Effect of Oncofid-P20 treatment on cell proliferation and viability
A431 human squamous carcinoma cells, a commonly model used for studies of skin cancer and normal keratinocytes (NHK) from healthy subjects were treated with a broad panel of Onco-P20 concentration (0.05-1 µg/mL) for 72 hours before MTT assay and cells count. MTT assay showed a strong reduction of NAD(P)H-dependent cellular mitochondrial activity in Onco-P20-treated carcinoma cell cultures that correlated to a significant reduction in the number of cancer cell at all the concentrations tested (Fig.1a and b). In contrast, at the experimental end point in NHK Onco-P20 treatment mildly impacted on metabolic function and on the number of cells resulted modestly reduced (Fig.1c and d). Hyaluronic acid, alone did nor exerted any significant modification of cell viability and proliferation indicating that this portion of the molecule is not implicated in the biological activity observed. Since in addition to cell count, microscopic observation of cell cultures highlighted marked differences comparing normal and tumor cells (Fig. 1e), we additionally evaluated cell cycle distribution demonstrating G2/M accumulation in both cell types and extensive cell death only in cancer cells (Fig. 1f). Significant increase in the number of apoptotic cells was confirmed by AnnexinV/PI staining proving the selective cytotoxic effect on tumor cells (Fig. 1g). Thus, while in carcinoma cultures metabolic activity measurement and the number of apoptotic cells correlated indicating that cell death is the main mechanism of Onco-P20 function, in normal keratinocyte, all the concentrations tested, the treatment reduced the metabolic activity without affecting cell viability. However, high doses of Onco-P20 (>0.25 µg/mL) impacted on keratinocyte proliferation capability as evidenced by cell accumulation in G2/M cell cycle phases.
Since established cell lines maintained in culture for a long period frequently present evidence of deviations from the phenotype of the originating tumor, we used a collection of specimens from patients well characterized for disease stage and follow up to isolate low passage tumor cell lines and patient-matched normal keratinocytes cultures. In this case, cytotoxic effect against short term carcinoma cell cultures showed an overall lighter intensity compared to A431 cells and the MTT assay, that does not discriminate between cytotoxic and cytostatic effect, displayed modest difference between normal and carcinoma cells (Fig. 2a and b). By contrast, AnnexinV/PI staining confirmed that Onco-P20 preferentially reduced viability of neoplastic cells (Fig. 2b). Since A431 cell cultures and fresh isolated carcinoma cells were cultured in non-overlapping growth condition (DMEM supplemented with FBS and chemically-defined medium M154 supplemented with HKGS respectively; see material and method section), we addressed the question if the observed differences could be attributed to culture medium composition or to distinctive biological properties. In this case, we used A431 and SCC1300-UC, one of the low passage SCC cell lines isolated in our laboratory capable to proliferate both in defined media (M154 plus HKGS) as well as in high calcium FBS-containing DMEM medium, to evaluate the impact of cell medium composition on Onco-P20 treatment. As shown in Suppl. Fig. 1, the sensibility to Onco-P20 treatment resulted attenuated in M154 defined medium compared to DMEM plus serum (Fig. S1a) and inversely correlated to the proliferation rate (Fig. S1b). In fact, cell culture composition influenced cell growth demonstrating a significant slower growth curve kinetic in chemically defined medium that correlated with lower pharmacological activity. However, responsiveness to Onco-P20 is also an intrinsic characteristic of different cell lines related to the CD44 level of expression that is not necessarily over-expressed in skin carcinoma compared to normal cells . In fact, FACS analysis confirmed that patient-derived carcinoma specimens present a moderate level of CD44 expression comprised between those of normal cells and one of A431 cells (Fig. 3a). The low presence of CD44 on the membrane could partially explain the higher response of A431 that could be still appreciated using identical growth condition (Fig. S1a). Accordingly, with the idea that the relative abundance of CD44 is not particularly elevated in the keratinocyte lineage, hyaluronan-conjugate formulation failed to demonstrate any significant gain of function in comparison to conventional free paclitaxel (Fig. S2).
CD44 overexpressing cells comprise not only tumor cells and cancer stem cells but also CAFs [40-42]. Strongly CD44-positive CAFs are particularly frequent in tumor hypoxic areas . Since also dermal fibroblasts physiologically express high level of this receptor (Fig.3b) we evaluated the effect of Onco-P20 on normal human fibroblasts (NHF). For this purpose, and in line with patient’s characteristics, we used adult fibroblasts isolated from photo-exposed area of geriatric donors. NHF treated with Onco-P20 (0.05-1.0 µg/mL) for 72 hours displayed attenuation of mitochondrial activity and a significant abatement in cellular proliferation compared to control cells (Fig.4a and b). Similar to normal keratinocytes, fibroblasts did not show acute cytotoxicity as measured by AnnexinV/PI double staining (Fig. 4c). Cell cycle analysis evidenced a marked accumulation in the G2/M phase (Fig. 4d). In addition, evidences of cell hypertrophy (a large cell morphology), a universal marker of in vitro cell senescence  appeared at days 3-5 (Fig.4e). Increase expression of p53 and p21, as well as, the presence of several promyelocytic leukemia protein (PML)-nuclear bodies, typical nuclear matrix structures implicated in the induction of the senescence process , confirmed diminished proliferative propensity of Onco-P20-treated cells (Fig. 3f and g). In contrast to p21, another important cell cycle checkpoint CDK inhibitor, p27 resulted decreased. Cyclin E, that is required and rate liming for S phase entry, is lightly downregulated. We further analyzed the protein level of cyclins in Oncofid-P20-treated cells. The level of cyclin D1 which promotes progression through G1/S phases was elevated, while cyclin B1, promoting G2/M transition, was barely present [45, 46]. According with previous studies [47, 48] the cyclins profile was consistent with G2-arrest. However, fibroblasts seem to be long-term arrested rather than irreversible senescence since after treatment removal cells resumed proliferation even in case of prolonged exposure to Onco-P20 (Fig. S4).
Finally, Onco-P20 activity was compared with that of equal amount of free paclitaxel. Following 72 hours drugs exposure, unconjugated paclitaxel exhibited a significant lower effect on NHF (Fig.5a). Thus, due to high level of CD44 expression on the membrane surface of mesenchymal cells, Onco-P20 exerts a stronger pharmacological activity then free paclitaxel. Moreover, to simulate drug availability in the case of topical application on patient’s skin, we exposed NHF to treatments (Onco-P20 or paclitaxel) for a defined short period (8 hours) before wash out and subsequent cell culture in drug-free medium. In this case, using the same end point of previous experiments we observed reduced proliferation exclusively in Onco-P20-treated cells, whereas paclitaxel failed to modified fibroblast proliferation (Fig.5b). By contrast, in carcinoma cells no differences were observed comparing Onco-P20 and paclitaxel using short period treatment (Fig. 5c). However, it is interesting to note that in this case both molecules exerted a mild effect suggesting or an augmented sensibility to taxol of carcinoma cells compared to fibroblasts or an accelerated trigger of drugs effect.
Onco-P20 deeply modifies fibroblasts gene and protein expression profile
Given that fibroblasts could impact keratinocytes homeostasis through diffusible molecules, we assessed the impact of Onco-P20 on the expression of secreted proteins involved in tumor progression. Under these circumstances, several growth factors were differentially expressed in Onco-P20-treated fibroblasts versus control samples (Table 1). Among these, HGF was strongly down-regulated both at mRNA and protein levels (Fig. 6a). Additionally, the high molecular weight of bFGF, frequently associated with a poor prognosis in various human cancer , was reduced by Onco-P20 (Fig. 6b), whereas its mRNA resulted unmodified. However, since high molecular weight isoforms of bFGF are associated to highly proliferative phenotype of fibroblast , it is possible that the observed difference reflects an autocrine control of cell growth. Similarly, KGF resulted decrease mostly at protein level. EGF transcript was near to the detection limit in most of the fibroblasts cultures and lightly reduced by the treatment. The IGF was unchanged whereas, members of insulin-like growth factor binding proteins (IGFBPs) superfamily, a group of secreted proteins capable to binding IGFs and to modulate the mitogenic, anti-apoptotic and metabolic actions of IGFs, were found strongly increased. In particular, IGFBP3, IGFBP4, IGFBP5, IGFBP6 and IGFBP7 mRNA were significantly up-modulated. Immunoenzymatic quantification of IGFBP4 and 6 confirmed gene expression analysis (Fig. 6a). Possible repression of neo-angiogenesis is supported by reduced expression of vascular endothelial growth factors (VEGF) at mRNA and protein level (Fig. 6b).
The Wnt/β-catenin signaling pathway, known to be activated in CAFs [51, 52], resulted down-modulated in Onco-P20-treated fibroblasts as demonstrated by decreased of β-catenin, the pivotal molecule of the Wnt signaling pathway (Fig. 6a). Accordingly, the expression of some negative regulators of the pathway, Wnt5a, DKK1 and SFRP2, were increased suggesting an autocrine mechanism of Wnt signaling regulation. By contrast, TGFb expression resulted unchanged. Among CAF’s markers, PDGFα and b were reduced whereas the expression of FAP1 and αSMA resulted increased and unmodified respectively. Strong increase of IL1α, IL1b, IL6 and IL8 argues for the activation of inflammatory pathways. To further investigate signal transduction pathways involved in cancer-fibroblasts cross-talk, we investigated a wide range of cytokines and chemokines using gene expression array cards and an antibody membrane array. Among the 93 mRNA studied, 38 were in most or all the sample undetectable and were discarded from the analysis, 13 resulted unmodified by the treatment, 2 lightly decreased and 40 enhanced (>1.5 fold-increase). Of these highly expressed transcripts 17 reached the statistical significance (p≤0.05 or p≤0.01) (Table 2). Most of the protein up-regulated, including KLK14 and members of MAP kinase family, such as MAPK8 (JNK1) and MAPK14 (p38α), are implicated in the IL1 pathway. The expression of IL1R1 in tumor microenvironment, the main receptor of IL1α and b, due to the interconnection with NF-KB and MAP kinase pathways plays contrasting roles in different tumor stage . Sustained PTGS2 (cox-2) expression confirmed pro-inflammatory profile of NHF Onco-P20-induced phenotype. Annexin1A (ANXA1), increased by Onco-P20, has been proposed as a tumor suppressor in head and neck squamous cell carcinoma . However, in prostatic cancer, ANXA1 exacerbated expression of this marker has been correlated to high amount of cancer stem cells . Interestingly, CD40 an antigen frequently loss by basal and squamous cell carcinoma during tumor escape from activated T cells, was augmented by Onco-P20 treatment. Intercellular adhesion molecule-1 (ICAM1, CD54), a receptor that support leukocytes accumulation in inflamed tissue, resulted strongly enhanced whereas vascular cell adhesion molecule-1 (VCAM), another mediator of leukocytes trafficking, was lightly increased. Additionally, high LTA4H that catalyzes the hydrolysis of epoxide LTA4 to LTB4, which mainly functions as a neutrophil, macrophage, and T lymphocyte chemoattractant, confirmed the inflammation-enhancing effect of Onco-P20. The antibody array used to quantified secreted inflammatory factors confirmed a significant arise in IL6 and IL8 production whereas IL1α and ILb appeared moderately increased at protein level. Additionally, the secretion of CCL11 (Eotaxin-1), a chemokine implicated in eosinophils recruitment , was up-regulated by Onco-P20 (Table 3). Lastly, CCL1, whose release by fibroblast has been linked to bladder cancer cells invasion , significantly diminished.
Secretome of Onco-P20-treated fibroblasts modulates carcinoma cells growth
Cancer and stromal cells communicate mainly by a complex bidirectional cross-talk that evolve during disease progression and the diffusion of soluble factors through basement membrane. To study the effect of Onco-P20-induced fibroblast phenotype on cancer cells simulating the in vivo situation, we used two different experimental systems: fibroblast pre-treated conditioned medium (CM) and the trans-well permeable support (TW) to growth tumor and stromal cells on two different but connecting monolayer. Taking in consideration that previous studies evidenced that mesenchymal cells are capable to release into the microenvironment functional active paclitaxel for 24 hours , in both cases we replaced Onco-P20 containing medium with fresh starved medium 48 hours before collect CM or start co-culture experiments. In either way, due to the absence of paclitaxel and serum, cell proliferation depends exclusively by autocrine and paracrine activity of both cell populations. At the experimental end point (72 hours), both systems clearly demonstrated that carcinoma cells dose-dependently slowed the proliferation rate in presence of Onco-P20-pretreated fibroblasts o their CM (Fig.7a-and b). The anti-proliferative effects observed depends on the dose of drug used to treat NHFs and on the relative fibroblasts-carcinoma ratio as demonstrated by the stronger reduction of A431 cells growth in the trans-well upper (smaller) compartment compared to the lower one. Further, co-culture data were confirmed using a patient derived model assembled with NAF, CAF and carcinoma cells from the same donor (Fig.7c) or NAF, CAF and A431 cell line for cases presenting unsuccessful isolation of tumor cells in vitro (Fig.7d). Overall, data demonstrated that the sensibility to Onco-P20 of CAFs is similar to normal fibroblasts.