Selective Loss of TGFβ Smad-Dependent Signalling Prevents Cell Cycle Arrest and Promotes Invasion in Oesophageal Adenocarcinoma Cell Lines

In cancer, Transforming Growth Factor β (TGFβ) increases proliferation and promotes invasion via selective loss of signalling pathways. Oesophageal adenocarcinoma arises from Barrett's oesophagus, progresses rapidly and is usually fatal. The contribution of perturbed TGFβ signalling in the promotion of metastasis in this disease has not been elucidated. We therefore investigated the role of TGFβ in Barrett's associated oesophageal adenocarcinoma using a panel of cell lines (OE33, TE7, SEG, BIC, FLO). 4/5 adenocarcinoma cell lines failed to cell cycle arrest, down-regulate c-Myc or induce p21 in response to TGFβ, and modulation of a Smad3/4 specific promoter was inhibited. These hyperproliferative adenocarcinoma cell lines displayed a TGFβ induced increase in the expression of the extracellular matrix degrading proteinases, urokinase-type plasminogen activator (uPA) and plasminogen activator inhibitor 1 (PAI-1), which correlated with an invasive cell phenotype as measured by in vitro migration, invasion and cell scattering assays. Inhibiting ERK and JNK pathways significantly reduced PAI and uPA induction and inhibited the invasive cell phenotype. These results suggest that TGFβ Smad-dependent signalling is perturbed in Barrett's carcinogenesis, resulting in failure of growth-arrest. However, TGFβ can promote PAI and uPA expression and invasion through MAPK pathways. These data would support a dual role for TGFβ in oesophageal adenocarcinoma.


INTRODUCTION
Transforming Growth Factor b (TGFb), and the components of its signal transduction pathway, are known to demonstrate tumour suppressor activity. In many human cancers, there are inactivating mutations in components of the TGFb pathway resulting in uncontrolled proliferation. In addition, through its actions on both the tumour cells and the surrounding stromal cells, TGFb can enhance invasion, motility and metastasis. TGFb can thus play a dual role in the initiation and malignant progression of human cancer [1,2].
TGFb signalling occurs through activation of type I and II trans-membrane serine/threonine kinase receptors, which leads to the phosphorylation of Smad 2 and 3. This complex, in association with Smad 4 translocates to the nucleus resulting in transcriptional activation of downstream targets [3]. The anti-proliferative effects of TGFb in normal epithelial cells are achieved through the inactivation of cyclin dependent kinases (cdk) 2, 4 and 6, which lead to cell cycle arrest. This is mediated by the rapid induction of the cdk4/6 inhibitor p15 INK4B , and the cdk2 inhibitor p21 CIP1/WAF1 . Another key event in TGFb mediated cell cycle arrest involves the repression of c-Myc. This transcription factor binds the promoters of p15 INK4B and p21 CIP1/WAF1 and suppresses their expression [4]. Rapid transcriptional down-regulation of c-Myc by TGFb relieves this repression, allowing binding of the heteromeric Smad complex [5]. The net effect of this is to halt the cell cycle in G 1 , and thus potently inhibit proliferation [6].
TGFb is frequently overexpressed by epithelial cancer cells which become unresponsive to its anti-proliferative effects and this leads to paracrine stimulation of stromal cells within the tumour microenvironment [6]. As a result there is stimulation of angiogenesis and upregulation of extracellular matrix (ECM) degrading proteinases. Expression of the urokinase-type plasminogen activator (uPA) is upregulated by TGFb [7], which through the formation of plasmin enables the tumour cell to penetrate the basement membrane. Overexpression of uPA has been observed in invasive malignancies of the breast, colon and stomach [8]. The uPA inhibitor plasminogen activator inhibitor 1 (PAI-1) is another TGFb target gene that is up-regulated in advanced cancers [8]. The paradox of an elevated expression level of PAI promoting tumour invasion is partially explained by the observation that this inhibitor can promote cell migration and angiogenesis, independent of the effects on plasmin activity [9,10].
In contrast to the plethora of information about mutations in the TGFb signal cascade in colon, gastric and pancreatic cancers [2,[11][12][13], the contribution of perturbed TGFb signalling in oesophageal adenocarcinoma has not been elucidated completely. Oesophageal adenocarcinoma usually arises through a multi-step sequence from Barrett's metaplasia to dysplasia and carcinoma [14]. Recent in vivo and in vitro evidence suggests that inactivation of SMAD4, lack of RUNX3 and inability to degrade SnoN may be some of the causes for the unresponsiveness of oesophageal cancer to TGFb anti-proliferative effect [15][16][17]. This therefore begs the question whether TGFb can promote invasion in this disease in the context of deranged SMAD4 signalling.
The signalling pathways by which TGFb exerts its effects on migration and invasion are gradually being elucidated. The activated TGFb receptor complex is now known to activate kinase pathways independently of Smad signalling and these have been shown to stimulate the expression of both uPA and PAI [18][19][20]. These pathways include the phosphoinositol-3 kinase (PI3K) [9], and the mitogen activated protein kinases (MAPK) particularly extracellular signal regulated kinase (ERK), [21] c-jun N-terminal kinase (JNK) [22], and p38 [23].
The specific aims of this study were therefore: 1). To determine whether TGFb can simultaneously affect proliferation (cell cycle, c-Myc and p21 expression) and invasion indices (ECM proteinase expression and functional invasive characteristics) in oesophageal adenocarcinoma cell lines; 2). To elucidate the cell signalling pathways involved in these responses.

Lack of anti-proliferative response to TGFb
First, the effect of TGFb on cell cycle progression and expression of cell cycle associated genes, p21 and c-myc, was assessed. The control experimental conditions were cells treated with serum free media (SF), which is known to induce cell cycle arrest, and cells treated with SF media followed by release into the cell cycle by complete media (C), ( Figure 1A and 1B). All cells maintained in serum free media for 24 h demonstrated an increase in the proportion of cells in G0/G1 ( Figure 1A, 1B). The cells were also arrested by stimulation with UV light as determined by a BrdU incorporation assay (data not shown). However, when cultured in the presence of TGFb (T) only 1/5 BE adenocarcinoma cell lines (OE33), exhibited an increase in the fraction of cells in G0/G1 (11% OE33 p,0.05), (Fig 1A, 1B). Since the inhibition of proliferation in response to TGFb would be expected to occur as a result of an up-regulation of p21 CIP/WAF1 and a decreased expression of c-Myc, the effect of TGFb on these mRNA expression levels was examined. Real time PCR demonstrated that only OE33 cells exhibited a significant increase in p21 and decrease in c-Myc expression (Figure 2A and 2B), (p,0.05 in both cases). TGFb is thus unable to significantly alter the expression of c-myc and p21 mRNA, and halt cell cycle progression in 4/5 BE invasive adenocarcinoma cell lines. This is in keeping with our previous data on cell proliferation measured using the MTT assay [17].

Impaired Smad-dependent transcriptional regulation
Nuclear translocation of Smad 2/3 ( Fig 3A) and phosphorylation of Smad 2 and 3 ( Fig 3B) were observed in all cell lines following stimulation with TGFb. Nuclear accumulation of Smad2/3 in BIC, which does not express Smad4, confirms early work suggesting that Smad4 is not necessary for Smad 2/3 nuclear translocation [3]. The nuclear translocation of Smad 2/3 and of the phosphorylation of Smad 2 and 3 attest to the functionality of the TGFb pathway. However, the cells remain unable to decrease their proliferation as a result of TGFb stimulation. Therefore, to determine whether the inability of TGFb to regulate the cell cycle is secondary to a lack of Smad dependent transcriptional regulation we used a (CAGA) 12 -luciferase reporter. The lack of Smad 4 expression in BIC makes this cell line the perfect negative control for transcriptional regulation of the TGFb pathway. BIC and SEG both display no transcriptional activity while the low transcriptional activity in TE7 and FLO may not be sufficient to trigger an anti-proliferative response. Only the TGFb responsive cells, OE33, demonstrated a marked induction of transcriptional activity in TGFb treated cells compared to control cells ( Figure 3C) (6865.5; p,0.01). Hence, the non-responsiveness to the antiproliferative effects of TGFb may be due, at least partly, to a lack of Smad-dependent transcriptional regulation, which might stem from Smad4 inactivation, the inability to degrade SnoN and possibly other unknown mechanisms [16,17].

MAPK dependent TGFb induction of genes associated with ECM regulation
The ability of TGFb to regulate the genes uPA and PAI, which are associated with ECM modulation, was then assessed. All cell lines, with the exception of BIC up-regulated either PAI and/or uPA in response to TGFb. (Figure 4A and 4B; OE33, FLO p,0.001; TE7, SEG p,0.05 for PAI and OE33, SEG, FLO p,0.01; TE7, p,0.05 for uPA). TGFb was thus able to up-regulate the expression of both genes associated with ECM modulation in 4/ 5 cell lines, including three lines (TE7, SEG and FLO) that lacked Smad-dependent transcriptional responses and growth suppression by TGFb.
To characterise the molecular basis for the differential responses of the cell lines to TGFb, we analysed some of the Smad independent kinase pathways known to activate genes associated with ECM. [22] Overall, the levels of phosphorylated JNK did not vary following TGFb treatment ( Figure 5). The levels of phosphorylated Akt and ERK1/2 were increased in TE7 (1.6 and 5.3 fold respectively) and SEG (1.7 and 1.8 fold respectively) following 4 hours of stimulation with TGFb ( Figure 5 and Figure S1 shows densitometry data). Only pAKT was increased in OE33 (1.8 fold) following treatment but very little variation was observed in BIC and FLO (figure S1). There was no change in total expression levels of these kinases (data not shown). BIC cells did not show an induction of uPA, PAI and Smad independent kinase pathways and were therefore not studied further.
Inhibition of PI3Kactivity by LY294002 resulted in a variable effect on PAI expression (Fig. 6A). Specifically, in OE33 and SEG, TGFb induced up-regulation of PAI mRNA was significantly diminished by LY294002 (OE33, SEG p,0.001). Inhibition of ERK activity by PD98059 significantly diminished TGFb induced up-regulation of PAI mRNA in all cell lines (OE33, TE7 p,0.05; SEG p,0.001; FLO p,0.01), ( Figure 6A). Following inhibition of JNK activity by SP600125 there was a reduction of the TGFb induced up-regulation of PAI by a minimum of 50% in all cell lines analysed (p,0.05 for all cell lines). These findings were generally confirmed for a second of each class of kinase inhibitor (wortmannin, U0126 and Curcumin), (table S1), although ERK inhibition with UO126 was less effective than PD98059 possibly because of differential specificity of these inhibitors for MEK1 and 2. A similar pattern was seen when uPA expression was assessed in inhibitor treated cells ( Figure 6B). Whereas, the effects of PI3K inhibition had a variable effect on uPA expression, inhibition of ERK by PD98059, and inhibition of JNK by SP600125 markedly reduced (by a minimum of 40%) the TGFb induction of uPA in all BE adenocarcinoma cells ( Figure 6B). Overall, the most consistent results were seen with the JNK inhibitors since both inhibitors (SP600125 and Curcumin) significantly reduced the up-regulation of PAI and uPA in all cell lines, (table S1).
Next we determined whether inhibition of these kinase pathways had an effect on PAI and uPA enzyme activity, as well as expression levels. Zymography and reverse zymography indicated that TGFb induces an increase in the proteolytic activity of uPA (increase in the white band intensity), and an increase in the ability of PAI to retard the action of uPA in all cells lines analysed (increase in the dark band intensity), (Fig 7A-D, compare lane C with T). The inhibition of PI3K activity by LY294002, and inhibition of ERK signalling by PD98059 resulted in a variable effect on uPA and PAI activity depending on the cell line (densitometry given as figure S2). Inhibition of JNK by SP600125 resulted in the most consistent decrease in activity of uPA and PAI ( Figure 7A-D, compare lane T with SP for each cell line with a variation between 0.1 and 0.5 fold on densitometric analysis, figure S2). The expression and the enzyme activity data thus suggests that activation of the ERK and JNK pathways is important for TGFb induced upregulation of PAI and uPA in BE adenocarcinoma cells.

TGFb increases invasive cellular properties via activation of kinase pathways
We determined the ability of TGFb to influence the invasiveness of cells using three in vitro assays: aggregation, invasion and wounding. An aggregation assay measures the ability of cells to grow in an anchorage-independent manner; invasion assays through matrigel measures the ability of cells to degrade the basement membrane and to migrate through it while the wounding assay only measures the ability of cells to migrate to occupy an empty space. Therefore a highly invasive cell in vitro should not form aggregates, should migrate through matrigel and to close up a wound. TGFb had significant effects on at least one assay for each of the adenocarcinoma cells lines, compared to the control (no TGFb), (Table 1). In order to examine the effect of kinase inhibitors these were only used in the cells lines exhibiting an increase in their invasive phenotype for a particular assay in response to TGFb. Therefore, aggregation was performed for TE7, SEG, FLO and OE33, invasion for TE7, SEG and OE33 and wounding for OE33. With regards to aggregation, the presence of all three kinase inhibitors diminished the cell scattering ability (a higher score denotes less scattering) of TGFb in all the cell lines to a greater or lesser extent ( Figure 8A). The increase in TGFbinduced invasion observed in TE7, SEG and OE33 cells was significantly inhibited by LY294002 in TE7 cells (p,0.05), by SP600125 in SEG (p,0.05) and by all three inhibitors in OE33 ( Figure 8B; p,0.001, LY294002; p,0.01, PD98059; p,0.001, SP600125). In OE33, the TGFb induced increase in wound healing over 24 h (p,0.05), is significantly diminished by all three kinase inhibitors ( Figure 8C; p,0.05). Overall inhibition of the JNK pathway with SP600125 had the most consistent effect seen in SEG, FLO and OE33 in all three of the assays.

DISCUSSION
These data suggest a dual role of TGFb in the regulation of both the proliferative and invasive properties of BE adenocarcinoma in vitro. 4/5 BE cell lines are unable to mount an anti-proliferative response to TGFb. However, the loss of TGFb responsiveness is selective, such that the same cell lines that fail to exhibit an anti-proliferative response demonstrate a TGFb dependent increase in the expression of PAI and uPA via PI3K, ERK and JNK ( Table 2). The effect of kinase inhibitors on expression levels and the enzyme activity of uPA and PAI suggest that the ERK and JNK MAPK pathways may play a role. TGFb induced activation of kinase pathways is functionally significant since their inhibition results in a reduction of cellular invasive properties in vitro. TGFb unresponsiveness in the BE adenocarcinoma cells could be attributed to an inability to down-regulate c-Myc, similar to breast cancer cells [24]. Given that all of the unresponsive cells in our panel do not demonstrate a TGFb induced increase in transcriptional activity of a TGFb reporter plasmid, defective Smad-mediated signal transduction is a possible underlying mechanism ( Figure 3). However, since the Smad3/4 reporter construct is not the native promoter and may miss possible   (2) or media with 10ng/mL TGFb (+) for 4 hours for phosphorylated ERK1/2, phosphorylated JNK or phosphorylated Akt. Immunoprecipitation was performed with total antibody for Akt, ERK and JNK, and then blots were probed with the phosphorylated antibody. doi:10.1371/journal.pone.0000177.g005 enhancer or suppressor elements this conclusion must be interpreted with caution. Furthermore, other mechanisms such as the inability to degrade SnoN have been described in oesophageal cells as a possible reason for a lack of responsiveness to TGFb [16].
In contrast, there was a modest TGFb dependent activation of kinase pathways (PI3K, ERK and JNK MAPK) in BE adenocarcinoma cells ( Figure 5). This is interesting since inhibition of PI3K, ERK and JNK MAPK decreased the TGFb induced expression of the ECM genes PAI and uPA ( Figure 6). The PAI promoter contains motifs complementary to the binding sequence of AP-1    [25,26], while the uPA promoter also contains two AP-1 binding sites [27]. Targets of the MAPK pathways ERK and JNK include the transcription factors c-Fos, c-Jun and ATF-2, which can undergo c-Fos/c-Jun or c-Jun/ATF-2 heterodimerisation to form AP-1. Activation of MAPK pathways can thus increase the expression of PAI and uPA [28,29]; however, we cannot rule out the contribution of other pathways such as matrix metalloproteinases which also contain AP-1 promoter sites and are involved in ECM remodelling during invasion.
The combination of uPA, uPA receptor and PAI is crucial in the modulation of ECM homeostasis and tumour invasion as demonstrated in breast and prostate cancer cells [30,31] and oesophageal adenocarcinoma cells made to over-express uPA in vitro [11]. Elevated levels of uPA and PAI can also enhance migration, through modulation of cell-to-matrix adhesion [7,32]. The involvement of PAI and uPA in invasion may explain the functional significance of the inhibition of kinase pathways on the invasive properties of oesophageal adenocarcinoma cells (Figure 8). It is worth noting that OE33 was the only cell line that had a demonstrable increase in the SMAD reporter construct as well as increased kinase activity in response to TGFbl. For this cell line it is not possible to rule out a contribution of Smad activation in the TGFbl response of OE33 cells although the invasive properties were significantly reduced by kinase inhibitors (Figure 8).
uPA and PAI expression through the activity of ERK and JNK may also be relevant for tumour progression in vivo. For example, in prostate, breast and pancreatic carcinoma cells, JNK, or the upstream activator of both kinases, MKK4, has been demonstrated to mediate growth of cells in vitro, and as xenografts in vivo [33,34]. Furthermore, over-expression of PAI and uPA has been shown to correlate strongly with tumour stage and to be of  independent prognostic significance in oesophageal adenocarcinoma [13,35].
In conclusion, we have shown that there is a divergent response to TGFb in cells from BE adenocarcinoma. The Smad dependent anti-proliferative response is lost; however, the ability of TGFb to induce ECM genes PAI and uPA prevails. This occurs predominantly through the ERK and JNK MAPK pathways, and thus contributes to the invasive properties of the oesophageal tumour cell. Such a differential response utilising different cell signalling pathways may allow TGFb to play a role in both the initiation and progression of oesophageal adenocarcinoma.

Cell lines and culture conditions
For this study cell lines derived from Barrett's adenocarcinoma were used. Barrett's associated adenocarcinoma cell lines OE33 (European Collection of Cell Cultures, Wiltshire, UK), TE7 (gift from T Nishihira, Kurokawa County Hospital, Japan), were maintained in RPMI-1640, and SEG, BIC and FLO (gift from D Beer, University of Michigan, MI, USA) were maintained in DMEM. All cells were supplemented with 10% (FBS), 100 U/mL penicillin, 100 mg/mL streptomycin and 2 mM glutamine.

Kinase inhibition
Cells were grown to 70% confluence in complete medium, which was then replaced with serum free medium for 3 h. Cells were pretreated for 1 h with inhibitors at the minimal concentration leading to kinase inhibition (data for determining optimal inhibitor concentrations not shown): PI3K inhibitors -Wortmannin (5 nM) and LY294002 (20 mM) (Calbiochem-Novabiochem, Nottingham, UK); ERK inhibitors -U0126 (inhibits MEK 1 and 2), (10 mM) and PD098059 (inhibits MEK 1), (20 mM) (Sigma-Aldrich, Dorset, UK); and JNK inhibitors -SP600125 (20 mM) (Sigma-Aldrich) and curcumin (10 mM) (Calbiochem-Novabiochem). Cells were then treated with 10 ng/mL TGFb1 for 24 h in the presence of inhibitors. This concentration was used as we have previously shown in dose response experiments that it is the optimum concentration to induce a decrease in proliferation [17].
Flow cytometry 10 4 cells were seeded in 6 well plates in complete growth medium, and all cells were synchronised overnight in serum-free medium. Cells were then either incubated in complete medium with or without 10ng/mL recombinant TGFb1 (Tebu-bio, Cambridgeshire, UK), or kept in serum free medium, for 24 h. Cells were harvested, fixed in 70% ethanol at 4uC and then incubated in PBS, 0.1% Tween 20 containing 2 mg/mL propidium iodide (Sigma-Aldrich) and 200 mg/mL RNAse A for 30 min at 37uC. DNA content was analysed by fluorescence associated cell sorting (FACS) using the BD LSR1 flow cytometer (BD Biosciences, Oxford, UK), and data captured using Cell quest software (BD Biosciences, Oxford, UK).

Immunofluorescence staining
Cells were seeded overnight onto sterilised coverslips in a 24 well plate. Media was removed and replaced with fresh media with or without 10 ng/ml TGFb1. Following a 6 h incubation, media was removed and the coverslips washed twice in 0.5 ml PBS. The coverslips were then stained with Smad2/3 antibody (clone E-20, Santa Cruz Biotechnology, Inc., Santa Cruz, CA, USA) following a procedure described previously [17].

Luciferase reporter assay
Cells were seeded in 24 well plates at a density of 5610 4 cells/well and left overnight. Using Fugene6 transfection reagent (Roche Applied Science, East Sussex, UK) cells were transfected with the reporter plasmid (CAGA) 12 -Luc, containing 12 repeats of the CAGA Smad3/4 binding sequence. Cells were also transfected with Renilla luciferase reporter plasmid to normalise transfection efficiency. After 24 h medium was replaced with serum free medium with or without TGFb1 (10 ng/mL). After a further 24 h luciferase activity was determined using the Dual Luciferase Reporter System (Promega, Southampton, UK) in a TD 20/20 Luminometer (Promega).

Quantitative real time PCR
Two micrograms of total RNA isolated using Trizol Reagent (Invitrogen, Paisley, UK) were reversed transcribed, and 1 mL of cDNA was amplified in a 50 ml volume containing 25 mL of 26 Quantitect SYBR Green PCR Master Mix (Qiagen, West Sussex, UK) and 0.2 mM of each primer (Table 3). Triplicate reactions were performed in a DNA Opticon using the conditions of initial enzyme activation of 15 mins at 95uC, followed by 30 to 35 cycles of 10 s at 95uC, 20 s at the annealing temperature and 20 s at 72uC. The melting curve was constructed for each primer to ensure reaction specificity. Following PCR, the threshold cycle (C T ) was obtained and relative quantities determined compared to b actin.

Zymography and reverse zymography
To assess uPA activity 50 mg of protein lysate were incubated with non-reducing buffer for 20 min at 37uC, then electrophoresed in a polyacrylamide gel containing 2 mg/mL a-casein (Sigma-Aldrich) and 5 mg/mL plasminogen (Sigma-Aldrich). Gels were washed 2 times for 30 min each in 50 mM Tris-HCl (pH 7.6), 2.5% Triton X-100, incubated for 16 h at 37uC in 50 mM Tris-HCl (pH 7.6) and then stained for 1 h in a solution of 30% methanol, 10% acetic acid and 0.5% Coomassie brilliant blue G250 and de-stained. Urokinase activity was detected as digested clear bands on a blue background. To determine PAI activity lysates were subjected to SDS-PAGE in gels containing 2 mg/mL a-casein, 5 mg/mL plasminogen and 0.8 U/mL human urokinase (Calbiochem-Novabichem). Gels were then washed and stained as above. PAI activity was detected as dark undigested bands against a light background.

Wound healing assay
Cells were grown in a 60 mm dish to confluency, and then wounded using a hand held engraving drill (Minicraft MB168) fitted with a cut down 10 mL pipette tip. Wound area was measured using an eyepiece graticule and calibrated using a haemocytometer grid, at 106 magnification. The wounded monolayer was washed with PBS, and then cultured in serum free medium, containing 10 ng/mL TGFb1 alone or with LY294002 (20 mM), PD98590 (20 mM) or SP600125 (20 mM) for 24 h. The wound area was again measured, the absolute change in wound size deduced and the % healing at 24 h then calculated. In each experiment 8 wounds were made and each experiment was repeated at least three times.

Invasion assay
Fifty mL of a 10 mg/mL solution of matrigel (BD Biosciences), containing 25 mg/mL plasminogen was applied to the upper compartment of a Transwell 24 well insert (BD Biosciences) of 8 mm pore size, and incubated at 37uC. Cells were washed three times in serum free RPMI, and then 1610 5 cells were resuspended in 200 ml of serum free RPMI containing 10 ng/mL TGFb1 alone or with LY294002 (20 mM), PD98590 (20 mM) or SP600125. Cells were then applied to the upper compartment of the invasion chamber. In the lower compartment 500 ml of serum containing medium was added and the cells were then incubated at 37uC for 24 h. Medium was then removed and the cells were then fixed for 20 min in methanol at 220uC. Chambers were stained in haematoxylin for 1 min and a cotton swab used to remove noninvasive cells in the upper compartment. Membranes were excised from the chamber and fixed on a microscope slide. Membranes were quantified by counting the number of cells that had invaded the lower chamber in 5 high-power fields. [11] Slow aggregation assay 100 mL of agar was solidified in 96 well plate formats. Cells were washed with Moscona's solution, trypsinised and then resuspended in serum free media. 2610 4 cells (100 ml) were seeded into each well and left for 1 h. 100 mL of serum free medium containing TGFb1 alone or with LY294002 (20 mM), PD98590 (20 mM) or SP600125 (20 mM). [36] Aggregates, for 10 replicates in three independent experiments, were scored after 24 h incubation at 37uC, 5% CO 2 using an inverted microscope. Replicates were scored 0 if no aggregates were present, 1 for small aggregates and 2 for large aggregates.

Statistics
The Kruskal-Wallis test was used to compare values between different data sets, and the Dunn's Multiple Comparison Test was used to identify specific differences, using Graphpad Prism software. A p,0.05 was necessary for statistical significance.

SUPPORTING INFORMATION
Table S1 Found