Sorafenib Inhibits Lymphoma Xenografts by Targeting MAPK/ERK and AKT Pathways in Tumor and Vascular Cells

The anti-lymphoma activity and mechanism(s) of action of the multikinase inhibitor sorafenib were investigated using a panel of lymphoma cell lines, including SU-DHL-4V, Granta-519, HD-MyZ, and KMS-11 cell lines. In vitro, sorafenib significantly decreased cell proliferation and phosphorylation levels of MAPK and PI3K/Akt pathways while increased apoptotic cell death. In vivo, sorafenib treatment resulted in a cytostatic rather than cytotoxic effect on tumor cell growth associated with a limited inhibition of tumor volumes. However, sorafenib induced an average 50% reduction of tumor vessel density and a 2-fold increase of necrotic areas. Upon sorafenib treatment, endothelial and tumor cells from SU-DHL-4V, Granta-519, and KMS-11 nodules showed a potent inhibition of either phospho-ERK or phospho-AKT, whereas a concomitant inhibition of phospho-ERK and phospho-AKT was only observed in HD-MyZ nodules. In conclusion, sorafenib affects the growth of lymphoid cell lines by triggering antiangiogenic mechanism(s) and directly targeting tumor cells.


Introduction
Sorafenib is an oral multitargeted inhibitor of kinases approved by the Food and Drug Administration for the treatment of patients with advanced renal cell carcinoma (RCC) and those with unresectable hepatocellular carcinoma (HCC) [1][2][3]. It is also approved by the European Medicines Agency for the treatment of patients with HCC and patients with advanced RCC with whom prior interferon-a-or interleukin-2-based therapy had failed or those considered to be unsuitable for such therapy. Sorafenib is currently undergoing phase II/III clinical evaluation in a wide variety of other solid as well as hematopoietic tumors [4][5][6][7].
In addition to targeting the receptor tyrosine kinases c-kit, Flt3, and RET [8,9], sorafenib at a clinically achievable plasma concentration of 10 mM, inhibits multiple additional receptor tyrosine kinases and intracellular kinases [10]. By inhibiting the Raf/mitogen-activated protein (MAP)/extracellular-signal regulated kinase (ERK) kinase (MEK)/ERK (Raf/MEK/ERK) signaling pathway sorafenib exerts direct antiproliferative, apoptotic and antiangiogenic effects on a variety of solid tumors as well as leukemic cell lines [11]. The antiproliferative activity of sorafenib varies from nanomolar to micromolar concentrations depending on the oncogenic signaling pathways driving proliferation. The proapoptotic effects of sorafenib may vary among cell lines and result from mechanism(s) of action that are only partially elucidated. A commonly observed feature is the inhibition of phosphorylation of the initiation factor eIF4E and downregulation of the antiapoptotic protein myeloid cell leukemia-1 (Mcl-1) [12,13], a Bcl-2 family member implicated in malignant hematopoietic cell survival. Downregulation of Mcl-1 by sorafenib is associated with the release of cytochrome c from mitochondria into the cytosol, caspase activation, and apoptotic cell death [14]. The ability of sorafenib to suppress tumor angiogenesis has been well established and is likely due to the potent inhibition of the proangiogenic vascular endothelial growth factor receptor (VEGFR1)-1, VEGFR-2, VEGFR-3, and platelet-derived growth factor receptor-ß [9]. In vivo, sorafenib exhibits antitumor activity against a variety of human xenograft models of multiple histological types, including melanoma, renal, colon, pancreatic, hepatocellular, lung, and thyroid carcinomas, as well as acute myelogenous leukemia [7,[15][16][17]. Furthermore, sorafenib produced partial tumor regressions in mice bearing PLC/PRF/5 HCC and induced tumor regression in a breast cancer model harboring B-Raf and K-Ras oncogenic mutations [18].
Several lines of in vitro evidence suggest that sorafenib might have a role in the treatment of lymphomas by overcoming the cytoprotective effects of ERK, and Mcl-1 and eventually targeting oncogenic signaling pathways driving lymphomagenesis [19][20][21][22][23].
Indeed, sorafenib-induced inhibition of the ERK1/2 pathway might results in Bcl-X L downregulation [24], thus mimicking rituximab-mediated effects on CD20-positive non-Hodgkin lymphoma cell lines [25]. Additionally, sorafenib might downregulate Mcl-1 [14], which is implicated in resistance to anticancer drugs and is overexpressed in a significant proportion of diffuse large B cell lymphoma (DLBCL) and follicular lymphoma (FL) [26], thereby restoring lymphoma cell sensitivity to apoptosis.
It was the aim of the present study to further investigate in preclinical models both in vitro and in vivo the antitumor activity and mechanism(s) of action of sorafenib using a panel of cell lines representative of the heterogeneity of lymphoproliferative disorders. Data reported herein demonstrate that pharmacologically achievable concentrations of sorafenib inhibit the growth of lymphoid cell lines of different histotypes, and suggest that both antiangiogenic mechanism(s) as well as direct targeting of tumor cells mediate the in vivo antilymphoma activity of sorafenib.

Reagents
Sorafenib (Bayer AG, Leverkusen, Germany, EU) was dissolved in 100% dimethylsulfoxide (DMSO) to obtain a stock solution (60 mg/ml). Working solutions (1006) were obtained by further diluting in RPMI-1640 and were added to in vitro cultures at 1% (v/v) to achieve a final DMSO concentration of 0.01%.

Apoptosis and viable cell counting
Apoptotic cells were detected by annexin-V/propidium iodide (PI) double staining and flow cytometry analysis, as previously described [27]. Viable cells, i.e. annexin-V 2 /PI 2 cells, were counted with Flow-Count beads (Beckman Coulter, Milano, Italy, EU) by flow cytometry. The proportion of viable cells was further determined by an WST-based vitality assay following the manufacturer's recommendations (Biovision, Milpitas, CA, USA).

Measurement of DYm
Mitochondrial membrane depolarization was determined by using the fluorescent probe TMRE and flow cytometry, as previously described [28]. The fluorescent dye TMRE is accumulated by mitochondria and as a results of mitochondrial membrane depolarization, a shift to the left in the emission spectrum by apoptotic cells can be detected.
In vivo activity of sorafenib in tumor-bearing non-obese diabetic/severe combined immunodeficient (NOD/SCID) mice Six-to eight-week-old NOD/SCID mice were purchased from Charles River (Milano, Italy, EU) and xenografted with SU-DHL-4V, Granta-519, HD-MyZ, or KMS-11 cell lines [27]. Mice were housed under standard laboratory conditions according to our institutional guidelines. Animal experiments were performed according to the Italian laws (D.L. 116/92 and following additions), which enforce the EU 86/109 Directive, and were approved by the institutional Ethical Committee for Animal Experimentation of the National Cancer Institute Foundation. Tumor cells (5610 6 cells/mouse) were inoculated subcutaneously (SC) in the right flank of each mouse. When tumor volume reached approximately 100 mg in weight (6-13 days after tumor inoculation), mice were randomly assigned to receive either a short-or long-term treatment with sorafenib diluted in DMSO (final concentration, 10% v/v) or control vehicle (DMSO 10%, v/ v). In preliminary experiments, mice injection with a 10% DMSO solution failed to affect tumor cell signaling likely due to an in vivo DMSO dilution as well as metabolism. The short-term treatment consisting of intraperitoneal (IP) sorafenib (90 mg/kg) or DMSO for 5 days was used to assess necrotic areas and tumor vascularity. The long-term treatment consisted of sorafenib (90 mg/kg) or DMSO 5 days per week for 3 weeks. The endpoint of the longterm treatment was tumor weight. The tumors were measured with calipers, and their weights were calculated using the formula: (a6b 2 )/2, where a and b represented the longest and shortest diameters, respectively. Antitumor efficacy was measured as tumor growth inhibition (TGI) defined as [12(T/C)6100], where T and C are the mean tumor weight in the treated and untreated control groups, respectively. Mice were monitored twice weekly and were killed by cervical dislocation when they showed signs of terminal illness, including hind leg paralysis, inability to eat or drink, and/ or moribund. Each experiment was performed on at least two separate occasions, using five mice per experiment.

Analysis of stained sections
Entire tissue sections were acquired at 206 magnification with an automatic high-resolution scanner (dotSlide System, Olympus, Tokyo, Japan) and subdivided into a collection of non-overlapping red, green, and Blue (RGB) images in TIFF format (final resolution, 3.125 pixels/mm). For necrosis quantification, images were acquired at 26 magnification without further subdivision. Image analysis was performed using the open source imaging software ImageJ (http://rsb.info.nih.gov/ij/). Routines for image analysis were coded in ImageJ macro language and executed on RGB images without further treatment. Per each experimental condition, at least three tissue sections from at least three different tumor nodules were analyzed. Necrotic areas was evaluated on TUNEL-stained sections as previously described [27]. Endothelial cells were analyzed on cryosections from in vivo biotinylated mice which were stained with HRP-conjugated streptavidin. Automatic routines were validated by comparing results with those obtained by visual counting of up to 10% of the total images.

Confocal microscopy
Confocal microscopy was performed as previously described [27]. To detect tumor vessels, frozen sections were incubated with Alexa Fluor 488-conjugated streptavidin. To detect pericytes and tumor vessels, frozen sections were double stained with Alexa Fluor 568-conjugated streptavidin and anti-mouse NG2. Formalin-fixed, paraffin-embedded tumor nodules were sectioned at 1 mm and double-stained with anti-mouse CD31 (Santa Cruz Biotechnology, Inc., Heidelberg, Germany, EU) and anti human/ mouse phospho-AKT (Ser473) or phospho-ERK1/2 (Cell Signaling). CD31 expression was revealed by appropriate Alexa Fluor 568-conjugated secondary antibody (Invitrogen) and tumor/ endothelial p-AKT or p-ERK1/2 expression was revealed by Alexa Fluor 488-conjugated secondary antibody (Invitrogen). Nuclei were detected incubating sections with TO-PRO-3 nuclear dye (Invitrogen). Fluorescence-stained sections were examined under an epifluorescent microscope equipped with a laser confocal system (MRC-1024, Bio-Rad Laboratories). Image processing was carried out with LaserSharp computer software (Bio-Rad Laboratories).

Statistical analysis
Statistical analysis was performed with the statistical package Prism 5 (GraphPad Software, San Diego, CA, USA) run on a Macintosh Pro personal computer (Apple Computer Inc.). To test the probability of significant differences between untreated and treated samples, the Student's t test for unpaired data (two-tailed) or two-way ANOVA were used, as appropriate. Tumor volume data and proteome profiler array data were statistically analyzed with two-way ANOVA, and individual group comparisons were evaluated by Bonferroni's multiple comparison test. Differences were considered significant if p#0.05.

Sorafenib reduces viable cell countings and triggers apoptosis
The effects of sorafenib on cell proliferation were investigated in vitro by performing dose-response and time-course experiments using four cell lines representative of lymphoproliferative disorders with different histotypes. Readout of these experiments included viable cell countings and a WST-based cell proliferation assay. Exposure to 5 mM sorafenib for 24 or 48 hours resulted in modest cytostatic effects ( Figure 1A, Figure S1). In contrast, in vitro exposure to a clinically achievable concentration of sorafenib (10 mM) [30] for 24-48 hours resulted in significant reductions of cell viability ( Figure 1B, Figure S1) (p#0.05 at least) with cytotoxic effects being detected in SU-DHL-4V, Granta-519, and KMS-11 cell lines and cytostatic effects being detected in HD-MyZ cell line ( Figure 1B).
To investigate whether sorafenib-induced antiproliferative effects involved apoptosis, annexin-V/PI double staining was performed ( Figure 1B, Figure S2). Again, exposure to 5 mM sorafenib had no effect on cell death, whereas exposure to 10 mM sorafenib significantly increased (p#0.05 at least) cell death in all cell lines but HD-MyZ cells. Sensitive cell lines showed a wide degree of variability to sorafenib-induced apoptosis, with mean cell death values of 35% (range, 30% to 40%) and 55% (range, 40% to 64%) following a 24-and 48-hour incubation, respectively ( Figure 1B). Upon sorafenib exposure, SU-DHL-4V and Granta-519 cell lines but not KMS-11 cell line showed evidences of caspase-independent apoptosis as suggested by the lack of caspase activation and PARP cleavage ( Figure 2A). To further examine the molecular mechanisms whereby sorafenib triggers apoptosis, we investigated whether loss of mitochondrial potential was involved in sorafenib-induced cell death. Indeed, a potent mitochondrial membrane depolarization could be detected in all apoptosis-prone cell lines ( Figures 2B-C), with loss of mitochondrial potential ranging from 58% to 94%.

Effects of sorafenib on primary DLBCL and MCL cells
Apoptotic activity of sorafenib was further evaluated by analyzing its effect on primary cells purified from the peripheral blood of DLBCL and MCL patients at diagnosis ( Table 1). While sorafenib-induced apoptosis could be detected in only 1 out of 3 MCL samples, apoptotic cell death was consistently observed in all DLBCL samples.

Sorafenib affects phosphorylation of MAPK and PI3K/Akt pathways
To analyze sorafenib's effects on MAPK and PI3K/Akt signaling pathways, we used a standard antibody array technology allowing simultaneous analysis of all three major MAPKs, including ERK1/ERK2, JNKs, and p38a, as well as the Akt isoforms, and the direct and indirect Akt targets GSK-3 a/b and p70 S6 kinase 1 (p70S6K). Sorafenib (10 mM, 2 hours) affected MAPK and PI3K/Akt signaling pathways in a rather heterogeneous manner ( Figure 3A-C). As demonstrated by reduced phosphorylation levels for ERK1/2 and p38a kinases, sorafenib downregulated MAPK signaling in all cell lines. Analysis of PI3K/ Akt signaling revealed significantly reduced phosphorylation levels of Akt1 and Akt2 in SU-DHL-4V and Granta-519 cell lines. These lines also showed reduced phosphorylation levels of the direct and indirect Akt targets GSK-3 a/b and p70S6K, revealing that sorafenib may severely affect the PI3K/Akt pathway in some lymphoma cell lines ( Figure 3A-B). In contrast, PI3K/Akt signaling was unaffected in the HD-MyZ, and KMS-11 cell lines.
Western blot analysis essentially confirmed the previous antibody array results, further showing that sorafenib treatment modulated MAPK and PI3K/Akt phosphorylation levels in a rather heterogeneous manner depending on the different target cells ( Figure 3C). Consistent with sorafenib's effect on the Raf/ MEK/ERK pathway, a time-dependent down regulation of MEK and ERK phosphorylation could be detected in SU-DHL-4V, HD-MyZ, and KMS-11 cell lines ( Figure 3C). Additionally,    Figure 3C). Upon sorafenib exposure, expression of Mcl-1 was markedly down regulated in SU-DHL-4V, HD-MyZ and KMS-11 cells, whereas no effect was observed in Granta-519 cells ( Figure 3C).

Sorafenib inhibits tumor cell proliferation and angiogenesis
NOD/SCID mice inoculated subcutaneously with SU-DHL-4V, Granta-519, HD-MyZ, or KMS-11 cells, received a 15-day treatment with sorafenib (90 mg/kg/day, 5 days per week for 3 weeks) or vehicle control. Treatment was started when tumor volumes reached approximately 100 mg in weight (i.e., 6-13 days after tumor inoculation). Sorafenib treatment resulted in modest effects on the growth of SU-DHL-4V, Granta-519, and KMS-11 xenografts, as shown by TGI values of 37%, 39%, and 53%, respectively, whereas it substantially affected the growth of HD-MyZ xenograft, resulting in a 71% TGI ( Figure 4A). These findings were paralleled by a strong decrease of Ki-67 expression in tumor cells ( Figure 4B), suggesting that sorafenib indeed inhibited tumor cell proliferation. However, such an antiproliferative effect was not associated with an increase of tumor cell apotosis, as evaluated by TUNEL staining (data not shown), suggesting that the effect of sorafenib on tumor cell growth was cytostatic rather than cytotoxic.
We then evaluated the effects of sorafenib on tumor vasculature by using a sulfo-NHS-LC-biotin-based in vivo assay that allows a detailed qualitative and quantitative analysis of tumor vasculature [29]. Tumor vessel density was assessed by quantifying the percentage of biotinylated cells, i.e., the percentage of the entire tissue section occupied by endothelium. Reduction of tumor vasculature was calculated within viable tumor areas only while excluding necrotic areas. Regardless of the lymphoma subtype, tumor vasculature in control mice was abundant, tortuous, and evenly distributed throughout the tumor, which thus appeared well vascularized ( Figure 5A). Extensive vascularization is likely due to the upregulation of VEGFR-2 expression by tumor cells and tumor endothelium and the concomitant production of VEGF by tumor cells (data not shown) [31].
In sorafenib-treated tumors, vessels appeared smaller in length, lacking in sproutings and much less arborized and dishomogeneously distributed throughout tumor tissue as compared to vehicletreated animals ( Figure 5A) [32], suggesting that in contrast with other antiangiogenic drugs, sorafenib has no effect on vessel normalization [33]. To further investigate this issue, tumor sections were co-stained for both tumor vasculature with Alexa Fluor 568-conjugated streptavidin, and pericytes with anti-NG2 antibody. After sorafenib treatment, confocal microscopy analysis showed a marked reduction of pericytes in SU-DHL-4V and HD-MyZ xenografts, whereas no difference was observed in Granta-519 and HD-MyZ xenografts, further supporting no activity of sorafenib on vessel normalization ( Figure 6).
In vivo anti-tumor mechanism(s) of sorafenib were additionally investigated by analyzing Akt and ERK1/2 phosphorylation in sorafenib-treated tumor cells and tumor vasculature (Figure 7). Tumor and vascular ERK1/2 phosphorylation was unaffected in SU-DHL-4V and Granta-519 cell lines, whereas it was strongly  (Figure 7). Notwithstanding the strong inhibition of ERK1/2 and/or Akt phosphorylation on vascular cells, we did not detect any increase in tumor endothelial cell apoptosis, as evaluated by CD31/TUNEL double-staining (data not shown).

Discussion
Data reported herein using four cell lines representative of different lymphoproliferative disorders show that sorafenib may exert antilymphoma activities that are likely mediated by a variety of mechanism(s), including antiproliferative effects on tumor cells as well as antiangiogenic effects. These findings significantly extend the spectrum of tumors that sorafenib targets and are consistent with previously reported results showing efficacy of sorafenib in multiple tumor xenograft models [7,[15][16][17].
In vitro, clinically achievable concentrations of sorafenib (5-10 mM) [30] induced a marked inhibition of MAPK and/or PI3K/Akt signaling as well as Mcl-1 in all cell lines, resulting in a significant reduction of cell proliferation and/or induction of apoptosis. Sorafenib affected both MAPK and PI3K/Akt signaling in essentially all analyzed cell lines, with PI3K/Akt signaling being markedly affected in SU-DHL-4V and Granta-519 cell line. In vivo, sorafenib has previously been reported to inhibit the growth of a wide variety of human tumor xenografts in mice [7,[15][16][17]. Our data obtained in four xenograft models of lymphoma demonstrate that the potent antiangiogenic activity of sorafenib, eventually combined with its marked inhibition of lymphoma cell  proliferation, induces a relevant tumor destruction as shown by large areas of tumor necrosis detected in sorafenib-treated mice. According to previously reported studies, an optimal biological dose of sorafenib (90 mg/kg/day) was choosen for our in vivo experiments [9,18,34]. Sorafenib dose resulting in antitumor activity in our xenograft models (90 mg/kg/day) roughly equals to 600 mg/day in humans and compares well with the standard sorafenib daily dose used in clinical trials. The treatment dose used in our experiments is expected to produce plasma drug concentrations in the range of pharmacologically achievable concentrations of sorafenib, i.e., 5 to 15 mM, that are consistently detected in patients receiving the standard sorafenib daily dose of 800 mg [4,5,17,30].
While a 5-day treatment with sorafenib induced an average 50% inhibition of tumor angiogenesis and extensive tumor necrosis, biologically significant effects on the volume of lymphoma xenografts could only be observed in one of four xenografts (i.e., HD-MyZ) and required a long-term sorafenib treatment. Indeed, the limited effects of sorafenib on the volume of lymphoma xenografts closely resemble the findings frequently reported in patients treated with sorafenib who may experience a clinical benefit in the absence of any tumor volume reduction [3,35].
Analysis of tumor xenografts showed a substantial reduction of Ki-67 expression, as well as inhibition of phospho-Akt and/or phospho-ERK1/2 levels, suggesting that sorafenib-mediated lymphoma growth inhibition also involves a direct inhibition of cancer cell proliferation. However, analysis of tumor nodules from mice receiving a 5-day sorafenib treatment revealed a marked reduction of tumor angiogenesis, eventually associated with inhibition of vascular phospho-Akt and/or phospho-ERK1/2, thus suggesting that inhibition of angiogenesis is likely to play a key role in triggering the antitumor activity of sorafenib. Indeed, angiogenesis inhibition almost certainly represents a primary effect of sorafenib treatment rather than being secondary to tumor cell loss and reduced production of angiogenic factors. In fact, inhibition of tumor angiogenesis was detected in non-necrotic areas, suggesting that inhibition of tumor vasculature was the first event triggered by sorafenib, subsequently leading to the generation of extensive tumor necrosis due to hypoxic conditions. Alternatively, if tumor cell loss would have preceded reduction of angiogenesis, tumor vessel density would not be expected to decrease and should result unchanged or possibly increased. Taken together, these data strongly support that sorafenib acts primarily as an antiangiogenic agent, as shown by the marked tumor vessel reduction, but it also directly affects lymphoma cells, as shown by Ki-67 inhibition.
It has been proposed that one of the mechanisms by which antiangiogenic agents mediate their effect is to transiently ''normalize'' tumor vasculature, resulting in attenuation of hyperpermeability, increased vascular pericyte coverage, reduction in tumor hypoxia and interstitial fluid pressure [36]. The results of our studies however, show that administration of sorafenib results in a decrease of vasculature, leading to elevated level of tumor necrosis with the absence of vessel normalization as detected by vessel morphological features as well as number of pericytes. Although vessels leakiness poses a challenge for drug delivery [37], at the same time, lack of pericytes makes the vessels susceptible to antiangiogenic therapies [38]. Sorafenib may therefore improve anti-angiogenic therapy in lymphoma by targeting both endothelial cells as well as their supporting pericytes.
The best in vivo antitumor response to sorafenib, i.e., a 70% TGI, was detected in NOD/SCID mice bearing HD-MyZ xenografts that showed a marked inhibition of both tumor and vascular ERK1/2 and Akt phosphorylation, suggesting that the concomitant inhibition of MAPK and PI3K/Akt signaling is required to achieve an optimal antilymphoma activity. In contrast, modest in vivo responses, ranging from 35% to 50% TGIs, were observed in mice bearing SU-DHL-4V, Granta-519 or KMS-11 xenografts that were characterized by a selective in vivo inhibition of either ERK1/2 or Akt signaling. Overall, MAPK, PI3K/Akt and Mcl-1 signaling pathways appear to be critical determinants mediating the in vivo antitumor response to sorafenib [12,14,39].
Human lymphomas harbor multiple genetic and epigenetic alterations leading to deregulation of multiple pathways and cellular processes. Our data as well as those of others [19][20][21][22] suggest that the therapeutic armamentarium today available for human lymphomas might also include sorafenib mainly due to its antiangiogenic properties along with its antiproliferative and signaling inhibitory activities eventually associated with down modulation of Mcl-1 expression [31,40,41]. As a single agent, sorafenib might preferentially act as an inducer of tumor necrosis, whereas it might enhance the efficacy of radio-chemotherapy due to its effects on vessel normalization. In conclusion, our results provide evidence that sorafenib may be a potentially interesting molecule for the treatment of lymphoproliferative disorders mainly due to its antiangiogenic properties.