Moonlighting role of PEDF in breast cancer

1 ABSTRACT 2 Background: Breast cancer is the leading cause of death among females in developed 3 countries. Although the implementation of screening tests and the development of new 4 therapies has increased the probability of remission, relapse rates still remain high. 5 Numerous studies have indicated the connection between cancer initiating cells and slow 6 cellular cycle cells, identified by their capacity to retain long labelling (LT+). 7 Methods: We have designed a transgenic protein consisting in the C-terminal part of this 8 protein, which acts by blocking endogenous PEDF in culture cell assays. Present work is 9 based in doses-response in vitro assays as well as flow cytometry analysis of surface 10 markers and cell cycle kinetic study of the tumour initiating cells. 11 Results: In this study we show that this type of cells is present not only in cancer cell lines 12 but also in cancer cells from patients with metastatic and advanced stage tumours. We also 13 present new assays showing how stem cell self-renewal modulating proteins, such as PEDF, 14 can modify the properties, expression of markers, and carcinogenicity of cancer stem cells. This protein has been involved in self-renewal in adult stem cells and has been described 16 as anti-tumoral because of its anti-angiogenic effect. However, we show that PEDF 17 enhances resistance in breast cancer patient cells in vitro culture by favoring a slow cellular 18 cycle population (LT+). The PEDF signalling pathway could be a useful tool for controlling 19 cancer stem cells self-renewal, and therefore control patient relapse. 20 Conclusions: We demonstrate that it is possible to interfere with the self-renewal capacity of 21 cancer stem cells, induce anoikis in vivo , and reduce resistance against Docetaxel treatment 22 in cancer patient cells in vitro culture. We have also demonstrated that this PEDF modified 23 protein produces a significant decrease in cancer stem cell markers. All these properties 24 make this protein a potential application in clinical cancer therapies via co-administration 25 with chemotherapy for relapse cancer treatment. 26

This protein has been involved in self-renewal in adult stem cells and has been described 16 as anti-tumoral because of its anti-angiogenic effect. However, we show that PEDF 17 enhances resistance in breast cancer patient cells in vitro culture by favoring a slow cellular 18 cycle population (LT+). The PEDF signalling pathway could be a useful tool for controlling 19 cancer stem cells self-renewal, and therefore control patient relapse. 20 Conclusions: We demonstrate that it is possible to interfere with the self-renewal capacity of 21 cancer stem cells, induce anoikis in vivo, and reduce resistance against Docetaxel treatment 22 in cancer patient cells in vitro culture. We have also demonstrated that this PEDF modified 23 protein produces a significant decrease in cancer stem cell markers. All these properties 24 make this protein a potential application in clinical cancer therapies via co-administration 25 with chemotherapy for relapse cancer treatment. 26

BACKGROUND 29
The incidence of breast cancer has increased in recent years, due to an aging population [1]. 30 In fact, breast cancer is the leading cause of death among females in developed countries, 31 although the implementation of screening tests and the development of anti-neoplasic 32 therapies such as Trastuzumab has increased the probability of a cure in those patients [2]. 33 However, the relapse rate remains very high in different breast cancer types [3] and this is 34 why further study of new pharmacological drugs and diagnostic methods is still needed. 35 Relapse is mainly due to the tumour cell population's resistance, characterised by their 36 capacity for self-renewal, resistance to drugs, and a slow cellular cycle [4,5]. This last 37 characteristic allows for detection of this population, since this produces long retained 38 labelling[6-8]. The auto-renewal capacity of these cells is essential for stem cells to be 39 maintained throughout the life of the organism. Pigment Epithelium-Derived Factor (PEDF) 40 be important not only as an anti-neoplasic agent but also as a regulator of self-renewal in 48 TICs and patient relapse. As has been widely described, PEDF is a pleiotropic molecule, 49 presenting two domains with clearly differentiated functions, an anti-angiogenic part and a 50 second domain with neurotrophic properties, each activating different signalling pathways. 51 By using this fragmentation of the molecule into these two domains we are able to take 52 advantage of their different effects on signalling pathways, including the carboxy-terminal 53 fragment´s inhibition of the crucial TICs population´s self-renewal ability, which thus hinders 54 tumour recurrence. This is why we suggest a new therapeutic mechanism that consists in 55 the co-administration of carboxy-terminal PEDF protein fragments and chemotherapy. 56 In order to detect TICs, four different epitopes were analysed; these epitopes are implicated 57 in the cells' different processes and have been previously related with cancer stem cells in 58 literature. BCRP1 is a drug transporter from the ABC transporter family, but while it is not 59 as ubiquitous as other family members, such as MDR1, it is commonly expressed in the 60 population [16,17]. EpCAM is a transmembrane glycoprotein that is involved in cell signalling, 61 migration, proliferation, and differentiation [18,19]. This whole process is closely related to 62 the epithelial-mesenchymal transition, essential in the metastatic mechanism in which TICs 63 could play an important role. CD133 is a pentaspan membrane glycoprotein that has been 64 used as a stem cell biomarker since its discovery in 1999, although its function is still 65 unknown. AC133 is a glycosylated-isoform of CD133, recognised by a specific antibody 66

METHYL PURPLE ASSAY 164
This method[30] is used to quantify surviving cells. 5 000 cells/well were seeded in 24-well 165 plates, in a final volume of 250 µL. The next day, cells were treated with increasing doses 166 of chemotherapeutic agents and stored in a humidified incubator at 37ºC, 5% CO2 for 4 days.

ANALYSIS OF CELL MORPHOLOGY 182
500 000 treated and control cells were seeded in p100 plates. After 24 hours, cells were 183 incubated for 30 minutes with Hoechst (5µg/mL). Ten random microscopy images were 184 taken (Motic AE31.Barcelona, Spain) using an ultraviolet light. Analysis of the cytoplasmic 185 area and the separation between cells was analyzed using the Image J application 186 (https://imagej.nih.gov/ij/). 187

STATISTICAL ANALYSIS 188
Data was analyzed with R software 3.5.1 version (https://www.r-project.org/). The statistical 189 analysis was carried out using Mann-Whitney U tests (one tailed, significance level=0.05). 190 The data is expressed as the mean plus the standard error (SE). At least n=3 independent 191 experiments were performed for every assay. The results obtained are considered 192 statistically significant when p<0.05(*), p<0.01(**) and p<0.005(***). Blood cell data was first 193 transformed into a quadratic variable to improve data homogeneity. Then, normality was 194 tested using the Sapiro-Wilk test, Q-Q plots, and Levene's test for homogeneity of variance. 195 Next, groups were compared using non-parametrical Kruskall-Wallis test by ranks and 196 Wilcoxon's not paired test to analyse data in pairs. Finally, logistic regression was used to 197 model the relationship between the number of cells and progression. The odds ratio was 198 used to strengthen the association between variables (Confidence Interval 95%). 199 Ascitic fluid cells from a patient with a metastatic adenocarcinoma (Pa00 cells) also 211 present long retaining labelling cells after 8 days in culture. A methyl purple assay proves 212 that LT-cells show the same growth rate as our control cells but have a higher growth rate 213 than LT+ cells ( Figure 1B). Those cells underwent a dose-response assay with Docetaxel 214 chemotherapy. As we can see in Figure 1C, even though LT+ cells grow less than LT-or 215 control cells, they are more resistant to chemotherapy than control or LT-cells (IC50 value 216 is double in the case of LT+ cells compared to control or LT-cells). Docetaxel treatment 217 eradicates 53% of LT-cells at 2 nM, while only 27% of LT+ were affected by the same 218

concentration. 219
Finally, Pa00 LT+ and LT-sorted cells were injected in nude mice to study the 220 carcinogenicity of those populations in vivo. This experiment shows that the tumour volume 221 is similar when injecting LT-cells and control non-separated cells but smaller when injecting 222 LT+ cells ( Figure 1D). In addition, Pa00 LT-cells failed to form tumours after cell injection. 223 That´s why the frequency of tumour formation is lower when injecting LT-cells than LT+ 224 cells or control non-separated cells ( Figure 1E). 225 In short, LT+ cells show a lower growth rate and are more resistant to chemotherapy 226 than LT-or control non-separated cells. In addition, in vivo assays reveal a decrease in the 227 frequency of tumour formation in LT-cells when compared to LT+ or control cells. 228 there are no significant differences between the nucleus size with and without treatment 252 ( Figure 2B). 253 We have also studied the growing pattern of chronically PEDF-treated patient cells. 254 Close to a 40% decrease in cell growth can be seen in treated cells compared to controls 255 ( Figure 2C). This slower growth rate of treated cells translates into a lower tumour volume 256 and less necrotic area when comparing with control untreated cells. The histological analysis 257 of the xenograft, showed that PEDF treated tumours present smaller necrotic areas than 258 control xenografts. PEDF treatment also produces a compact growth of the tumoral cells 259 with dense cytoplasm and compact external matrix compared to control tumours ( Figure 2D-260 E). 261 These cells were also checked in a dose-response assay, and despite their low 262 growth rate, PEDF treated cells exhibit higher IC50 values and resistant population, as 47% 263 of PEDF treated cells survived, which means there were 16% more resistant cells than in 264 the control. This result implies a higher drug resistance to Docetaxel than control cells 265 ( Figure 2F). 266 We carried out the next assay to prove that the slow growth observed in vitro and in 267 vivo is due to the appearance of a higher number of slow-cycle cells after PEDF treatment. 268 The quantification of LT+ cells after 3DIV revealed a significant increase after PEDF 269 treatment compared to the control ( Figure 2G). A dose-response assay was performed, and 270 the results demonstrate that the resistance to docetaxel is 38% higher in PEDF treated cells 271 at 4 nM than in the control group ( Figure 2H). 272 To sum up, PEDF treated cells show changes in the cytoplasm size, a decrease of cell 273 cycle kinetics, an increase in drug resistance, and the capacity to produce tumours with not 274 only a lower growth rate but also lower tumour volume and fewer necrotic areas. 275 shows an anoikis effect ( Figure 3B) and morphological changes that can be measured 299 by cytoplasmic and nuclear areas and cell distances ( Figure 3C). Nuclear and 300 cytoplasmic areas are reduced 29% (Nuclear area: control 110±2 µm 2 , CTE 81±2.34; 301 Cytoplasmic area control: 556±18 µm 2 , CTE: 394±17 µm 2 ) while cell distances double 302 their size (control 12±1 µm 2 , CTE: 23.0±0.2 µm 2 ). 303 The next approach was a dose-response assay with Docetaxel. The result 304 demonstrates that CTE-PEDF treated cells are less resistant to the drug than control 305 cells. Just the CTE-PEDF treatment produces a 20% reduction of the initial population 306 ( Figure 3D). The IC50 value is double in the control group when compared to CTE-PEDF 307 treated cells ( Figure 3E). 308 The results demonstrate that CTE-PEDF produces anoikis in cancer patient cells and 309 reduces the drug resistance to Docetaxel. 310

CTE-PEDF depletes CSC expression markers. 318
We already know the effect produced in cells decreasing their resistance and 319 inducing anoikis as is show in previos results, when cultures are treated with CTE-PEDF 320 but, what we are now considering is whether this effect is produced because of a 321 reduction in CSC number and thus, in CSC expression markers. To solve this question, 322 Pa00 cells were injected with CTE-PEDF or PBS in nude mice. Tumours were dissected 323 and dissociated to study stemness marker expression by flow cytometry. Four different 324 epitopes, that have been previously related with cancer stem cells, were analysed: 325 BCRP1, EpCam, CD133 and AC133. CTE-PEDF treatment produces a significant 326 reduction in all studied markers in patient tumour cells Pa00. The same experiment was 327 performed with cells chronically treated in culture with CTE-PEDF, and the same 328 reduction in marker expression was observed (Figure 4). 329 Breast cancer is the most common cancer in women worldwide. We study how to prevent 340 relapse, taking breast cancer as an example because of its impact on the population. It's 341 important to distinguish tumour initiating cells, responsible for tumour formation, and also 342 probably for tumour relapse, and this is why we have focused on marker expression to detect 343 this kind of cells (TICs). A combination of different epitopes found in the literature was 344 applied to detect these cells [34][35][36]. Later, our objective was to discover a possible new 345 treatment for patients in which TICs were detected. We demonstrate that PEDF (and Ascitic fluid cells from a patient with a metastatic adenocarcinoma (Pa00 cells) have 365 been used in this paper parallel to the use breast cancer cell lines. Ascitic fluid should be an 366 acellular liquid resulting from inflammation events. After metastasis progression, ascitic fluid 367 could contain tumour cells that grow in culture conditions [44,45]. These cells also present 368 long retained labelling after 8 days in culture, so we sorted these cells according to their 369 DDAO content. This experiment showed that LT+ cells grow more slowly than LT-cells and 370 respond less to chemotherapy than control or LT-cells. We postulated that this is the reason 371 shown data, leads us to consider that the difference between these two cell types is the 385 initial tumour capacity. This data correlates with previous data from research groups that 386 correlate chemorresistance, tumorigenicity potential, and slow-cycling in some tumour 387 cells [39][40][41]. 388 All this data supports the hypothesis that LT+ cells present slow cell cycle division, higher 389 chemotherapy resistance, and higher frequency of tumour formation than LT-or control cells. 390 These characteristics confirm the idea that LT+ cells could be involved in relapse and 391 metastasis progression in breast cancer. 392 indicating higher resistance to chemotherapy. One possible explanation for this effect could 398 be that the low growth rate leads to more time for these cells to repair chemotherapy damage, 399 even though another hypothesis could contribute to this stage, such as higher expression of 400 ABC drug transporters in these cells. 401 The PEDF stem cell self-renewal modulator protein modify the carcinogenicity of cancer 428 stem cells and could be a useful tool to control its self-renewal and therefore control patient 429 relapse. We have designed a transgenic peptide derived from PEDF to interfere with the 430 self-renewal capacity of cancer stem cells, inducing anoikis in vivo and reducing resistance 431 in cells from cancer patients. We have also shown that this PEDF-derived protein produces 432 a significant decrease in cancer stem cell markers, making this protein a potential tool for 433 delaying patient relapse.