TIMP-2 secreted by monocytes is a potent suppressor of invadopodia formation in pancreatic cancer cells CURRENT STATUS: ACCEPTED

Background Monocytes are a major component of the tumor microenvironment (TME) in pancreatic ductal adenocarcinoma (PDAC). However, the complex interactions between tumor cells and monocytes and their role in tumor invasion have not been fully established. Methods In this in vitro study, to specifically test the impact of interaction on invasive potential, two PDAC cell lines PaTu8902 and CFPAC-1 were selected on their ability to form invasive adhesions, otherwise known as invadopodia. Co-culture experiments were performed using undifferentiated THP1 monocytes. Results When the PDAC cells were co-cultured with undifferentiated THP1 monocytes invadopodia formation was significantly suppressed. Moreover, conditioned media of THP1 cells (CM) was also able to suppress invadopodia formation. Further investigation revealed that both tissue inhibitor of metalloproteinase (TIMP) 1 and 2 were present in the CM. However, suppression of invadopodia formation was found that was specific to TIMP2 activity. Conclusions Our findings indicate that TIMP2 levels in the tumour microenvironment may have prognostic value in patients with PDAC. Furthermore, activation of TIMP2 expressing monocytes in the primary tumour could present a potential therapeutic opportunity to suppress cell invasion in PDAC.

2 Abstract Background Monocytes are a major component of the tumor microenvironment (TME) in pancreatic ductal adenocarcinoma (PDAC). However, the complex interactions between tumor cells and monocytes and their role in tumor invasion have not been fully established. Methods In this in vitro study, to specifically test the impact of interaction on invasive potential, two PDAC cell lines PaTu8902 and CFPAC-1 were selected on their ability to form invasive adhesions, otherwise known as invadopodia. Co-culture experiments were performed using undifferentiated THP1 monocytes. Results When the PDAC cells were co-cultured with undifferentiated THP1 monocytes invadopodia formation was significantly suppressed. Moreover, conditioned media of THP1 cells (CM) was also able to suppress invadopodia formation. Further investigation revealed that both tissue inhibitor of metalloproteinase (TIMP) 1 and 2 were present in the CM. However, suppression of invadopodia formation was found that was specific to TIMP2 activity. Conclusions Our findings indicate that TIMP2 levels in the tumour microenvironment may have prognostic value in patients with PDAC.
Furthermore, activation of TIMP2 expressing monocytes in the primary tumour could present a potential therapeutic opportunity to suppress cell invasion in PDAC.

Background
Pancreatic ductal adenocarcinoma (PDAC) is a devastating disease characterised by an aggressive biological tumour behaviour leading to a high mortality rate [1]. The invasive character of PDACs leads to a rapid progression of the disease [2]. In order to invade the surrounding tissue or to create distant metastases, cancer cells need to break through the basement membrane and degrade the extracellular matrix. One identified strategy is to utilise actin-rich membrane protrusions called invadopodia [3] that can degrade extracellular matrix. Initiating mechanisms to suppress such formation is a potential therapeutic target [4]. Unlike podosomes -which are present in normal myeloid cells -invadopodia are specific for invasive cancer cells [5]. The primary purpose of invadopodia is the targeted secretion of matrix metalloproteases (MMPs) to degrade the extracellular matrix (ECM) [6]. Membrane-type 1 metalloprotease (MT1-MMP, also known as MMP14), MMP9 and MMP2 have been identified as the most important MMPs in invadopodia functionality [7,8].
3 MMPs can be inhibited by tissue inhibitors of metalloproteinases (TIMPs). Of the four TIMPs [9,10], TIMP1 and 2 are the best-studied. TIMP1 has a relatively low affinity for membrane-bound MMPs [11] whereas TIMP2 is a strong inhibitor of MT1-MMP [12]. Very little is known about the effects of TIMPs on invadopodia formation, although one study in human breast cancer cells suggested TIMP2 is specifically able to decrease the formation of invadopodia [13].
In recent years, the tumour microenvironment (TME) has become of major interest to researchers and has recently been suggested to play a specific role in influencing the formation of invadopodia [14].
Around 90% of the PDAC mass are stromal cells, only 10% are carcinoma cells [15]. The main components of the TME are cancer-associated fibroblasts (CAF) and tumour-associated macrophages (TAMs). The pancreatic TME is rich in TAMs but undifferentiated monocytes [16] are also present at detectable levels. The interaction between monocytes / macrophages and the tumor cells is complex with both pro-and anti-tumorigenic effects reported [17].
This study aimed to establish whether an interaction between PDAC cells and undifferentiated monocytes presented pro-or anti-tumorigenic responses.

Cell culture
Patu8902 and Capan2 were obtained from "Deutsche Sammlung von Mikroorganismen und Cells were maintained in RPMI-1640 media supplemented with 10% v/v FBS with 1 mM penicillin/streptomycin. For inhibitor treatment cells were incubated with 10 mM GM6001 as previously described [39]. All cell lines were regularly screened for mycoplasma contamination using DAPI staining.

THP-1-conditioned media (CM)
THP-1 cells were seeded in serum-free, antibiotic-free RPMI media at a density of 200,0000 cells/ml and incubated at 37°C and a 5% CO 2 atmosphere. After 24 hours, the cell suspension was centrifuged at 1,500 rpm at room temperature for 10 minutes. CM was either directly used for experiments (fresh CM), stored at -20°C and rewarmed to 37°C (frozen CM) or heat-inactivated at 95°C for 10 mins (boiled CM).

Antibodies
Anti-TIMP1 antibody (D10E6) produced in rabbit was purchased from Cell Signalling Technology, U.S.A., anti-TIMP2 antibody produced in rabbit antibody (SAB4502972) was purchased from Sigma.

Invadopodia Assay
For the invasion assays, the QCM™ Gelatin Invadopodia Assay (Red) (Chemicon® / Millipore) was used. Briefly, coverslips were inverted onto poly-L-lysine in deionized water for 20 minutes at room temperature (RT). The slides were then washed with PBS three times before incubation with glutaraldehyde: PBS for 15 minutes at RT. After washing three times with PBS, each coverslip was placed on gelatin in PBS in a 1:5 ratio of fluorescently-labelled -unlabelled gelatin and incubated for 10 minutes at RT and subsequently washed in PBS three times. The. Patu8902 and CFPAC-1 cells were detached using non-enzymatic Cell Dissociation Solution (Sigma Aldrich UK), resuspended in DMEM F-12 growth media (10% FBS, 1 mM penicillin/streptomycin) and seeded onto the prepared coverslips.

5
For co-culture experiments cells were seeded in the presence of control media (serum-free RPMI media and DMEM-F12 media in a 1:1 ratio), in the presence of THP1-CM (frozen CM mixed with DMEM-F12 media with 10% FBS and 1% P/S in a 1:1 ratio) or in the presence of 50,000 THP1 cells. Alternatively, PDAC cells were co-cultured with THP1 cells for 24 hours prior to dissociation and seeding on prepared coverslips.

Western Blot
Serum-free THP-1 conditioned media (CM), serum-free RPMI containing recombinant TIMP1 and TIMP2 (expressed in CHO cells Sigma Aldrich UK) in various concentrations ranging from 5 to 6000 ng/ml, respectively, and serum-free RPMI control media, respectively, were filled in a Spin-X® UF concentrators (Spin-X UF 6 10K MWCO, Corning) and centrifuged at 4,000 rpm for 18 minutes at RT.
Subsequently, gel sample buffer was added to the concentrated CM and the sample was boiled at 90°C for 3 min. Equal amounts of protein were electrophoresed on 10% SDS-polyacrylamide gels then transferred to nitrocellulose membranes as described elsewhere [40]. Nitrocellulose were incubated with primary antibodies using the recommended concentrations and HRP-conjugated secondary antibodies (Dako Ltd).

Micro Array
THP1 CM was screened for proteins using the RayBio® C-Series Human Cytokine Antibody Array C5 (RayBiotech Norcross, USA) according to the manufacturer's protocol.

Image analysis
Images were analysed using ImageJ 1.51h (National Institutes of Health, USA). For gelatine degradation analysis, the total amount of degradation per image was measured in a total of 10 images per tested condition. The amount of degradation was computed automatically using ImageJ.

Statistical analyses
For data collection and statistical analysis, Microsoft Excel (Microsoft, Redmond, USA) and Prism 5.0 (GraphPad Software, La Jolla, CA, USA) were used. To test for significant differences, the two-tailed Student's t-tests was used. Data are presented as mean ± standard error of the mean (SEM). A difference was considered significant at p < 0.05.

Pancreatic cancer cells produce invadopodia
As compared to the well-described characteristics of invadopodia formation in breast cancer cells,  Table I). Our screen revealed that PaTu8902 and CFPAC-1 constitutively form a large number of invadopodia. To confirm our observations, invadopodia areas of matrix degeneration were co-localised with cortactin-positive puncta as a marker of invadopodia ( Figure 1A). We found that PDAC cells require longer incubation times than previously reported for breast cancer cells [18] to generate quantifiable degradation activity and thus quantification of activity was based on the total area of gelatin degradation per field of view. PaTu8902 and CFPAC-1 cells revealed high levels of degradative activity ( Figure 1B).

Monocyte co-culture suppresses invadopodia driven matrix degradation
We performed a number of different co-culture experiments with eGFP-tagged THP1 cells (THP1; a commonly used cell line model for undifferentiated monocytes) and PaTu8902 cells (Figure 1C&D) or CFPAC-1 cells (Figure 1E&F). Either the PDAC cells and THP1 cells were cultured together prior to the invadopodia assay, or PDAC cells and THP1 cells were cultured together during the invadopodia assay, or conditioned medium from THP1 cells was added to the PDAC cells during the invadopodia assay. In all co-culture conditions, gelatin degradation was reduced compared to control in both cell lines (Figure 1C-F). Given that exposure to conditioned medium was sufficient to suppress invadopodia activity we explored this phenomenon in more detail in both cell lines. The gelatin degradation assays were repeated using a control condition (control media) and three different CM conditions. To assess the nature of the inhibitory factor, CM was either boiled, used immediately after collection (fresh CM) or freeze/thawed. Incubation of cells with fresh or freeze/thawed CM significantly reduced invadopodia formation whereas incubation with boiled CM was unable to significantly reduce activity compared to control cells (Figure 1G-J). These findings suggest that the factors secreted by THP1 cells to suppress invadopodia actively are likely to be protein based.

Tissue inhibitor of metalloproteinases 1 and 2 are secreted by THP1 monocytes
To further elucidate the nature of THP1 CM suppression of invadopodia activity the THP1 CM was screened against a microarray of selected hormones and cytokines (Figure 2A). The only proteins detected in significant concentrations were RANTES, Interleukin 8 (IL-8) as well as tissue inhibitor of metalloproteinases 1 and 2 (TIMP1 and TIMP2). Both RANTES [21] and IL-8 are suggested to rather promote than reduce cancer cell invasion, so we focused on the further examination of TIMP1 and TIMP2. Performing a western blot with specific antibodies we were able to detect both TIMP1 and TIMP2 in the CM samples consistent with our array and determine that the concentration of both TIMP1 and TIMP2 in CM was between 5 and 50 ng/ml (Figure 2B-D, for full-length gels and blots see

Inhibition of invadopodia formation is driven specifically by TIMP2
To assess the specific effects of TIMPs on invasion, the invasion assays were repeated adding commercial recombinant TIMP1 (rTIMP1) or recombinant TIMP2 (rTIMP2), respectively at two different 8 concentrations, 5 ng/ml and 50 ng/ml, to cover the range of likely concentration in conditioned medium ( Figure 2C&E). Whilst the presence of rTIMP1 had no impact on degradative ability ( Figure   3A&B), the presence of rTIMP2 was able to significantly reduce gelatin degradation in the treated cells even at a concentration of 5 ng/ml ( Figure 3A&C). Moreover, similar results were obtained when CFPAC-1 cells were treated with rTIMP2 but not with rTIMP1 (Figure 3D-F). Our results suggest a specific inhibition of invadopodia activity in PDAC cells exposed to low concentrations of TIMP2.
However, we cannot rule out a global deleterious effect of rTIMP2 exposure on cell behaviour. To address this issue we tested whether cells could recover invadopodia activity if the rTIMP2 was removed. As a control, responses were compared to incubation with a pharmacological MMP inhibitor GM6001 which is known to transiently inhibit invadopodia formation until washed out [22] ( Figure   4A&B). Subsequently, we tested the recovery of cells incubated with rTIMP2 for 5 hrs prior to removal ( Figure 4C&D). Similar results were obtained when CFPAC-1 cells were examined in the same protocol ( Figure 4E-H). Thus, the inhibition of invadopodia activity in the presence of rTIMP2 is a specific response.

Discussion
There is evidence suggesting there is a strong link in some tissue types between the ability of cancer cells to form invadopodia in vitro and their invasive potential in vivo [23]. However, only one previous study screened PDAC cell lines for invadopodia formation [20]. Although in this case, the authors reported invadopodia prevalence in most cell lines, this was not comparatively quantified and degradation activity was not assessed. Indeed, we found that some of the PDAC cell lines were extremely inconsistent in invadopodia formation and could not be relied upon for reproducible studies. Other reports focus on one PDAC cell line examining specific pathways [24,25]. We have now identified two PDAC cell lines that reproducibly potentiate significant invadopodia activity and can be confidently used to further studies in this area.
Using these cell lines we proceeded to investigate how the TME might impact on invasive activity. The TME has a particular role in PDAC -not only because the major part of PDACs consists of fibroblast and monocytes/macrophages [16] but also because there is a complex interaction between 9 monocytes/macrophages and PDAC cells potentially modifying the invasive potential of PDAC cells [17]. Monocytes and macrophages in the TME appear to have both pro-and antitumor effects which are suggested to be due to different cytokine profiles depending on the differentiation and polarization [26,27]. Most studies examining mononuclear cells as part of the TME are focusing on macrophages [28, 29], however, there is evidence that monocytes are present in the TME of PDAC [16]. Nevertheless, the prognostic value of monocyte infiltration of the PDAC TME has not been extensively explored.
Taken together, the results of current studies suggest that pro-inflammatory features of the TME determined by the polarization and pro-inflammatory cytokine profile leads to a more aggressive and invasive tumour behaviour and thus decreases prognosis. This is the first study to show that monocytes -which are an essential part of the TME [16] -have the ability to reduce the formation of invadopodia in PDAC cells and thus highly significantly decrease their invasive migratory behavior, which would inhibit their metastatic potential.
Our results indicate that the suppressive activity of undifferentiated monocytes of the PDAC cells might be mediated via secretion of TIMP2 but not TIMP1. This is consistent with a report in breast cancer cells that suggested TIMP2 suppressed invadopodia formation. However, in this study, activity was measured by positive cortactin staining but not matrix degradation [13]. The specific inhibitory effect of TIMP2 might be due to the differential inhibition of TIMP1 and TIMP2. Nonetheless, we believe that there are other mechanisms of how undifferentiated monocytes can inhibit invadopodia formation as well since we found that a monocyte/PDAC cell co-culture prior to the actual invadopodia assay was also able to reduce cancer cell invasion ( Figure 1C and D). These mechanisms could include transcriptional changes in the cancer cells, e.g. upregulation of S100A8 and S100A9 which is known to increase the invasive potential of cancer cells [34].
Interestingly, in human hepatocellular carcinoma (HCC) samples, downregulation of TIMP2 expression was significantly associated with liver invasion and poorer survival outcomes [35]. Currently, PDAC tumours are considered to be immunologically "cold" [36]. We would suggest that an aspiration to increase the immune infiltrate in the PDAC setting [37] should consider specifically increasing the presence of TIMP2 secreting monocytes.
In conclusion, our findings are indicative that TIMP2 could both be a potential prognostic marker and a therapeutic target in PDAC. A high ratio of TIMP2-secreting monocytes could be associated with reduced metastatic potential and better prognosis. Furthermore, the application of synthetic TIMP2 agonists could possibly lead to a reduction of cancer cell invasion in vivo.

Competing interests
The authors declare no competing interests.