Potent in vitro and in vivo activity of sorafenib in multiple myeloma: induction of cell death, CD138-downregulation and inhibition of migration through actin depolymerization


Correspondence: Professor Dr Monika Engelhardt, MD, Department of Haematology and Oncology, Freiburg University Medical Centre, Hugstetter Str. 55, Freiburg 79106, Germany

E-mail: monika.engelhardt@uniklinik-freiburg.de


Despite considerable advances, multiple myeloma (MM) remains incurable and the development of novel therapies targeting the interplay between plasma cells (PCs) and their bone marrow (BM) microenvironment remains essential. We investigated the effect of various agents in vitro on the proliferation, phenotype, morphology, actin polymerization and migration of MM cells and, in vivo, the tumour growth of L363-bearing non-obese diabetic severe combined immunodeficient mice with a deficient interleukin-2 receptor gamma chain (NSG). In vitro, we observed a dose-dependent cytotoxicity with bortezomib and sorafenib. Using RPMI8226 cells co-expressing histone 2B-mCherry and cytochrome c-GFP, bortezomib- and sorafenib-induced apoptosis was confirmed, and both agents combined showed synergism. Sorafenib induced CD138-downregulation and abolished CXCL12-induced actin polymerization. L363 cells expressed CCR4 and CCR5 and migrated to their common ligand CCL5. Chemotaxis to BM stroma cells was notable and significantly reduced by sorafenib. Downregulation of phospho-ERK appeared relevant for the inhibition of actin polymerization and chemotaxis. Sorafenib alone, and combined with bortezomib, showed substantial antitumour activity in L363-bearing NSG. Correspondingly, sorafenib induced clinical responses in MM-/AL-amyloidosis patients. We conclude that, in addition to the cytotoxic and anti-angiogenic effects of sorafenib, blocking of MM cell migration and homing represent promising mechanisms to interrupt the interplay between PCs and their supportive microenvironment.

Despite considerable advances, multiple myeloma (MM) remains an incurable disease. The interaction between plasma cells and the bone marrow (BM) microenvironment plays a pivotal role in MM pathogenesis (Hideshima & Anderson, 2011). Targeting this interplay represents an essential therapeutic strategy to overcome drug resistance. Currently, various agents are under investigation. In addition to next generation proteasome inhibitors and immunomodulatory drugs, interest in the use of tyrosine kinase inhibitors has also increased.

In the present study, we focussed on the effects of sorafenib in comparison to and in combination with other antimyeloma agents. Sorafenib is an oral multikinase inhibitor that targets several cancer-specific pathways and directly affects cell proliferation, cell survival and neovascularization (Wilhelm et al, 2006; Ramakrishnan et al, 2010). Sorafenib inhibits phosphorylation of cell surface and intracellular kinases (Fan et al, 2012). The Ras/Raf/MEK/ERK pathway is critical for the proliferation of MM cells, and as ERK represents a cytoskeletal signalling-related protein (Ngo et al, 2009), its blockage may also compromise MM cell adhesion and migration. Sorafenib was shown to induce dose-dependent growth inhibition of MM cells and synergism when combined with bortezomib, dexamethasone and rapamycin (Ramakrishnan et al, 2010). It reduced MM cell proliferation, even in the presence of BM stroma cells (BMSCs), overcoming the protective effect of the BM microenvironment (Ramakrishnan et al, 2010). Moreover, sorafenib showed potent anti-vascular endothelial growth factor receptor-2 (VEGFR2) activity and induced a significant decrease in the secretion of vascular endothelial growth factor (VEGF) and interleukin 6 (IL-6), thereby exerting anti-angiogenic effects. We have demonstrated an increase of VEGFR2+ cells in BM samples of MM patients as compared to those with monoclonal gammopathy of undetermined significance or healthy donors. In addition, VEGFR2+ cells were particularly elevated in patients with symptomatic MM (Udi et al, 2011). Besides its presence on mature endothelial cells, VEGFR2 appears to be expressed on cancer stem cells (Beck et al, 2011; Benitah, 2011). For various tumours, it has been postulated that cancer stem cells both secrete VEGF and express VEGFR, which enables stimulation of cell proliferation through an autocrine mechanism. Therefore, blocking VEGFR2 in MM may prevent angiogenesis, interrupt this autocrine loop and inhibit tumour growth (Beck et al, 2011; Benitah, 2011).

Although the survival of MM cells depends on the selective homing to the BM, and chemokines and the corresponding receptors play a pivotal role therein (Moller et al, 2003), the extent to which sorafenib affects the interplay between plasma cells and their microenvironment is not fully understood. In a systematic approach, we here examined the in vitro activity of sorafenib and other antimyeloma agents on cell death, phenotype, morphology, the Ras/Raf/MEK/ERK signalling pathway, actin polymerization and chemotaxis. We confirmed the antimyeloma activity of sorafenib in vivo using L363-bearing non-obese diabetic severe combined immunodeficient mice with a deficient interleukin-2 receptor gamma chain (NSG) and in a pilot study in MM-/AL-amyloidosis patients.

Materials and methods

Multiple myeloma cell lines (MMCLs), culture conditions and antimyeloma agents

L363, U266 and RPMI8226 cells (Oncotest, Freiburg) were cultured in 24-well plates at 37°C in a humidified incubator containing 5% CO2, at a concentration of 1 × 105 cells/ml, with RPMI 1640 medium, 10% fetal calf serum (FCS) and 0·2% penicillin/streptomycin (Zlei et al, 2007). Sorafenib was provided via material transfer agreement by Bayer HealthCare (Leverkusen, Germany) and lenalidomide from Celgene (Summit, NJ, USA). Epigallocatechin gallate (EGCG) was provided by Sigma-Aldrich (München, Germany).

Cytotoxicity analysis

MM cells were cultured with antimyeloma agents for 3 and 6 days as described (Zlei et al, 2007). For the single agent use, bortezomib and sorafenib were logarithmically increased (1, 10, 100 nmol/l and 1, 10, 100 μmol/l, respectively). In the combination studies, bortezomib was used at 2, 5 and 10 nmol/l, and sorafenib at 1, 5 and 10 μmol/l. Thalidomide was applied at 1, 10 and 100 μg/ml and lenalidomide at 1, 10 and 100 μmol/l. EGCG was used at 100–500 μmol/l. Cytotoxicity was assessed with propidium iodide (PI) staining and by flow cytometry on a FACSCalibur (Becton Dickinson, BD, Franklin Lakes, NJ, USA) before (day 0) and on day 3 and 6 after culture (Hideshima et al, 2001a). Viable cells were quantified by trypan blue dye exclusion. In order to determine whether the cytotoxic effect of sorafenib and bortezomib was caused by induction of apoptosis, we generated RPMI8226 cells that expressed a red chromatin marker (histone 2B fused to monomeric [m] Cherry; H2B-mCherry) and green fluorescent protein-tagged cytochrome c-GFP (Cyt. c-GFP) (Goldstein et al, 2000, 2005). For the detection of chromatin condensation and fragmentation and cytochrome c release (Bossy-Wetzel et al, 1998; Renz et al, 2001), we performed confocal laser microscopy. Images were taken with the objectives PL APO 20x NA 0·7 and HCX PL APO lbd.BL 63x NA 1·2 W.

Immunocytochemical staining

To evaluate morphological and phenotypical changes, MM cells were grown with 1, 10 and 100 μmol/l sorafenib for 3 days. Cells were fixed with 4% paraformaldehyde for 20 min at 4°C and permeabilized with 0·2% Triton X-100 at room temperature for 5 min. Staining was performed with an anti-CD138-fluorescein isothiocyanate (FITC)-labelled monoclonal antibody, Phalloidin-Alexa594 (F-actin) and 4',6-diamidino-2-phenylindole (DAPI; DNA), and cells were analysed with a Leica TCS SP2 AOBS confocal microscope.

Chemotaxis assay

Chemotaxis was performed as described (Burger et al, 2005). With or without sorafenib and cell growth for 3 days, 1 × 106 cells were placed in the upper chambers of 96-well chemotaxis plates (5 μm pore size) and allowed to migrate toward various chemoattractants in the lower chambers. After 3 h of incubation at 37°C and 5% CO2, the number of cells that migrated into the lower chambers was assessed by flow cytometry.

Actin polymerization assay

Actin polymerization was investigated as described (Burger et al, 2005). Cells (1·5 × 106/ml) were suspended in RPMI1640 with 0·5% bovine serum albumin at 37°C and stimulated with 100 ng/ml CXCL12. After 15, 60, 120 and 300 s of stimulation, 400 μl of cell suspension was added to 100 μl of a solution containing 50 μg/ml FITC-labelled phalloidin, 50 mg/ml α-lysophosphatidylcholine (both Sigma-Aldrich) and 37% formaldehyde in phosphate-buffered saline. Intracellular F-actin was measured by flow cytometry as mean fluorescence intensity and represented as percentages of F-actin relative to baseline.

Western blot analysis

After 3, 6 and 24 h of treatment with 1 and 10 μmol/l sorafenib, total protein was isolated using immunoprecipitation buffer. Equal amounts of protein were loaded and separated by 12% sodium dodecyl sulphate-polyacrylamide gel electrophoresis and transferred to polyvinylidene difluoride membranes. Western blot analysis was performed using antibodies (Abs) recognizing the phosphorylated and unphosphorylated forms of p44/42 mitogen-activated protein kinase (MAPK; Cell Signaling Technology, Beverly, MA, USA) after stripping the membrane with Re-Blot Plus Strong Solution (Millipore, Billerica, MA, USA). Immunoreactive bands were visualized using a horseradish peroxidase-linked secondary antibody and the enhanced chemiluminescence (ECL) system (ECL Plus Western Blotting Detection System; GE Healthcare, Pittsburgh, PA, USA). The intensity of phosphorylated ERK1/2 was normalized and quantified to the corresponding ERK1/2 and β-actin signal.

Treatment of L363-bearing mice with antimyeloma agents

In a first experiment, 24 NSG mice received intratibial (i.t.) injections of 5 × 105 L363 cells: six mice were analysed on day 0 to determine MM cell engraftment before therapy. Seven days after injection, 18 mice were stratified into three treatment groups of six animals each: control mice (Group A1) received aqua at 10 ml/kg/d intraperitoneally on days 7–11 and 14–18, Group B1 received 0·7 mg/kg/d bortezomib (iv) on days 7, 11 and 18 and Group C1 received sorafenib orally at 100 mg/kg/d on days 7–18. Tumour cell engraftment was determined by flow cytometry. In a second experiment, mice received i.t.-injected L363 cells and were stratified into four groups of six mice per group receiving aqua (Group A2), 75 mg/kg/d sorafenib as scheduled previously (Group B2), 0·7 mg/kg/d bortezomib on days 7, 11, 14 and 18 (Group C2) or sorafenib plus bortezomib (Group D2). Tumour growth was monitored by fluorescence-based-in-vivo-imaging (IVI) using an hCD138-Ab coupled with Alexa750. Cell engraftment was assessed in the BM and distant organs (spleen and liver). Myeloma growth inhibition was calculated as the reduction of tumour volumes compared to untreated mice. All experimental procedures were approved by the Institutional Ethical committee (University of Freiburg) and conducted according to protocols approved by the National Directorate of Veterinary Services (Germany).

Sorafenib treatment in MM patients

Sorafenib was applied in a pilot run-in-study of three symptomatic MM patients. Patient characteristics are summarized in Table 1. One patient had concomitant AL-amyloidosis with periorbital, cutaneous, gastrointestinal and cardiac amyloid involvement. Fluorescence in situ hybridization (FISH) revealed deletion 13q14 in Patient 1 and a normal karyotype in the others. Sorafenib doses were 400 mg/day. Remission was determined according to the International Myeloma Working Group (IMWG) response criteria (Kyle & Rajkumar, 2009). The analysis was carried out according to the guidelines of the Declaration of Helsinki and good clinical practice. All patients gave their written informed consent for institutional-initiated research studies and analyses of clinical outcome studies conforming to our institutional review board guidelines.

Table 1. Patient characteristics, therapy response and side effects under sorafenib treatment
 Patient 1Patient 2Patient 3
  1. MM, multiple myeloma; ISS, International Staging System; FISH, Fluorescence in situ hybridization; TP, total protein; SFLC, serum free light chains; FUO, fever of unknown origin; Max, maximal; IgG/IgA, Immunoglobulin G or A; NK, normal karyotype (no evidence of del13q14, del17p13, t(4;14), t(14;16), t(11;14) or hyper-/hypodiploidy; SD, stable disease; CTCAE, Common Terminology Criteria for Adverse Events v4·0 (http://www.eortc.be/services/doc/ctc/CTCAE_4.03_2010-06-14_QuickReference_5x7.pdf).

Patient age (years)746664
Gender (male vs. female)mmf
MM typeIgA-KappaIgG-Kappa+AL-amyloidosisIgG-Kappa
Disease stage
Durie & SalmonIII AIII AII A
Cytogenetics (by FISH)del13q14NKNK
Pre-sorafenib therapy (cycles/lines)10/528/70/0
Fold change TP decline−1·05−1·07−1·07
Fold change IgG/IgA decline1·10−1·08−1·09
Fold change SFLC decline−1·37−1·08−1·19
Fold change Hb increase1·141·061·06
Side effectsrecurrent FUO (CTCAE grade 2)leg cramps (CTCAE grade 1)xerositis, hand foot skin reaction, leg cramps (CTCAE grade 1)
Best response under sorafenib treatmentSDSDSD
Sorafenib treatment duration (months)2368

Statistical analysis

Data regarding cytotoxicity and cell viability between treated and untreated cells were compared using the Wilcoxon signed rank test. A P value of <0·05 was considered as statistically significant. The effect of combining bortezomib and sorafenib was investigated using the median effect method of Chou and Talalay and analysed with Calcusyn software (Chou, 2010). A combination index (CI) of 1 indicated an additive effect, below 1 synergism and above 1 antagonism (Chou, 2010). We calculated the Spearman correlation coefficient (r) to describe the association between sorafenib dose and the number of migrated cells.


Different antimyeloma agents induce cell death

In a systematic approach, the cytotoxic effect of sorafenib compared to reference agents like bortezomib, thalidomide, lenalidomide and EGCG was investigated. For the cytotoxicity assays, we assessed all 3 MMCLs in analogy to previous studies from our group that showed distinct growth kinetics (Zlei et al, 2007). Treatment of L363 cells with 10 and 100 μmol/l sorafenib induced a significant increase in PI+ cells, both on day 3 and 6 (n = 11; P = 0·001 for all comparisons; Fig 1A). As expected, viable cells significantly decreased with 10 and 100 μmol/l compared to untreated cells on day 3 and 6 (n = 11; P = 0·001 for all comparisons; Fig 1B). Similarly, in U266 and RPMI8226, PI+ cells significantly increased with 10 and 100 μmol/l sorafenib. Even 1 μmol/l of sorafenib induced a significant reduction of viable cells in both cell lines on day 3 and 6.

Figure 1.

Effect of sorafenib on cell death in L363 cells. Significantly increased propidium iodide (PI)-positive cells were observed with 10 and 100 μmol/l sorafenib, on both days 3 (d3) and 6 (d6), as compared to the control (A). Viable cells showed a significant decrease with 10 and 100 μmol/l on d3 and d6 (B). Values represent mean ± standard deviation (SD) from 11 experiments. *P < 0·05. Effect of bortezomib on cell death in L363 cells. Significantly increased PI-positive cells were observed with 100 nmol/l bortezomib, on d3 and d6, as compared to the control (C). Cell viability significantly decreased with 100 nmol/l, on both d3 and d6 (D). Values represent mean ± SD from 6 experiments. *P < 0·05. Effect of thalidomide on cell death in L363 cells. No significant increase in PI-positive cells was observed after 3 and 6 days of culture with thalidomide (E). Viable cell numbers only insignificantly decreased with 10 and 100 μg/ml (F). Values represent mean ± SD of 9 experiments. Effect of lenalidomide on cell death in U266 cells. Significantly increased PI-positive cells were observed on d3 with 100 μmol/l lenalidomide and on d6 with 10 and 100 μmol/l, as compared to the control (G). Significant decreases in cell viability were observed with 10 and 100 μmol/l on d6, but not on d3 (H). Similar results regarding PI-positivity were observed in L363 cells, with 10 and 100 μmol/l, on d6 (data not shown). Values represent mean ± SD of 9 experiments. *P < 0·05. Effect of EpiGalloCatechin-3-Gallate (EGCG) on cell death in L363 cells. Significantly increased PI-positive cells were seen with 250 μmol/l and 500 μmol/l of EGCG on d3 (I). Significant decreases in viable cells were observed with 250 and 500 μmol/l EGCG on d3, but not on d6 (J). Values represent mean ± SD of 6 experiments. *P < 0·05.

PI+ cells substantially increased with 100 nmol/l bortezomib in L363 cells, both on day 3 and 6 (n = 6, P = 0·03; Fig 1C). Similar cytotoxic effects were observed in U266 and RPMI8226.

In contrast to both sorafenib and bortezomib, exposure of L363 cells to 1, 10 and 100 μg/ml thalidomide did not substantially enhance PI+ cells (n = 9; Fig 1E). Viable cell numbers were also less affected (n = 9; Fig 1F). This was similarly observed in U266 and RPMI8226 and supports the lesser in vitro cytotoxicity of thalidomide, most likely due to the requirement for in vivo metabolism (Zhu et al, 2011), and additional immunomodulatory effects rather than cytotoxicity alone (Teo et al, 2005).

Although lenalidomide induced more substantial cytotoxicity and cell decline of L363 and RPMI8226 cells, this effect was not significant. In U266, however, 100 μmol/l lenalidomide induced a significant increase of PI+ cells on day 3 as compared to the control (n = 9; P = 0·004; Fig 1G). More significant cytotoxicity was induced on day 6 with 10 μmol/l and 100 μmol/l (P = 0·004 for both comparisons), and viable cells significantly decreased with 10 and 100 μmol/l on day 3 (P = 0·039 and P = 0·011, respectively) and with 100 μmol/l on day 6 (n = 9; P = 0·02; Fig 1H).

Treatment of L363 cells with EGCG for 3 and 6 days resulted in an increase of PI+ cells, reaching significance for 250 μmol/l and 500 μmol/l on day 3 (P = 0·031 for both comparisons) and for 500 μmol/l on day 6 (n = 6; P = 0·031; Fig 1I). Viable cells were also substantially reduced on day 3 and day 6 (n = 6; P < 0·05; Fig 1J).

Assessment of early and late apoptosis after treatment with sorafenib and bortezomib

We next assessed whether the cytotoxic effects of sorafenib and bortezomib were mediated through the induction of apoptosis, and whether both substances caused translocation of cytochrome c from the mitochondrial intermembrane space into the cytosol, an early event in the apoptotic process. Chromatin condensation and nuclear fragmentation were analysed as morphological hallmarks of apoptosis at later stages (Bossy-Wetzel et al, 1998; Renz et al, 2001; Ziegler & Groscurth, 2004). In RPMI8226 cells coexpressing histone 2B-mCherry and cytochrome c-GFP and consistent with our PI results, the cytotoxic effect of sorafenib and bortezomib was prominent and apoptosis was induced. The occurrence of apoptosis was particularly evident in chromatin condensation and fragmentation, indicating late apoptosis. The transition of the intracellular distribution of cytochrome c-GFP from a punctuate pattern, when localized in the mitochondria, into a diffuse, cytosolic pattern reflected early apoptosis (Fig 2A,B).

Figure 2.

Apoptotic changes induced by exposure of RPMI8226 cells to sorafenib and bortezomib. Confocal images of RPMI8226 cells co-expressing fluorescently labelled cytochrome c-GFP (green) and histone 2B-mCherry (red) showing early and late apoptosis after treatment with sorafenib 1 and 10 μmol/l and bortezomib 10 and 25 nmol/l for 3 days. The release of cytochrome c was evident, as the freckled pattern (filled white arrow) of cytochrome c-GFP became diffuse (framed white arrows). Chromatin condensation and fragmentation (white arrow heads), as a typical marker for late apoptosis, were observed. Combined effect of sorafenib and bortezomib on cell death in RPMI8226 cells. RPMI8226 cells were cultured with bortezomib and sorafenib for 3 days. Bortezomib enhanced sorafenib-induced cell death (C) showing a synergistic effect for the combination of bortezomib 10 nmol/l and sorafenib 10 μmol/l (CI=0·804; normalized isobologram; (D) Values represent mean ± SD of 4 experiments. *P < 0·05.

Cytotoxic effects of combination treatment

Given that sorafenib, bortezomib and EGCG efficiently induced cell death as single agents, we analysed their combined activity. In RPMI8226, both without and with 1, 5 and 10 μmol/l sorafenib, we observed a dose-dependent enhancement in PI+ cells after adding bortezomib (2, 5 and 10 nmol/l). PI-positivity with sorafenib 5 μmol/l and bortezomib 10 nmol/l, and with sorafenib 10 μmol/l combined with bortezomib at 5 or 10 nmol/l, was significantly higher compared to untreated cells (n = 4; P < 0·05 for all three comparisons; Fig 2C). The combination of sorafenib and bortezomib, not at all tested concentrations, but when applied at 10 μmol/l and 10 nmol/l, respectively, reached a CI of 0·80, indicating synergy (Fig 2D).

Additionally, we tested the combination of bortezomib and EGCG, shown to act in an antagonistic way in previous reports. With increasing concentrations of EGCG, bortezomib-induced cell death was maintained and a plateau cytotoxic effect induced (n = 3; Figure S1).

Sorafenib downregulates CD138 expression, induces actin depolymerization, cell shape changes and inhibits migration

In the absence of sorafenib, L363 cells revealed a substantially higher CD138 expression compared to U266 (Fig 3A), nevertheless, cytotoxicity and down-regulation of CD138 was observed in both cell lines with sorafenib concentrations as low as 1 μmol/l. Moreover, sorafenib concentrations as low as 1 μmol/l, induced evident cell shape changes, cell size alterations and an impressive reduction in the red immunostaining of actin filaments, suggesting actin depolymerization (Fig 3A).

Figure 3.

(A) Sorafenib downregulated CD138-expression and induced actin depolymerization and cell shape changes in MM cells. L363 and U266 cells were treated with sorafenib 1, 10 and 100 μmol/l for 3 days. Confocal images after immunocytochemistry staining with CD138-FITC (green), Phalloidin-594-Alexa (actin filaments, red) and DAPI (DNA, blue) showed CD138-downregulation, changes in cell size and reduction of F-actin content with sorafenib concentrations as low as 1 μmol/l. (B) Chemotaxis of L363 cells to M210B4 supernatant. L363 cells displayed a significant chemotaxis to M210B4 supernatant as compared to the control. Values represent mean ± SD of 6 experiments. *P < 0·05. (C) Chemotaxis of L363 cells to various chemokines. The chemotaxis of L363 cells to CCL5 at 50 ng/ml was significantly increased as compared to the control, whereas the exposure to CXCL12 at 10 and 100 ng/ml and CCL3 at 100 ng/ml did not induce substantial migration. Values represent mean ± SD of 15 experiments with CXCL12 10 and 100 μmol/l, 9 experiments with CCL3 and 6 with CCL5. *P < 0·05. (D) Chemokine receptor expression in L363 cells. Expression of various chemokine receptors was investigated by flow cytometry. The filled histograms represent the isotype-matched controls. L363 cells showed low levels of CXCR4, but were positive for CCR4 and CCR5, two specific receptors for CCL5, supporting the fact that M210B4-triggered chemotaxis of L363 cells was induced by CCL5 (representative experiment). (E) Chemotaxis of L363 cells to M210B4 in the presence of AMD3100. In the presence of the CXCR4 inhibitor AMD3100, chemotaxis of L363 cells to M210B4 supernatant did not significantly decrease, supporting the involvement of chemokines other than CXCL12. Values represent mean ± SD of 6 experiments. *P < 0·05. (F) Chemotaxis of L363 cells to M210B4 using various sorafenib concentrations. L363 cells were cultured for 3 days in the absence and presence of increasing sorafenib doses (1-50 μmol/l), assessing then M210B4 induced chemotaxis. A dose-dependent inhibition of chemotaxis was observed, also at low sorafenib concentrations that did not induce viability decreases. Data represent average fold-change from baseline of 6 experiments. *P < 0·05.

For the chemotaxis assays, we chose the suspension L363 cell line, rather than the more adherent cell lines U266 and RPMI8226. Chemotaxis of L363 cells towards the M210B4 supernatant was significant (n = 6; P < 0·05; Fig 3B). We used two strategies to identify which chemokines and receptors were responsible for this: 1. analysis of chemotaxis with single chemokines: CXCL12 at 10 and 100 ng/ml (n = 15), CCL3 at 100 ng/ml (n = 9) and CCL5 at 50 ng/ml (n = 6; Fig 3C), and 2. analysis of the expression of the chemokine receptors CXCR4, CCR1, CCR3, CCR4 and CCR5. L363 cells significantly migrated to CCL5 (P = 0·004), whereas chemotaxis with CXCL12 and CCL3 alone was less substantial (Fig 3C). Accordingly, L363 cells expressed low CXCR4; but both CCR4 and CCR5, corresponding to the chemotaxis towards their common ligand CCL5 (Fig 3C,D). To further clarify the role of CXCL12 and CXCR4, we investigated the M210B4-induced chemotaxis in the absence and presence of the CXCR4-inhibitor AMD3100. After addition of AMD3100, chemotaxis of L363 cells did not significantly decrease, supporting the involvement of chemokines other than CXCL12 in M210B4-induced chemotaxis (Fig 3E). Chemotaxis to M210B4 was substantially decreased when sorafenib was used (Fig 3F). Based on our cytotoxicity results, this reduction with 1 and 5 μmol/l sorafenib occurred due to the inhibition of cell migration rather than sorafenib-induced cytotoxicity (n = 6; P = 0·04 and P = 0·02, respectively; Fig 3F). Of note, we obtained a Spearman correlation coefficient between sorafenib dose and the number of migrated cells of = −0·74 (P < 0·0001).

Sorafenib inhibits CXCL12-induced actin polymerization

Reorganization of the actin cytoskeleton is a crucial and early occurring event for chemotaxis and homing of MM cells to the BM (Menu et al, 2002). Based on the reduction in the fluorescent staining of actin filaments with Phalloidin-Alexa594, we examined whether actin polymerization was hampered by sorafenib. We first measured intracellular F-actin after stimulation with CXCL12, CCL5 or M210B4-supernatant and achieved the best actin-polymerization responses with CXCL12. As depicted in Fig 4A, a significant and transient increase in F-actin occurred after 15 s of stimulation with CXCL12 in the absence of sorafenib or with low sorafenib concentrations (1 μmol/l), whereas with 10 μmol/l sorafenib, CXCL12-induced actin polymerization was abolished.

Figure 4.

(A) Inhibition of CXCL12-induced actin polymerization. L363 cells were incubated for 3 days with sorafenib 1 or 10 μmol/l. Intracellular F-actin was measured using FITC-labelled Phalloidin at 15, 60, 120 and 300 s after stimulation with 100 ng/ml CXCL12. Results are presented as percentages of intracellular F-actin relative to baseline content. Sorafenib at 10 μmol/l notably inhibited CXCL12-induced actin polymerization. (B) Effect of sorafenib on ERK1/2 phosphorylation. L363 cells were cultured with sorafenib at 1 μmol/l for 6 h and 24 h and at 10 μmol/l for 6 h. Sorafenib significantly decreased phosphorylation of ERK1/2 at 1 μmol/l after 3, 6 and 24 h in a time-dependent manner, which was even more notable after 3 and 6 h with 10 μmol/l. (C) Quantification of bands from Western blot. The intensity of each phosphorylated ERK1/2 band was normalized to the corresponding ERK1/2 and β-actin bands and the result expressed in function of untreated control cells.

Sorafenib downregulates ERK1/2 phosphorylation

Because the MAPK pathway in MM cells is involved in cell proliferation, actin polymerization and cell migration, we investigated changes in the phosphorylation levels of ERK1/2 proteins by Western blotting. L363 cells were incubated without and with 1 or 10 μmol/l sorafenib for 3, 6 and 24 h. ERK1/2 was constitutively active in L363 cells and this phophorylation potently inhibited by sorafenib in a dose- and time-dependent manner. We observed a progressive downregulation of ERK1/2 phosphorylation with 1 μmol/l sorafenib after 3, 6 and 24 h, which was even more distinctly downregulated with 10 μmol/l (Fig 4B,C).

Tumour growth inhibition in L363-bearing NSG after bortezomib and sorafenib treatment

L363-bearing NSG mice were treated with bortezomib or sorafenib to compare their in vivo antimyeloma activity. Bortezomib and sorafenib were administered with maximum tolerated doses as determined in titration studies (Figure S2 for bortezomib and Figure S3 for sorafenib). L363 cells reliably engrafted the BM (take rate = 100%) and distant organs, such as spleen (38%) and, to a lesser extent, the liver (8%): at 7, 14, 21 and 28 days after implantation, tumour cells were increasingly detectable via fluorescence-activated cell sorting (FACS) and IVI (Fig 5A,B, Figure S4). During the observation period (day 0–28), all mice developed MM symptoms, such as hind limb pareses, body weight loss and bone lesions, which were delayed in all treatment groups compared to control animals. FACS analyses confirmed that primary tumour growth was markedly reduced by sorafenib and, to a lesser extent, by bortezomib (Fig 5A and Figure S3). In a second experiment, IVI with hCD138-Ab was used to determine tumour engraftment and treatment efficacy (Fig 5B). The BM engraftment was markedly reduced by bortezomib and sorafenib single use as well as their combination, resulting in a substantially reduced tumour load of 7%, 13% and 5% compared to the untreated control animals, respectively (Fig 5B).

Figure 5.

(A) Chemosensitivity of L363 cells in i.t. injected mice treated with control vehicle, bortezomib and sorafenib determined by flow cytometry. Flow cytometry analyses were performed using anti hCD138-antibody. Mice were treated with control vehicle (0·9% NaCl, intraperitoneally days 7-11, 14-18; black), 0·7 mg/kg/d bortezomib i.v. (days 7, 11, 18; purple) or with 75 mg/kg/d sorafenib p.o. (days 7-18; red). BM engraftment was markedly reduced by sorafenib and less so by bortezomib. (B) Chemosensitivity of L363 cells in i.t.-injected mice treated with vehicle, bortezomib and sorafenib mono- and combined therapy. Mice were treated with control vehicle (0·9% NaCl, p.o. days 7-18; black), 0·7 mg/kg/d bortezomib i.v. (days 7, 11, 15, 18; purple), 100 mg/kg/d sorafenib p.o. (d7-18; red) or 0·7 mg/kg/d bortezomib i.v. (d7, 11, 15, 18) and 100 mg/kg/d sorafenib p.o. (days 7-18; pink). BM engraftment was reduced by bortezomib and more notably by sorafenib monotherapy as well as their combination, resulting in a reduction in tumour load of 7%, 13% and 5% compared to the untreated control animals, respectively (P < 0·05). (C) Clinical response parameters after sorafenib treatment. The measurement of MM response parameters before and after sorafenib treatment revealed a decline in total protein (TP), paraprotein levels (immunoglobulin G or A [IgG/IgA]) and serum free light chains (SFLC), whilst haemoglobin (Hb) levels notably increased as an indicator of therapy response. Results are expressed as mean fold change (±SD) from baseline levels.

Clinical response of MM-/AL-amyloidosis patients treated with sorafenib

MM specific parameters after sorafenib treatment revealed a decline of the total protein (TP), paraprotein (IgG/IgA) and serum free light chains (SFLC) in all patients (Fig 5C). Haemoglobin levels increased as an additional determinant of sorafenib response (Fig 5C) and due to secondary erythrocytosis (Alexandrescu et al, 2008), which was verified histologically in BM specimens. Moreover, the AL-amyloidosis patient displayed substantial loss of peripheral oedema (body weight loss of 10 kg in 2 weeks), reversal of dyspnoea and improvement of his general condition (Karnofsky Performance Status: 60%–>80%). The best response under sorafenib treatment according to the IMWG response criteria was stable disease (SD) in all patients. Sorafenib was generally well tolerated and severe adverse events (Common Terminology Criteria grade 3/4) were not observed. Of interest, the patient with MM and AL-amyloidosis was initially treated with sorafenib alone and, after the described response to sorafenib monotherapy, with the combination of sorafenib and bortezomib (Table 1).


The present study tested the ability of sorafenib to target MM cells and the interaction of MM and BMSCs. We investigated in vitro effects on proliferation, phenotype, signalling pathways, actin polymerization and chemotaxis, as well as cytotoxic interactions in combination with bortezomib. We observed a significant cytotoxicity after sorafenib treatment, confirming growth inhibition with concentrations ranging from 0·01 to 10 μmol/l in MM cell lines and patient BM specimens, irrespective of the presence of BMSCs (Ramakrishnan et al, 2010). As in previous analyses (Zlei et al, 2007), we focused on day 3 and 6 to evaluate antimyeloma agent effects and noted potent cytotoxic activity for 1 to 100 μmol/l sorafenib at both earlier (d3) and later (d6) time points. Our in vitro concentrations of 1 and 10 μmol/l sorafenib administered on day 0 correspond to in vivo doses of 120 and 1200 mg, respectively. These in vitro doses are well comparable or below the medication doses of 400-600 mg administered once or twice daily leading to cumulative in vivo doses of 1200 mg and 2400 mg after 3 days of sorafenib with 400 mg, and 1800 mg and 3600 mg with 600 mg, respectively (Rini, 2006; Strumberg et al, 2007; Miller et al, 2009).

Our systematic analysis included reference antimyeloma agents and demonstrated impressive cytotoxicity of sorafenib compared to bortezomib or IMIDs. Our slightly higher effective EGCG concentrations as compared to others seem most likely to be related to different assessment schedules and methods to determine cytotoxicity (Shammas et al, 2006). Our later time points allowed MM cells to partly metabolized EGCG, so that higher concentrations were needed for significant effects. In support of prior observations regarding the in vitro effect of bortezomib (Hideshima et al, 2001a,b, 2009; Edwards et al, 2009), we observed considerable cell death at similar concentrations. Chauhan et al (2008) treated MMCLs and patient plasma cells with lower bortezomib doses and showed no significant growth inhibition with concentrations ranging from 1 to 10 nmol/l. MMCLs tested in previous reports differ from ours, as do in vitro culture durations used to assess bortezomib-induced cytotoxicity. Thus, our observations widen the present knowledge on the in vitro antimyeloma activity of bortezomib. We observed additive effects for all sorafenib and bortezomib combinations and synergism for sorafenib at 10 μmol/l and bortezomib at 10 nmol/l, in line with Ramakrishnan et al (2010) who demonstrated synergism for sorafenib combined with bortezomib, dexamethasone or rapamycin in other MMCLs.

In vivo experiments from our group (Schüler et al, 2009) and others (LeBlanc et al, 2002) illustrate the antimyeloma activity of bortezomib in immunodeficient mice, showing tumour growth inhibition, apoptosis, reduced microvessel density and prolonged survival (LeBlanc et al, 2002). The promising in vitro and in vivo data when applied alone and its synergistic effect with sorafenib in various tumour cell lines (Yu et al, 2006), including MM (Ramakrishnan et al, 2010) makes bortezomib an attractive combination partner for sorafenib. Complementary to our in vitro analysis, and given the fact that humanized mouse models are widely pursued for antimyeloma strategies (Calimeri et al, 2011; DeWeerdt, 2011; Tassone et al, 2012), we tested the in vivo efficacy of sorafenib in L363-bearing NSG. L363 cells reliably engrafted the BM and, to a lesser extent, the spleen and liver. Tumour cells were detectable via FACS and fluorescence-based IVI. During the observation period, all mice developed MM symptoms, which were delayed in all treatment groups. FACS analyses and fluorescence IVI confirmed that primary tumour growth and myeloma dissemination were markedly reduced by sorafenib. Consistent with our in vitro results, the combination of bortezomib and sorafenib seemed more effective in reducing myeloma tumour load in NSG than if both agents were used alone, albeit single bortezomib and sorafenib schedules and doses were already that effective so that it was challenging to induce even better results with their combination. The observed synergy between bortezomib and sorafenib was of particular interest and could be the basis of future clinical trials. Moreover, based on our in vitro and in vivo evidence, we also assessed sorafenib in symptomatic MM and AL-amyloidosis patients. We observed paraprotein decline in all patients and an impressive organ response in the AL-amyloidosis patient. Sorafenib was very well tolerated and induced SD in all patients. Even though our data was based on 3 patients only and has to be extended to larger patient cohorts and phase I/II studies, our clinical results are encouraging, and represent a proof of concept for sorafenib single and combination use in MM.

Using RPMI8226 cells expressing histone 2B-mCherry and cytochrome c-GFP (Goldstein et al, 2000, 2005), we also proved that the cytotoxic effect of sorafenib was mediated through apoptosis by inducing cytochrome c release. The release of cytochrome c consistently precedes annexin V binding and PI accumulation in the nucleus and represents a valuable early marker of apoptosis, allowing its detection either at earlier stages or at lower concentrations of a specific antimyeloma agent.

We observed downregulation of phospho-ERK with sorafenib, which appeared responsible for the inhibition of MM cell proliferation, actin polymerization and M210B4-triggered chemotaxis. The ability of sorafenib to downregulate ERK phosphorylation has been described (Ramakrishnan et al, 2010), but whether this effect influences other cellular processes, such as MM cell adhesion and migration, was not previously assessed. We detected CD138-downregulation and cell shape changes with sorafenib concentrations as low as 1 μmol/l. CD138 downregulation has previously been observed under various antimyeloma agent treatment and is postulated to occur due to cytotoxicity, dedifferentiation of plasma cells and CD138 internalization (Zlei et al, 2007).

The observed cell shape changes seemed to be a consequence of cytoskeletal reorganization. We showed inhibition of CXCL12-induced actin polymerization in L363 cells treated with sorafenib. Confocal images of MM cells treated with sorafenib concentrations as low as 1 μmol/l showed impressive reduction in the red immunostaining of actin filaments, further supporting our hypothesis that sorafenib may abrogate actin polymerization and through this impair MM cell motility (Menu et al, 2002). Similarly, investigations in Waldenstrom macroglobulinaemia showed CXCL12-induced actin polymerization and its inhibition by the Src tyrosine kinase inhibitor AZD0530 (Ngo et al, 2009). Of interest was that chemotaxis of L363 cells to the M210B4-supernatant was significant, supporting the involvement of chemokines, such as CCL5. Consistent with the migration to their common ligand CCL5, L363 cells clearly expressed CCR4 and CCR5, but low CXCR4 levels, which is in line with the tendency of plasma cell leukaemia (PCL) cells to disseminate into the periphery rather than to home to the BM (Badr et al, 2011). Corresponding to the low CXCR4 expression, migration of L363 cells to CXCL12 was not substantial. Moreover, we did not observe a significant reduction in the M210B4-triggered chemotaxis with AMD3100, supporting the concept that CCL5 and other chemokines, rather than mainly CXCL12, are involved in M210B4-induced chemotaxis. After treatment with sorafenib at concentrations that did not affect cell growth, we observed a dose-dependent inhibition of M210B4-triggered chemotaxis. This has been described for thalidomide (Fuchida et al, 2008) and the phytochemical compound thymoquinone (Badr et al, 2011), but not for sorafenib.

In conclusion, we here demonstrate the potent antimyeloma activity of sorafenib single use and combined with bortezomib in vitro, in vivo and clinically in a preliminary pilot study of three MM/AL-amyloidosis patients. To the best of our knowledge, this is the first study demonstrating the cytotoxic effect of sorafenib in systematically performed in vitro and in vivo analyses as well as on the chemotaxis of myeloma cells. We postulate the reduced ERK-phosphorylation and consequent inhibition of actin polymerization to be the underlying mechanisms for the abrogation of MM cell chemotaxis. One characteristic of MM cells is their homing to the BM, thus blocking of cells represents a promising target to interrupt the interplay between malignant plasma cells and their protective microenvironment. The observed synergy between bortezomib and sorafenib is of interest and needs to be tested in subsequent phase I/II trials.


The authors greatly thank Dr. Antonia Müller, University of Stanford and University of Zürich and PD Dr. Meike Burger, PD Dr. Katja Zirlik, Dr. Martina Kleber and Heike Reinhardt (all University of Freiburg), for useful discussions and critical comments on the manuscript. We are also very thankful for valuable recommendations from Prof. Dr. Hans Messner, Toronto, Prof. Dr. Hans Stauss, London, Prof. Dr. Hubert Serve, Frankfurt and Prof. Dr. Dieter Wolf, La Jolla. We are indebted to the anonymous internal reviewers of our department for their review and comments and diligent expert comments of the anonymous external reviewers. This work was supported by the Biothera foundation.

Author contributions

Josefina Udi, Julia Schüler and Dagmar Wider performed the research. Monika Engelhardt and Ralph Wäsch designed research study. Josefina Udi, Julia Schüler, Dagmar Wider, Julie Catusse, Gabriele Ihorst, Dominik Schnerch, Marie Follo and Monika Engelhardt analysed the data. Monika Engelhardt, Josefina Udi and Ralph Wäsch wrote the paper.

Conflicts of interest

All authors declare no conflicts of interest.