Main

The interaction of tumor cells with their microenvironment is essential for their growth and survival in vivo, as well as their localization and ability to metastasize to other organs. Growth of B-cell lymphomas is primarily restricted to lymphoid organs such as lymph nodes, spleen and bone marrow; multiple myeloma cells expand in the bone marrow and cannot grow in other organs. Several cytokines or adhesion molecules which are involved in the cross-talk between stroma and lymphoma cells have been identified: IL-6 (refs. 1,2), IL-7 (ref. 3), VEGF4, CXCR-4 (ref. 5), RANTES5, MIP-1 (ref. 5), JAGGED-1 (ref. 6) and IGF-1 (ref. 7).

Recent studies suggest that stromally produced hedgehog (Hh) proteins—indian (Ihh), sonic (Shh) and desert (Dhh)—have a role in the proliferation of hematopoietic stem cells as well as lymphoid cells8,9,10,11,12. Shh is produced by follicular dendritic cells in the lymph node and is necessary for survival, proliferation and antibody production of germinal-center B cells9. Hh proteins are expressed in all lymphoid organs (Ihh in bone marrow8, Shh in lymph nodes and spleen9) and are either secreted as soluble ligands by stromal cells or mediate the interaction between two cell types through direct cell contact. Hedgehog proteins activate the transmembrane receptor Patched (PTC), which alleviates PTC-mediated suppression of SMO, a seven-transmembrane-helix protein. SMO activation triggers a cascade of intracellular events leading to transcription of target genes regulated by GLI1, GLI2 and GLI3 (glioma-associated oncoproteins 1–3), such as PTCH1, GLI1 and cyclin D1 (CCND1) (refs. 13–15). Constitutive activation of the hedgehog pathway by point mutations in PTCH1 or SMO have been associated with medulloblastomas16,17,18, basal cell carcinomas19 and rhabdomyosarcomas20, and autocrine or paracrine pathway activation has been described for other solid tumors21,22,23,24,25.

Results

Stroma-produced Hh promotes survival

To identify growth factors that are important for the survival of B-cell malignancies we used Eμ-Myc mice26, a well established model for mouse B-cell lymphomas. Eμ-Myc mice overexpress Myc in the B-cell lineage under the control of the immunoglobulin heavy chain (Igh) enhancer and develop malignant B-cell lymphomas within one year after birth26. We isolated lymphoma cells from bone marrow, lymph nodes and spleens of mice with either weight loss of more than 15% or visibly enlarged lymph nodes. These Myc-lymphoma cells were propagated on bone marrow or spleen stromal cells from diseased mice for 2 weeks and then transferred to stromal cells isolated from Cdkn2a−/− mice27 and maintained in vitro. A total of 32 Myc-positive primary lymphoma cell cultures were established and were characterized by flow cytometry (B220, CD19, CD138), BCL6 immunohistochemistry and H&E staining (Supplementary Fig. 1 online). We observed a broader than reported range of lymphoma types, including slowly growing CD138+ plasmacytomas in 40% of mice with weight loss as the leading symptom (Supplementary Fig. 1). This phenotype has recently been described in another Myc transgenic mouse strain, MycHis (ref. 28). All mice with visibly enlarged lymph nodes had a pre-B-cell lymphoma phenotype.

In vitro proliferation and survival of lymphoma and multiple myeloma cells was dependent on the presence of a stromal layer, as shown by induction of apoptosis within 24 h in cells that were removed from the stroma (Supplementary Fig. 2 online). Growth of lymphoma cells in the absence of a stromal layer could be sustained for at least 2–3 d by addition of conditioned media produced by stromal cells, indicating that soluble factors secreted by stromal cells contribute to the survival of lymphoma cells. To identify these factors, we grew lymphoma cells without stroma in the presence of different growth factors. No effect on lymphoma growth was seen after stimulation with IL-3, IL-11, GM-CSF, SCF or Wnt3a, and only very limited growth was induced by stimulation with IL-6 and IL-7. In contrast, stimulation with recombinant Shh or Ihh increased the number of viable lymphoma cells about threefold 2 d after removal of the stromal cells (Fig. 1a). Three other Myc-lymphomas representing different types of lymphoma or multiple myeloma (Myc-Ly3, Myc-Ly6, Myc-Ly9) showed similar effects upon stimulation with recombinant mouse Shh (Supplementary Fig. 3 online), and all cultures tested also showed prolonged survival upon treatment with the Smoothened (Smo) agonist purmorphamine29 (Supplementary Fig. 3). The proliferative effect of Shh and of stroma-conditioned media could be inhibited by abrogation of Hh signaling with cyclopamine, an alkaloid which binds to Smo and stabilizes its inactive conformation30 or by disruption of Shh binding to its receptor Ptch using the Hh-specific neutralizing antibody 5E1 (ref. 31) (Fig. 1b,c). Cyclopamine (5 μM) treatment of lymphoma cells (Myc-Ly6) cultivated on an intact stromal layer resulted in a complete clearance of lymphoma cells within 48 h, while growth of the stromal layer was not visibly affected (Fig. 1d). For quantification of the effect we developed a stroma (unluciferased)-lymphoma (luciferased) cocultivation assay in 96-well format with a luciferase readout to assess lymphoma growth. Luciferase counts were reduced in a dose-dependent manner in cultures treated with cyclopamine (IC50 0.5–2 μM) or the synthetic Smo antagonist SANT-1 (ref. 32) (IC50 1–4 μM), but were not reduced in cells treated with tomatidine, an alkaloid structurally similar to cyclopamine but devoid of Smo-binding activity33 (Fig. 1e). Twenty-three of 31 (74.1%) tested Myc-lymphoma cultures responded to Hh pathway inhibition with cyclopamine (>80% growth inhibition at 5 μM). B-cell lymphomas (B220+CD138−) and plasmablastomas (B220+CD138+) had a response rate of 80% (8/10) each and plasmacytomas (B220−CD138+) 66% (6/9) (Table 1). Stromal growth was not inhibited by either of these compounds up to a concentration of 10 μM, as demonstrated in an independent experiment using the mitochondrial potential dye Alamar Blue as a viability readout (Fig. 1f).

Figure 1: Hedgehog ligands provided by stromal cells from lymphoid organs are survival factors for lymphoma and multiple myeloma cells.
figure 1

(a) Alamar Blue assay after 48 h of cultivation of lymphoma cells in medium containing RPMI 1640, 10% FBS and various cytokines at concentrations of 0, 1, 5, 10 and 50 ng/ml shows increased number of viable cells in Shh- and Ihh-stimulated cells. (b) Alamar Blue viability assay showing inhibition of growth of lymphoma cells in the presence of 200 ng/ml Shh over 48 h when the Hh pathway is inhibited with 5E1 monoclonal antibody (10 μg/ml) to Hh or cyclopamine (5 μM). (c) Alamar Blue assay showing decrease in number of viable cells upon Hh pathway inhibition by cyclopamine in lymphoma cells cultivated in stromal supernatant. (d) Clearance of lymphoma cells from the stromal layer after 48 h of cyclopamine treatment (5 μM). Scale bars, 100 μm. (e) Treatment of luciferased lymphoma cells (Myc-Ly6, Myc-Ly1) growing on an unluciferased stromal layer with cyclopamine, SANT-1 and tomatidine (0, 0.5, 1, 5 and 10 μM). Luciferase readout with Bright-Glo luminescence reagent after 48 h. (f) Alamar Blue assay after 48 h treatment of stromal cells with cyclopamine, SANT-1 and tomatidine treatment. (g) Left, expression of Hh protein in bone marrow stroma of different mouse strains (antibody recognizes Ihh and Shh). Right, RT-PCR for Shh, Ihh and Dhh. (h) Viability, assessed by trypan blue exclusion, of multiple myeloma and non-Hodgkin's lymphoma samples grown on the stromal cell line M2-10B4 with different concentrations of cyclopamine for 72 h.

Full size image
Table 1 Myc-lymphomas: phenotype and cyclopamine response
Full size table

Hh protein expression could be detected in bone marrow stroma from three different mouse strains (BALB/c, C57BL/6, Cdkn2a−/−) using a polyclonal antibody that recognizes all three known family members (Fig. 1g). Further analysis by RT-PCR demonstrated exclusive expression of Ihh RNA in bone marrow stroma, whereas lymph node and spleen stroma from the same mice showed expression of Shh transcripts (Fig. 1g and Supplementary Fig. 4 online). Stroma from all three organs supported proliferation of lymphoma cells independently of whether the stroma was taken from diseased or healthy animals (Supplementary Fig. 4). The Cdkn2a−/− bone marrow stromal cells used for long-term lymphoma cultures also showed high expression of Ihh and were characterized as mesenchymal stem cells (Sca+CD45−). Reduction of Ihh expression in these stromal cells by expressing an Ihh short hairpin RNA construct with a lentiviral vector substantially decreased growth of lymphoma cells (Supplementary Fig. 5 online), indicating that Hh ligands may be important growth factors for lymphomas.

To address whether Hh signaling also has a role in human lymphoma and multiple myeloma, we isolated CD138-positive cells by magnetically activated cell sorting (MACS) separation from bone marrow of subjects with multiple myeloma and CD19-positive cells from lymph nodes of those with non-Hodgkin's lymphoma (NHL). Primary multiple-myeloma and NHL samples were cultivated on the stromal cell line M2-10B4 (ref. 34), which also expresses Ihh (Supplementary Fig. 4). Cell cultures were treated with cyclopamine for 72 h and viability was assessed by trypan blue exclusion in a 96-well plate. All five NHL and four multiple myeloma samples (Supplementary Table 1 online) responded to Hh pathway inhibition (Fig. 1h), implying an important role for hedgehog signaling in lymphoma and multiple myeloma in humans as well.

Hh pathway inhibition induces apoptosis in lymphoma cells

Cyclopamine treatment of Myc-lymphoma cells led to induction of apoptosis within 24 h, as measured by Annexin V staining. (Fig. 2a). Cell-cycle distribution of the remaining viable cells showed a reduction of the G2/M population from 20.4% to 13.3% after 24 h of treatment (Fig. 2b), similar to effects described for Hh pathway inhibition in T cells and in small cell lung cancer cell lines22,35.

Figure 2: Hh pathway inhibition induces apoptosis in stroma-dependent lymphoma cells.
figure 2

(a) Annexin V staining after treatment of lymphoma cells with 5 μM cyclopamine for 0 h, 24 h and 48 h. The percentage of Annexin V-positive cells as a percentage of the total population is indicated (horizontal bar) in each graph. (b) Cell cycle distribution of gated viable cells after treatment with cyclopamine for 0 h and 12 h (SubG1 phase excluded). (c) RT-PCR analysis of transcripts from Hh pathway members and Hh targets in splenic lymphocytes (lymph.) and various mouse lymphoma cells. (d) RT-PCR analysis of hedgehog pathway members and target genes in samples from subjects with multiple myeloma (MM) and NHL. (e) Immunohistochemical staining for Hh ligands Gli1 and Smo in spleen and bone marrow infiltrated with lymphoma cells. Scale bars, 10 μm. (f) Downregulation of Gli1 and Bcl2 protein relative to β-actin control in lymphoma cells after 24 h treatment with cyclopamine. (g) Ptch1 and Bmi1 transcript levels after 18 h of cyclopamine treatment (5 μM) assessed by quantitative RT-PCR.

Full size image

RNA extracted from mouse lymphomas and from human NHL and multiple myeloma samples directly after MACS separation showed transcripts of all essential Hh pathway members, including high levels of GLI1 and PTCH1, which represent direct target genes of the hedgehog pathway itself36 (Fig. 2c,d). Immunohistochemical staining of spleens of diseased mice showed expression of Gli1, Gli2 and Smo in infiltrating lymphoma cells, whereas dendritic cells and mesenchymal stem cells expressed hedgehog ligands (they were positive with an antibody recognizing Ihh and Shh) (Fig. 2e), confirming that the activation of the hedgehog pathway in lymphoma cells is dependent on its interaction with stromal cells.

Cyclopamine treatment of lymphoma cells resulted in a reduction of Gli1 protein and Ptch1 transcripts within 24 h (Fig. 2f,g). Bcl2, a known target of Hh signaling in T lymphocytes37, was also downregulated in lymphoma cells after cyclopamine treatment (Fig. 2f). Bmi1, a member of the Polycomb group of transcriptional repressors which was shown to be essential for lymphoma development in transgenic Myc mice38, was upregulated in all Myc-lymphomas compared with splenic B-cells (Fig. 2c). This confirms recently published results that Bmi1 is a direct transcriptional target of c-Myc39. Cyclopamine treatment of lymphoma cells did not significantly reduce Bmi1 transcript levels, indicating that Bmi1 regulation in lymphoma cells is not dependent on Hh signaling (Fig. 2g).

Ectopic Hh signaling induces lymphoma growth in skin

To further establish the role of the Hh pathway in B-cell malignancies, we overexpressed wild-type SMO and the constitutively active SMO mutant SMOW535L (ref. 16) in mouse lymphoma and multiple myeloma cells (Fig. 3a) using a retroviral pMSCV construct containing the gene for green fluorescent protein (GFP) under the control of an internal ribosome entry site (IRES) sequence. Overexpression of wild-type SMO and SMOW535L enhanced Gli1 and Ptch1 transcription (Fig. 3b), induced cell proliferation about twofold compared with expression of the empty vector control (Supplementary Fig. 6 online) and shifted IC50 values for cyclopamine treatment from 1 to 3 and 3.5 μM respectively. Purmorphamine, known to compete with cyclopamine for Smo binding (Supplementary Fig. 6), gave similar results.

Figure 3: Constitutive Hh pathway activation in Myc-lymphoma cells induces highly proliferative lymphomas in the skin.
figure 3

(a) Overexpression of Smo and SmoW535L in Myc-Ly6 cells detected by western blot analysis. (b) Gli1 and Ptch1 transcript levels in cells overexpressing wild-type Smo or SmoW535L. Expression is normalized to Gapdh and then compared with expression in Ly6 cells. (c) BALB/c nude mice 15 d after subcutaneous injection of 1 × 106 Myc-Ly6 cells overexpressing GFP alone, wild-type Smo plus GFP, or SmoW535L plus GFP. (d) Assessment of tumor growth by three-dimensional tumor measurements. n = 8 mice per group. (e) TaqMan PCR for Gli1 transcript levels in Ly6 tumors, Ly6 Smo tumors and Ly6 SmoW535L tumors. The average value for three tumors of each tumor type was normalized to Gli1 mRNA levels in Ly6 tumors.

Full size image

We injected 1 × 106 Ly6 GFP control cells, Ly6 SMO wild-type cells or Ly6 SMOW535L cells subcutaneously into BALB/c nude mice. Fifteen d after implantation we saw only limited growth in Ly6 GFP control cells, potentially owing to a lack of essential growth factors such as Shh in the subcutaneous environment. In contrast, cells overexpressing wild-type SMO or SMOW535L developed highly proliferative tumors in the skin (Fig. 3c,d) with elevated Gli1 transcript levels indicating active Hh signaling in these tumors in vivo (Fig. 3e). These results indicate that activation of the Hh pathway in lymphoma cells by either overexpression of wild-type SMO (enhancing the amount of unblocked SMO in the cell membrane) or constitutively active SMO induces lymphoma growth and enables lymphoma cell expansion even in an unnatural environment such as the skin.

Gli1, Fused and Bcl2 function as downstream effectors

To establish that the effect of cyclopamine is specific for hedgehog pathway inhibition, we overexpressed the downstream effectors FUSED and GLI1 in different lymphoma cells. FUSED overexpression led to a nearly complete and GLI1 overexpression to a partial inhibition of apoptosis induction by cyclopamine (Fig. 4a,b), with the IC50 in both cases shifted beyond 10 μM. These data indicate that the effect of cyclopamine on lymphoma cells is directly linked to hedgehog pathway inhibition.

Figure 4: Overexpression of hedgehog pathway members Fused and Gli1 or Bcl2, but not loss of the Cdkn2a-Arf locus, can inhibit cyclopamine-induced apoptosis in Myc-lymphomas.
figure 4

(a) Overexpression of Gli1 and Fused in Myc-Ly6 partially rescues lymphoma cells from cyclopamine (5 μM)-induced apoptosis, shown by images after 48 h treatment. Scale bar, 10 μm. (b) Cocultivation assay of luciferased Myc-Ly6 cells overexpressing different Hh pathway members and stromal cells. 48 h treatment with cyclopamine. Luciferase assay (Bright-Glo) in 96-well format as viability readout for lymphoma cells. RLU, relative light units. (c) Cyclopamine (5 μM) treatment of Tg(Myc/GFP)Trp53−/− lymphoma cells and Tg(Myc/GFP)Cdkn2a−/− lymphoma cells for 48 h induces apoptosis, whereas Tg(Myc/GFP)Tg(Bcl2) cells are resistant to cyclopamine treatment as shown by fluorescence imaging of GFP-positive cells. There is partial resistance of Tg(Myc/GFP)Bax−/− and Tg(Myc/GFP)Casp3−/− cells. Scale bars, 10 μm. (d) Luciferase assay after 48 h treatment of lymphoma-stroma cocultures with cyclopamine. (e) Stromal withdrawal for 96 h induces apoptosis in Tg(Myc/GFP)Trp53−/− cells and Tg(Myc/GFP)Cdkn2a−/− cells. Tg(Myc/GFP)Tg(Bcl2) cells grow independently of stroma, as shown by light microscopy. Scale bars, 100 μm.

Full size image

To further investigate the effect of Hh signaling on pathways regulating apoptosis and cell cycle, we generated lymphoma cultures from several different genetic backgrounds with modifications in apoptotic or senescence pathways. We extracted bone marrow from the following mouse strains: C57BL/6 wild-type, Tg(Bcl2) (transgenic for Bcl2; ref. 40), Bax−/− (ref. 41), Casp3−/− (lacking caspase 3; ref. 42), Cdkn2a−/− (ref. 28) and Trp53−/− (lacking the p53 tumor suppressor43). The bone marrow samples were infected with a pMSCV Myc-GFP retrovirus, creating Tg(Myc/GFP) lines. Transformed lymphoma cultures were obtained using Whitlock-Witte culture conditions. Bone marrow from all genetic backgrounds tested except C57BL/6 wild-type gave rise to transformed lymphoma cultures. We transferred these lymphomas to stroma from Cdkn2a−/− mice for long-term culture. Treatment of Tg(Myc/GFP)Trp53−/− and Tg(Myc/GFP)Cdkn2a−/− lymphoma cells with cyclopamine-induced apoptosis within 48 h (Fig. 4c,d). In marked contrast, no reduction of viable cells was detected in Tg(Myc/GFP)Tg(Bcl2) lymphoma cells. Tg(Myc/GFP)Bax−/− cells and Tg(Myc/GFP)Casp3−/− cells showed a partial resistance to cyclopamine-induced apoptosis (Fig. 4d). Stromal removal resulted in apoptosis induction in Tg(Myc/GFP)Trp53−/− lymphoma cells and Tg(Myc/GFP)Cdkn2a−/− lymphoma cells, whereas Tg(Myc/GFP)Tg(Bcl2) lymphoma cells were not affected (Fig. 4e). In summary, these results imply that the apoptotic cell death seen in lymphoma cells after stromal removal or after treatment with cyclopamine is mediated through the classical mitochondrial apoptosis pathway (Bcl2) and does not involve regulation of p53 and/or the Cdkn2a locus or Bmi1.

Hedgehog pathway inhibition inhibits lymphoma growth in mice

To verify the importance of the Hh pathway for lymphoma growth in vivo, we injected 1 × 106 lymphoma cells expressing luciferase into syngeneic C57BL/6 mice. Starting on day 2 after injection, mice were treated with either vehicle control or cyclopamine (100, 50 or 25 mg per kilogram body weight twice a day (b.i.d.) for a maximum of 21 d). Twelve d after injection, all mice in the control group showed high luminescence in the lymph nodes and spleens, whereas mice treated with cyclopamine 50 or 100 mg/kg b.i.d. had only minimal signs of disease (Fig. 5a). The median survival of mice in the control group was 16 d, whereas survival of cyclopamine treated mice was increased to 20 d in the lowest dosage group (25 mg/kg b.i.d.) and 30 to 35 d with cyclopamine doses of 50 and 100 mg/kg b.i.d. (Fig. 5b). Two additional Myc-lymphoma cell lines (one plasmablastoma and one B-cell lymphoma) were responsive in vivo under similar conditions, but one lymphoma with endogenous insensitivity to cyclopamine treatment in vitro and one lymphoma overexpressing Fused showed only minimal response to cyclopamine treatment at 50 mg/kg b.i.d., as shown by survival curves (Fig. 5c, Supplementary Fig. 7 online).

Figure 5: Hedgehog pathway inhibition abrogates lymphoma expansion in vivo.
figure 5

(a) Injection of 1 × 106 luciferased Myc-Ly9 cells into C57BL/6 mice and start of treatment with cyclopamine (25, 50 and 100 mg/kg b.i.d.) or vehicle control day 2 after injection. Bioluminescence imaging of mice 12 d after injection. (b) Survival curves for mice injected with 1 × 106 luciferased Myc-Ly9 cells that were treated with vehicle or cyclopamine (25, 50 and 100 mg/kg b.i.d.) for a maximum of 21 d. (c) Survival curves for mice injected with 1 × 106 Myc-Ly7 (in vitro resistant) or Myc-Ly6 (in vitro–responsive) lymphoma cells, treated with vehicle control or cyclopamine for a maximum of 45 d. (d) Coinjection of stroma and 1 × 106 luciferased lymphoma cells, Tg(Myc/GFP)Trp53−/− or Tg(Myc/GFP)Tg(Bcl2), intraperitoneally into BALB/c nude mice. Bioluminescence imaging after treatment with cyclopamine 50 mg/kg b.i.d. for 7 d. (e) Injection of 1 × 106 luciferased Myc-Ly9 cells in C57BL/6 mice. Bioluminescence imaging after 3 d treatment with cyclopamine 50 mg/kg b.i.d. in mice with already fully developed lymphomas. (f) Spleen weight and liver weight comparison after a 3-d treatment with cyclopamine. (g) H&E staining of spleens isolated from vehicle- and cyclopamine-treated mice. Ki67 staining indicates highly proliferative cells. Scale bars, 100 μm.

Full size image

To validate the role of Bcl2 as a downstream target of the hedgehog signaling pathway in vivo, we injected 1 × 106 luciferased Tg(Myc/GFP)Tg(Bcl2) lymphoma cells and 1 × 106 Tg(Myc/GFP)Trp53−/− lymphoma cells mixed with unluciferased stromal cells intraperitoneally into Nod/Scid (nonobese diabetic, severe combined immunodeficient) nude mice. Cyclopamine treatment significantly reduced expansion of Tg(Myc/GFP)Trp53−/− lymphoma cells, whereas it had no effect on expansion of Tg(Myc/GFP)Tg(Bcl2) lymphoma cells, indicating that Bcl2 is an important downstream target of the Hh pathway in vitro and in vivo (Fig. 5d).

To determine the importance of Hh signaling in established lymphomas, cyclopamine treatment was started 10 d after cell injection, a time when lymphomas were detectable in all mice by bioluminescence imaging. Cyclopamine treatment with 50 mg/kg b.i.d. for 3 d significantly reduced light emission, markedly decreased the infiltration of the spleen and reduced lymphoma mass in lymph nodes and other organs compared with results in the control group (Fig. 5e). Spleen weight after a 3-day cyclopamine treatment was decreased by nearly 50% compared with that in the vehicle control group and liver weight in the treated mice was reduced to an average weight of around 1,000 mg (Fig. 5f). Histopathological analysis of spleens isolated from vehicle- and cyclopamine-treated mice showed a reduction of lymphoma cells in the spleen after 3 d of treatment with cyclopamine (Fig. 5g). Kinesin 67 (Ki67) staining demonstrated a strong decrease of highly proliferative cells in the spleen compared with numbers in the vehicle group (Fig. 5f).

Discussion

Data presented in this work provide new insight in the interaction between lymphoma or multiple myeloma cells and their microenvironment mediated by hedgehog ligands in vitro and in vivo. Despite accumulation of a variety of genetic lesions, the majority of human NHL and multiple myeloma cells remain dependent on their microenvironment during the course of disease, and therefore long-term cultivation of early stage NHL and multiple myeloma cells without stromal support is rarely possible in vitro. Lymphoma cells are mainly restricted to lymphoid organs, whereas multiple myeloma cells are restricted to the bone marrow and only lose their dependency on this specific microenvironment in the latest stages of disease. Although several solid tumors depend on hedgehog pathway activation either by point mutations in Smo16,17,18, Ptch19 or Sufu20 or by autocrine activation of the hedgehog pathway21,22,23,24, we show that lymphoma and plasmacytoma cells are dependent on hedgehog ligands secreted by stromal cells in lymphoid organs. Overexpression of activated SmoW535L in lymphoma cells induced highly proliferative lymphomas in the skin, whereas wild-type lymphoma cells showed only very limited growth in this unnatural environment. These experiments indicate that Hh pathway activation is critical for lymphoma expansion in vivo and that hedgehog ligand secretion might be one of the limiting factors that determine the localization of lymphoma and multiple myeloma growth.

The molecular mechanisms of hedgehog signaling in lymphoma cells differ from those in other tumors, but show a potential role of hedgehog signaling in the transformation process of Myc-induced lymphomagenesis. Reciprocal chromosomal translocations of Myc, or c-Myc overexpression by other mechanisms, are widely acknowledged as initiating events in human Burkitt's lymphoma44,45 and other lymphoma types, including multiple myeloma46. c-Myc overexpression not only induces proliferation, but also has a proapoptotic effect by upregulating p53 and neutralizing Bcl2 by inducing activity of Bim (a member of the Bcl2 superfamily). Either loss of p53 or overexpression of Bcl2 can induce lymphomagenesis and rescue Myc-overexpressing cells from induction of apoptosis in vivo49. In the experiments shown here, Hh pathway inhibition downregulates Bcl2 expression in stroma-dependant lymphoma cells and induces apoptosis in Trp53−/− and Cdkn2a−/− lymphomas, whereas overexpression of Bcl2 rescues lymphoma cells from cyclopamine-induced apoptosis, suggesting Bcl2 as a crucial target of Hh signaling in lymphomas. Therefore, stromally expressed Hh ligands can provide the necessary environment for the survival of c-Myc–overexpressing cells by counteracting the proapoptotic effect of c-Myc through upregulation of the key antiapoptotic factor Bcl2, and can rescue Myc-pretransformed cells from apoptosis. The continued requirement for Hh activation in Tg(Myc/GFP)Trp53−/− and Tg(Myc/GFP)Cdkn2a−/− cells indicates that targeting the Hh pathway might be beneficial even in lymphomas harboring mutations in the genes encoding p53, p16INK4a or p14ARF.

In summary, we demonstrate that Ihh and Shh are ligands produced by stromal cells in the bone marrow, the spleen and the lymph nodes that allow survival and expansion of lymphoma cells in vitro and in vivo. These findings implicate the Hh signaling pathway as a crucial link in B-cell malignancies, and pharmacological inhibition of Hh signaling could represent a new treatment possibility for NHL and multiple myeloma.

Methods

Genetically altered mice and cultivation of primary cells.

We maintained and genotyped Eμ-Myc mice26, Bax−/− mice43, Trp53−/− mice45, Tg(Bcl2) mice42, C57BL/6 mice (all Jackson Laboratory), Cdkn2a−/− mice27 (US National Cancer Institute mouse repository) and Casp3−/− mice44 as previously described. Eμ-Myc mice were monitored for signs of disease, including development of visible lymphomas or weight loss of more than 15%. Mice that developed disease were killed, bone marrow, spleen and lymph nodes were extracted and lymphoma cells were propagated under Whitlock-Witte culture conditions with addition of IL-3 to the media for one week and stepwise withdrawal of IL-3 thereafter. After 2–3 weeks we transferred lymphoma cells to bone marrow stroma from Cdkn2a−/− mice for maintained growth. In order to generate lymphomas with defined genetic background, we extracted bone marrow from Trp53−/− mice, Casp3−/− mice, Bax−/− mice, Cdkn2a−/− mice and Tg(Bcl2) mice and infected cells with a pMSCV Myc-IRES-GFP retroviral construct. Propagation of transformed lymphocytes was performed under Whitlock-Witte culture conditions and lymphomas were maintained on stroma from Cdkn2a−/− mice.

Cell culture experiments with mouse lymphomas.

Lymphoma cells were infected with a retrovirus containing a pMSCV IRES puro-luc sequence and selected for 7 d with puromycin on puromycin-resistant Cdkn2a−/− stroma. We obtained cyclopamine from Toronto Research Chemicals and SANT-1 from EMDBioscience. Both were dissolved as × 1,000 stocks in DMSO. SCF was obtained from RDI; all other cytokines were obtained from R&D Systems. 5E1 monoclonal antibody to the N-terminal domain of Sonic hedgehog (ShhN) was from the Developmental Studies Hybridoma Bank and was used at a concentration of 10 μg/ml. For cytokine stimulation and Shh inhibition in supernatants, we seeded lymphoma cells into 96-well plates at a density of 20,000 cells per well and measured the mitochondrial activity of viable cells with an Alamar Blue assay as described by the manufacturer (Promega).

For Shh pathway inhibition on stroma, we seeded 6,200 stromal cells per well in a 96-well plate. After 24 h, we seeded 20,000 luciferased lymphoma cells per well, and 2 h later we added compound or DMSO. Luciferase assays were performed using Bright-Glo reagent (Promega) as described by the manufacturer.

To generate cells overexpressing Hh pathway members, lymphoma cells were infected with a retroviral pMSCV IRES-GFP construct carrying SMO, SMOW535L, GLI1 or FUSED or empty vector and then sorted for GFP-positive cells.

Primary human NHL and multiple myeloma samples.

We isolated NHL cells from lymph nodes from untreated human subjects with NHL via MACS separation using magnetic beads coated with CD19 protein. We isolated multiple myeloma samples from the bone marrow of subjects with multiple myeloma using CD138 magnetic beads. We added 20,000 NHL or multiple myeloma cells to 4,000 cells of the mouse stromal cell line (M2-10B4) in a 96-well plate and treated the cells with different concentrations of cyclopamine. The number of viable NHL or multiple myeloma cells was evaluated by trypan blue exclusion.

Immunohistochemistry.

We performed single-color diaminobenzidine immunoperoxidase staining on paraffin sections using antibodies to Ki67 (NeoMarkers), Gli1 (N-16, Santa Cruz Biotechnology), Smo (H-300, Santa Cruz Biotechnology) and Hh (H-160, Santa Cruz Biotechnology) according to the manufacturers' recommendations.

Western immunoblot.

We sonicated cell lysates in 2% SDS, 50 mM Tris HCl, pH 8. Western blots using rabbit or goat polyclonal antibodies to Shh-N, c-Myc, Smo or Gli1 were performed as recommended by the manufacturer. ShhN (H-160), c-Myc (9E10) and Smo (H-300) antibodies were obtained from Santa Cruz Biotechnology; Gli1 (ab7523) antibody was obtained from Abcam.

RT-PCR and quantitative PCRs.

Total cellular RNA was treated with DNase, reverse transcribed and amplified for 33 cycles at an annealing temperature of 55 °C. PCR primers were obtained from IDT. Quantitative PCR was assessed by TaqMan PCR (summary of primers and probes in Supplementary Methods online).

Mouse experiments.

Systemic disease. C57BL/6 mice 6–8 weeks of age were injected intravenously with 1 × 106 luciferased lymphoma cells. Treatment with 25 mg/kg b.i.d., 50 mg/kg b.i.d. or 100 mg/kg b.i.d. cyclopamine or vehicle control started on day 2 after injection or when lymphomas were already developed. Cyclopamine doses of 100 mg/kg b.i.d. were administered stepwise over a period of 20 min to avoid seizures. We measured bioluminescence using a Xenogen imaging system 10 min after injection of Renilla luciferase substrate.

Subcutaneous tumors. We injected 1 × 106 lymphoma cells subcutaneously in the right flank of BALB/c nude mice. Tumor size was measured by three-dimensional tumor measurement twice a week and a photograph was taken day 15 after injection.

Intraperitoneal lymphoma model. We injected 1 × 106 luciferased lymphoma cells together with 1 × 106 unluciferased stromal cells intraperitoneally in BALB/c nude mice (6–8 weeks of age). Cyclopamine treatment was started 7 d after injection for 7 d.

All animal experiments were approved by the IACUC committee at the Genomics Institute of the Novartis Research Foundation and were in accordance with the US National Institutes of Health Statement of Compliance with Standards for Humane Care and Use of Laboratory Animals. Patient samples were used with approval of the ethics committee of the University of Freiburg and under informed consent of the patients.

Note: Supplementary information is available on the Nature Medicine website.