Anti‐angiogenic effects of the blue‐green alga Arthrospira platensis on pancreatic cancer

Abstract Arthrospira platensis, a blue‐green alga, is a popular nutraceutical substance having potent antioxidant properties with potential anti‐carcinogenic activities. The aim of our study was to assess the possible anti‐angiogenic effects of A platensis in an experimental model of pancreatic cancer. The effects of an A platensis extract were investigated on human pancreatic cancer cells (PA‐TU‐8902) and immortalized endothelial‐like cells (Ea.hy926). PA‐TU‐8902 pancreatic tumours xenografted to athymic mice were also examined. In vitro migration and invasiveness assays were performed on the tested cells. Multiple angiogenic factors and signalling pathways were analysed in the epithelial, endothelial and cancer cells, and tumour tissue. The A platensis extract exerted inhibitory effects on both migration and invasion of pancreatic cancer as well as endothelial‐like cells. Tumours of mice treated with A platensis exhibited much lesser degrees of vascularization as measured by CD31 immunostaining (P = .004). Surprisingly, the VEGF‐A mRNA and protein expressions were up‐regulated in pancreatic cancer cells. A platensis inhibited ERK activation upstream of Raf and suppressed the expression of ERK‐regulated proteins. Treatment of pancreatic cancer with A platensis was associated with suppressive effects on migration and invasiveness with various anti‐angiogenic features, which might account for the anticancer effects of this blue‐green alga.

forms a molecule called C-phycocyanin exerting potent biological effects, including anticancer activities. 11 However, molecular mechanisms accounting for these antiproliferative effects are largely unknown.
We recently reported the potent suppressive effects of A platensis on the growth and proliferation of experimental pancreatic cancer. 4 The RAS-regulated RAF-MEK1/2-ERK1/2 pathway, with possible impacts on angiogenesis in the cancer tissue, 12,13 is dysfunctional in pancreatic cancer. 14,15 In fact, anti-angiogenic therapeutic approach targeting the vascular endothelial growth factor (VEGF) or the epidermal growth factor receptor (EGFR) signalling has become a promising strategy in the treatment of pancreatic cancer 16,17 with the aim to modulate protein kinase B (AKT) and extracellular signal-regulated kinase (ERK) (pAKT and p-ERK) pathways dysregulated in these cancers. 18 Thus, the aim of this current study was to evaluate the possible anti-angiogenic effects of A platensis to account for the antiproliferative effects of this alga.

| Materials
The A platensis was purchased from Martin Bauer GmbH (Vestenbergsgreuth, Germany). The water extract of both A platensis and phycocyanobilin was prepared as has been previously described elsewhere. 4 The cell culture media and non-essential amino acids (NEAAs) were obtained from Sigma-Aldrich, and the other cell culture components were from Biosera (Nuaille, France). The serine/ threonine phosphatase and protease inhibitor cocktails were purchased from either Sigma-Aldrich or Serva. The Geltrex™ LDEV-Free Reduced Growth Factor Basement Membrane Matrix was purchased from Thermo Fisher Scientific. The recombinant growth factors and inhibitors were procured as follows: rVEGF, rEGF (epidermal growth factor), rAREG (amphiregulin, autocrine mitogen related to EGF), rHGF/ SF (hepatocyte growth factor/scatter factor), PD 0325901 (all from Sigma-Aldrich), erlotinib (Cell Signaling Technology), vatalanib and axitinib (Selleck Chemicals) and bevacizumab (LGM Pharma). Unless otherwise specified, all other common chemicals were from Sigma-Aldrich.

| Tumour tissue from in vivo experiments
Pancreatic cancer xenografts (PA-TU-8902 cells) from our previous study on mice treated with biologically relevant doses of A platensis extract 4 were used for the Western blot, immunohistochemical staining, angiogenic proteome and mRNA expression analyses. In these studies, tumour sizes were significantly smaller as early as the third day after initiation of the A platensis extract treatment reaching only 40% of the size of untreated animals in 2 weeks of treatment. 4 The mice were killed after 2 weeks of intragastric administration of a water suspension of freeze-dried A platensis (0.5 g/kg once daily); after, the tumour tissue specimens were sampled and stored at −80°C until analysed.

| Cell viability assays
The effect of growth factors (VEGF; EGF; AREG at concentrations of 0.1, 1, 10, 50, 100 μg/L) on the viability of PA-TU-8902 pancreatic cancer and EA.hy926 endothelial-like cells was measured by a MTT viability assay.

| Tube-like formation assay
Immortalized EA.hy926 cells that retain several endothelial characteristics were used to determine the effect of A platensis on angiogenesis. These EA.hy926 endothelial-like cells (2.5 × 10 4 cells per well) pretreated with a water extract of A platensis (0.3 g/L) for 24 hours were seeded in a 96-well plate covered with a Geltrex™ basement membrane matrix, with reduced growth factors in DMEM supplemented with 0.5% serum in either the presence or absence of a water extract of A platensis. The formation of tube-like structures was inspected, and photographs were taken after 24 hours using an Olympus TL4 microscope (Shinjuku, Tokyo, Japan). The images were analysed by the Angiogenesis Analyzer tool in ImageJ software (NIH), and the total length of the tube-like structures was calculated as the length of the tubes relative to the control cells.

| Wound-healing assay
PA-TU-8902 or EA.hy926 cells were seeded into 12-well plates at a density of 2 × 10 5 cells/well in complete DMEM and cultured to 100% confluence. The cells were starved in a low-serum medium 6 hours prior to the experiment. The confluent cell monolayer was then scratched with a pipette tip and washed three times with Hank's solution to remove cell debris. The cells were incubated at 37°C for 24 hours in a low-serum medium (DMEM supplemented with 0.5% serum) with a water extract of A platensis. To quantify cell migration, images of the wound were taken 0 and 24 hours after being scratched (Olympus TL4 microscope, Tokyo, Japan), and the images were then processed with ImageJ software. The rate of cell migration was then calculated as the area filled by cells migrating into the denuded area (in square pixels) and plotted as the average from at least 3 independent experiments in quadruplicate.

| Cell migration and invasion assay
Cultrex ® in vitro angiogenesis and endothelial cell invasion assays (Trevigen) were used for the cell migration and invasiveness studies.
The EA.hy926 cells were first starved in a low-serum medium for 6 hours and then pretreated with a low-serum medium for the next 6 hours with the water extract of A platensis (0.3 g/L). The cells were harvested, re-suspended in a serum-free medium with the water extract of A platensis (0.3 g/L) and then seeded onto the upper insert of a 96-well plate (2 × 10 4 cells per well) prepared according to the manufacturer's instructions. The membranes of the wells on the upper insert plate were covered with the basement extracellular membrane extract depending on whether being tested for invasion or migration. Cells that penetrated across the membrane were detected after 24 hours by measurement of the fluorescence signal (excitation 485 nm, emission 520 nm) after calcein-AM internalization using a microplate reader (Infinite ® 200, Tecan).

| Western blot analyses
Samples were separated by SDS-PAGE, blotted onto a nitrocellulose or PVDF membrane, blocked for 1 hours with 5% non-fat dry milk in TBS-Tween (0.05%-0.1%) or in PBS-Tween and then probed with the corresponding primary and secondary antibodies (Table S1). The chemiluminescent signal of horseradish peroxidase (HRP)-conjugated antibodies was detected on film (CD31, VEGF-A) or digitally processed by a Fusion For CD31/VEGF-A, using the Western blot analyses from pancreatic cancer xenografts, the excised tissue was homogenized in a RIPA lysis buffer. The chemiluminescent process and quantification of the immunoreactive bands on the exposed films were carried out as have previously been described. 20 For the EGF receptor (EGFR) protein expression, PA-TU-8902 cells were incubated for 24 hours in a medium with or without A platensis. Ten minutes before cell lysis in RIPA buffer, supplemented with phosphatases and proteases inhibitors, rEGF or rAREG was added at a final concentration of 50 ng/mL. Quantification and statistical analyses of the Western blots were performed from at least three independent experiments run in triplicate.

| Immunohistochemistry
Cryosections of tumour tissue were fixed in acetone at −20°C for 30 minutes. Slides were blocked with 10% normal goat serum

| Quantitative real-time PCR
Total RNA was extracted using a PerfectPure™ RNA Cell Isolation Kit (5 PRIME GmbH, Hilden, Germany). The cDNA was prepared using a High-Capacity cDNA Reverse Transcription Kit (Applied Biosystems). The cDNA was mixed with SYBR GREEN master mix (Applied Biosystems), plus with specific primers for genes of interest (Table S2), and then examined by RT-PCR on a ViiA™ cycler (Applied Biosystems).

| VEGF-A, AREG and EGF protein determination
For measuring the effect of A platensis extract on VEGF-A, AREG and EGF production, PA-TU-8902 cell media were collected after 24-hour incubation, with or without water extract of A platensis

| Statistical analyses
The statistical significance of the differences between variables was evaluated by t test or Mann-Whitney rank-sum test. Differences between multiple groups were assessed by ANOVA or Kruskal-Wallis rank-sum test with post hoc Dunn's test. Depending on their normality, data are presented as the mean ± SD, or the median and IQ range. Differences were considered statistically significant when the p-values were <.05.

For comparison of the effects of A platensis in cells stimulated
with either HGF/SF or 4HT ( Figure 6), data from two independent experiments were analysed, each performed in duplicates. Twoway ANOVA was performed to assess an additive batch effect between the experiments. Twofold changes in ERK phosphorylation with false discovery rate (FDR) <0.05 were considered statistically significant. Batch-corrected values were plotted ( Figure 6), with the median of the controls set to zero. All analyses were performed in R.

| Arthrospira platensis inhibits migration of both pancreatic cancer and endothelial cells
The migration capacity of cancer cells is an important factor for tumour progression. Thus, we assessed the effect of A platensis with regard to this phenomenon. We observed that the A platensis water extract (0.3 g/L) significantly suppressed wound healing in a scratch assay performed on PA-TU-8902 pancreatic cancer cells (P < .001, Figure 1A), as well as (to a lesser extent) also on EA.hy926 endothelial-like cells ( Figure 1B). The potential of A platensis extract to inhibit cell migration of EA.hy926 cells was also confirmed in the Boyden chamber transwell migration assay as well as in the invasion assay. Migration was significantly suppressed (by 47%, P < .001, Figure 1C); also, the drop in invasiveness of the cells was remarkable and reached borderline significance (P = .062, Figure 1C).

| Arthrospira platensis inhibits angiogenesis in vitro
As tumour progression is dependent upon angiogenesis, we investigated the possible role of A platensis on the functional capacity of endothelial cells. Treatment with the A platensis extract (0.3 g/L) significantly reduced the total length of the tube-like structures of EA.hy926 cells (to approximately 50% that of the control cells, P < .05, Figure 1D); although, as mentioned above, the viability of the endothelial cells was not compromised.

| Arthrospira platensis decreases vascularization in human pancreatic cancer xenografts
Due to the beneficial effects of A platensis on angiogenesis observed in the in vitro assays, we focused on an evaluation of the vascularization of tumours, as measured by the expression of endothelial marker CD31. 21 Indeed, as compared to the controls, the expression of CD31 was more than 2X lower in the tumours of animals treated with A platensis (45% of controls, P < .01, Figure 2A).

| Arthrospira platensis stimulates VEGF-A production in human PA-TU-8902 pancreatic cancer cells
Although angiogenesis is a complex process, being regulated and modified by multiple factors, VEGF-A is considered the key player. 22 However, despite the significantly lower expression of the CD31 capillary marker in the tumour mass of animals treated with A platensis (Figure 2A), the VEGF-A protein expression was surprisingly higher in tumour tissues, reaching borderline significance (by 48%, P = .056, Figure 2B). Immunohistochemical staining of tumours These results were confirmed in the in vitro experiment with the PA-TU-8902 cells exposed to A platensis (0.3 g/L). The VEGFA mRNA expression was significantly up-regulated after both 1 and 24 hours of A platensis exposure, by 50% and 49%, respectively (P < .05 for both comparisons, Figure 3A). This overexpression was reflected by increased VEGF-A secretion after 24 hours of exposure to A platensis extract (by 47%, P < .01, Figure 3B). A similar stimulatory effect on VEGF-A production after just 5 hours of exposure was not only observed for A platensis, but also for phycocyanobilin and vatalanib (a selective VEGF receptor (VEGFR) inhibitor), while no effect was found for axitinib (a less specific tyrosine kinase inhibitor) ( Figure 3C).
The effect of A platensis extract on VEGF-A production by PA-TU-8902 cells was long-standing, as increased VEGF-A production persisted for as long as 12 hours after exchanging the culture media with one not containing A platensis extract (data not shown).
To assess whether up-regulated VEGF-A production in PA-TU-8902 cells was also associated with changes in mRNA expressions of VEGFR, we analysed both VEGFR1 and VEGFR2 mRNA levels on exposure of these cancer cells to the A platensis extract. However, no changes in VEGFR1 or VEGFR2 were detected regardless of the exposure status, suggesting that VEGF-A overproduction by PA-TU-8902 cells exposed to A platensis extract did not have any positive autocrine effect on VEGFR expression. Similarly, no change in VEGFR1 protein expression was found in the tumour tissue excised from the mice treated A platensis extract, as well as in EA.hy926 endothelial-like cells exposed to A platensis extract (data not shown).
We also determined whether VEGF-A overproduced upon exposure to A platensis can have paracrine effects on the proliferation of pancreatic cancer cells. To do this, we tested the viability of PA-TU-8902 cells exposed to recombinant VEGF-A (within a concentration range of 10-100 ng/mL). However, no changes in the cell viability/proliferation status were observed at any of the concentrations used (data not shown).  (Table 1b). Among them, VEGF-A and AREG were also present, confirming our results from the xenograft tumours. We also detected EGF, an additional ligand of EGFR. As VEGF and EGF signalling pathways are inter-related in pancreatic carcinogenesis, 23 and AREG is an important prognostic factor of pancreatic cancer, 24 we next focused on the possible role of the AREG pathway in the A platensis-mediated therapeutic effects.

| The role of EGF/AREG in A platensismediated therapeutic effects
The expression of mRNA of AREG (an EGFR ligand) in PA-TU-8902 cells treated with A platensis extract was up-regulated to 205% ( Figure 4A, P < .05). In line with this result, increased production of the AREG protein in treated pancreatic cancer cells was also observed, and this trend was consistent even when A platensis extract was co-administered with the majority of other angiogenesis modulating compounds ( Figure 4B). Recombinant VEGF-A or bevacizumab did not influence the level of AREG in the control cells; however, a significant increase in AREG was observed in A platensis-treated cells. On the other hand, erlotinib (an EGFR inhibitor) decreased the AREG level in both the control cells (by 41%, P < .05) and the A platensis-treated cells (by 36%, P < .05). Simultaneously, recombinant EGF markedly increased the AREG level in the control cells (to 142%, P < .01); and an additional exposure to A platensis further increased its production ( Figure 4B). Interestingly, no EGF was detected in the medium of control cells or in cells treated with A platensis extract.
In contrast to both VEGF-A and AREG having no effect, recombinant EGF added to the culture medium consistently increased proliferation of PA-TU-8902 cells ( Figure 4C). Interestingly, treatment with A platensis extract (0.3 g/L) eliminated this effect of EGF ( Figure 4C).
Exposure of PA-TU-8902 cells to exogenous EGF (50 mg/L) did not have any effect on either VEGF-A mRNA or protein expression.
A platensis extract did not affect EGFR protein expression in PA-TU-8902 cells after 24 hours ( Figure 4D). In contrast, EGF treatment led to down-regulation of EGFR protein expression, whereas AREG had the opposite effect (this effect was co-stimulated by A platensis extract, Figure 4D).

| Arthrospira platensis interferes with ERK activation and potentiates AKT activation
In pancreatic cancers, KRAS is almost invariably mutated, where it activates downstream ERK and PI3K/AKT signalling pathways to promote proliferation, survival and invasion of cancer cells. 25,26 Due to the key roles of the ERK signalling pathway in tumour cell progression, the effects of A platensis on ERK phosphorylation were also analysed in PA-TU-8902 cells that harbour Ras (G12V) activating mutation. One-hour exposure of PA-TU-8902 pancreatic cancer cells to A platensis extract led to a significant decrease in ERK phosphorylation ( Figure 5A,B).
Similar effect of A platensis extract on ERK inhibition was also observed in PANC-1 and MiaPaCa-2 pancreatic cancer cell lines, while no inhibitory effect was seen in BxPC-3 cells ( Figure 5E). Surprisingly, exposure of PA-TU-8902 pancreatic cancer cells to A platensis extract led to a slight but reproducible increase in phosphorylation of AKT ( Figure 5C). The up-regulation of PI3K activity and AKT phosphorylation could be a consequence of the inhibition of the ERK pathway. 27 We thus examined whether pharmacological inhibition of the ERK pathway activates AKT in PA-TU-8902 cells. Indeed, we found that the cells treated with MEK inhibitor PD0325901 increased the phosphorylation of AKT ( Figure 5D). These data indicate that A platensis extract induces AKT phosphorylation at least partly through the negative crosstalk with the ERK pathway and possibly also independently of ERK by so far unknown mechanism.

| Arthrospira platensis inhibits ERK activity upstream of Raf and affects the expression of ERKregulated proteins
ERK activation is in direct association with high malignant potential of pancreatic cancer cell invasion of cancer cells. 25 To gain further insight into the molecular mechanisms of an inhibitory function of A platensis, we examined whether A platensis was able to modulate this signalling system in response to different stimuli. We utilized epithelial MDCK cells stably expressing a conditional active version of Raf (ΔRaf-1:ER). 19 In these cells, sustained ERK activity can be induced by two different means. ERK can be activated by extracellular HGF/SF via its cognate receptor, c-Met (c-mesenchymal-epithelial transition factor). In fact, the HGF/Met signalling pathway is importantly involved in pathogenesis of pancreatic cancer, and HGF/Met inhibitors are believed to become in near future crucial therapeutics for pancreatic cancer management. 28 Importantly, HGF/SF then activates the signalling components, including Ras, that are required for the signal transduction from the c-Met receptor towards the ERK pathway ( Figure 6A). ERK can also be activated more directly by 4HT, which converts ΔRaf-1:ER protein to an active form, and then subsequently activates the ERK pathway in a Ras-independent manner ( Figure 6A). Although this model is not related to pancreatic cancer, examining the effect of A platensis on HGF/c-Met-Ras and Raf-mediated ERK activation could help to assess the mechanisms of A platensis impact on the ERK pathway.
When the cells were treated with HGF/SF, A platensis inhibited both acute and sustained ERK activation ( Figure 6B and Figure S1).

| D ISCUSS I ON
Despite the medical progress achieved over the last few decades, cancer still poses a major threat to worldwide public health, with increasing incidence rates in most countries. 30 Diet is among the most important factors affecting the risks of cancer. 31 Besides the role of the composition of macro-nutrients, specific food components have been shown to contribute substantially to the prevention of carcinogenesis, modulating various stages in this process including apoptosis, angiogenesis and metastasis. 32 In fact, angiogenesis plays a crucial role in tumour progression, making it an attractive target for novel cancer therapies. 12 In our study, we have explored the potential anti-angiogenic activities of an extract of A platensis, a freshwater blue-green alga commonly used as a nutraceutical. 33 Indeed, the apparent antiproliferative effects of A platensis seen in our previous studies 4 might at least partially be due to the lowered vascularization of the pancreatic tumours, as evidenced by decreased expression of CD31, together with a lowered invasion potential of the treated cells. Surprisingly, this phenomenon was accompanied by an increased production of VEGF-A and other angiogenic factors. Tumour angiogenesis is a complex process, which involves a highly regulated orchestration of multiple signalling pathways, and whose impairment certainly evokes an array of feedback mechanisms, as can clearly be documented by the differential expressions of the multiple angiogenic factors seen in our study.
The relationship between CD31 in tumours and VEGF-A expressions is not obvious. Although a positive (and expected) association was reported in one study on patients with pituitary adenomas, 21 this has not been confirmed by others in the same tumours, 34 nor in osteosarcoma 35 and/or in breast cancer. 36,37 On the other hand, VEGF-A overexpression is a potent prognostic marker in patients with pancreatic cancer, indicating increased malignancy of the tumours 38 ; further, the same association as with the prognosis of pancreatic cancer has also been reported for AREG. 24 We have shown proof of an increased production of VEGF-A in several independent experiments, both in vivo and in vitro, as well as at both the mRNA and the protein level. The same up-regulation was also observed for other angiogenic factors, including AREG acting as a ligand for EGFR. This seems to be important, as the EGFR pathway plays an important F I G U R E 3 VEGF-A production by human PA-TU-8902 pancreatic cancer cells exposed to Arthrospira platensis and various anti-angiogenic drugs: (A) VEGF-A mRNA expression after 1-and 24-h exposure to A platensis extract. B, VEGF-A protein production upon 24-h exposure to A platensis extract. C, VEGF-A protein production upon 5-h exposure to various anti-angiogenic drugs. The VEGF-A mRNA level was determined by RT-qPCR. The concentrations of VEGF-A secreted into the medium of PA-TU-8902 cells were detected by ELISA. The following concentrations of anti-angiogenic factors were used: A platensis extract (0.3 g/L), PCB (125 μmol/L), axitinib (50 nmol/L) and vatalanib (1 μmol/L). Data expressed as % of control values and represent median ± IQ range. p P = .054 vs control; *P < .05 vs control; **P < .01 vs control. A pl., A platensis; PCB, phycocyanobilin role in Ras-mutated pancreatic cancers, with direct impacts on the ERK and AKT signalling pathways. 18 One could logically speculate that the overexpression of VEGF-A observed in our study was due to a positive feedback mechanism caused by VEGFR inhibition. In fact, an untouched VEGF-A production was observed with vatalanib, a selective VEGFR inhibitor, in both experimental and clinical studies on various cancers including pancreatic carcinoma. [39][40][41] In our study, vatalanib treatment TA B L E 1 The effect of Arthrospira platensis extract on the expression of angiogenic proteins in human PA-TU-8902 pancreatic cancer xenografts (a) and in vitro cultured cells (b)  reported for AREG in human chondrosarcoma cells. 45 Notably, AREG was also up-regulated in our A platensis-treated pancreatic cancer cells.
One of the anticancer mechanisms of A platensis treatment seems to be inhibition of the ERK signalling pathway. The aberrant activation of the ERK signalling pathway, composed of the Raf, MEK and ERK protein kinases, is directly associated with the high malignancy potential of numerous cancers, including pancreatic ductal carcinomas. 25 The ERK pathway is prototypically activated by small GTPase Ras, which directly associates with RAF, and promotes its activation. 46 Ras mutational activation is also the main oncogenic signalling pathway in pancreatic cancer, 47 and the PA-TU-8902 pancreatic cancer cells used in our study also contain the Ras(G12V) oncogenic mutation. 48 Our results from the MDCK cell line expressing conditionally active Raf deletion mutant show that A platensis inhibits ERK activation by Ras(G12V), but not Raf. In addition, we found that A platensis extract does not affect ERK activity in BxPC3 pancreatic cancer cells that contain constitutively active B-RAF due to in-frame deletion. 49 As active Ras promotes Raf activation by direct The elevated expression of VEGF-A and amphiregulin, together with inhibition of the mitogenic RAF-MEK-ERK signalling cascades seen upon treatment with A platensis extract, was surprising, as ERK signalling is generally believed to induce VEGF-A production. 56 We speculate that there are pathways parallel to ERK that promote VEGF-A expression. The primary candidate for the pathway acting parallel to ERK is the PI3K-AKT pathway. 57 In accord with our results, it has been shown that the inhibition of the ERK pathway resulted in up-regulation of PI3K activity and

DATA AVA I L A B I L I T Y S TAT E M E N T
All data generated or analysed during this study are included in this published article (and its supplementary information files).