Cardiovascular Research (2014) 104, 303–314 doi:10.1093/cvr/cvu210

In vitro and in vivo characterization of the actin polymerizing compound chondramide as an angiogenic inhibitor Magdalena H. Menhofer 1, Dominik Bartel 1, Johanna Liebl 1, Rebekka Kubisch1, Johanna Busse2, Ernst Wagner 2, Rolf Mu¨ller3, Angelika M. Vollmar 1, and Stefan Zahler 1* 1 Department of Pharmacy, Pharmaceutical Biology, University of Munich, Butenandtstrasse 5-13, Munich 81377, Germany; 2Department of Pharmacy, Pharmaceutical Biotechnology, University of Munich, Butenandtstrasse 5-13, Munich 81377, Germany; and 3Helmholtz Institute for Pharmaceutical Research Saarland, Helmholtz Centre for Infection Research and Department of Pharmaceutical Biotechnology, Saarland University, Saarbru¨cken, Germany

Received 28 May 2014; revised 14 August 2014; accepted 7 September 2014; online publish-ahead-of-print 19 September 2014 Time for primary review: 35 days

Aims

Inhibiting angiogenesis is a major approach in tumour therapy. To combat angiogenesis, the tubulin cytoskeleton has emerged as an interesting target in many pre- and clinical studies. Contrarily, the actin cytoskeleton has been largely neglected as a potential drug target in angiogenesis. However, due to the development of drug resistances, new therapeutic strategies are always needed in tumour treatment. Therefore, the therapeutic potential of actin-binding small molecules is of particular interest. ..................................................................................................................................................................................... Methods We investigate the impact of chondramide (Ch), an actin polymerizing myxobacterial compound, on angiogenesis and and results underlying signalling. Chondramide treatment not only reduces the migration of endothelial cells but also the maturation of endothelial tube networks on matrigel. These observations can partly be explained by a disintegration of stress fibres due to aggregation and subsequent accumulation of actin in cellular structures known as ‘aggresomes’. Chondramide treatment impairs the maturation of focal adhesions and reduces the amount of active b1 integrin at the cell surface. Accordingly, signalling events downstream of focal adhesions are reduced. Thus, we observed that the activity of Src and downstream factors Rho-GTPases Rac1 and Rho is reduced upon Ch treatment. In vivo, Ch was well tolerated in mice and vascularization of a tumour xenograft as well as of the developing retina was significantly reduced. ..................................................................................................................................................................................... Conclusion Chondramide diminishes angiogenesis via two ways: (i) the disintegration of stress fibres and (ii) the reduction of promigratory signals. Our findings highlight Ch as a novel class of therapeutic lead compound with anti-angiogenic potential.

----------------------------------------------------------------------------------------------------------------------------------------------------------Keywords

Chondramide † Actin † Angiogenesis

1. Introduction Angiogenesis, the development of new blood vessels from pre-existing ones, contributes to many diseases like artherosclerosis, autoimmune diseases, age-related macular degeneration, and tumour growth.1 In the development of therapies inhibiting angiogenesis, the VEGF2 or its receptor are currently the main targets in order to reduce vessel formation.3 However, this highly specific approach faces the problem of resistances and unexpected side effects4 raising the need for new targets and strategies.5 In the process of angiogenesis, proliferation and cell migration of endothelial cells represent key steps that contribute to new vasculature.

Thereby, the endothelial cell migration is a process that requires highest plasticity of the cells and, thus, its internal machinery. Therefore, a functional actin- and microtubule-cytoskeleton are a prerequisite for cell dynamics and motility.6 Microtubules have been investigated as target in anti-angiogenic therapy and microtubule binding drugs have found their way into the clinic years ago.7 In contrast, only little is known about the actin cytoskeleton as therapeutic target in spite of its well-described role in angiogenesis. In the process of endothelial cell migration, actin builds the most fundamental migratory structures. At the leading edge of a migrating cell, lamellipodia, rims of dense actin network, are built for cellular protrusion.8 At the cell rear, endothelial cells build long bundles of actin fibres that allow the cell,

* Corresponding author. Tel: +49 89 2180 77196; Fax: +49 89 2180 77170; Email: [email protected] Published on behalf of the European Society of Cardiology. All rights reserved. & The Author 2014. For permissions please email: [email protected].

304 together with myosins, to retract the cell body into the direction of migration.8,9 These structures are mainly regulated by Rho-GTPases with Rac1 inducing lamellipodia and Rho regulating stress fibre contractility.10 The Rho-GTPases in turn are activated via the focal adhesions. At focal adhesions, which comprise distinct and complex adhesive structures, signals from outside the cell are transmitted via integrins to the inside.11 Integrin activation leads to the accumulation and activation of signal transducing proteins like the focal adhesion kinase and Src which in turn activate Rho-GTPases. The activation and accumulation of signal transducing proteins occurs during the maturation of focal adhesions. For this maturation process, a contractile force from actin stress fibres is required.12 As actin plays such a central role in endothelial cell migration and angiogenesis, we hypothesize that actin represents a promising target to inhibit angiogenesis. In our study, we address the actin cytoskeleton in endothelial cells using the natural compound chondramide (Ch). The cyclodepsipeptide Ch is of myxobacterial origin and known to bind fibrilous actin and induce actin polymerization in vitro.13 We demonstrate that Ch inhibits angiogenesis in vitro and in vivo, and we provide first insights into the underlying signalling.

2. Methods 2.1 Material Chondramides were isolated as previously described.14 For this study, pure (. 90%) ChA or ChB were used. The purity of compounds was analysed by HPLC and mass spectrometry as described14 and is shown in Supplementary material online, Figure SA and B for ChB. ChA and B were dissolved and stored in DMSO and diluted in growth medium for experiments containing DMSO at a maximum of 0.1% (v/v). Jasplakinolide (Jk) was purchased from Enzo Life Sciences (Lo¨rrach, Germany) and Cytochalasin D (CytoD) from AppliChem (Darmstadt, Germany).

2.2 Cell culture Endothelial cells were cultured under constant humidity at 378C and with 5% CO2 in an incubator (Heraeus, Hanau, Germany). As cell culture medium, endothelial cell growth medium (ECGM, Promocell, Heidelberg, Germany) was used. The human microvascular endothelial cell line CDC/EU. HMEC-1 (HMEC-1) was kindly provided by Centers for Disease Control and Prevention (Atlanta, GA, USA) and used until passage 12. Human umbilical vein endothelial cells (HUVECs) were purchased from Promocell (Heidelberg, Germany) and used for experiments up to passage 4.

2.3 Proliferation assay HMEC-1 cells or HUVECs (1.5 × 103 cells/well) were seeded in a 96-well plate. After 24 h, cells were treated with the indicated concentrations of Ch and incubated for 72 h. Finally, cells were washed with PBS, incubated with 100 mL/well Crystal Violet solution (0.5% Crystal Violet, 20% methanol in H2O) for 10 min, washed, and dried. For solvation of Crystal Violet, 100 mL/well ethanol/Na-citrate solution (50% ethanol, 50% 0.1 M Na-citrate in H2O) were added, incubated for 5 min and measured at 540 nm using a microplate reader (Sunrise, Tecan, Ma¨nnedorf, Switzerland).

M.H. Menhofer et al.

the total image area. For chemotaxis assay, 50 × 103 HMEC-1 cells were seeded in a m-slide chemotaxis (ibidi GmbH, Munich, Germany). After 2 h, a gradient between 0 and 30% (v/v) FCS was applied according to the manufacturer’s instructions. Images were taken every 10 min for 20 h using an open U-iMIC microscope (TILL Photonics, GmbH, Gra¨felfing, Germany), ×10 objective. Images were analysed employing the ImageJ plugin Chemotaxis tool. Cell velocity, the accumulated distance (both as measures of overall motility), the direct distance between starting point and end point (Euclidean distance as a measure of directionality), as well as the centre of mass (representing the averaged point of all cell endpoints, and, thus, also indicating directionality) were measured in this assay.

2.5 Tube formation assay 11 × 103 HMEC-1 or HUVECs were seeded on matrigel (MatrigelTM , Schubert & Weiss-OMNILAB, Munich, Germany) in an angiogenesis slide from ibidi (Munich, Germany), treated as indicated and incubated for 15 h. Images were taken using the TILLvisION system. Analysis of images was performed by Wimasis GmbH (Munich, Germany). For time lapse imaging, the open U-iMIC microscope was used with a stage incubator. As parameters of tube formation, tube length, number of nodes, number of tubes, and of total loops were analysed. For visualization of apoptotic cells in this setting, samples were incubated with 10 mg/mL propidium iodide (PI) for 20 min and images using the TILLvisION system.

2.6 Fluorescence imaging For staining of actin combined with a-tubulin, subconfluent HMEC-1 were treated with DMSO or Ch, washed and fixed with aceton. For adhesion experiments, HUVECs were pretreated as indicated, seeded freshly for 30 min and fixed with 4% (v/v) paraformaldehyde. F-actin was stained with rhodamin– phalloidin (1:400, R 415, Molecular Probes/Invitrogen) and nuclei with bis-benzimide H33342 (Sigma-Aldrich, St Louis, MO, USA). The following antibodies were used: a-tubulin and Integrin a2b1 (ab18251, ab30483, Abcam), LAMP-1 (H4A3, Developmental Studies Hybridoma), Rab5A (S-19, sc-309, Santa Cruz), p21-Arc (612234, BD Biosciences), pp(T18S19)-MLC and Cortactin (3674, 3503, Cell Signaling), g-tubulin (T 6557, Sigma), Integrin a5 (AB1928, Millipore, Upstate), VE-cadherin (sc-9989, Santa Cruz). Images were obtained using a Zeiss LSM 510 META confocal microscope (Zeiss, Jena, Germany). In case of Vinculin visualization, HUVECs were transfected with pLifeAkt-RFP (ibidi, Munich, Germany) and pEGFP-Vinculin (50513, Addgene, Inc., Cambridge, MA, USA) 24 h before experiment and images were taken using a.Leica TCS SP 8 SMD confocal microscope (Leica, Mannheim, Germany). For other life cell imaging, cells were transfected with pLifeAkt-GFP (ibidi). Images were taken using an open U-iMIC microscope (Fey, Munich, Germany).

2.7 Adhesion assay Pretreated HUVECs were trypsinized, suspended in ECGM containing Ch or DMSO (solvent control), and allowed to adhere on fibronectin or collagen for 30 min. Cells were fixed with 4% (v/v) paraformaldehyde and stained for F-actin. Images were taken (×10 objective) and counted for adhering cells. Cell morphology was distinguished in normal spreading cells (normal), irregular shaped cells, showing non-symmetrical spreading (irregular), and completely rounded cells (globular).

2.4 Migration assays For scratch assay, confluent HMEC-1 cells or HUVECs were scratched using an automated custom-made device and treated as indicated. Cells were allowed to migrate for 13 h, then fixed with 4% (v/v) paraformaldehyde and images were taken using TILLvisION system (Lochham, Germany) in connection with an Axiovert 200 microscope (Zeiss, Jena, Germany). Images were analysed by Wimasis GmbH (Munich, Germany). Migration was quantified as the percentage of the cell covered area compared with

2.8 FACS analysis Pretreated HUVECs were trypsinized, washed, and stained for total integrin b1 (Santa Cruz Biotechnology Inc., Santa Cruz, CA, USA) or active integrin b1 (12G10, 30394, Abcam Inc., Cambridge, MA, USA), respectively. Samples were analysed by flow cytometry using a FACSCanto II (Becton Dickinson, Heidelberg, Germany). The mean fluorescence was calculated using the Diva-Software of BD Bioscience (San Jose, CA, USA). Measurement of

305

Chondramide inhibits angiogenesis

apoptotic cells was performed as described15 followed by flow cytometry. Data were analysed using flow cytometry analysis software FlowJo 7.6.

2.9 Western blotting Western blot analysis was performed as described previously.16 The following antibodies were used: FAK, p(Y397)FAK, GAPDH, MLC2 (sc-1688, sc-11765-R, sc-69778, sc-15370, Santa Cruz Biotechnology), p(S19)MLC2, Src, p(Y416)Src (3675, 2110, 6943, Cell Signaling, Danvers, MA, USA), Rac (23A8, 05-389, Merck Millipore, Darmstadt, Germany), Rho (16116, Thermo Scientific, Bonn, Germany). For quantification, blots were scanned and the grey value of bands was quantified with an ImageJ plugin.

2.10 Rho-GTPase pull-down experiments Activation of Rac1 and Rho were induced by freshly seeding HUVECs for 30 min followed by lysis and a pull-down assay according to the manufacturer’s protocol (Thermo Scientific, Bonn, Germany).

2.11 In vivo tumour angiogenesis Mice were housed in individually ventilated cages, under specific pathogen-free conditions, with a 12 h day/night cycle and food and water ad libitum. 5 × 106 MDA-MB-231 cells were injected subcutaneously into the flank of SCID mice (C.B-17/IcrHanwHsd-Prkdcscid, Harlan, USA). Cells were injected in a mixture of 1:1 PBS/matrigel (BD Biosciences). Mice (10 per group) were treated with 0.75 mg/kg Ch i.p. or equal amounts of DMSO (control group). Treatment was started on Day 9 after tumour inoculation and was continued thrice a week. On Day 36, mice were sacrificed through cervical dislocation and tumours were harvested. All animal procedures were approved and controlled by the local Ethics Committee and carried out according to the guidelines of the German law of protection of animal life. Frozen sections: tumours were excised, embedded in Medium Neg50 (Thermo Scientific), frozen, and sectioned in 10 mm slices using a Microm AM 500 (Thermo Scientific). Cryo sections were stained for CD31 (anti-rat CD31 antibody, 553370, BD Biosciences, San Jose, USA, 1:100, plus Alexa Fluor 546-Goat-Anti-Rat IgG, 11081 Molecular Probes/Invitrogen) (1:400), for proliferating cells (Ki67, ab15580, abcam, plus Alexa Fluor 647, chicken anti-rabbit IgG, 21443, Molecular Probes/Invitrogen) and for apoptotic cells using ApopTagw Fluorescein In Situ Apoptosis Detection Kit (Merck Millipore, Darmstadt, Germany) according to manufacturer’s instructions. Fluorescent images were taken using ×10 objective and Zeiss or Leica confocal microscope. Paraffin sections: Tumours were removed, fixed in 4% paraformaldehyde (pfa) for 24 h, left in 1% pfa, embedded into paraffin, and sections (10 mm) were prepared. Paraffin sections were stained for Hif1a using anti-Hif1a antibody (1:100, 610958, BD Biosciences) and Vectastain Universal Elite ABC Kit (PK-6200, Vector Labs, Burlingame, CA, USA). Images were taken with a BX41 microscope and a DP25 camera from Olympus (Munich, Germany).

2.12 Retina assay Chondramide B (1 mg/kg) or DMSO equivalent was injected i.p. in pups of Black/6 mice at Day P1, P2, and P3. On Day P4, pups were sacrificed by decapitation, retinas were prepared and stained according to Pitulescu et al. 17 In brief, eyes were removed, fixed in 4% (v/v) pfa (2 h, RT), and retinae were prepared. After blocking (2 h, RT in PBS containing 10% Triton X100 and 1% BSA), retinae were stained for isolectin B4 (IB4, Alexa 488 conjugated, Millipore) and VE-Cadherin (CD144, 1:100, 555289, BD Biosciences). Pictures were taken with a.Leica TCS SP 8 SMD confocal microscope. The numbers of vascular sprouts per 1000 mm perimeter of the vascular front were counted by using Image J. The number of vascular branching points per 0.25 mm2 of tissue area was also quantified with ImageJ as previously described.17 All animal procedures were approved and controlled by the local Ethics Committee and carried out according to the guidelines of the German law of protection of animal life.

2.13 Statistical analysis The number of independently performed experiments (n) and the statistical tests used are stated in the respective figure legend. Graph data represent mean + SEM. Statistical analysis was performed with the software GraphPad Prism Version 5.04 (GraphPad Software, Inc., La Jolla, CA, USA). In brief, non-parametric tests were used: Mann– Whitney test for comparison of two groups, and Kruskal – Wallis test with Dunns test as a post hoc for comparison of more than two groups. Statistical significance is assumed if P ≤ 0.05. For most tests, three independent replicates (n ¼ 3) were used in triplicate, respectively. For the chemotaxis assay, n ¼ 4 was used. For the in vivo tumour assay, n ¼ 10 mice per group were used.

3. Results 3.1 Ch inhibits proliferation of endothelial cells at nanomolar concentrations Initially, we examined the effect of Chs on endothelial cell proliferation. Ch inhibits the proliferation of HMEC-1 cells at nanomolar concentrations with an IC-50 value of 82 nM for ChA and 36 nM for ChB, respectively (Figure 1A). The effect of ChB on HUVECs is nearly identical (Figure 1A). Thus, both substances are very potent; however, ChB shows a slightly higher activity on proliferation than ChA. Interestingly, for both substances, the curve showed an extreme sigmoidal slope meaning no effect up to 30 nM followed by a complete drop to nearly zero at around 100 nM indicating this as a functionally relevant concentration range (see Supplementary material online, Figure S1C –E). Further, we compared the activity of Ch to the known actin-targeting substances Jk and CytoD. These showed an IC-50 of 27 nM (Jk) and 52 nM (CytoD), respectively (Figure 1A). ChA and B, thus, have a comparable potency compared with established actin-targeting compounds.

3.2 Ch inhibits endothelial cell migration but not specifically chemotaxis The migration of endothelial cells represents a major step in the process of angiogenesis. Therefore, we examined the effect of Ch on the migration of endothelial cells. First, a wound healing assay was performed. The migration of HMEC-1 cells into the scratch was inhibited in a concentration-dependent manner up to an inhibition by nearly 50% at 200 nM (Figure 1B) and the migration of HUVECs was inhibited to 50% at 100 nM for ChB (Figure 1C). Secondly, the impact on chemotactic properties in the migration towards a gradient of growth factors was tested. In the chemotaxis assay, which is performed on single cells, migration was already inhibited at 100 nM, as seen by a reduction of the mean velocity and the accumulated distance to ,50% (Figure 1D). The chemotactic ability was, however, not affected specifically as the centre of mass towards the gradient was only reduced when cell migration was inhibited. Additionally, the overall passed distance (accumulated distance) and the distance between start and end point (Euclidean distance) were either not (30 nM), or both reduced (100 nM) indicating that Ch has an overall effect on cell motility but not on the directionality of the movement.

3.3 Ch affects tube formation To study the process of angiogenesis in vitro, a tube formation assay on matrigel was performed. The tube structure when compared with the DMSO control could not be established with increasing concentrations of Ch (see Supplementary material online, Figure S2A). In the presence of 100 nM Ch, an initial network is still established up to 4 h of incubation.

306

M.H. Menhofer et al.

Figure 1 Chondramide inhibits proliferation and migration of endothelial cells. (A) Half maximal inhibitory concentrations (IC-50) on proliferation are shown for ChA and B as well as the actin-targeting substances Jasplakinolide (Jk) and Cytochalasin D (CytoD) (n ¼ 3). (B and C) Chondramide inhibits migration of HMEC-1 (B) and HUVECs (C ) in a concentration-dependent manner. Left panel: Representative images (out of triplicates in three independent experiments each) are shown. The area covered with cells is displayed in green. Right panel: Images were analysed on cell covered area. (Kruskal – Wallis test with Dunns test as post hoc, *P , 0.05 vs. DMSO, n ¼ 3. (D) Chondramide inhibits migration per se but does not diminish chemotactic properties of HMEC-1 specifically. Upper panel: The analysis of one representative experiment (out of triplicates in four independent experiments each) is shown. Lower panel: Quantitative evaluation of the parameters centre of mass, accumulated distance, euclidean distance, and mean velocity. (Kruskal – Wallis test with Dunns test as post hoc, *P , 0.05 vs. DMSO, n ¼ 4 in triplicates).

However, the maturation of the network, which is characterized by a reduction of tubular loops and tube number, as well as by an increase of tube length in controls, does not occur after treatment with 100 nM Ch (Figure 2A). This leads to a fragmented phenotype of the tubular network with a higher number of shorter tubes. At 300 nM, cells could not build a network at all (see Supplementary material online,

Figure S2A). To evaluate the influence of cell death on the tube-forming process, we stained tubes with PI. In the staining, PI-positive cells were seen to a similar degree in all experimental groups, indicating no effect of Ch via cell death in this setting (see Supplementary material online, Figure S2A). Additionally, HUVECs cultured under normal 2D-conditions were tested for apoptosis induction via measuring the

Chondramide inhibits angiogenesis

307

Figure 2 Chondramide disturbs tube formation of HMEC-1 in vitro. (A) Chondamide alters tube maturation on matrigel. Upper panel: Time series of images representative for n ¼ 3 experiments in triplicates each are shown. Cell covered areas recognized by the software are indicated in blue, tubes in pink. Loops are indicated by yellow numbers. The white arrows indicate sites of tube fusion during network maturation. Scale bar: 1 mm. Lower panel: Image analysis after 15 h. Images of tubes were analysed on their respective number of tubes, the mean tube length, number of branching points, and number of loops. (Mann – Whitney test, *P , 0.05 vs. DMSO, n ¼ 3 in triplicates each). (B) Cell – cell contacts are disturbed in Chondramide-treated tubes. After tube formation assay (15 h), tubes were stained for F-actin (red), VE-cadherin (green), and nuclei (blue). Yellow indicates an overlap of actin and VE-cadherin. Images representative for n ¼ 3 experiments in triplicates each are shown. Bar represents 20 mm (n ¼ 3).

PI positive cells using FACS analysis (see Supplementary material online, Figure S2B). Nuclear fragmentation could only be seen at 300 nM Ch and higher. Further, we elucidated the integrity of the formed tubes and stained cell –cell junctions via VE-Cadherin (Figure 2B). Here, we could see that in tubes formed in the presence of 100 nM Ch, the cell –cell contacts were clearly disrupted indicating that Ch interferes with the stability of tubes and development of cell –cell contacts.

3.4 Ch leads to actin aggregation forming aggresomes To get a first insight into the intracellular effect of Ch, staining of the cytoskeleton was performed. The staining of the actin cytoskeleton revealed an aggregation of the fibrilous actin by Ch (Figure 3A). At 30 nM Ch, aggregation was only seen close to the nucleus (arrow) and starting around 4 h after treatment. However, at 100 nM Ch, already after 2 h

308

M.H. Menhofer et al.

Figure 3 Chondramide induces actin aggregation and disassembly of stress fibres leading to aggresome formation. (A) Proliferating HMEC-1 were treated with 30 or 100 nM ChB for the indicated time periods, fixed and stained for F-actin (upper panel) or a-tubulin (lower panel). Arrows indicate actin aggregates. Scale bar represents 20 mm. Representative images out of two independent experiments performed in triplicates are shown. (B) Actin nucleation occurs at stress fibres after 2 – 3 h followed by aggregate transportation to the perinuclear region. Time lapse imaging of transfected HMEC-1 (pLifeakt-GFP) treated with 30 nM ChB for indicated time periods was performed. One representative time series was chosen out of three independent experiments. See also Supplementary material online, Movie. Scale bar represents 50 mm.

aggregation occurred in the cytoplasm followed by a complete retraction of the cells. In contrast, microtubules did not show any specific difference. At 30 nM and also at 100 nM microtubules changed only due to deformation of the whole cell. These retracted cells were still viable as seen by a PI staining, a complete recovery of the cells after substance removal and nuclear fragmentation assay (see Supplementary material online, Figure S3). In comparison to cells treated with Ch for 8 h, and then immediately stained (Figure 3A, top right), the cells after Ch removal still show the massive perinuclear accumulation of actin, but have recovered stress fibres and lamellipodia to some degree. Following the aggregation via time lapse imaging, we could see that also at 30 nM Ch

aggregates arise from stress fibres after 2–4 h and are then transported proximally to the nucleus (Figure 3B, and Supplementary material online, Movie S1). This aggregate formation is time and concentration dependent (see Supplementary material online, Figure S4). In order to characterize these aggregates, we stained for markers of aggresomes, which have previously been described in connection to misfolded proteins and actin-binding compounds.18,19 Under Ch treatment, 80% of the cells show an aggregate in close proximity to the microtubule organizing centre (MTOC) (Figure 4A). In LAMP-1 and Rab5 stainings, we could see no overlap indicating neither a lysosomal nor an endosomal localization of the aggregates (Figure 4B). Finally, staining of the two actin-binding

309

Chondramide inhibits angiogenesis

Figure 4 Chondramide leads to aggresome formation. (A) Chondramide induced aggregates are located in proximity to the MTOC. HMEC-1 were treated with 30 nM ChB, stained for g-tubulin (green, white arrows) and counted. Cells were distinguished in two groups: (i) aggregates close to the MTOC or (ii) not. Left: Representative images from three independent experiments performed in triplicates are shown. Right: The number of cells with or without close proximity of aggregate and MTOC was counted (Mann – Whitney test, *P , 0.05 vs. cells with close proximity, n ¼ 3). (B) F-actin aggregates are neither lysosomal nor endosomal. Cells were stained via LAMP-1 (green) for lysosomes or Rab5 (green) for endosomes, respectively. Arrows indicate aggregates. (C) F-actin aggregates contain actin-related proteins. Cells were stained for F-actin (arrows indicate aggregates), nuclei, and the actin-binding proteins p21-arc or Cortactin (green), respectively. The overlap of actin with p21-arc or Cortactin is shown in yellow. For all stainings, representative images were chosen out of three independent experiments. Scale bars represent 20 mm.

proteins p21-Arc and Cortactin showed a clear colocalization with the aggregate (Figure 4C). These findings indicate that Ch induces aggresome formation in endothelial cells.

3.5 Ch reduces adhesion on collagen and the maturation of focal adhesions One major step in migration and maturation of tubes is the attachment to surfaces. Thus, the influence of Ch on the adhesion of endothelial cells to ECM proteins was tested. On collagen, a clearly decreased number of cells was able to adhere when treated with 100 nM Ch, whereas on fibronectin no change in cell number could be seen (Figure 5A). However, on both substrates, cells were not able to spread and showed a globular morphology at 100 nM (Figure 5B and C ). Additionally, half of the cells adhering on fibronectin showed an irregular shape meaning disturbed spreading (Figure 5B). So, obviously, Ch on the one hand reduces adhesion of cells to collagen and on the other hand also affects spreading on both matrices. To elucidate this in more detail, we analysed the influence of Ch on focal adhesions by visualizing vinculin in spreading cells (Figure 5D). The analysis of the number of mature focal adhesions per cells showed a significant reduction in Ch-treated cells. In analogy, the

vinculin-substrate connecting integrins, as well as associated actin structures, showed changes in their local distribution irrespective of the substrate (see Supplementary material online, Figure S5). 30 nM were used for stimulation in this case, as cells at 100 nM did not adhere or were completely rounded.

3.6 Ch affects integrin associated signalling by inhibiting Src activity Focal adhesions play a major role in migratory signalling. Since these were influenced by Ch, we analysed downstream signalling pathways. First, FACS analysis on total and active integrin b1 was performed (Figure 6A). Both, total integrin b1, and a high affinity (activated) conformation of integrin b120 were less present at the cell surface under Ch treatment hinting towards a trafficking problem but not a problem of integrin activation. Next, the major downstream complex consisting of FAK and Src was tested for its activity via the phosphorylation status of the respective components. The autophosphorylation site Y397 of FAK was unchanged under Ch treatment (Figure 6B, Supplementary material online, Figure S6 for quantification). However, the Src autophosphorylation site Y416 showed decreased phosphorylation after incubation with

310

M.H. Menhofer et al.

Figure 5 Chondramide reduces adhesion and spreading of endothelial cells on collagen and the maturation of focal adhesions. (A) Chondramide reduces adhesion on collagen. HUVECs were pretreated with ChB and seeded on fibronectin or collagen, respectively. After 30 min adhesion, cells were fixed, stained, and counted on four images per sample and experiment (Kruskal – Wallis test with Dunns test as post hoc, *P , 0.05 vs. DMSO, n ¼ 3). (B and C) Chondramide-treated cells cannot spread after attachment on fibronectin (B) and collagen G (C). Images (triplicates of three independent experiments) were quantified for normal spreading cells (normal), irregular shaped cells, showing non-symmetrical spreading (irregular), and completely rounded cells (globular). (Kruskal – Wallis test with Dunns test as post hoc, *P , 0.05 vs. DMSO, n ¼ 3). (D) Chondramide-treated cells show less mature focal adhesions. HUVECs were transfected with EGFP-Vinculin and pLifeAct-RFP and treated with Ch for 4 h. Cells were seeded freshly on collagen, allowed to spread for 1 h and images were taken. Representative images out of three independent experiments are shown. Bars represent 25 mm. For quantification 17 cells per condition were counted on their respective number of focal adhesions. (Mann – Whitney test, *P , 0.05 vs. DMSO).

Ch. Finally, the Src-dependent phosphorylation of the FAK (Y576/577) was tested and showed less phosphorylation after Ch treatment. These data indicate a reduction in the activity of the Src kinase upon Ch treatment. Inhibiting Src, using 10 mM Saracatenib, reduced the migration of endothelial cells down to 60% of the control (Figure 6C). At the same concentration, the autophosphorylation of Src at Y416 is reduced (see Supplementary material online, Figure S7) indicating its importance for migration in the setting chosen.

assays were performed (Figure 6D, Supplementary material online, Figure S6 for quantification). Both Rac1 and Rho showed reduced activity upon Ch treatment. Concomitant with that, the downstream target to Rho, MLC2, showed less phosphorylation at S19 as seen in western blot analysis (Figure 6E, Supplementary material online, Figure S6 for quantification) and fluorescent staining (see Supplementary material online, Figure S8). Thus, targeting the actin cytoskeleton has an effect on the upstream effectors of active Rac1 and Rho.

3.7 Ch reduces Rho-GTPase activity

3.8 Ch decreases angiogenesis in vivo

The FAK/Src complex regulates the activity of the Rho-GTPases Rac1 and Rho. To test whether their activity is reduced by Ch, pull-down

The in vivo effect of Ch on angiogenesis was investigated in two different settings: tumour angiogenesis in a murine xenograft model, and

Chondramide inhibits angiogenesis

311

Figure 6 Chondramide affects signalling pathways associated with integrins and Rho-GTPases activity. (A) Chondramide reduces total and active b1 integrin on the cell surface. HUVECs were pretreated with ChB (4 h) and harvested for FACS analysis (Kruskal – Wallis test with Dunns test as post hoc, *P , 0.05 vs. DMSO, n ¼ 3). (B) Chondramide has no effect on autophosphorylation of FAK at Y397 but reduces autophosphorylation of Src at Y416 and the Src-dependent phosphorylation of FAK Y576/577. HUVECs were pretreated with ChB (4 h, 30 nM), seeded freshly, and harvested for western blot analysis. Representative western blot images of three independent experiments are shown. (C) Specific inhibition of Src using Saracatenib reduces migration of HUVECs. Upper panel: Representative images of scratch assay are shown (three independent experiments performed in triplicate). The area covered with cells is displayed in green. Lower panel: All images were analysed on cell covered area. (Kruskal – Wallis test with Dunns test as post hoc, *P , 0.05 vs. DMSO, n ¼ 3). (D) Treatment with Chondramide (4 h, 30 nM) reduced the amount of active GTP-bound Rac1 and Rho (one representative blot each is shown for n ¼ 3 independent experiments). (E) Phosphorylation of MLC2 at S19 is reduced upon chondramide treatment. One representative western blot out of three independent experiments of control and Chondramide-treated HUVECs for pS19-MLC2 and tot-MLC2 is shown. GAPDH serves as loading control. For quantification of the western blots, see Supplementary material online, Figure S6.

developmental angiogenesis in the mouse retina. In the xenograft tumour model with MDA-MB-231 cells, Ch treatment has previously been shown to reduce tumour volume over time.21 Tumour tissues from this experimental series were used to analyse the degree of

vascularization of treated and untreated tumours by staining for CD31. The amount of microvessels per picture was reduced in Ch-treated tumours compared with control (Figure 7A, left panel). In good accordance with this reduced microvessel density, Hif1a positive

312

M.H. Menhofer et al.

Figure 7 Chondramide reduces angiogenesis in vivo. (A) Tumourangiogenesis. Ten mice per group with subcutaneous tumours were treated thrice a week with Chondramide or solvent i.p. for 35 days. (A, left panel) Explanted tumours were stained for CD31. Four images per tumour (×10 objective) were taken and vessels were counted. (40 images per group, n ¼ 10, Mann– Whitney test, *P , 0.05 vs. control). (A, right panel) Tumour sections were stained for Hif1a (red). Four images per tumour (×4 objective) were taken and the Hif-positive area was calculated relative to the tumour area (40 images per group, i.e. four per animal, n ¼ 10, Mann– Whitney test, *P , 0.05 vs. control). Scale bar: 100 mm. Co-staining of tumour sections with a vascular marker (CD31) and a marker for cell death (TUNEL) or proliferation (Ki67) revealed no substantial vascular cell death or proliferation either in controls or the treated animals (A, lower panel). Images representative for 40 images per group (10 animals each) are shown. (B) Chondramide affects angiogenesis in mouse retina. Three pups per group were injected with on Day P1, P2, and P3 with either 1 mg/kg ChB or DMSO equivalent. Pups were sacrificed on Day 4 and retinae (six per group) were prepared as described and stained for Isolectin B4. Four images per retina of the sprouting front were taken and the number of sprouts relative to the front line was counted and calculated; one image representative for the respective group is shown. (Mann – Whitney test, *P , 0.05 vs. control), the scale bar represents 100 mm.

Chondramide inhibits angiogenesis

areas (i.e. putatively hypoxic regions) were increased in Ch-treated tumours (Figure 7A, right panel). In the Hif staining, no differences concerning size of necrotic areas could be seen between treated and untreated animals (data not shown). Additionally, apoptotic (TUNEL) and proliferating (Ki67) cells were stained. In these samples no overlap with vessels could be seen either for control or for Ch-treated tumours (Figure 7Aiii). This indicates that the reduced microvessel density is not primarily due to higher apoptosis or reduced proliferation in the tumour vasculature at the time of analysis. To test the activity of Ch on angiogenesis in a physiological setting, the developing retina was used as a second in vivo model. Retinae from control and Ch-treated mice were stained for IB4 and analysed for the number of vascular sprouts per vessel length. This parameter was clearly decreased in the retinae of treated mice (Figure 7B). In contrast to the tube formation assay, no clear effect of Ch treatment on the localization of VE-Cadherin staining was observed (see Supplementary material online, Figure S9, upper panel), which might be due to a lower mechanical stability of tubes when compared with in vivo vessels. Currently, we can only speculate that this discrepancy could be due to a stabilization of vessels by mural cells (e.g. pericytes, smooth muscle cells) in vivo. The number of vascular branching points was not significantly reduced by Ch (see Supplementary material online, Figure S9, lower panel).

4. Discussion On first sight, the actin cytoskeleton seems an obvious target to inhibit angiogenesis, as actin-dependent migration of endothelial cells is a key step in the angiogenic cascade.22 However, to date only few investigations have been published targeting the actin cytoskeleton directly in angiogenesis. E.g. for the actin stabilizer Cytochalasin D in vitro studies were made showing an inhibition of cell growth, migration, and tube formation23 of endothelial cells as well as inhibition of angiogenesis in a chorioallantoic membrane (CAM) assay.24 Another actin-stabilizing drug is Latrunculin, whose derivative 15-O-methyllatrunculin B showed promising anti-angiogenic activity in a CAM assay.25 To our knowledge, the actin nucleating Jk has never been tested for an anti-angiogenic effect possibly due to high toxicity in vivo.26,27 Thus, actin-binding compounds have not been investigated in detail concerning their mechanism of action and potency in angiogenic therapy. Our study shows promising anti-angiogenic effects of Ch in vitro and in vivo. In fact, we could show a pharmacological reduction of tumour blood vessels in vivo at a non-toxic concentration and reduced sprout formation in retina angiogenesis. Furthermore, we could demonstrate that major hallmarks of angiogenesis, namely the proliferation, the migration, and the in vitro tube formation of endothelial cells, are inhibited by Ch in a low nanomolar range. The most obvious effect of Ch in endothelial cells which could be responsible for an anti-migratory effect is the aggregation of fibrilous actin. This is in line with another study applying Ch in Pt K2 potoroo cells13 and is also known for other actin-binding substances.19 These actin lumps induced by actin-binding compounds are called ‘aggresomes’, a cellular stress response to misfolded proteins.18 Criteria for aggresomes are the localization close to the MTOC and for actin toxins the inclusion of actin-binding proteins.28,29 These features are shared with the Ch-induced aggregates indentifying those as aggresomes. Thus, Ch induces aggresomes as a stress response in endothelial cells. The aggregation of actin by Ch is first seen at stress fibres and results in a disintegration of these bundles. This is in accordance with findings for Jk29 as well as fluorescence

313 images of fluorescently labelled Ch which preferentially stained stress fibres.30 The functionality of stress fibres, however, is important for further structures and signalling cascades in the cell. So, functional contractility from stress fibres is needed for the maturation of focal adhesions in endothelial cells.12 Thus, the disintegration of stress fibres could be the reason for the impaired maturation of focal adhesions observed after Ch treatment. In addition to this, the reduced amount of integrins at the cell surface may contribute to fewer focal adhesions. Focal adhesions represent a major platform for integrin outside-in signalling, which is important in spreading and migration.11 Consequently, the disturbance of these adhesive structures has a striking impact on important signalling pathways for migration. One central complex in this signalling cascade is built by the focal adhesion kinase (FAK) and Src. Under Ch treatment, the autophosphorylation of Src at Y416, needed for its own activation,31 and the Src-dependent phosphorylation of FAK at Y576/57732 are reduced. Both phosphorylation sites are an indication of reduced Src activity. Interestingly, it has been reported that the translocation of Src to focal adhesions is actin-dependent in swiss 3T3 cells.33,34 Thus, in addition to a reduced activation of Src due to less integrins, the translocation to the membrane and FAK could be impaired resulting in a reduced autophosphorylation. This could also explain why the autophosphorylation of Src but not the one of FAK is disturbed. Most importantly, also direct pharmacological inhibition of Src reduced the migration of endothelial cells. This Src-dependent inhibition underscores the importance of Src for endothelial cell migration. Corresponding to a reduced activity of Src and thus the whole complex FAK/Src, further downstream signalling would be expected to be diminished.11 Indeed, the downstream activated Rho-GTPase Rac1 is reduced in activity at same conditions as Src. Although the FAK/Src complex mediates a transient inhibition of Rho activity, a later Src-dependent Rho activation is discussed.11 This activation of Rho is supposed to initiate after the spreading phase due to mechanical stimuli and could be reduced under Ch treatment as Ch-treated cells showed defects in spreading and finally lead to the lower levels in Rho-GTP. Obviously, Ch has an overall inhibitory effect on the Rho-GTPases and thus on very important pro-migratory signals. Taken together, Ch probably inhibits migration in two interconnected ways: first, the reduction of contractility by disintegration of stress fibres itself, and second, as a downstream consequence of failing mechanical input, by the resulting reduction in pro-migratory signalling cascades, like Src and Rho-GTPases. The therapeutical relevance of this study could be shown in a xenograft tumour model. Here, colleagues showed a reduced tumour growth over time21 and we could now demonstrate a decreased vessel density in Ch-treated tumours. Further, the larger tissue areas showing nuclear presence of Hif1a in treated tumours indicate elevated hypoxia in Ch-treated tumours supporting the data of reduced vasculature in the treated tumours. Though the reduced vessel density does not seem to originate from enhanced apoptosis or reduced proliferation of endothelial cells at the time of analysis, it might well be that these processes have played a role at earlier stages of tumour growth in our in vivo model (especially in the light of the pronounced anti-proliferative effects of Ch in vitro). The results from the developing retinae also reflect the complex kinetics of vascular morphogenesis: on the one hand, the number of vascular branchings was only slightly reduced by Ch; on the other hand, the formation of vascular sprouts was clearly diminished. As the formation of sprouts is highly dependent on actin dynamics and endothelial cell migration, we suggest that this latter effect mainly results from anti-migratory effects of Ch, as already seen

314 in vitro. The lack of a clear effect on branching could be due to an interference of two factors: inhibition of migration, and hampered network maturation (as observed in tube formation in vitro). With our investigations, we provide insights into the mechanism how Ch diminishes the migration of endothelial cells, and thereby reduces angiogenesis in vivo. This work adds to our knowledge about actin polymerizing compounds for anti-angiogenic therapy. Most importantly, it supports the idea of targeting actin in anti-angiogenic therapies and reveals Ch as an interesting structure for further development.

Supplementary material Supplementary material is available at Cardiovascular Research online. Conflict of interest: none declared.

Funding This work was supported by a grant from the German Research Foundation (DFG FOR 1406).

References 1. Folkman J. Angiogenesis: an organizing principle for drug discovery? Nat Rev Drug Discov 2007;6:273 –286. 2. Ferrara N. VEGF as a therapeutic target in cancer. Oncology 2005;69(Suppl. 3):11 –16. 3. Gotink KJ, Verheul HM. Anti-angiogenic tyrosine kinase inhibitors: what is their mechanism of action? Angiogenesis 2010;13:1 –14. 4. Chen HX, Cleck JN. Adverse effects of anticancer agents that target the VEGF pathway. Nat Rev Clin Oncol 2009;6:465 –477. 5. Giuliano S, Pages G. Mechanisms of resistance to anti-angiogenesis therapies. Biochimie 2013;95:1110 –1119. 6. Bayless KJ, Johnson GA. Role of the cytoskeleton in formation and maintenance of angiogenic sprouts. J Vasc Res 2011;48:369 –385. 7. Schwartz EL. Antivascular actions of microtubule-binding drugs. Clin Cancer Res 2009;15: 2594 –2601. 8. Pollard TD, Cooper JA. Actin, a central player in cell shape and movement. Science 2009; 326:1208 –1212. 9. Lamalice L, Le Boeuf F, Huot J. Endothelial cell migration during angiogenesis. Circ Res 2007;100:782 –794. 10. Ridley AJ. Rho GTPases and cell migration. J Cell Sci 2001;114:2713 –2722. 11. Huveneers S, Danen EH. Adhesion signaling—crosstalk between integrins, Src and Rho. J Cell Sci 2009;122:1059 –1069. 12. Wolfenson H, Henis YI, Geiger B, Bershadsky AD. The heel and toe of the cell’s foot: a multifaceted approach for understanding the structure and dynamics of focal adhesions. Cell Motil Cytoskeleton 2009;66:1017 – 1029. 13. Sasse F, Kunze B, Gronewold TM, Reichenbach H. The chondramides: cytostatic agents from myxobacteria acting on the actin cytoskeleton. J Natl Cancer Inst 1998;90:1559–1563. 14. Herrmann J, Huttel S, Muller R. Discovery and biological activity of new chondramides from Chondromyces sp. Chembiochem 2013;14:1573 –1580.

M.H. Menhofer et al.

15. Nicoletti I, Migliorati G, Pagliacci MC, Grignani F, Riccardi C. A rapid and simple method for measuring thymocyte apoptosis by propidium iodide staining and flow cytometry. J Immunol Methods 1991;139:271 – 279. 16. Liebl J, Weitensteiner SB, Vereb G, Taka´cs L, Fu¨rst R, Vollmar AM, Zahler S. Cyclindependent kinase 5 regulates endothelial cell migration and angiogenesis. J Biol Chem 2010;285:35932 –35943. 17. Pitulescu ME, Schmidt I, Benedito R, Adams RH. Inducible gene targeting in the neonatal vasculature and analysis of retinal angiogenesis in mice. Nat Protocol 2010;5:1518 –1534. 18. Johnston JA, Ward CL, Kopito RR. Aggresomes: a cellular response to misfolded proteins. J Cell Biol 1998;143:1883 –1898. 19. Kazami S, Usui T, Osada H. Actin stress fiber retraction and aggresome formation is a common cellular response to actin toxins. Biosci Biotechnol Biochem 2011;75:1853 –1855. 20. Mould AP, Garratt AN, Askari JA, Akiyama SK, Humphries MJ. Identification of a novel anti-integrin monoclonal antibody that recognises a ligand-induced binding site epitope on the beta 1 subunit. FEBS Lett 1995;363:118 –122. 21. Foerster F, Braig S, Moser C, Kubisch R, Busse J, Wagner E, Schmoeckel E, Mayr D, Schmitt S, Huettel S, Zischka H, Mueller R, Vollmar AM. Targeting the actin cytoskeleton: selective anti-tumor action via trapping PKCepsilon. Cell Death and Disease 2014;5: e1398. 22. Thoenes L, Gunther M. Novel approaches in anti-angiogenic treatment targeting endothelial F-actin: a new anti-angiogenic strategy? Curr Opin Mol Ther 2008;10:579 –590. 23. Mabeta P, Pepper MS. A comparative study on the anti-angiogenic effects of DNAdamaging and cytoskeletal-disrupting agents. Angiogenesis 2009;12:81–90. 24. Melkonian G, Munoz N, Chung J, Tong C, Marr R, Talbot P. Capillary plexus development in the day five to day six chick chorioallantoic membrane is inhibited by cytochalasin D and suramin. J Exp Zool 2002;292:241–254. 25. El Sayed KA, Youssef DT, Marchetti D. Bioactive natural and semisynthetic latrunculins. J Nat Prod 2006;69:219–223. 26. Scott VR, Boehme R, Matthews TR. New class of antifungal agents: jasplakinolide, a cyclodepsipeptide from the marine sponge, Jaspis species. Antimicrob Agents Chemother 1988; 32:1154 –1157. 27. Schweikart K, Guo L, Shuler Z, Abrams R, Chiao ET, Kolaja KL, Davis M. The effects of jaspamide on human cardiomyocyte function and cardiac ion channel activity. Toxicol In Vitro 2013;27:745 –751. 28. Garcia-Mata R, Bebok Z, Sorscher EJ, Sztul ES. Characterization and dynamics of aggresome formation by a cytosolic GFP-chimera. J Cell Biol 1999;146:1239 –1254. 29. La´zaro-Die´guez F, Aguado C, Mato E, Sa´nchez-Ruı´z Y, Esteban I, Alberch J, Knecht E, Egea G. Dynamics of an F-actin aggresome generated by the actin-stabilizing toxin jasplakinolide. J Cell Sci 2008;121:1415 – 1425. 30. Milroy LG, Rizzo S, Calderon A, Ellinger B, Erdmann S, Mondry J, Verveer P, Bastiaens P, Waldmann H, Dehmelt L, Arndt HD. Selective chemical imaging of static actin in live cells. J Am Chem Soc 2012;134:8480 –8486. 31. Roskoski R Jr. Src kinase regulation by phosphorylation and dephosphorylation. Biochem Biophys Res Commun 2005;331:1–14. 32. Schlaepfer DD, Hauck CR, Sieg DJ. Signaling through focal adhesion kinase. Prog Biophys Mol Biol 1999;71:435 –478. 33. Fincham VJ, Unlu M, Brunton VG, Pitts JD, Wyke JA, Frame MC. Translocation of Src kinase to the cell periphery is mediated by the actin cytoskeleton under the control of the Rho family of small G proteins. J Cell Biol 1996;135:1551 –1564. 34. Timpson P, Jones GE, Frame MC, Brunton VG. Coordination of cell polarization and migration by the Rho family GTPases requires Src tyrosine kinase activity. Curr Biol 2001;11: 1836– 1846.