(2005) 19, 223–229 & 2005 Nature Publishing Group All rights reserved 0887-6924/05 $30.00 www.nature.com/leu Role of B-cell -associated molecules and lipid rafts in CD5-induced of B CLL cells

Y Renaudineau1,SNe´dellec1, C Berthou2, PM Lydyard3, P Youinou1 and J-O Pers1

1Laboratory of Immunology, Brest University Medical School, Brest, France; 2Department of Hematology, Brest University Medical School, Brest, France; and 3Department of Immunology and Molecular Pathology, Royal Free and University College Medical School, London, UK

A total of 40 patients with B-CLL were investigated for CD5- Lyn resides constitutively in membrane microdomains re- triggered apoptosis and categorized as 20 resistant (group I) ferred to as lipid rafts (LRs), enriched in ganglioside M1 (G ), and 20 sensitive patients (group II). The densities of surface IgM M1 and clustering transduction . Following crosslinking, a (sIgM) and CD5 were lower in group I than group II, as were the 7 percentages of CD79b þ , CD38 þ , and Zap70-expressing B subset of BCRs partition into these LRs, where CD79a/CD79b cells. CD5 signaling was mediated through the BCR in group II undergoes tyrosine (Y) phosphorylation by Lyn. Once recruited B cells, as established by coimmunoprecipitation of CD5 and to phosphorylated CD79a/CD79b, Syk is also Y phosphorylated CD79a and tyrosine phosphorylation of CD79a. Following by Lyn, and the SHP-1 is displaced from the BCR. SHP-1, which colocalization of CD5 and sIgM in membrane lipid rafts (LRs), is linked to the CD79a/CD79b subunits through CD5, negatively Syk became associated with these molecules, whereas SHP-1 was uncoupled from CD5. Nonresponsiveness to CD5 cross- regulates . linking in group I was ascribed to three possible abnormalities, Our hypothesis that heterogeneous responses of malignant and defined as three subgroups of patients. In subgroups Ia cells to CD5 engagement reflect differences in components of and Ib, CD5 and sIgM colocalized within the LRs. SHP-1 the signaling machinery was strengthened by the discovery that remained attached to the BCR in subgroup Ia, but not in groups of B-CLL samples express different , most notably subgroup Ib, where signal transduction was associated with an among those modulated during BCR signaling.8 We provide excess of truncated CD79b. In subgroup Ic, CD5 and sIgM segregated into different LRs, resulting in no signaling of evidence for the delivery of the CD5-induced apoptosis signal apoptosis. through CD79a/CD79b, and the translocation of this dimer to Leukemia (2005) 19, 223–229. doi:10.1038/sj.leu.2403601 the LRs, along with CD5. Published online 16 December 2004 Keywords: CD5; chronic lymphocytic leukemia; BCR; lipid rafts; apoptosis Materials and methods

Patients and controls Introduction A total of 40 untreated patients fulfilling the criteria for the 9 In B-CLL, accumulation of B cells reflects their resistance to diagnosis of B-CLL were enrolled in the study, and their informed consent was obtained according to our institutional apoptosis, and one of their salient features is the expression of 10 CD5.1 This leukemia comprises several entities, as indicated by review board. According to the Binet classification, 19 the molecular complexity of the disease.2 To confirm such a patients were stage A, 16 B, and five C. Blood was also heterogeneity, our finding that CD5 crosslinking resulted in obtained from 10 healthy volunteers, and tonsils from 14 apoptosis of B cells3 led us to examine CD5-induced apoptosis children. of B-CLL cells. The patients were classified into those resistant 4 and those sensitive to CD5-triggered apoptosis. At that time, we and reagents were unable to identify the reasons for such differences in the responses. The antibodies (Abs) were from Beckman-Coulter (Villepinte, CD5 is associated with the CD79a/CD79b transduction France), unless otherwise stated. F(ab’) UCHT2 anti-CD5 (gift partner of surface IgM (sIgM) in the vicinity of the BCR. Weak 2 from Professor P Beverley, London, UK) were crosslinked on the expression of sIgM has indeed been ascribed to a reduced B-CLL cell surface with sheep F(ab’) anti-mouse F(ab’) expression of CD79b in B cells from more than 90% of the 2 2 (Jackson, West Grove, PA, USA). Anti-CD3, -CD4, -CD8 and patients, due to mutations or alternative splicing of the CD79b -CD56 monoclonal Abs (mAbs) were used to enrich B cells, of .5 In the latter case, the lack of exon 3 engenders shortened which the purity was determined by phycoerythrin (PE)- or FITC- variants of the transcript and truncated (TR) isoforms of the anti-CD19, combined with FITC-anti-CD3 or PE-anti-CD5, . Since defective transduction has been reported, it has respectively. FITC-rabbit F(ab’) anti-human IgM (Jackson) was been suggested that intracellular signaling might need to be 2 associated with PE-anti-CD79b, anti-CD38, or anti-CD5 mAbs. attenuated in B-CLL cells to allow survival in the face of constant B cells were permeabilized with 70% methanol prior to self-antigen stimulation.6 The simplest mechanism to prevent B- incubation with anti-Zap70 mAb (clone 29: Becton-Dickinson, cell responses is to uncouple the BCR from tyrosine kinases, Le Pont de Claix, France), or control IgG , and their binding such as Syk and Lyn. 1 detected using FITC-sheep F(ab’)2 anti-mouse IgG (Jackson). Cell lysates were precipitated with Leu-1 anti-CD5 mAb Correspondence: Dr P Youinou, Laboratory of Immunology, Brest (Becton-Dickinson), which recognizes epitope 1 of CD5, University Medical School Hospital, BP 824, F-29 609 Brest Cedex, 11 France; Fax: þ 33 298 22 38 47; E-mail: [email protected] whereas UCHT2, which recognizes epitope 2, was used in Received 31 March 2004; accepted 11 October 2004; Published Western blotting (WB) experiments. Anti-CD79a and anti- online 16 December 2004 phospho-Y (PY) mAbs (Santa-Cruz, Santa Cruz, CA, USA) were CD5-induced apoptosis in B-CLL Y Renaudineau et al 224 also used in immunoprecipitation (IP) experiments. Other 721C for extension. A final 10-min extension cycle was reagents included anti-Lyn, anti-Syk, anti-PY (all from Santa- performed with 35 cycles for CD79b and 30 for GAPDH. The Cruz), anti-SHP-1 (Becton-Dickinson) and anti-CD79a mAbs. intensity of each CD79b signal was normalized to that of Biotinylated (biot) goat anti-mouse IgG and anti-rabbit IgG Abs, GAPDH. Simultaneous amplification of two different transcripts and HRP- streptavidin were from Amersham-Pharmacia (Saclay, in the same reaction provided a good estimate of their relative France). Biot-cholera toxin B (CTB) is specific for GM1 (Sigma, St frequencies. The results were expressed as TR/FL ratios. Louis, MO, USA), and its binding indicated the quality of The same 21 patients were also evaluated by quantitative isolated LRs. In the confocal microscopy (CM) studies, LRs were PCR, which was carried out in 10 ml mixtures containing 50 ng s labeled with Alexa Fluor (AF) 594-CTB (Molecular Probes, template cDNA, 1 Â Sybr Green PCR master mix, and 500 nM 0 0 Eugene, OR, USA), combined with FITC-goat F(ab’)2 anti-human of each primer. We used the 3 primer 5 -CCTCGCAGCGTC- IgM (Sigma) to detect translocation of sIgM into the LRs, or with ACTATGTCCTC-30 located in exon 6 for both isoforms of 0 0 FITC-sheep F(ab’)2 anti-mouse F(ab’)2 to crosslink the CD5 Ab. CD79b, and a pair of 5 primers: 5 -CACAGAGCTGCGAGT- CD5 staining was also combined with rhodamine-anti-human CATGGGATTCAG-30 encompassing exons 2 and 3 of the FL IgM (Jackson) to demonstrate colocalization of CD5 and sIgM isoform, or 50-CGGTACCGGAATCCCAAAGGATTCAG-30 en- within the LRs. compassing exons 2 and 4 of the TR isoform. The transcripts were cloned using the p-TrueBlue ligation . Amplification conditions consisted of one cycle at 501C for 2 min, one cycle at Cell culture 951C for 10 min, and 40 cycles at 951C for 15 s and 601C for 1 min. Peripheral blood mononuclear cells (PBMCs) were separated by Ficoll–Hypaque density gradient centrifugation. Those from patients were treated with a mixture of anti-CD3, -CD4, -CD8, Determination of apoptosis and -CD56 mAbs, and the B cells negatively selected using immunomagnetic anti-mouse IgG Ab-coated beads (BioAd- For a given patient, six wells of flat-bottomed 96-well plates vance, Emerainville, France). Normal PBMCs and tonsillar cell (Nunc, Roskilde, Denmark) were coated with 50 mg/ml of sheep suspensions were T-cell-depleted by two cycles of rosetting with F(ab’)2 anti-mouse F(ab’)2 for 24 h at 41C, followed by 10 mg/ml sheep erythrocytes, and their B cells purified as above. This of F(ab’)2 anti-CD5. Each of these six coated wells was procedure yielded cell populations containing more than 96% B arbitrarily paired with an uncoated well. Each sample was cells, of which, for the B-CLL patients, 77–99% expressed CD5, seeded at 5 Â 105 cells per well in 0.2 ml of RPMI medium 1640 as shown by FITC-anti-CD19 and PE-anti-CD5 staining. FITC- with 10% FCS. Apoptosis was established through the binding of anti-CD3 staining revealed no residual T cells. annexin V (AV). An aliquot of cells was stained with FITC-AV and propidium iodide (PI) after 24 h, and another after 36 h. Apoptotic cells were defined as those positive for AV but Flow cytometry negative for PI, and necrotic cells as those positive for AV and PI. Significances of AV binding increases from six uncoated B cells were incubated with various double combinations of wells to six paired coated wells were determined using the mAbs for 30 min on ice, washed twice, and analyzed on an Wilcoxon test for paired data: when P was 40.05, the patient’s EPICS Elite flow cytometer (Beckman-Coulter). The percentages sample was assigned to the resistant group, and when P was of positive cells and mean fluorescence intensities were o0.05, it was assigned to the sensitive group. Apoptosis was compared with isotype controls. The number of molecules per confirmed by cell morphology, enumeration of hypoploid cells cell was quantified by the amount of Ab binding to the cells at and DNA fragmentation analysis, using methods that we have saturating concentrations, using the Quantumt Simply Cellulars previously described.3,4 kit (Flow Cytometry Standards Corps, San Juan, PR, USA).

Immunoprecipitation and Western blotting CD79b transcript analysis Two aliquots of 1.5 Â 107 cells were left on ice for 30 min: the Since anti-CD79b mAbs from commercial sources cannot first one was incubated with 15 mg/ml UCHT2 F(ab’)2 and the distinguish full-length (FL) and TR variants of CD79b,12 RT- second with medium alone. Both were washed and warmed to PCR was required. mRNA was purified from two tonsillar 371C for 5 min. Before IP with Leu-1 anti-CD5 or anti-CD79 and seven PBMC preparations, and from 13 CD5-resistant and mAb, the cells were treated for 30 min at 41C with 1% Brij 96 eight CD5-sensitive randomly selected B-CLLs. Purified mRNA (Sigma) in lysis buffer (20 mM Tris-HCl, pH 7.5, 140 mM NaCl, was incubated with oligo (dT) primers, deoxynucleotides mix 1mM EDTA) with 1 mM PMSF, 10 mg/ml aprotinin, and 1 mM and 200 U of RT from Moloney murine leukemia virus for sodium orthovanadate. For IP with anti-PY, 1% Triton X-100 50 min at 421C, and 2 U of RNase H for another 20 min at 371C. (Sigma) was substituted for Brij 96 to lyse the cells. All these Two pairs of primers were designed: for the CD79b gene,5 preparations were centrifuged at 10 000 g for 10 min at 41C, and 50-GTGACCATGGCCAGGCTGGCGTTGT-30 plus 50-CCATC their supernatants cleared with protein G-coated beads, and CATGTGTGGGGACGGATC-30; and for the GAPDH gene, with mouse IgG. Protein G was conjugated to magnetic beads 50-CTTAGCACCCCTGGCCAAGG-30 plus 50-CTTACTCCTTG (Miltenyi, Paris, France). A measure of 2 mg of the precipitating GA-GGCCATG-30. We obtained PCR products of 978 bp for Ab was added to the lysates, along with 50 ml of protein G- FL-CD79b, 602 bp for its TR isoform, and 542 bp for GAPDH. microbeads. After incubation on ice for 30 min, the lysate was PCR amplification was carried out, after the initial denaturation passed over a ‘m column’ placed in the magnetic field of the step for 4 min at 941C, and the addition of 2.5 U of Taq mMACS separator. This was rinsed 4 times with lysis buffer and polymerase (Genaxis, St Cloud, France). The cycles were 30 s at once with 20 mM Tris-HCl. For SDS-PAGE, the column-bound 941C for denaturation, 60 s at 551C for annealing and 60 s at proteins were incubated for 5 min with 20 mlof951C hot

Leukemia CD5-induced apoptosis in B-CLL Y Renaudineau et al 225 1 Â SDS gel loading buffer. IP proteins were then eluted using 50 ml of the same buffer. The proteins were resolved by 10% SDS-PAGE, blotted onto a PVDF membrane, and unbound sites blocked overnight with 1% gelatin. The strips were probed, developed with biot-goat anti-mouse IgG, plus HRP-streptavi- din, and analyzed using the ECL detection system (Amersham- Pharmacia).

LR studies

To determine the role of LRs, the cells were incubated with 10 mM of methyl-b-cyclo-dextrin (MCD, Sigma) for 30 min at 371C in serum-free medium, before being cultured. The LRs were isolated based on their insolubility in nonionic detergents and buoyant density on sucrose gradient. Cells (7 Â 107 ) were washed with 25 mM Tris-HCl, 150 mM NaCl and 5 mM EDTA (TNE), and lysed for 30 min on ice in 1% Triton X-100 in TNE buffer with protease inhibitors. The lysates were centrifuged at 500 g. In all, 1 ml of the supernatant was mixed with 1 ml of 85% sucrose in TNE, overlaid with 3 ml of 35% sucrose and 1.5 ml of 5% sucrose in TNE, and centrifuged at 180 000 g for 12 h at 41C. The LR band was visible at the interface of 35 and 5% sucrose, and the nonraft membrane (NRM) fraction at the bottom of the tube. In all, 20 ml of each fraction was fractionated by SDS-PAGE, transferred onto PVDF membranes, blotted, and revealed with biot-goat anti-mouse IgG and HRP-streptavidin. Three double-staining combinations were used for CM. In the first combination, B cells were incubated for 30 min on ice with AF 594-CTB to target red GM1 to the LRs and FITC-rabbit F(ab’)2 anti-human IgM to stain sIgM green, along with crosslinking F(ab’)2 mouse anti-CD5. One aliquot of cells was left at 41Casa control for nonactivation of CD5, and a second warmed to 371C for 5 min to allow activation through CD5. Both were washed and analyzed for double-stained cells with or without CD5 activation. In the second combination, B cells were incubated Figure 1 Spontaneous and CD5-induced apoptosis in B cells from with AF 594-CTB and crosslinking F(ab’)2 mouse anti-CD5. One 40 patients. (a) B-CLL B cells were cultured for 24 or 36 h in the aliquot was left at 41C, and another warmed to 371C for 5 min. absence (open bars) or in the presence (filled bars) of F(ab’)2 anti-CD5 Both were incubated with FITC-sheep F(ab’) anti-mouse F(ab’) crosslinked with F(ab’)2 anti-F(ab’)2. Apoptotic cells were annexin V 2 2 (AV) þ /propidium iodide (PI)À, and necrotic cells AV þ /PI þ . Two for 30 min on ice to induce crosslinking of CD5. In the third groups of patients were identified: 20 patients were resistant to CD5- combination, B cells were incubated with rhodamine-rabbit triggered apoptosis and necrosis (group I), and 20 were sensitive F(ab’)2 anti-human IgM and crosslinking F(ab’)2 mouse anti- (group II). (b) A representative example of each group is shown. CD5. One aliquot was left at 41C, and, again, another warmed to 371C for 5 min. Both were incubated with FITC-sheep F(ab’)2 anti-mouse F(ab’)2 for 30 min on ice to visualize CD5. The cells delineated: no changes were seen with cells from 20 cases were fixed in 3% paraformaldehyde for 10 min and centrifuged (group I), following CD5 crosslinking. In the remaining 20 at 300 g onto microscope slides for analysis. (group II), this treatment increased the percentage of AV þ /PIÀ cells from 9.471.1% (spontaneous) to 23.972.3% at 24 h (P ¼ 0.0007), and from 16.271.6 to 32.871.8% at 36 h Statistical analysis (P ¼ 0.0008), and those of AV þ /PI þ cells from 4.070.6 to 14.771.9% at 24 h (P ¼ 0.0007), and from 10.171.3 to Results were expressed as means7s.e.m., and significances 30.272.1% at 36 h (P ¼ 0.0004). Figure 1b shows a representa- determined using the w2-test with Yates correction when tive example of each group. AV-binding cells were increased as required, and the paired Wilcoxon test and the Mann–Whitney early as 18 h in culture in 15 group II patients, whereas this U-test for unpaired data. occurred later in another five, where differences became significant only after 36 h. These five patients were thus resistant at 24 h, and became sensitive to CD5-triggered apoptosis at Results 36 h. Apoptosis resistance or susceptibility to CD5-stimulation was stable for at least 3 years, as determined from serial blood CD5-induced apoptosis samples from 12 group I and 14 group II patients. In addition to AV binding, cell morphology, enumeration of hypoploid cells, To confirm that CD5 did not induce apoptosis in all samples, B and DNA fragmentation analysis were used to confirm apoptosis cells from 40 patients were incubated with or without F(ab’)2 in six group I and nine group II randomly selected patients. anti-CD5, and the CD5-added apoptosis measured after 24 and Advanced stage disease was less frequent in the resistant than in 36 h. Figure 1a shows that two groups of patients were the sensitive group of patients: 14 were at stage A and six at

Leukemia CD5-induced apoptosis in B-CLL Y Renaudineau et al 226 Table 1 Phenotype of B-CLL B cells in relation to sensitivity to CD5-induced apoptosis

Group I Group II Difference ( I vs II) Control PBMCs Tonsils

CD79b 5.971.1% (n ¼ 20) 16.971.5% (n ¼ 20) Po10À4 93.171.5% (n ¼ 10) 88.273.2% (n ¼ 5) CD38 6.171.5% (n ¼ 20) 30.273.6% (n ¼ 20) Po10À6 59.577.5% (n ¼ 5) 69.073.1% (n ¼ 14) Zap70 10.771.6% (n ¼ 20) 65.675.7% (n ¼ 20) Po10À3 0.770.2% (n ¼ 4) ND Means7s.e.m. are presented. PBMC: peripheral blood mononuclear cell, ND: not determined.

stage B in group I, while five were at stage A, 10 at stage B and five at stage C in group II (P ¼ 0.0006).

Relationships between CD5-mediated apoptosis and cell markers

The most direct approach to distinguish patient samples showing differences in their response to CD5 crosslinking was to analyze their phenotype. The reduction in expression of sIgM was more pronounced in group I than in group II B-CLLs (20.073.4 vs 33.272.7 Â 103 molecules per cell, P ¼ 0.0074). In contrast, CD5 was overexpressed by B cells in all patients, but less so in group I than in group II (35.873.3 vs 51.974.3 Â 103 molecules per cell, P ¼ 0.0066). The percentages of CD79b- and CD38- positive cells were also lower in group I than in group II (Table 1). Three additional observations are worthy of note. Firstly, when patients with 20% or more B cells expressing CD38 were considered positive,13 and those with less than 20% considered negative, fewer CD38- positive patients were seen in group I than in group II: 1/20 vs 16/20 (Po10À5). Secondly, less Zap7014 was expressed in group Figure 2 Coimmunoprecipitation of BCR-associated molecules by I than in group II (10.771.6 vs 65.675.7%, P ¼ 0.0013), and in anti-CD79a, anti-CD5, or anti-phosphotyrosine (PY) Abs, with and 7 without prior engagement of CD5. (a) B cells from tonsils and group II CD38-negative than in CD38-positive group II cells (46.7 9.0 patients. The strips were probed, the Abs revealed by goat anti-mouse 7 vs 78.9 3.5%, P ¼ 0.0064). Thirdly, the TR-CD79b/FL-CD79b IgG, plus HRP-streptavidin, and the blots developed using an ECL kit. ratios (supplementary Figure 1) were higher in group I than in (b) B cells from group I patients. Three subgroups were identified, and group II (1.5570.23 vs 0.8170.05, P ¼ 0.0145). These results a representative example of each is presented. Cells are CD5- were verified by quantitative PCR: there were increasing crosslinked (‘yes’) or not (‘no’). amounts of TR products in group I B cells, whereas FL products predominated in group II B cells. subgroup Ib, although SHP-1 was uncoupled from CD5 (compare lane 8 with 7, or lane 10 with 9), the cascade was CD5-induced apoptosis requires an intact BCR signaling not initiated (compare lane 12 with 11). This blockade was pathway associated with an excess of the TR form of CD79b, relative to its FL variant (TR-CD79b/FL-CD79b ¼ 2.6070.70 in five sub- The suggestion, based on alterations in surface markers, that group Ib vs 1.1070.11 in six subgroup Ia, P ¼ 0.0139). This CD5 signaling involves CD79 was confirmed by IP. It appeared characteristic does not prove any inhibiting effect of the TR form that CD5 was associated with CD79a, since these two of CD79b,12 as there was no relationship between transcription molecules coprecipitated each other in CD5-stimulated B cells of either variant and the membrane expression of the molecule: from four tonsils and seven group II patients (lanes 2 and 6, and 6.972.3% of B cells expressed CD79b in subgroup Ia vs lanes 4 and 8, respectively, in Figure 2a). SHP-1 and Lyn were 2.971.3% in subgroup Ib. Upon CD5 stimulation, inspite of the also linked to the BCR, provided CD5 was not engaged. segregation of CD5 and CD79a, Syk was not recruited. In seven Following engagement of CD5, SHP-1 was dissociated from patients, termed subgroup Ic, CD5 was not linked to CD79a, CD5, Syk recruited by the BCR (compare lane 6 with 5, and lane since these two molecules did not precipitate together (see lanes 8 with 7) and CD79a and Syk Y phosphorylated (compare lane 14 and 13, and lanes 16 and 15). Therefore, Syk was not 10 with 9, and lane 12 with 11). The resistance of group I cells recruited, SHP-1 not dissociated, and CD79a not Y-phosphory- suggests that their early signal transduction pathway is faulty. lated. Although CD5 and CD79a were associated, as shown by their co-IP in 13 group I patients (see lanes 1 and 2, or 3 and 4 in Figure 2b), CD5 engagement did not Y phosphorylate CD79a Involvement of LRs (compare lane 6 with 5, or lane 12 with 11). In seven of these 13 patients, termed subgroup Ia, CD5 Cholesterol-depleted LRs did not transduce the CD5 message, engagement did not displace SHP-1 from CD5 (compare lane 2 confirming that their role in initial signaling was critical. Owing with 1, or lane 4 with 3 in Figure 2b). In six patients, termed to previous incubation of the cells with MCD, the percentage of

Leukemia CD5-induced apoptosis in B-CLL Y Renaudineau et al 227 CD5-induced AV-binding cells diminished from 35.372.1 to In subgoup Ia, CD79a and part of CD5 colocalized with LRs, but 5.671.2% (P ¼ 0.0096) in tonsillar CD5 þ B cells, and from still SHP-1 was linked to CD5 (compare lane 2 with lane 1, and 52.774.6 to 12.673.9% (P ¼ 0.0042) in group II patient B cells. lane 4 with lane 3). In subgroup Ib, CD5 and CD79a were also Therefore, we examined the association of CD5 with other BCR- together within the LRs, whereas SHP-1 was excluded (compare related molecules. One may predict that B cells from tonsils and lane 6 with lane 5, and lane 8 with lane 7). In subgroup Ic, CD5 group II patients, where CD5 coprecipitated CD79a, would also moved to LRs, but, based on the absence of Syk, no apoptosis colocalize CD5 and CD79a in their LRs. These were isolated message was delivered. Again (Figure 3c), these findings were from B cells of three group II patients, and their position in the confirmed by CM examination of CD5 or sIgM, combined with sucrose gradient determined by the presence of GM1 (Figure 3a). the LR marker. In the seven subgroup Ic patients, CD5 and sIgM Following CD5 engagement, CD5 and CD79a were associated were not in the same LRs (lower 3 panels), although CD5 with GM1, Syk and Zap70 recruited to the LRs, and SHP-1 left in crosslinking encouraged CD5 (upper three panels), as well as the NRMs. Association of CD5 and sIgM within the LRs was then sIgM (middle 3 panels), to partition into LRs. Hence, CD5 and analyzed by CM (supplementary Figure 2): overlay of the green- sIgM segregated into different LRs. stained CD5, or green-stained sIgM, with the red-stained LRs was seen as yellow, demonstrating that CD5 and sIgM were within the same LRs. Discussion Next, B cells from the 20 group I patients were investigated for association of CD5 and the BCR relative to the LRs (Figure 3b). We have confirmed that two groups of B-CLL patients can be distinguished based on the response of their B cells to CD5 crosslinking. 4 Those from half of them resisted CD5-induced apoptosis (group I), while those from the remainder (group II) underwent more apoptosis with than without anti-CD5. One explanation could be differences in the numbers of CD5 and IgM molecules expressed by the different groups. Although the intensity of CD5 staining was higher in leukemic than in normal B cells, it was less so in group I than in group II. That of sIgM was lower in leukemic than in normal B cells, but more so in group I than in group II. The magnitude of the signal could thus be related to the number of CD5 molecules to which the Ab binds. Such differences could reflect the stages of differentiation or the levels of activation of the two categories of leukemic B cells, or, alternatively, that groups of leukemic cells responding differ- ently in these experiments derive from separate normal CD5 þ B-cell populations that we have postulated to exist.15 However, that this could explain differences in the responses to CD5 engagement was discounted by more in-depth analysis of CD5 in relation to other molecules. A high expression level of CD38 characterized those patients sensitive to CD5 cross- linking. This confirms the involvement of CD38 in BCR- triggered apoptosis,12,16 the behavior of CD38 and CD5 as inducible activation markers,17 and the relationship between Figure 3 Involvement of the lipid rafts (LR) in early signaling 16 mediated through CD5 crosslinking of B cells. (a) LRs of B cells from 3 CD38 expression and early signaling through sIgM (Table 2). group II B-CLL patients were separated from the nonraft membrane However, while the combined stimulation of B-CLL cells with (NRM) fraction. Following electroelution, they were blotted with biot- anti-CD38 mAb and IL-2 has been shown to prolong survival,18 anti-GM1 Ab, or with nonbiot anti-CD5, anti-CD79a, anti-Syk, anti- CD5 crosslinking favors apoptosis. Both of these in vitro SHP-1 or anti-Zap70 Abs, all these five Abs being revealed by biot- phenomena are associated with a more agressive course of the goat anti-mouse IgG. Cells are CD5-crosslinked (‘yes’) or not (‘no’). (b) LRs of B cells from 20 group I patients were isolated and blotted. Cells disease. However, the role of CD5 in leukemic cells in vivo is are CD5-crosslinked (‘yes’) or not (‘no’). Representative examples are the central issue. Importantly, the common denominator of shown. (c) Confocal microscopy analysis of B cells from one these two opposing observations is the competency of the BCR representative B-CLL patient from subgroup Ic. transducing machinery. Its effectiveness appears to be more

Table 2 Summary of abnormalities of early signaling in B-CLL B cells in response to crosslinking of CD5

B-cell samples Association of Phosphorylation Colocalization of TR-CD79b/ FL- Association of Presence of CD79b with of CD79a CD79a and CD5 CD79ba SHP-1 with CD5/ Syk in lipid CD5 in lipid rafts CD79a rafts

Tonsils (n ¼ 3) + + + 0.470.1b À + B-CLL subgroup Ia (n ¼ 7) + À + 1.170.1 + À B-CLL subgroup Ib (n ¼ 6) (+) À (+) 2.670.7 ÀÀ B-CLL subgroup Ic (n ¼ 7) ÀÀ À1.370.1 ÀÀ B-CLL group II (n ¼ 20) + + + 0.870.1 À + aTR-CD79b and FL-CD79b are the truncated and the full-length transcripts of CD79b, respectively. bMean7s.e.m. (+) denotes a weak response.

Leukemia CD5-induced apoptosis in B-CLL Y Renaudineau et al 228 frequent in the worse cases. Using an anti-CD5 mAb is an Acknowledgements artificial way of stimulating the molecule. Yet, none of the five ligands for CD5 have been shown to be functional in vitro,18 so This study was supported by the Acade´mie Nationale Franc¸aise de there is no way of providing a physiological stimulus via CD5.19 Me´decine, the Conseil Re´gional de Bretagne, and the Commu- The IgVh gene mutational status is another prognostic indicator, naute´ Urbaine de Brest, France. We thank Professor P Beverley, such that patients with mutated IgVh genes survive longer than London, UK for providing reagents, and appreciate the secretarial those using unmutated IgVh genes.20 Patients with a severe assistance of C Se´ne´ and S Forest. course exhibit an intact sIgM transduction pathway16 and possess CD38, suggesting that the ability to respond to external signals by prolonged survival of the cells contributes to poor Supplementary Information prognosis.20 Since CD5 is part of the BCR, and CD5-mediated apoptosis is associated with high CD38 expression, group II Supplementary Information accompanies the paper on the patients should use unmutated IgVh genes. Our finding of Leukemia website (http://www.nature.com/leu). an elevated Zap70 expression in group II B cells is consistent with this prediction, since it correlates closely with unmutated IgVh genes.21 Thus, aberrant expression of Zap70 could References contribute to the high sensitivity of group II B cells to CD5- triggered apoptosis. This view is supported by the observation 1 Lavabre-Bertrand T, Janossy G, Exbrayat C, Bourquard P, Duperray that CD5 is associated with Y-phosphorylated Zap70 in C, Navarro M. Leukemia-associated changes identified by quanti- .22 tative flow cytometry.II. CD5 over-expression and monitoring in B-CLL. Leukemia 1994; 8: 1557–1563. We next established that the CD5-induced early signaling 2 Dighiero G. Unsolved issues in CLL biology and management. events proceeded through CD79a/CD79b by showing the Leukemia 2003; 17: 2385–2391. mutual co-IP of CD5 and CD79, followed by Y phosphorylation 3 Pers JO, Jamin C, Le Corre R, Lydyard PM, Youinou P. Ligation of of CD79a. Lyn was constitutively associated with the BCR, CD5 on resting B cells, but not on resting T cells, results in while, in response to CD5 crosslinking, Syk was recruited, and apoptosis. Eur J Immunol 1998; 28: 4170–4176. SHP-1 uncoupled from CD5. This is consistent with the 4 Pers JO, Berthou C, Porakishvili N, Burdjanadze M, Le Calvez G, Abgrall JF et al. CD5-induced apoptosis of B cells in some patients dissociation of SHP-1 in CD5-negative B cells, following with chronic lymphocytic leukemia. Leukemia 2002; 16: 44–52. 23 activation, Furthermore, our studies on LRs showed that 5 Alfarano A, Indraccalo S, Circosta P, Minuzzo S, Vallario A, CD5, CD79a and Syk, but not SHP-1, cohabit in the LRs Zamarchi R et al. Alternatively spliced form of CD79b may after CD5 activation. Indeed, for tonsil and group II B-CLL B account for altered B-cell receptor expression in B-chronic cells, CD5, seen initially outside the LRs, moved there, lymphocytic leukemia. Blood 1999; 93: 2327–2335. along with CD79 and sIgM, while SHP-1 was left behind. 6 Bernal A, Pastore RD, Asgary Z, Keller SA, Cesarman E, Liou HC et al. Survival of leukemic B cells by engagement of the antigen Lyn phosphorylates CD79a/ CD79b, which then pushes Syk into receptor. Blood 2001; 98: 3050–3057. being translocated to the LRs. 7 Cheng PC, Dykstra ML, Mitchell RN, Pierce SK. A role for lipid The results of the IP and LR studies warranted division of rafts in B cell antigen receptor signaling and antigen targeting. J Exp group I into three subgroups. In subgroup Ia, although CD5 and Med 1999; 190: 1549–1560. sIgM colocalize within the LRs, the apoptosis signal was not 8 Rosenwald A, Alizadeh AA, Widhopf G, Simon R, Davis RE, Yu X delivered: SHP-1 failed to separate from CD5 in the LRs, so that et al. Relation of phenotype to immunoglobulin genotype in B cell chronic lymphocytic leukemia. J Exp Med 2001; the proximity of SHP-1 facilitates inhibition of signaling. In 194: 1639–1647. contrast, SHP-1 was not found in LRs in subgroup Ib B cells, but 9 Matutes E, Catovsky D. The value of scoring systems for the this detachment of SHP-1 from CD5 was not sufficient for signal diagnosis of biphenotypic and mature B-cell disorders. transduction. The percentage of CD79b-expressing B cells was Leuk 1994; 13 (Suppl 1): 11–14. higher in group II, compared with group I patients whose B cells 10 Binet JL, Auquier A, Dighiero G, Chastang C, Piguet H, Goasguen J contained more copies of the TR-CD79b transcript. This et al. A new prognostic classification of chronic lymphocytic leukemia derived from a multivariate survival analysis. Cancer disproportionate expression of the TR form of CD79b, relative 1981; 48: 198–206. to its FL variant, was associated with the blockade. Such an 11 MacKenzie LE, Lydyard PM. Epitope mapping of the CD5 excess in TR-CD79b was found in the non-responders in molecule. In: Knapp W, Do¨rken B, Rieber EP, Stein H, Gilks general, but more so in subgroup Ib patients.5 Although the WR, Schmidt RE, von dem Borne AEGKr (eds) Leucocyte Typing alternatively spliced variant might not be functional, this IV. Oxford: Oxford University Press, 1989, pp 336–337. truncated molecule could act as an inert decoy, or as a protein 12 Cragg MS, Chan HT, Fox MD, Tutt A, Smith A, Oscier DG et al. 12 The alternative transcript of CD79b is overexpressed in B-CLL and with directly opposing functions. In subgroup Ic, CD5 was inhibits signaling for apoptosis. Blood 2002; 100: 3068–3076. separated from CD79. Following crosslinking, CD5 and BCR 13 Ibrahim S, Keating M, Do KA, O’Brien S, Huh YO, Jilani I et al. were targeted to LRs, but did not partition into the same LRs. The CD38 expression as an important prognostic factor in B-cell driving forces for LR localization are unknown, but evidence is chronic lymphocytic leukemia. Blood 2001; 98: 181–186. accumulating for the existence of distinct subsets of LRs.24 14 Chen L, Widhopf G, Huynh L, Rassenti L, Rai KR, Weiss A et al. Transduction of the apoptosis signal from CD5 proceeds through Expression of ZAP-70 is associated with increased B-cell receptor signaling in chronic lymphocytic leukemia. Blood 2002; 100: CD79a/CD79b, which suggests a strong association of CD5 and 4609–4614. sIgM within the same LRs. 15 Youinou P, Jamin C, Lydyard PM. CD5 expression in human B-cell We have established that CD5 and the BCR colocalize in LRs populations. Immunol Today 1999; 20: 312–316. for the early signaling events leading to apoptosis, and described 16 Zupo S, Isnardi L, Megna M, Massara R, Malavasi F, Dono M et al. several mechanisms that could underlie defective signaling from CD38 expression distinguishes two groups of B-cell chronic CD5 in B-CLL cells. Since CD5-mediated signaling might be lymphocytic leukemias with different responses to anti-IgM antibodies and propensity to apoptosis. Blood 1996; 88: important for its effectiveness against B-CLL in vivo, our in vitro 1365–1374. 4 findings offer a rationale for the observation that not all patients 17 Zupo S, Dono M, Massara R, Taborelli G, Chiorazzi N, Ferrarrini respond to treatment with anti-CD5 mAb. M. Expression of CD5 and CD38 by human CD5-B cells:

Leukemia CD5-induced apoptosis in B-CLL Y Renaudineau et al 229 requirement for special stimuli. Eur J Immunol 1994; 24: genes, inferior clinical outcome, and distinct gene expression 1426–1433. profile. Blood 2003; 101: 4944–4951. 18 Deaglio S, Capobianco A, Bergui L, Du¨rig J, Morabito F, Du¨hrsen 22 Gary-Gouy H, Lang V, Sarun S, Boumsell L, Bismuth G. In vivo U et al. CD38 is a signaling molecule in B-cell chronic association of CD5 with tyrosine-phosphorylated ZAP-70 and p21 lymphocytic leukemia cells. Blood 2003; 102: 2146–2155. phospho-zeta molecules in human CD3+ thymocytes. J Immunol 19 Berland R, Wortis HH. Origins and functions of B-1 cells with 1997; 159: 3739–3747. notes on the role of CD5. Annu Rev Immunol 2002; 20: 253–300. 23 Gary-Gouy H, Bruhns P, Schmitt C, Dalloul A, Dae¨ron M, Bismuth 20 Hamblin TJ, Davis Z, Gardiner A, Oscier DG, Stevenson FK. G. The pseudo-immunoreceptor tyrosine-based activation motif of Unmutated Ig VH genes are associated with a more aggressive CD5 mediates its inhibitory action on B-cell receptor signaling. J form of chronic lymphocytic leukemia. Blood 1999; 151: Biol Chem 2000; 275: 548–556. 1848–1854. 24 Roper K, Corbeil D, Huttner WB. Retention of prominin 21 Wiestner A, Rosenwald T, Barry TS, Wright G, Davis RE, in microvilli reveals distinct cholesterol-based lipid micro- Henrickson SE et al. ZAP-70 expression identifies a chronic domains in the apical plasma membrane. Nat Cell Biol 2000; 2: lymphocytic leukemia subtype with unmutated immunoglobulin 582–592.

Leukemia