright, E. and Harbin, A., J. Immunol. 1985. 14: 2550.

23 Miller, M. H., Urowitz, M. B. and Gladman, D. D., Arthritis Rheum. 1983. 26: 1181.

24 Ballem, P. J., Segal, G . M., Stratton, J. R., Gernsheimer, T., Adamson, J. W. and Slichter, S . J., J. Clin. Invest. 1987. 80: 33.

Lambert, P. H., J. Immunol. 1987. 139: 1426.

H., J . Immunol. 1989, in press.

J. Immunol. 1983. 131: 2165.

munol. 1986. 136: 2348.

Invest. 1985. 76: 685.

P. H., Immunology. 1988. 65: 501.

1981. 18: 85.

could play an important role in the induction of autoan- tibodies. In this context, platelet membrane antigens could be involved in the activation of autoreactive B cells in tolerant mice.

The present model of thrombocytopenia associated with the lupus-like syndrome occurring after induction of neonatal tolerance t o alloantigens, could represent a useful tool for understanding the pathogenesis of the thrombocytopenia in man. This is particularly the case for thrombocytopenia com- plicating SLE [23] and for chronic idiopathic thrombocy- topenia [24].

We wish to thank Dr Shozo Izui (University of Geneva) and Dr Raf- faele D’Amelio (Laboratory of Immunology, Divisione Aerea, Roma) for helpful discussions. We also thank Ms Monika Berney and Mr Guy Brighouse for very useful technical assistance.

Received February 8, 1989; in revised form June 1, 1989.

5 References

1 Billingham, R. W. and Brent, L., Philos. Trans. R. SOC. London

2 Hard, R. C. and Kullgren, B., Am. J. Pathol. 1970. 59: 203. 3 Goldman, M., Feng, H., Engers, H., Hochmann, A., Louis, J.

4 Simpson, E., Hopp, O., Harrison, S., Mosier, M., Melief, D. and

5 Luzuy, S. , Merino, J., Engers, H., Izui, S. and Lambert, P. H.,

Ser. B 1959. 242: 439.

and Lambert, P. H., J. Immunol. 1983. 131: 251.

Cantor, K. H., Immunology 1974. 27: 289.

J. Immunol. 1986. 136: 4420.