Stromal endothelial cells establish a bidirectional crosstalk with chronic lymphocytic leukemia cells through the TNF-related factors BAFF, APRIL and CD40L

Abstract

Chronic lymphocytic leukemia (CLL) is a clonal B cell disorder of unknown origin. Accessory signals from the microenvironment are critical for the survival, expansion and progression of malignant B cells. We found that the CLL stroma included microvascular endothelial cells (MVECs) expressing BAFF and APRIL, two TNF family members related to the T cell-associated B cell-stimulating molecule CD40 ligand (CD40L). Constitutive release of soluble BAFF and APRIL increased upon engagement of CD40 on MVECs by CD40L aberrantly expressed on CLL cells. In addition to enhancing MVEC expression of the CD40 receptor, leukemic CD40L induced cleavases that elicited intracellular processing of pro-BAFF and pro-APRIL proteins in MVECs. The resulting soluble BAFF and APRIL proteins delivered survival, proliferation, Ig gene-remodeling and differentiation signals by activating CLL cells through TACI, BAFF-R and BCMA receptors. BAFF and APRIL further amplified CLL cell survival by up-regulating the expression of leukemic CD40L. Inhibition of TACI, BCMA and BAFF-R expression on CLL cells, abrogation of CD40 expression in MVECs, or suppression of BAFF and APRIL cleavases in MVECs reduced the survival and diversification of malignant B cells. These data indicate that BAFF, APRIL and CD40L form a CLL-enhancing bidirectional signaling network linking neoplastic B cells with the microvascular stroma.

Introduction

Chronic lymphocytic leukemia (CLL) is an incurable disease of unknown etiology characterized by progressive accumulation of clonal B lymphocytes with mature morphology and phenotype (1). The clinical course of CLL is variable and correlates with the mutational status of immunoglobulin (Ig) heavy chain variable (V_H) genes and the percentage of leukemic cells expressing CD38 and zeta-chain-associated protein 70 (ZAP-70) (2-4). Presence of unmutated V_H genes and expression of CD38 and ZAP-70 are biological variables associated with worse prognosis. B cell-intrinsic factors such as abnormal expression of anti-apoptotic NF-κB and Bcl-2 proteins and accumulation of heterogeneous genomic lesions regulate the survival, expansion and clonal evolution of CLL cells (5, 6). B cell-extrinsic factors such as antigenic stimulation may also be important, as leukemic B cells from unrelated groups of CLL patients express restricted Ig VDJ genes encoding stereotyped antigen-binding V_H regions (7-9).

Accessory signals from the microenvironment further contribute to the pathogenesis of CLL (10, 11). Indeed, stromal cells such as nurse-like cells, mesenchymal cells, follicular dendritic cells and macrophages recruit CLL cells by secreting CXCL12 and CXCL13 chemokines and enhance CLL cell survival by releasing B cell-activating factor of the TNF family (BAFF or BLyS) and a proliferation-inducing ligand (APRIL) (12-14). These tumor necrosis factor (TNF) family members are usually expressed by myeloid and epithelial cells and engage BAFF-receptor (BAFF-R or BR3), transmembrane activator and CAML interactor (TACI) and B cell maturation antigen (BCMA) on B cells (15-21). The ensuing recruitment of TNF receptor-associated factor (TRAF) adaptor proteins leads to the activation of NF-κB, a transcription factor that regulates the survival, proliferation, diversification and differentiation of B cells (19, 20, 22). A similar pathway becomes activated after engagement of the CD40 receptor on B cells by CD40 ligand (CD40L), a TNF family member typically expressed by T cells but also by CLL cells (5, 23).

CD40L, BAFF and APRIL likely represent an important component of CLL proliferation centers, which consist of pseudo-follicles containing focal aggregates of activated and proliferating CLL cells. Proliferation centers also include T cells expressing CD40L and stromal macrophages and nurse-like cells releasing BAFF and APRIL (13, 24). Remarkably, malignant B cells shape the cellular composition of proliferation centers to generate a CLL-supportive microenvironment. Indeed, CLL cells release chemokines such as CCL22 to attract CD40L-expressing T cells and produce angiogenic molecules such as vascular endothelial growth factor (VEGF) to stimulate the formation of microvessels (24, 25). In addition to favoring leukemic dissemination, microvessels deliver survival signals to CLL cells through unknown factors released by endothelial cells (26).

We found that the CLL stroma contained abundant microvascular endothelial cells (MVECs) that constitutively expressed the B cell-stimulating factors BAFF and APRIL. MVECs increased the release of soluble BAFF and APRIL after engagement of CD40 on MVECs by CD40L on CLL cells. Signals from CD40 induced furin and TNF-α converting enzyme (TACE or ADAM17), which cleaved inactive BAFF and APRIL precursors into active soluble proteins. By activating NF-κB through TACI, BAFF-R and BCMA, endothelial BAFF and APRIL delivered survival, proliferation, activation and Ig DNA-modifying signals to CLL cells. Endothelial BAFF and APRIL also increased the expression of CD40L on CLL cells, indicating that BAFF, APRIL and CD40L link malignant B cells with stromal MVECs through an integrated bidirectional signaling network. Interruption of this network by specific inhibitors may be useful for the treatment of CLL.

Materials and Methods

Patients and samples

Leukemic cells were isolated from the peripheral blood of 21 CLL patients (Table 1). The diagnosis of CLL was made according to standard criteria (27). All blood samples were obtained by phlebotomy upon request of consent according to a protocol approved by the Institutional Review Board (IRB) of Weill Medical College of Cornell University. Normal peripheral blood B cells were isolated from buffy coats provided by the New York Blood Center. Frozen tissues samples from lymph nodes, bone marrow and spleen of 10 healthy controls and 10 CLL patients were obtained from tissue repositories according to a protocol approved by the IRB of Weill Medical College of Cornell University.

Table 1.

Genotypic and phenotypic features of CLL cases.

ID	IgVH	Zap-70	CD38	CD40L	TACI	BCMA	BAFF-R
011	UM*, VH4-31	+	+	+	−	+	+
014	UM VH1-31	+/−‡	−	+	−	−	+
021	MU§, VH3-48	+	+	+/−	+/−	−	+
025	MU, VH2-5	+/−	−	−	−	+	+
031	MU, VH3-74	+	−	+	+	−	+
045	UM, VH1-69	+	−	+	+	+/−	+
048	MU, VH4-4	+	+/−	+	+	+	+
051	UM, VH2-5	+	−	−	+/−	−	+
062	N/A¶	N/A	N/A	+	+	+	+
067	N/A	+/−	−	+	+	−	+
078	MU, VH3-30	+/−	−	−	−	+	+
085	MU, VH2-5	N/A	N/A	−	−	−	+
090	UM, VH3-7	−	−	+	−	−	+
091	N/A	N/A	−	+	−	−	+
095	UM, VH3-11	−	+	+	−	−	+
098	UM, VH1-69	+	+	+/−	+	+	+
107	MU, VH6-1	+	+/−	+	+	−	+/−
112	N/A	+	+	−	−	+	+
123	N/A	+	+	+	+/−	−	+
126	MU, VH4-34	+/−	−	+	+	+/−	+
128	MU, VH1-18	+	+	+	+	−	+

^*UM, unmutated (≤ 2% mutations compared to putative germline VDJ gene).

^§MU, mutated (> 2% mutations compared to putative germline VDJ gene).

^¶N/A, not available.

^‡Intermediate expression (< 15% but > 5% of the cells).

Cells

Normal and CLL B cells were MACSorted from peripheral blood mononuclear cells (PBMCs) by negative selection using a B-cell isolation kit (Miltenyi Biotec, Bergish-Gladbach, Germany). Marginal zone CD19⁺IgD^lowCD27⁺ B cells, follicular naïve CD19⁺IgD^highCD27⁻ B cells, and memory CD19⁺IgD⁻CD27⁺ B cells were sorted from splenic mononuclear cells using a BD FACSAria III (BD Biosciences, San Diego, CA). Sorting was preceded by labeling of splenic mononuclear cells with fluorescein (FITC)-, phycoerythrin (PE)- or allophycocyanin (APC)-conjugated mouse monoclonal antibodies (MoAbs) to human CD19 (BD Pharmingen, San Diego, CA), IgD (Southern Biotech, Birmingham, AL) and CD27 (BD Pharmingen). Human umbilical vein endothelial cells (UVECs) and lymphatic microvascular endothelial cells (LMVECs) are commercially available (Lonza, Walkersville, MD). Splenic microvascular endothelial cells (SMVECs) were MACSorted from enzymatically digested splenic cell suspensions stained with mouse MoAbs to CD31 and mannose receptor (BD Pharmingen).

Cultures and reagents

Endothelial cells were propagated in endothelial basal medium EBM-2 supplemented with endothelial growth medium EGM-2MV SingleQuots (Lonza). Endothelial cells were seeded with EBM-2 plus EGM-2MV medium in 12-well plates at a concentration of 1.5 × 10⁴ cells/well for 48 hours prior to co-culture with CLL cells. The EBM-2 plus EGM-2MV medium was mixed with standard RPMI medium supplemented with 10% FCS (1:4, vol/vol) after seeding CLL cells at 1 × 10⁶ cells/ml. This mix is referred to as endothelial medium throughout the text and was used in all cultures involving CLL cells unless specified otherwise. Conditioned endothelial medium was obtained after culturing SMVECs for 4 days in endothelial medium. Some endothelial cell-CLL cell co-cultures were performed in a 24-well transwell system (Corning, Lowell, MA). Recombinant BAFF, APRIL and CD40L MegaLigand (Alexis Biochemicals, San Diego, CA) were used at 100 ng/ml. TACI-Ig and CD40-Ig decoy receptors and Fc5 or Ig (IgG1) controls (ZymoGenetics, Inc., Seattle, WA) were used at 5 μg/ml. Furin convertase inhibitor (peptidyl choromethylketone) and TNF-α processing inhibitor-2 (TAPI-2) were used at 100 μM and 5 mM, respectively (Enzo Life Sciences, PA), whereas IKK inhibitor III (BMS-345541) was used at 1 μM (Calbiochem, San Diego, CA). After a 3-hour incubation with IKK inhibitor III (an NF-κB inhibitor) or control DMSO vehicle, CLL cells were washed twice with endothelial medium and then co-cultured with MVECs.

Flow cytometry

Cells were stained with FITC-, PE-, APC-, or cyanine-3-conjugated mouse MoAbs to human CD5, CD19, CD38 (BD Pharmingen), BAFF-R, BAFF, CD31 (eBiosciences, San Diego, CA), CD40, CD40L (Ancell, Bayport, MN), CD54, CD102, CD144 (BioLegend, San Diego, CA), CD105 (ImmunoTools, Friesoythe, Germany), IgG, IgA or control isotype-matched reagents (Southern Biotech). Biotinylated antibodies to TACI (ZymoGenetics) were stained with PE-conjugated streptavidin (BD Pharmingen). A goat antibody to BCMA (Santa Cruz Biotechnologies, Santa Cruz, CA) was stained with an appropriate PE-conjugated secondary antibody (BD Pharmingen). Dead cells were excluded from analysis using 7-amino-actinomycin D. Events were acquired using a FACScalibur or LSRII (BD Biosciences) and were analyzed by FlowJo software (Tree Star, Ashland, OR).

Immunofluorescence and histology

Endothelial cell lines were grown on glass cover-slides in endothelial growth medium until reaching confluence. Tissues and cells were fixed and washed as described elsewhere (15) and then stained with the following unconjugated or conjugated primary antibodies to various human antigens: goat polyclonal antibody (PoAb) to IgD (Southern Biothech); mouse MoAb to Pax-5 (Santa Cruz); rabbit PoAb to Pax-5 (NeoMarkers); mouse MoAb to Ki-67, Factor VIII, CD31, CD3 and elastase (Dako, Carlsbad, CA); rabbit PoAb to Von Willebrand factor (Dako); rabbit PoAb anti-APRIL (ProSci, Poway, CA); mouse MoAb to CD3, CD11c, CD21, CD206 (BD Pharmingen); goat PoAb to AID (Santa Cruz); rabbit PoAb to furin (Santa Cruz Biotechnologies); rabbit PoAb to CD63 (Novus Biologicals, Littleton, CO); mouse MoAb to CD68 (Abcam, Cambridge, MA); mouse MoAb to CD40L (Ancell); and rabbit PoAb to TNF-α converting enzyme (TACE) (Sigma). Primary antibodies were labeled with an appropriate Alexa 488-, 546- or 647-conjugated or cyanin 5-conjugated secondary antibodies (Jackson Immunoresearch Laboratories, West Grove, PA). Cell nuclei were visualized with DAPI, 4′,6-diamidine-2′-phenylindole dihydrocloride (Boehringer Mannheim, IN). Coverslips were applied with SlowFade reagent (Molecular Probes, OR). Images were acquired with a Zeiss Axioplan 2 microscope (Atto Instruments).

Cell viability and proliferation assays

Cell viability was evaluated through a trypan blue exclusion test. To measure apoptosis, cells were double stained with FITC-conjugated Annexin V and propidium iodide (Calbiochem) and analized by flow cytometry. Cell proliferation and/or survival were monitored using a yellow tetrazolium 3-(4, 5-dimethylthiazolyl-2)-2 , 5-diphenyltetrazolium bromide (MTT) cell proliferation assay (Trevigen, Gaithersburg, MD). In this assay, dehydrogenases expressed by metabolically active cells generate NADH and NADPH that convert MTT into formazan. Briefly, 100-μl aliquots of culture supernatant were transferred into a 96-well plate and incubated overnight with 20 μl of MTT reagent at 37° C and 5% CO₂, followed by a 2-hour incubation with 100 μl of detergent reagent. Absorbance readings were performed at 570 nm in a microplate reader. Cell proliferation was further evaluated through a thymidine incorporation assays as described in published reports (28). The proliferation index was calculated according to the formula: mean of cpm of triplicates with stimulus/mean of cpm of triplicates without stimulus. Carboxyfluorescein diacetate succinimidyl ester (CFSE) was used to trace cell division as instructed by the manufacturer (Invitrogen).

ELISA

Human IgG, IgA, IgM, BAFF and APRIL were measured by ELISA as described elsewhere (29).

RT-PCR, quantitative RT-PCR, and Southern blot analysis

RNA was extracted with the RNeasy Mini Kit (Qiagen) and cDNA was synthesized as described (29). Germline Iγ1-Cγ1, Iγ2-Cγ2 and Iγ3-Cγ3 transcripts as well as Iγ1/2-Cμ, Iγ3-Cμ and Iα1-Cμ switch circle transcripts were RT-PCR amplified and were hybridized with appropriate radiolabeled probes by Southern blot analysis as described (15). TACI forward 1′-CAGA CAACTCGGGAAGGTACC-1′ and reverse 5′-GCCACCTGATCTGCACTCAGCTTC-1′, BAFF-R forward 5′-ACCGCGGGACTGAAAATCT-1′ and BAFF-R reverse 1′-CACGCTTATTTCTGCTGTTCTGA-1′, BCMA forward 1′-GCTTGCATACCTTGTCAA CTTCGATG and BCMA reverse 1′-GAATCGCATTCGTTCCTTTCACTG-1′, CD40 forward 1′-CACACTGCCACCAGCACAA-1′ and CD40 reverse 1′-GCCTTCTTCACAGGTGCAGAT-1′, AID forward 5′-AGAGGCGTGACAGTGCTACA-1′ and AID reverse 5′-TG TAGCGGAGGAAGAGCAAT-1′, BAFF forward 1′-GGGAATTCGATGACTCCACAG AAAGGGAGCAGTCA-1′ and BAFF reverse 1′-GGGCGGCCGCTTACAGCAGTTTCAA TGCACCAAAAAA-1′, APRIL forward 1′-CTGCACCTGGTTCCCATTAAC-1′ and APRIL reverse 1′-AAGAGCTGGTTGCCACATCAC-1′, TACE forward 1′-GAGGCGATTAATG CTACTTGCA-1′ and TACE reverse 1′-TGGAGGCGGGCACTCA-1′, furin forward 1′-GGCCTGCTCGTCCACACT-1′ and furin reverse 1′-TCGTCACGATCTGCTTCTCATT-1′ and β-actin forward 5′-GCTTGCATACCTTGTCAACTTCGATG-1′ and β-actin reverse 1′-GA ATCGCATTCGTTCCTTTCACTG-1′ were used to quantify specific mRNAs by RT-PCR as previously reported (29).

Immunoblots

Equal amounts of total protein lysate, cytoplasmic extract or nuclear extract were fractioned onto a 10% sodium dodecyl sulfate-polyacrylamide gel and transferred onto nylon membranes (BioRad, Hercules, CA). After blocking, membranes were probed with primary antibodies to APRIL (ProSci), BAFF (Millipore, Bedford, MA), TACE, furin or β-actin (Santa Cruz Biotechnology) as reported (29). Then, membranes were washed and incubated with an appropriate secondary antibody (Santa Cruz Biotechnology). Proteins were detected with an enhanced chemiluminescence detection system (Amersham, Little Chalfont, UK).

RNA interference

The following small interfering RNA (siRNA) duplexes were used after validation with three distinct siRNAs for each target gene of interest: GGAAATCAAATTCAGCTCTTT siRNA to BAFFR; CAGCGGAGTGGAGAAGTTGAA siRNA to TACI (Ambion, Austin, TX); ACCATTAAAGGACGAGTTTAA siRNA to BCMA (Qiagen, Valencia, CA); and AGGGCACCUCAGAAACAGATT siRNA to CD40 (Quiagen). The specificity of each siRNA was validated using a scrambled siRNA as control. Transfection of specific or control (scrambled) siRNA in CLL cells was performed as recommended by the manufacturer (Ambion). Briefly, 5 μM siRNA was incubated with 100 μl of nucleofection mix (Amaxa, Walkersville, MD) and 100 μl of CLL cell suspension (3 × 10⁶ cells) for 2 min before transfection using the Amaxa nuclefector program U-15. After 24 h, cells were seeded in 1 ml of complete medium. Transfection efficiency was monitored after 48 hours by measuring CLL cells positive for a control green fluorescent protein (GFP)-expressing plasmid through flow cytometry. Specific suppression of TACI, BCMA and BAFF-R gene expression was validated through quantitative RT-PCR (QRT-PCR). These measurements demonstrated highly specific targeting of TACI, BCMA or BAFF-R expression by distinct siRNAs. Transfection of CD40 or control siRNA in endothelial cells was performed using a SuperFect protocol (Qiagen). Briefly, endothelial cells were cultured on a monolayer with complete endothelial growth medium until optimal confluency was reached. 5 nM of siRNA duplexes and 7.5 μl of HiPerFect Transfection Reagent were diluted in 100 μl of endothelial cell culture without serum. The mixture was incubated at room temperature to allow formation of transfection complexes and further added to the cells drop-wise. Transfection efficiency was monitored by measuring the expression of CD40 through QRT-PCR after 48 hours.

Statistical analysis

One-way ANOVA followed by one-tailed unpaired Student’s t-test was utilized to determine statistical significance.

RESULTS

MVECs promote CLL cell survival

To elucidate the composition of the CLL stroma, we immunostained lymph node, spleen and bone marrow tissues from CLL patients for molecules typically expressed by stromal cells. Immunohistofluorescence showed that all CLL tissues included microvessels expressing CD31 (PECAM-1, platelet endothelial cell adhesion molecule-1) and containing factor VIII and von Willebrand (vW) factor coagulation proteins (Fig. 1A-B). MVECs from these microvessels established an intimate contact with malignant B cells expressing IgD (a molecule associated with mature B cells and CLL cells) and generally were more abundant than macrophages, follicular dendritic cells, dendritic cells (DCs) and neutrophils expressing CD68 (Fcγ receptor), CD21 (complement receptor 2), CD11c (integrin α X) and elastase (a granular protease), respectively.

FIGURE 1.

MVECs infiltrate CLL tissues and promote malignant B cell survival and proliferation. A. Immunofluorescence analysis of CLL lymph node (LN; one of 5 cases), spleen (SP; one of 3 cases) and bone marrow (BM; one of 3 cases) tissue samples stained for CD11c, CD21, CD31, CD68, CD206, elastase, factor VIII, vW factor or CD206. DAPI (blue) counterstains nuclear DNA. Original magnification, x10. B. CLL LN and SP tissue samples (one of 5 and 3 cases, respectively) stained for IgD (green), factor VIII or CD206 (red), and DAPI (blue). Original magnification, x40. C. CLL and normal LN tissue samples (5 cases in each group) stained for CD54, CD105, CD144 (red), and DAPI (blue). Original magnification, x40. D. Flow cytometric analysis of CD31, CD54, CD102, CD105, CD144 and CD206 expression by UVECs, SMVECs and LMVECs. Grey solid profile, isotype-matched control. E. Time course analysis of viable trypan blue-negative CLL cells (one of 5 cases) exposed to standard medium (SM, orange line), endothelial medium (EM, black line), conditioned medium from endothelial cell cultures (CM, blue line), or SMVECs (red line). F. MTT proliferation assay with CLL cells from mutated (MU, left panel) or unmutated (UM, right panel) samples (5 cases in each group) co-cultured with endothelial medium alone (control), UVECs or SMVECs. Results are expressed as fold induction release of MTT-derived formazan by proliferating CLL cells. Data are from one of three or five experiments yielding similar results (A-E) or summarize three different experiments (F). Error bars, s.e.m.; * P < 0.05 (ANOVA followed by one-tailed unpaired Student’s t-test).

In CLL spleens, MVECs expressed CD206 (mannose receptor, a molecule highly expressed by sinusoidal vessels) and were in close proximity with neoplastic B cells. Similar to MVECs from normal lymphoid tissues, MVECs from CLL tissues expressed CD54 (ICAM-1, intercellular adhesion molecule-1), CD102 (ICAM-2, intercellular adhesion molecule-2), CD105 (endoglin) and CD144 (VE-cadherin) (Fig. 1C). Flow cytometry showed that model MVECs from healthy individuals expressed a phenotype similar to that of MVECs from CLL patients (Fig. 1B-D and Supplementary Fig. S1). Indeed, UVECs, SMVECs and LMVECs expressed CD31, CD54, CD102, CD105 and CD144 as MVECs from CLL patients did (Fig. 1D). As in SMVECs from CLL patients, SMVECs further expressed CD206, whereas UVECs and LMVECs did not (Fig. 1D).

Given their phenotypic similarity with MVECs from CLL tissues, SMVECs, LMVECs and UVECs were utilized in a simplified co-culture model to establish the role of MVECs in the survival of malignant B cells. In the presence of SMVECs, approximately 90% of CLL cells survived after 7 days of culture as established by a canonical viability assay (Fig. 1E). The pro-survival effect of SMVECs was partly dependent on cell-to-cell contact, because conditioned medium from SMVEC cultures induced the survival of only 40% of CLL cells after 7 days. Of note, SMVECs and conditioned medium induced stronger survival signals than endothelial medium or standard medium alone, which indeed only induced 30% and 5% CLL cell survival, respectively. The partial pro-survival activity of endothelial medium alone was likely mediated by VEGF, a known CLL cell survival cytokine (30, 31). Compared to endothelial medium alone, MVECs increased CLL survival and/or proliferation as measured in an MTT assay in which metabolically active cells convert MTT into formazan (Fig. 1F).

In agreement with these findings, proliferation centers from CLL tissues included factor VIII-positive MVECs that were in close contact with actively cycling IgD-positive malignant B cells expressing the proliferation-associated nuclear molecule Ki-67 (Fig. 2A). The possible proliferation-inducing activity of MVECs was suggested by means of CFSE dilution assays, which documented the division of a fraction (from 5% to 15% of total B cells) of the leukemic clone upon exposure of CLL cells to UVECs or SMVECs but not endothelial medium alone (not shown). Thymidine incorporation assays demonstrated induction of DNA replication in CLL cells exposed to SMVECs but not endothelial medium alone (Fig. 2B). Unlike CLL cells, nonmalignant B cells expressing IgM and IgD did not augment DNA replication in response to SMVECs, suggesting that MVECs require interaction with a CLL-specific factor(s) to induce the proliferation and/or survival of leukemic B cells. These data show that MVECs are a major component of the CLL stroma and indicate that MVECs deliver survival and possibly proliferation signals to CLL B cells.

FIGURE 2.

MVECs promote division, activation and differentiation of CLL cells. A. Immunofluorescence analysis of CLL lymph node (LN) tissue samples (one of 5 cases) stained for Ki-67 (green or red), factor VIII (red) and IgD (blue or green). DAPI (blue) counterstains nuclei. Original magnification, x10 (left panel) or x63 (right panels). Arrowheads indicate proliferating CLL cells. B. 3HTdR incorporation of B cells from CLL patients or healthy individuals (5 cases in each group) cultured for 4 days with endothelial medium alone (control) or SMVECs. C. Southern blot analysis of germline Iγ1-Cγ1, Iγ2-Cγ2 and Iγ3-Cγ3 transcripts as well as Iγ1/2-Cμ, Iγ3-Cμ and Iα-Cμ switch circle transcripts RT-PCR amplified from CLL cells (one of 5 cases with little or no constitutive CSR) cultured for 4 days with endothelial medium alone (control), UVECs or SMVECs. β-actin is a loading control. bp, base pairs. D. QRT-PCR analysis of AID mRNA from CLL cells (5 cases) cultured for 4 days as in C. RE, relative expression as compared to freshly isolated CLL cells. E. CLL LN tissue (one of 5 cases) stained for AID (green), factor VIII (red) and Pax5 (blue). Arrowheads point to an intercellular bridge that connects factor VIII-positive MVECs with an AID-positive CLL cell, AID being a hallmark of ongoing B cell activation. F. ELISA of IgM, IgG and IgA secreted by CLL cells (5 cases) cultured for 7 days as in C. G. Flow cytometric analysis of IgG and IgA on CLL cells (one of 5 cases) cultured with endothelial medium alone (red open profile) or UVECs (blue open profile) for 7 days. Grey solid profile, isotype-matched control. H. Flow cytometric analysis of CD19 and CD38 on CLL cells (one of 5 cases) cultured for 7 days as in C. Data are from one of five experiments yielding similar results (A, C, E, G, H) or summarize three different experiments (B, D, F). Error bars, s.e.m.; * P < 0.05 (ANOVA followed by one-tailed unpaired Student’s t-test).

MVECs trigger CLL cell activation and differentiation

In addition to active proliferation, a fraction of the CLL clone shows ongoing Ig DNA remodeling via class switch DNA recombination (CSR) (32, 33). This process is highly dependent on activation-induced cytidine deaminase (AID), a DNA-editing enzyme expressed by actively proliferating CLL cells in response to poorly understood signals generated by the activated microenvironment (32-37). To further elucidate the stimulating activity of MVECs on the leukemic clone, we determined whether MVECs induce AID expression and CSR in CLL cells (32). RT-PCR followed by Southern hybridization analysis showed that, compared to CLL cells exposed to endothelial medium alone, CLL cells incubated with SMVECs induced or enhanced the expression of molecular hallmarks of ongoing CSR from IgM to IgG1/IgG2, IgG3 or IgA (15), including switch circle Iγ1/2-Cμ, Iγ3-Cμ and Iα-Cμ transcripts, respectively (Fig. 2C). SMVECs also up-regulated the expression of germline Iγ1-Cγ1, Iγ2-Cγ2 and Iγ2-Cγ2 transcripts, an early event in IgM to IgG1, IgG2 and IgG3 CSR, respectively (Fig. 2C). In the majority of cases, CLL cells constitutively expressed germline Iα-Cα transcripts and MVECs did not further augment this expression (not shown).

SMVECs also up-regulated the expression of transcripts for AID in CLL cells (Fig. 2D). This in vitro finding correlated with tissue data showing MVEC-proximal CLL cells that expressed AID (Fig. 2E), a hallmark of ongoing B cell activation and Ig gene diversification (32). ELISA and flow cytometric studies showed that SMVECs also induced secretion of class-switched IgG and IgA as well as unswitched IgM proteins (Fig. 2F), expression of surface IgG and IgA proteins (Fig. 2G), and up-regulation of surface CD38 (Fig. 2H), a microenvironment-inducible activation molecule that activates neoplastic B cells by binding to CD31 on MVECs (38-41). UVECs induced similar activation and differentiation signals to CLL cells. Similar to endothelial medium alone, a CLL cell survival factor such as the cytokine IL-4 did not induce CLL cell proliferation, AID expression or Ig secretion in spite of enhancing the viability of CLL cells (data not shown). These findings demonstrate that MVECs provide powerful stimulating signals to CLL cells, including AID-inducing and CSR-activating signals.

MVECs release BAFF and APRIL

Myeloid and stromal cells enhance the survival of peripheral B cells and under specific conditions elicit AID expression and CSR by releasing BAFF and APRIL (20). These immune mediators are also involved in the pathogenesis of CLL (13, 42-46), which prompted us to determine the involvement of BAFF and APRIL in the activation of CLL cells by MVECs. ELISAs demonstrated constitutive BAFF and APRIL release by MVECs, including SMVECs, LMVECs and UVECs (Fig. 3A). Contact-dependent exposure of MVECs to CLL cells further increased endothelial secretion of BAFF and APRIL, whereas contact-independent exposure of MVECs to CLL cells in a transwell system or exposure of MVECs to conditioned medium from CLL cell cultures did not increment steady state BAFF and APRIL secretion (Fig. 3A).

FIGURE 3.

MVECs enhance BAFF ad APRIL release by interacting with CLL cells. A. ELISA of BAFF and APRIL secreted by UVECs, SMVECs or LMVECs cultured for 7 days in the absence (left panel) or presence (right panel) of CLL cells (3 cases) or CLL cell-derived conditioned medium (CM) (3 cases). Cultures were performed in plates with or without a transwell system (TW, no TW, respectively). B. Immunofluorescence analysis of BAFF or APRIL (green), CD31, vWF, CD63 or TGN-46 (red), and DAPI (blue) in MVECs. Original magnification, x40. C. Immunofluorescence analysis of CLL lymph node (LN; 5 cases) and spleen (SP; 3 cases) tissue samples stained for BAFF or APRIL (green), factor VIII or CD206 (red) and DAPI (blue). Original magnification, x40. D. Flow cytometric analysis of BAFF on SMVECs. Gray open profile, isotype-matched control; red open profile; BAFF. E. Time course analysis of viable trypan blue-excluding CLL cells from unmutated (UN) or mutated (MU) samples (one of 5 cases in each group) cultured with endothelial medium in the presence or absence of BAFF or APRIL. F. Formazan release assay with CLL cells (5 cases) cultured for 4 days as in D. G. QRT-PCR analysis of AID mRNA from CLL cells (5 cases) cultured for 4 days as in D. RE, relative expression compared to freshly isolated CLL cells. H. Flow cytometric analysis of surface IgA on CLL cells (one of 5 cases) cultured for 4 days with endothelial medium alone (red open profile) or BAFF (blue open profile). Gray solid profile, isotype-matched control. I. ELISA of IgM, IgG and IgA secreted by CLL cells cultured for 7 days as in E. Data are from one of three or five experiments yielding similar results (B-E and H) or summarize three different experiments (A, F, G and I). Error bars; s.e.m.; * P < 0.05 (ANOVA followed by one-tailed unpaired Student’s t-test).

Fluorescence microscopy and confocal microscopy confirmed the constitutive expression of BAFF and APRIL proteins by MVECs, including SMVECs from CLL tissues, and further showed that these MVECs stored BAFF and APRIL proteins in intracellular endosomal and granular structures that also contained the tetraspanin protein CD63 and the coagulation proteins factor VIII and vW factor (Fig. 3B-C). Flow cytometry and QRT-PCR assays demonstrated that MVECs did not express membrane-bound BAFF (Fig. 3D) and yet contained more BAFF and APRIL transcripts than myeloid cell types with well-defined B cell-licensing functions (20), including DCs and macrophages (Supplementary Fig. S2). Compared to CLL cells exposed to endothelial medium alone, CLL cells exposed to endothelial medium supplemented with BAFF or APRIL showed increased proliferation and/or survival, AID transcription as well as surface IgA expression and IgA secretion (Fig. 3E-I). BAFF and APRIL also increased the secretion of IgM and IgG (Fig. 3I). These data indicate that MVECs promote CLL cell survival and diversification through both contact-dependent and contact-independent mechanisms, including BAFF and APRIL.

MVECs stimulate CLL cells through BAFF and APRIL

Published data show that BAFF and APRIL deliver survival signals to normal and malignant B cells through BAFF-R, plasma cell survival signals through BCMA, and CSR signals through TACI (19, 20, 44, 45, 47-49). TACI can also deliver B cell survival and proliferation signals (20, 50). As shown by flow cytometry, BAFF-R was detected on malignant B cells from all 21 CLL cases analyzed, whereas TACI and BCMA were found on malignant B cells from 12 and 9 CLL cases, respectively (Fig. 4A and Table 1). BAFF-R expression was generally more elevated than TACI and BCMA expression (Fig. 4A). QRT-PCR analysis demonstrated that TACI, BCMA and BAFF-R transcripts increased after incubation of CLL cells with CD40L (Fig. 4B), a molecule aberrantly expressed by CLL cells and a powerful inducer of BAFF and APRIL release by myeloid cells (5, 15, 44).

FIGURE 4.

MVECs stimulate CLL cells by engaging TACI, BCMA and BAFF-R through BAFF and APRIL. A. Flow cytometric analysis of BAFF-R, TACI and BCMA (red lined open profile) proteins on CLL cells from UN (upper row) or MU (lower row) samples (one of 4 cases in each group; see Table 1). Gray solid profile, isotype-matched control. B. Time course QRT-PCR analysis of BAFF-R, TACI and BCMA mRNAs from CLL cells (3 cases) exposed to CD40L. RE, relative expression compared to freshly isolated CLL cells. C. Viable trypan blue-negative CLL cells (one of 3 cases) cultured in endothelial medium with or without UVECs or SMVECs and in the presence or absence of control Fc5 or TACI-Ig for 0 or 144 hours. D. QRT-PCR analysis of AID mRNA from CLL cells (3 cases) cultured for 4 days in endothelial medium with SMVECs and control Fc5 or TACI-Ig. RE, relative expression compared to freshly isolated CLL cells. E. ELISA of IgM and IgA secreted by CLL cells (3 cases) cultured for 7 days as in D. F. QRT-PCR of BAFF-R, TACI and BCMA mRNAs in CLL cells (3 cases) nucleofected with scrambled, BAFF-R-, TACI- or BCMA-targeting siRNAs. RE, relative expression compared to non-nucleofected CLL cells. G. Viable trypan blue-negative CLL cells (3 cases) nucleofected with scrambled, BAFF-R-, TACI- or BCMA-targeting siRNAs and exposed to SMVECs for 4 days. H. QRT-PCR analysis of AID mRNA from CLL cells (3 cases) treated and cultured as in G. RE, relative expression compared to non-nucleofected CLL cells. Data are from one of eight (A) or three (C) experiments yielding similar results or summarize three different experiments (B and D-H). Error bars, s.e.m.; * P < 0.05 (ANOVA followed by one-tailed unpaired Student’s t-test).

A soluble TACI-Fc5 decoy receptor, which prevents binding of BAFF and APRIL to B cells, decreased CLL cell survival, AID expression as well as IgM and IgA secretion as induced by SMVECs or UVECs (Fig. 4C-E). In these co-cultures, an Fc5 molecule was used as control (Fig. 4C-E). CLL cells transfected with siRNAs specifically suppressing the expression of either BAFF-R or TACI or BCMA receptor showed not only decreased survival, but also attenuated expression of AID as induced by SMVECs (Fig. 4F-H). These data indicate that MVECs elicit survival and activation signals in CLL cells through a mechanism that involves engagement of multiple BAFF and APRIL receptors, including BAFF-R, TACI and BCMA receptors.

MVECs express CD40 and enhance CD40L expression on CLL cells

Stromal cells enhance BAFF and APRIL release in response to various immune signals, including CD40L (15), an NF-κB-inducible T cell molecule aberrantly expressed by CLL cells (5, 23, 51). Given the ability of CLL cells to up-regulate endothelial release of BAFF and APRIL in a contact-dependent manner, we wondered whether this up-regulation involved engagement of CD40 on MVECs by CD40L on CLL cells. Fluorescence microscopy showed that factor VIII-positive MVECs from CLL or nonmalignant lymphoid tissues expressed the CD40 receptor (Fig. 5A). In addition, CD40L was detected in CLL cells expressing the B cell-specific membrane protein IgD and nuclear protein Pax5, but lacking the T cell-specific membrane protein CD3 (Fig. 5B). Some of these CD40L-expressing CLL cells also contained the proliferation-associated nuclear protein Ki-67 and seemingly occupied CLL proliferation centers (Fig. 5B). In nonmalignant lymphoid tissues, only some germinal center B cells expressed CD40L (Fig. 5C). These CD40L-expressing B cells had a germinal center phenotype as they expressed AID but lacked IgD (Fig. 5C).

FIGURE 5.

MVECs express CD40 and up-regulate CD40L on CLL cells. A. Upper panels: immunofluorescence analysis of CLL lymph node (LN; 5 cases) or normal tonsillar (TO; 5 cases) tissues stained for CD40 (green), factor VIII (red), and IgD (blue). Bottom panels: flow cytometric analysis of CD40 (red open profile) on UVECs, SMVECs and LMVECs. Gray solid profile, isotype-matched control. Original magnification, x10 (upper panels) and x40 (bottom panels). B. Upper panels: CLL bone marrow (BM; 3 cases) stained for IgD or Pax5 (green), CD40L (red or blue), CD3 (yellow), and Ki-67 (blue). Inset in large left panel shows merged IgD and CD40L. Original magnification, x10 (left panels) or x63 (right panels). Bottom panels: CD40L (red open profile) on CLL cells (3 cases). Gray solid profile, isotype-matched control. Numbers indicate specific CLL cases. C. Left panels: normal tonsillar (TO; 5 cases) tissue stained for IgD (green), AID (red) and CD40L (blue). Original magnification, x10 (upper panel) or x20 (lower panel). Right panels: CD40L (red open profile) on naïve, marginal zone or memory B cells from a healthy spleen (3 cases). Gray solid profile, isotype-matched control. D. QRT-PCR of CD40L mRNA in naïve, marginal zone or memory B cells from healthy spleens (3 cases), freshly isolated CLL cells (6 cases, each analyzed in triplicates), or CLL cells exposed to SMVECs (2 cases, each analyzed in triplicates) for 4 days. CLL cells were pre-incubated with control vehicle (DMSO) or an NF-κB inhibitor (IKK inhibitor III) for 3 hours, washed, and then co-cultured with SMVECs. Results are normalized to freshly isolated naïve B cells. Numbers indicate CLL cases. E. QRT-PCR of CD40 mRNA in UVECs, SMVECs or LMVECs incubated with endothelial medium alone (control) or CD40L. F. Flow cytometric analysis of CD40 on SMVECs cells incubated with control Ig or CD40-Ig in the presence of CD40L^high CLL cells (blue open profile; one of three cases) or absence of CD40L^high CLL cells (red open profile) for 48 hours. Gray solid profile, isotype-matched control. Data are from one of six experiments yielding similar results (A-C, E, F) or summarize three different experiments (D). Error bars, s.e.m.; * P < 0.05 (ANOVA followed by one-tailed unpaired Student’s t-test).

Flow cytometry confirmed the expression of CD40 on MVECs as well as the expression of CD40L on malignant B cells from some CLL cases but not on naïve, marginal zone and memory B cells from healthy donors (Fig. 5A-D and Table 1). CD40L expression on malignant B cells ranged from high to low among CLL cases and did not seem to preferentially target unmutated versus mutated cases or Zap-70-positive versus Zap70-negative cases (Table 1). QRT-PCR identified CD40L transcripts in malignant B cells from both CLL cases with high CD40L protein expression and low CD40L protein expression (Fig. 5D), indicating that transcription of the CD40L gene is not always associated with surface CD40L protein expression. The QRT-PCR assay was highly specific, because, unlike CLL cells, nonmalignant naïve, MZ and memory B cells expressed little or no CD40L transcripts. The expression of CD40L transcripts in CLL cells further increased upon exposure to SMVECs, but this increase did not occur after a short pre-incubation of CLL cells with a specific inhibitor of the NF-κB pathway (Fig. 5D). We next wondered whether CD40L on CLL cells exerted a positive feedback on CD40 expression by MVECs. Exposure of MVECs to recombinant CD40L induced more expression of CD40 transcripts in MVECs (Fig. 5E). Similarly, exposure of MVECs to CD40L-positive CLL cells up-regulated the expression of CD40 on MVECs, but this up-regulation did not occur in the presence of a soluble CD40-Ig decoy receptor that inhibits engagement of CD40 by CD40L (Fig. 5F). These data show that endothelial NF-κB-inducing factors up-regulate CD40L expression on CLL cells and show that leukemic CD40L establishes a positive feedback by enhancing CD40 expression on MVECs.

MVECs release BAFF and APRIL in response to CD40L

Considering that CD40L is a powerful inducer of BAFF and APRIL (15, 52), we wondered whether engagement of CD40 on MVECs by CD40L on CLL cells elicits endothelial release of BAFF and APRIL. QRT-PCR and ELISA assays demonstrated that SMVECs, UVECs and (not shown) LMVECs up-regulated the expression of BAFF and APRIL transcripts as well as the release of soluble BAFF and APRIL proteins as early as 3 hours after incubation with CD40L or CLL cells expressing CD40L (Fig. 6A-B). Soluble BAFF and APRIL derive from processing of pro-BAFF and pro-APRIL proteins by furin convertase (53-55). This trans-Golgi network protease cleaves the membrane-proximal stalk region of pro-BAFF between arginine-133 and alanine-134 and the membrane-proximal stalk region of pro-APRIL between arginine-104 and alanine-105 (Fig. 6C). Alternative processing of BAFF has also been reported (55). Accordingly, the furin cleavage site of pro-BAFF and pro-APRIL was found to be partially overlapping with an alanine134-valine-134 motif resembling the alanine-76-valine-77 motif utilized by the disintegrin metalloprotease TACE to cleave the pro-TNF protein (56). Thus, we hypothesized that CD40L may stimulate BAFF and APRIL cleavage in MVECs via both furin convertase and TACE proteins.

FIGURE 6.

MVECs enhance cleavase-dependent processing of BAFF and APRIL in response to CD40L. A. Left panel: time course QRT-PCR analysis of BAFF and APRIL mRNAs from SMVECs incubated with CD40L. Right panel: QRT-PCR analysis of BAFF mRNA from SMVECs exposed to CLL cells with low or high CD40L expression (3 cases in each group). RE, relative expression compared to unstimulated SMVECs. B. ELISA of BAFF and APRIL from UVECs or SMVECs incubated for 0, 3 or 6 hours with CD40L. C. Furin cleavage sites (blue arrows) and putative TACE cleavage sites (green arrows) of pro-BAFF and pro-APRIL proteins. The TACE cleavage site (red arrow) of the pro-TNF protein is shown for comparison. Numbers indicate amino acid residues in the membrane-proximal stalk region of BAFF, APRIL and TNF. D. Immunoblot of full-length pro-BAFF and pro-APRIL proteins from total lysates of UVECs and SMVECs incubated with medium alone (control) or CD40L for 6 hours. β-actin is a loading control. kDa, kilodaltons. E. ELISA of BAFF and APRIL in the supernatants from SMVECs incubated with or without CD40L and in the presence or absence of vehicle (DMSO), furin inhibitor (choromethylketone, CMK) or TACE inhibitor (TAPI-2). F. Immunofluorescence analysis of CLL lymph node (LN; 3 cases) stained for TACE or furin (green), Factor VIII (red), and IgD (blue). Original magnification, x10. G. QRT-PCR of TACE and furin in UVECs and SMVECs incubated for 1 or 5 hours with CD40L. Results are normalized to β-actin mRNA. RE, relative expression compared to unstimulated cells. H. Immunoblot of TACE and furin from SMVECs incubated with CD40L for various time points. Actin is a loading control. In mid panel, 96, 90 and 60 kDa bands correspond to precursor, mature and cleaved furin proteins. I. Viable trypan blue-excluding CLL cells (one of 3 cases) cultured with SMVECs, UVECs or MVECs in the presence or absence of control Ig or CD40-Ig for 0, 24 or 48 hours. J. Viable CLL cells (one of 3 cases) cultured with SMVECs, nucleofected with scrambled (scr) or CD40 siRNA for 0, 24 or 48 hours. Data are from one of three experiments yielding similar results (C, D, F and H-J) or summarize three different measurements (A, B, E and G). Error bars, s.e.m.; * P < 0.05 (ANOVA followed by one-tailed unpaired Student’s t-test).

Consistent with this possibility, immunoblotting showed decreased intracellular amounts of pro-BAFF and pro-APRIL proteins upon exposure of MVECs to CD40L (Fig. 6D). Furthermore, ELISA showed decreased CD40L-induced released of soluble BAFF and APRIL proteins from SMVECs exposed to either a peptidyl chloromethylchetone that specifically inhibits furin convertase or a hydroxamate-based inhibitor that specifically inactivates TACE (Fig. 6E). A combination of these inhibitors completely abrogated BAFF and APRIL release by CD40L-stimulated SMVECs (data not shown). In agreement with these findings, immunohistofluorescence analysis and confocal microscopy showed that MVECs from CLL tissues contained abundant TACE and furin proteins, which were predominantly localized in perinuclear compartments together with APRIL and (not shown) BAFF (Fig. 6F and Supplementary Fig. S3). Furthermore, QRT-PCRs and immunoblotting showed rapid up-regulation of the expression of furin and TACE transcripts as well as furin and TACE proteins in SMVECs or UVECs exposed to CD40L (Fig. 6G-H).

The importance of CD40L in the generation of active BAFF and APRIL by MVECs was further documented by experiments showing that the pro-survival activity of SMVECs or UVECs on CLL cells decreased when SMVECs or UVECs were either made deficient of CD40 through RNA interference (Fig. 6I and Supplementary Fig. S4A) or exposed to a CD40L inhibitor such as CD40-Ig (Fig. 6J). Accordingly, interruption of CD40 signaling in MVECs attenuated the release of BAFF and APRIL elicited by CD40L-expressing CLL cells (Supplementary Fig. S4B). These data indicate that MVECs enhance CLL cell survival by releasing more BAFF and APRIL in response to leukemic CD40L, possibly as a result of an increased intracellular processing of BAFF and APRIL pro-proteins through furin and TACE cleavases.

MVECs up-regulate CD40L on CLL cells via BAFF and APRIL

Given that MVECs up-regulate CD40L expression on CLL cells via NF-κB, we hypothesized that this up-regulation could involve MVEC production of NF-κB-inducing factors such as BAFF and APRIL. Consistent with this possibility, flow cytometry showed that blockade of BAFF and APRIL by TACI-Ig inhibited not only the up-regulation of class-switched IgG and IgA isotypes, but also the up-regulation of CD40L on CLL cells exposed to SMVECs (Fig. 7A). This inhibition was specific, because blockade of BAFF and APRIL by TACI-Ig did not decrease TACI, BCMA and BAFF-R expression on CLL cells exposed to SMVECs. In agreement with these findings, CLL cells up-regulated CD40L expression on CLL cells exposed to BAFF or APRIL (Fig. 7B). This inhibition was reversed by pre-incubating CLL cells with a specific NF-κB inhibitor (data not shown). The functionality of CD40L on CLL cells was further evaluated by verifying its ability to deliver autocrine survival signals independently of MVECs. Consistent with this possibility, a CD40-Ig decoy receptor attenuated the viability of CLL cells cultured in the absence of MVECs (Fig. 7C). These data indicate that leukemic CD40L requires endothelial BAFF and APRIL signals for its up-regulation by MVECs and promotes CLL cell expansion and differentiation not only by inducing paracrine release of BAFF and APRIL from MVECs, but also by stimulating CLL cells in an autocrine manner (Fig. 7D).

FIGURE 7.

MVECs up-regulate CD40L expression on CLL cells through BAFF and APRIL. A. Flow cytometric analysis of CD40L, IgG, IgA, BAFF-R, TACI and BCMA on CLL cells (one of 6 cases) incubated with endothelial medium alone (red open profile, also shown in center and right panels), SMVECs plus control Fc5 (blue open profile), or SMVECs plus TACI-Ig (blue open profile) for 7 days. Gray solid profile, isotype-matched control. B. Flow cytometric analysis of CD40L on CLL cells (one of 6 cases) incubated with endothelial medium alone (red open profile; also shown in center and right panels) or endothelial medium supplemented with BAFF or APRIL (blue open profile) for 7 days. Gray solid profile, isotype-matched control. C. Viable trypan blue-excluding CLL cells (one of 6 cases) cultured with control Ig or CD40-Ig for 0, 24 or 48 hours. Data in A-C are from one of six experiments yielding similar results. D. Model summarizing crosstalk between MVECs and CLL cells via CD40L, BAFF and APRIL.

Discussion

We found that stromal MVECs delivered survival, proliferation, Ig gene remodeling and differentiation signals to CLL cells by releasing BAFF and APRIL. Engagement of CD40 on MVECs by CD40L aberrantly expressed on CLL cells enhanced BAFF and APRIL release through induction of furin and TACE cleavases, which elicited processing of BAFF and APRIL pro-proteins into soluble molecules. In addition to stimulating CLL cells through TACI, BCMA and BAFF-R, endothelial BAFF and APRIL augmented leukemic expression of CD40L, suggesting that MVECs and malignant B cells establish a bidirectional crosstalk via a complex signaling network involving multiple TNF family members.

CLL is characterized by a dynamic balance between malignant B cells circulating in the blood and malignant B cells located in permissive niches of lymphoid organs (57, 58). Circulating CLL cells show resistance to apoptosis but little or no proliferation, whereas a significant fraction of tissue-based CLL cells show active proliferation, suggesting that the tissue microenvironment provides a distinct set of signals to the leukemic clone (10, 11). In this regard, the CLL stroma has been proposed to provide Ig-driven antigenic signals by facilitating the interaction of stereotypic leukemic Igs with self-antigens exposed on apoptotic cells (59). The CLL stroma also contains nurse-like cells, macrophages and follicular dendritic cells that provide accessory survival signals, including BAFF and APRIL (13, 14, 60, 61). Moreover, the CLL stroma contains T cells expressing CD40L (24), a B cell-stimulating factor structurally and functionally related to BAFF and APRIL (19, 20).

Our findings indicate that MVECs constitute a key component of the CLL stroma by establishing an intimate crosstalk with neoplastic B cells through BAFF, APRIL and CD40L. CLL tissues contained MVECs that were proximal to malignant B cells expressing molecular hallmarks of ongoing proliferation and Ig gene remodeling, including Ki-67 and AID, respectively. Consistent with these data, MVECs stimulated leukemic cell survival, activation and differentiation, including CD38 expression, AID expression, CSR and antibody production, through both contact-dependent and contact-independent mechanisms, including soluble BAFF and APRIL as well as membrane-bound CD40L molecules. That MVECs may play a pivotal role in the pathogenesis of CLL is in agreement with evidence showing that malignant B cells from aggressive CLL cases express CD49d, a receptor for vascular adhesion molecule-1 (VCAM-1 or CD106) expressed by endothelial cells (62). It is also in agreement with data showing interaction of leukemic CD38 with the endothelial adhesion molecule CD31 (39) and with additional studies demonstrating that CLL cells produce angiogenic factors such as VEGF and generate survival signals via VEGF receptors (25, 30).

In mice, BAFF from radiation-resistant stromal cells is mandatory for the survival of peripheral B cells (63), but the nature of BAFF-producing stromal cells remains unclear. We found that MVECs contained more BAFF transcripts than DCs and macrophages and released large amounts of BAFF and APRIL, both constitutively and after activation. Tonic release of BAFF and APRIL may involve CD63-expressing endosomes widely dispersed in the cytoplasm of MVECs, whereas inducible release of BAFF and APRIL may involve perinuclear Weibel-Palade bodies containing vW factor and factor VIII. These endothelial storage granules release their content in response to various signals, including CD40L (64). In agreement with this, engagement of CD40 on MVECs by CD40L aberrantly expressed on CLL cells augmented endothelial processing of pro-BAFF and pro-APRIL proteins into soluble molecules through a mechanism involving furin and TACE cleavases, which were constitutively stored in Weibel-Palade bodies. Of note, CD40L rapidly increased furin and TACE transcription and expression by MVECs. This process may be important for MVECs to rapidly undergo post-degranulation replenishment of furin and TACE proteins. The relative contribution of furin and TACE to BAFF and APRIL cleavage remains uncertain, but one possibility is that these proteases form alternative proteolytic pathways to maximize BAFF and APRIL release by MVECs (54, 55). Consistent with this interpretation, the furin cleavage site in the stalk region of pro-BAFF and pro-APRIL proteins was found to be adjacent to a putative TACE cleavage site.

In addition to augmenting CLL cell survival and diversification, MVECs increased CD40L expression on CLL cells. Indeed, unlike nonmalignant B cells, leukemic cells from several CLL cases aberrantly expressed surface CD40L and contained CD40L transcripts. MVECs up-regulated the expression of CD40L through a BAFF- and APRIL-dependent mechanism that required NF-κB, a transcription factor that plays a crucial pathogenetic role in CLL (5, 45, 46). Indeed, CLL cells exposed to MVECs down-regulated CD40L expression upon pre-incubation with an NF-κB inhibitor or in the presence of TACI-Ig, a decoy receptor that prevents binding of BAFF and APRIL to B cells. Conversely, soluble BAFF and APRIL up-regulated CD40L expression on CLL cells.

In agreement with previously published findings (5), leukemic CD40L delivered autocrine CD40-dependent survival signals to CLL cells. Moreover, leukemic CD40L increased CD40 expression in MVECs as well as TACI, BCMA and BAFF-R expression in neoplastic B cells, suggesting that CD40, TACI, BCMA and BAFF-R form a CLL-enhancing signaling network that bidirectionally links the microvascular stroma with the leukemic clone. Consistent with this possibility, published data show that CD40L and BAFF cooperatively enhance B cell survival (44, 65). Remarkably, TACI and BCMA contributed as much as BAFF-R to the survival and activation of CLL cells exposed to MVECs. While the important function of BAFF-R and TACI in nonmalignant and malignant B cells is well established (19, 20, 43-45, 47-49, 66-71), less is known about the function of BCMA. This receptor delivers survival signals to non-malignant plasma cells (72, 73), but growing evidence indicates its additional involvement in B cell and plasma cell tumors (44, 45, 47, 49, 74).

Blockade of endothelial BAFF and APRIL by TACI-Ig or blockade of leukemic CD40L by CD40-Ig reduced the survival, activation and diversification of CLL cells co-cultured with MVECs. The inhibitory effect of TACI-Ig on leukemic AID expression as induced by MVECs may have clinical relevance, because Ig and non-Ig gene lesions induced by dysregulated AID expression may play a role in the survival and clonal evolution of CLL cells (6, 33, 75). In conclusion, our findings show that stromal MVECs establish a bi-directional interplay with CLL cells through multiple functionally related members of the TNF family, including BAFF, APRIL and CD40L. Blocking these molecules with specific inhibitors may be beneficial for the treatment of CLL.

³Abbreviations used in this paper

AID

activation-induced cytidine deaminase

APRIL

a proliferation-inducing ligand

BAFF

B cell-activating factor of the TNF family

CLL

chronic lymphocytic leukemia

LMVEC

lymphatic microvascular endothelial cell

CD40L

CD40 ligand

SMVEC

splenic microvascular endothelial cell

UVEC

umbilical vein endothelial cell