Identification of changes in biomarkers relevant for breast cancer biology occurring in a novel 3D-Biosilk model

(1)

IN

DEGREE PROJECT BIOTECHNOLOGY, SECOND CYCLE, 30 CREDITS

STOCKHOLM SWEDEN 2021,

Identification of changes in

biomarkers relevant for breast

cancer biology occurring in a

novel 3D-Biosilk model

EMMY STÅHL

KTH ROYAL INSTITUTE OF TECHNOLOGY

SCHOOL OF ENGINEERING SCIENCES IN CHEMISTRY, BIOTECHNOLOGY AND HEALTH

(2)

Identification of changes in

biomarkers relevant for breast

cancer biology occurring in a

novel 3D-Biosilk model

Emmy Ståhl

Supervisor: My Hedhammar

Co-supervisor: Caterina Collodet

Kungliga Tekniska Högskolan, KTH Royal Institute of Technology

School of Engineering Sciences in Chemistry, Biotechnology and Health Stockholm 2021

(3)

1

Abstract

Breast cancer is the most common cancer among women. It is a heterogenous and complex disease composed of several subtypes, each with distinct morphological and clinical implications [1]. To model and study cell biology, tissue morphology, molecular mechanisms and drug actions, cell cultures are canonically used [2]. Today two-dimensional (2D) models are still widely the preferred method for culturing cells in vitro [3]. A drawback with 2D models is that the microenvironment in these models does not mimic the in vivo structure of tumors and tissues, lacking three-dimensional (3D) cell-cell and cell-extracellular matrix (ECM) interactions [2]. Due to the disadvantages of 2D models, 3D cultures have become an increasingly interesting alternative to solve the need for a reliable preclinical model for drug testing and the study of cancer biology.

To develop a relevant tool for cancer research, the laboratory of professor My Hedhammar is currently establishing a 3D model of breast cancer. In such novel model, Biosilk is used as scaffold to grow immortalized cell lines representative of the three major classes of breast cancer (i.e. MCF-7 (luminal-like), SKBR-3 (HER2-overexpression) and MDA-MB-231 (triple- negative)). Since transcriptional signatures can be used to classify and study breast cancers, it is important to investigate if and how growth in 3D-Biosilk can impact gene expression profiles. The hypothesis tested in this study was that cells cultured in 3D-Biosilk have differences in expression of biomarkers relevant to breast cancer biology, when compared to the same cell lines cultured in 2D. To examine this, 3D-Biosilk models were created and evaluated to ensure their quality and reproducibility, for instance, the scaffold structure was monitored by brightfield microscopy, the construct’s area was measured with ImageJ, staining with phalloidin confirmed the presence of cells as well as their attachment to the construct, and Alamar blue was used to assess the cellular metabolic activity. Differences in gene expression of target genes were investigated using reverse transcription quantitative PCR (RT- qPCR), which revealed statistically significant changes depending on whether the cells were cultivated in 2D or a 3D-Biosilk model. For cell line MDA-MB-231 three genes were found, for SKBR-3 two genes were found and for MCF-7 four genes were found. The expression of one gene which was found downregulated in MCF-7 cultured in 3D-Biosilk (i.e. ZO-1) was validated at protein level by immunofluorescence. In conclusion, cultivating cells in 3D-Biosilk indicates a more aggressive phenotype.

Keywords: 3D-Biosilk, breast cancer, 3D cell culture, recombinant spider silk, ZO-1

(4)

2

Sammanfattning

Bröstcancer är den vanligaste formen av cancer som drabbar kvinnor. Det är en heterogen och komplex sjukdom som består av flera undergrupper, var och en med distinkt morfologi och kliniska implikationer [1]. För att modellera och studera cellbiologi, vävnadsmorfologi, molekylära mekanismer och läkemedels effekter används cellkulturer [2]. Idag är två- dimensionella (2D) modeller fortfarande den mest använda metoden för att odla celler in vitro [3]. En nackdel med 2D-modeller är att mikromiljön i dessa modeller inte imiterar in vivo strukturen av tumörer och vävnader, då de saknar tre dimensionella (3D) cell-cell och cell- extracellulär matrix (ECM) interaktioner [2]. På grund av nackdelarna med 2D-modeller, har 3D-modeller blivit mer intressanta som alternativ för att lösa behovet av en pålitlig preklinisk modell för läkemedelstestning och för studier av cancerbiologi.

För att utveckla ett redskap som är relevant för cancerforskning etablerar professor My Hedhammars laboratorium en 3D-modell av bröstcancer. I en sådan ny modell används Biosilk som byggnadsställning för att odla odödliga cellinjer som är representativa för de tre huvudklasserna av bröstcancer (i.e. MCF-7 (luminal-lik), SKBR-3 (HER2-överuttryckt) och MDA- MB-231 (trippel-negativ)). Eftersom transkriptions signaturer kan användas för att klassificera och studera bröstcancer är det viktigt att undersöka om och hur tillväxt i 3D-Biosilk kan påverka genuttrycksprofiler. Hypotesen som testades i denna studie var om cellkulturer i 3D- Biosilk kan ha signifikanta skillnader i uttryck av biomarkörer, relevanta för bröstcancerbiologi, vid jämförelse av samma cellinje kultiverad i 2D. För att testa detta utvärderades kvalitén och reproducerbarheten av 3D-Biosilk konstruktionen med hjälp av olika kvalitetstester.

Strukturen granskades med brightfield mikroskopi, arean av konstruktionen mättes med ImageJ, infärgning med phalloidin bekräftade cellnärvaro och cellvidhäftning till modellen.

Alamar blue utfördes för att bedöma den cellulära metaboliska aktiviteten i modellen.

Förändringarna av målgenernas genuttryck undersöktes med kvantitativ omvänd transkription PCR (RT-qPCR) och detta påvisade en statistiskt signifikant skillnad i genuttrycket beroende på om cellerna odlats i 2D- eller 3D-Biosilk modeller. I cellinje MDA-MB-231 hittades tre gener, i cellinje SKBR-3 hittades två gener och i cellinje MCF-7 hittades fyra gener.

Genuttrycket för en av dessa gener i cellinje MCF-7, som var kultiverad i 3D-Biosilk, var ned- reglerad (i.e. ZO-1). Detta kunde valideras på proteinnivå med immunofluorescens.

Sammanfattningsvis, celler odlade i 3D-Biosilk visar på en mer aggressiv fenotyp.

Nyckelord: 3D-Biosilk, bröstcancer, 3D cellkultur, rekombinant spindeltråd, ZO-1

(5)

3

1 Introduction

1.1 How breast cancer affects society

Breast cancer is the most common type of cancer affecting women, causing almost one in four female cancer cases worldwide every year [4]. In 2018 there were more than 2 million estimated new cases of breast cancer [4, 5]. When writing this report, there are currently 10.643 ongoing trials for breast cancer drugs [6]. Developing new cancer drugs is expensive, the cost of developing a new drug varies but it is estimated to cost around $2.6 billion from discovery until sales of a finished product. However, many of the developed drugs fail during clinical trial phase III [7], due to several factors which plays crucial roles for why a drug would fail during clinical trials, e.g. lack of efficiency, safety concerns, funding of the drug development, etc. [8].

1.2 Breast cancer classification

Breast cancer is a heterogenous and complex disease composed of several subtypes, each with distinct morphological and clinical implications. To tailor a patient´s therapy, breast cancers are routinely classified by assessing the expression of the three histological markers human epidermal growth factor receptor 2 (HER2), progesterone receptor (PR), and estrogen receptor (ER). In the past decades, several studies have identified collections of genes, whose changes in expression correlate with different prognosis for breast cancer patients. These results indicate that, to complement the histological evolution, patients can be additionally stratified through analysis on their gene expression [1, 3]. Sørlie et al. did pioneering work by analyzing the transcriptome of breast carcinomas, which resulted in the subtyping of patients into five groups, luminal A and luminal B subtypes (ER-positive/HER2-negative), basal-like subtype (ER-negative/HER2-negative), HER2-positive and normal breast-like tissue [1, 9, 10].

1.2.1 Breast cancer cell lines

Cell lines have been used as a tool for in vitro study of biological functions and phenomena for more than half a century. Currently, there are at least 92 well-documented breast cancer cell lines routinely used to better understand tumor biology [11].

In the herein presented study, three breast cancer cell lines were used being MCF-7, SKBR-3, and MDA-MB-231. The MCF-7 is the most commonly used and studied breast cancer cell line within research communities, and was established in 1973 by H. Soule and co-workers [12]

(7)

5 from the adenoma of a 69-years-old Caucasian female [13]. It has an epithelial-like morphology and grows in monolayers [14]. It belongs to the Luminal A subtype (ER⁺, PR^+/- and HER2^-), and often has a good response to endocrine therapy or chemotherapy [3]. The second cell line used herein, SKBR-3, was derived in 1970 by G. Trempe and L.J. Old from a malignant adenocarcinoma of a 43-year-old Caucasian female [15]. This cell line over-express the HER2 gene product, (ER^-, PR^- and HER2⁺) and is often used as a positive control when screening for HER2 [16]. It has been seen to respond to treatment with Trastuzumab or chemotherapy [3].

The third cancer cell line, MDA-MB-231, belongs to the claudin-low group, which is a subgroup of the basal-like subtype (ER^-, PR^- and HER2^-). It was derived from an adenocarcinoma of a 51- year-old Caucasian female [17] and is a highly aggressive, invasive and poorly differentiated type of cancer [18]. It has showed an intermediate response to chemotherapy [3].

Scientists are currently developing novel breast cancer in vitro models, aiming to support the research field with clinically relevant tools highly characterized. An example of such new models is given by the Wood cell line, produced by Cellaria Biosciences and derived from the infiltrating ductal and lobular carcinoma of the breast of a Caucasian patient [19].

1.3 Why 3D models?

Despite major advances in the cancer field, there is still a need to develop reliable preclinical models for drug discovery and the study of breast cancer biology. Cell cultures are canonically used as models to study cell biology, tissue morphology, molecular mechanisms and drug actions [2]. This in vitro model is beneficial since it provides an infinite supply of a relatively homogenous cell population capable of self-replication in culture medium. 2D cultures are still widely the preferred method for culturing cells in vitro [3]. Cell growth in 2D models occur in monolayers at the bottom of a plastic support (e.g. petri dish), where each cell has unlimited access to nutrients and oxygen in the medium. The model is simple, with low maintenance cost and the ability to perform functional tests. However, the microenvironment in these models does not mimic the in vivo structure of tumors or tissues, lacking 3D cell-cell and cell- extracellular matrix (ECM) interactions [2]. Due to the disadvantages of 2D models, 3D cultures have become an increasingly interesting alternative. An ideal 3D model would include cell-cell and cell-ECM interactions, have a tissue-specific stiffness, a combination of tissue-specific cells and matrix, and a gradient access to oxygen and nutrients [20].

(8)

6 1.3.1 Different types of models

Techniques to obtain 3D cultures can be divided into three groups, non-scaffold, anchorage- independent and scaffold-based [20]. The scaffold-based technique mimics cell-ECM interactions which are known to play an important role in drug resistance of tumors [20], the cells are allowed to grow in all directions on a supporting material [21]. Hydrogels are an example of a scaffold-based technique and consists of a network of cross-linked polymer chains or complex protein molecules of synthetic or natural origin swollen with water. Due to high water content, the matrix obtains biophysical properties similar to natural soft tissues and can support cell adhesion and sequestration [22-24]. Depending on which polymer is being used, different types of hydrogels can be obtained, ECM protein-based, natural, and synthetic hydrogels. The cells can grow on the surface or embedded in the hydrogel [23]. The stiffness of hydrogels affects the phenotype and growth of cancer cells [25].

Cells growing without a supporting scaffold, called a scaffold-free or non-scaffold model, are based on cells spontaneous self-assemble into clusters or spheroids, which is a natural phenomenon [21, 25]. Hanging drop is a technique based on the assembly of single cells into spheroids cultivated in a droplet were single cells are allowed to aggregate into a spheroid [21].

This new 3D-Biosilk model is somewhere between a spheroid and a scaffold-based model, the cells grow within a scaffold which is free-floating in a medium suspension.

1.3.2 Biosilk

Biosilk consists of a recombinantly produced spider silk protein, 4RepCT, derived from the spider Euprosthenops australis [26]. It is recombinantly produced in E. coli, purified using affinity chromatography and functionalized to harbor the cell adhesion motif from fibronectin (FN). The FN-4RepCT, referred to as Biosilk, has the ability to self-assemble into microfibers, allowing cells to adhere to the 3D structure and proliferate in an ECM-like environment [27].

Furthermore, Biosilk can easily be sterilized, is well tolerated in vivo and of non-animal origin [28].

1.3.3 Comparison between Biosilk and hydrogel

A study published 2019 by Johansson et al. indicated that cells grown in an alginate hydrogel have a limited proliferation, rounded morphology, and no spreading, however cells grown in 3D-Biosilk had a clear expansion with elongated morphology spread out within the Biosilk

(9)

7 network. In the report it was hypothesized that several integrins needs to be assembled for the formation of focal adhesions to trigger organization of the cytoskeleton, so cells can achieve accurate morphology. They could see that focal adhesion points were promoted in a 3D-Biosilk scaffold. Hydrogels with RGD motif and thin alginate chains (i.e. one saccharide unit thick) could not support the gathering of several integrins which in turn prohibits the formation of focal adhesion points important for cell growth [27].

(10)

8

2 Materials & Methods

2.1 Cell culture

Three cell lines, MDA-MB-231, SKBR-3 and MCF-7, were obtained from American Type Culture Collection (ATCC) and kept in T-flask culture with a media mixture of DMEM low glucose (Thermo Fisher, 11574446), supplemented by 10% fetal bovine serum (Thermo Fisher, 16140- 071) and 1% penicillin-streptomycin 10.000 U/ml (Thermo Fisher, 11548876).

The Wood cell line, provided by Cellaria, was cultured in RETM basal medium supplemented with 5% heat-inactivated fetal bovine serum (Hyclone, SH3007103HI), 3% RETM supplement, 1% penicillin-streptomycin 10.000 U/ml, and Cholera toxin with a final concentration of 0,025 µg/ml (EMD Millipore, 227036).

Regular splitting of the cell cultures with TrypLE (Thermo Fisher, 12605-028) was performed when cells reached a maximum confluency of around 80%. Medium was changed every second day and all cell cultures were kept in a cell incubator at 37°C and 5% CO2. Cell counting was performed using a Bürker chamber (0,100 mm; 0,0025 mm²; 10⁴).

2.2 3D-Biosilk model

2.2.1 Creating the 3D-Biosilk model

The 3D-Biosilk model was created using FN-Biosilk (3 mg/ml in PBS, endotoxin level below 200 EU/ml). A schematic overview of the process of creating a 3D-Biosilk model can be found in figure 1. For each 3D-Biosilk, 10.000 cells from a cell concentrate of 5.900 cells/µl were gently mixed with 8,3 µl FN-Biosilk to form a 10 µl droplet. Such droplet was placed on a Teflon mold anchored at the bottom of a 24 well-plate. A 3D structure was formed by rapidly pipetting air into the droplet, this was followed by an incubation for 20 minutes at 37°C to allow for the stabilization of the 3D-Biosilk. The 3D-Biosilk model was transferred, using a metal spoon, from the 24 well-plate to a hydrophobic 96 well-plate (Sarstedt, 83.3924.500). To facilitate the transfer, medium was added to both the 24 well-plate and the 96 well-plate. The medium was removed and a 3D-printed cap was placed on the plate. The 3D-printed cap was then connected to a VACUSAFE aspiration system, allowing to create a pressure difference and burst the air bubbles in the 3D-Biosilk. Notably, the pressure difference was applied and quickly released by disconnecting the cap, such procedure was repeated twice.

(11)

9 Fresh medium was added, and the cells were left to grow in the incubator for 7 days, during this time the medium was changed every other day. 2D controls were created by plating cells on tissue culture treated (TCT) plates.

Figure 1: Schematic representation of the formation of 3D-Biosilk.

2.2.2 Alamar blue viability assay

Both the 2D control and the 3D-Biosilk were stained with Alamar blue Cell Viability Reagent (Thermo Fisher, #DAL 1100) at three time-points during a 7-day period, usually day 1, 3 or 4 and 7. Alamar blue was diluted 1:10 in the culturing medium. The old media from the 2D and 3D models were removed and 180 µl Alamar blue dye mixture was added to the models, three wells with only Alamar blue was used as blank. The plates were incubated for two hours at 37°C. After incubation, 100 µl of media was transferred to a new 96 well-plate (Greiner, 655161) the plate was read using a plate reader (CLARIOstar, BMG Labtech) with an excitation and emission wavelength of 544 nm and 595 nm respectively. Fluorescence intensities of each well were obtained and the data analyzed with Excel, where the mean value of the blank was subtracted from each fluorescence intensity. For each condition, the average value and the standard deviation were calculated and plotted for comparison of 2D and 3D cultures.

2.2.3 3D-Biosilk area

A Nikon stereo microscope (SMZ745T) was used to take photographs of the 3D models. The photographs were analyzed with ImageJ to calculate the area of the 3D-Biosilk model. The scale was set to 0,126 pixels/µm corresponding to a scale bar obtained from the microscope.

2.2.4 Staining with phalloidin and DAPI

Both 2D and 3D models were washed with PBS and cells were fixed using 4%

paraformaldehyde in PBS for 15 minutes, afterwards the cells were washed three times with PBS and then kept in PBS (Mg+, Ca+) until cell staining. For staining, all steps were performed

(12)

10 using 200 µl solution per well, under gentle agitation. The fixed cells were permeabilized with 0,2% Triton X-100 in PBS for 15 minutes at room temperature (RT). This was followed by two washes in wash buffer, Tween-20 (0,1%) in PBS, and by blocking with 1% bovine serum albumin (BSA) in PBS for 20 minutes done at RT. Samples were then incubated, protected from light, for 40 minutes with phalloidin Alexa Fluor 488 (Invitrogen, A12379) diluted 1:400 in PBS, to stain actin-filaments. An incubation step of 15 minutes with Sudan Black (Sigma Aldrich, 199664-25G) 0,3% (w/v) working solutions dissolved in 70% ethanol was added for the 3D- Biosilk models to quench the autofluorescence signal from the silk [29]. Both models were washed three times with wash buffer before the nuclei were stained with 4’,6-Diamidine-2’- phenylindole dihydrochloride ((DAPI) Sigma Aldrich, 10236276001) diluted 1:1000 in PBS for 5 minutes. One final wash was done before imaging with the ANDOR camera, Leica DMI6000B.

2.2.5 Immunofluorescence staining

Previously fixed cells were permeabilized with 0,2% Triton X-100 in PBS for 15 minutes at RT, followed by blocking with 10% goat serum in PBS for 1 hour. An overnight incubation in a humidified chamber at 4°C was done using the primary antibodies ZO-1 (Invitrogen, 33-9100) diluted 1:50 and ERα (Invitrogen, MA5-14501, clone SP1) diluted 1:150 in a mixture of 50%

block buffer and 50% wash buffer. The following day, three 5 minute washes with washing buffer were performed before 1 hour incubation at RT shielded from light with Alexa 488- labeled anti-mouse secondary antibody (Invitrogen, A32723) and Alexa 546-labeled anti rabbit (Invitrogen, A11010) diluted 1:500 in a mixture of 50% block buffer and 50% wash buffer. After the incubation with secondary antibodies, 3D-Biosilk models were washed once before an incubation step with Sudan Black (Sigma Aldrich, 199664-25G) for 15 minutes. Both models were then washed three times with wash buffer, before the nuclei were stained with DAPI diluted 1:1000 in PBS for 5 minutes. Two final washes were performed before imaging with the ANDOR camera. 2D models fixed on coverslips were mounted on microscopy slide before imaging.

2.3 Sample preparation and RT-qPCR 2.3.1 RNA extraction and cDNA synthesis

Medium was removed and cells were washed once with PBS, afterwards cells were lysed through addition of cell lysis buffer for 2D (RNeasy minikit, QIAGEN 74104) or 3D (RNeasy microkit, QIAGEN 74034) to obtain a volume of 350 µl total. 2D cultures were collected by

(13)

11 scraping the bottom of the 6 well-plate with a cell scraper, the cells were collected in an eppendorf tube. For the 3D cultures 5 or 3 Biosilks were pulled together for lysis done at day one and day seven, respectively. A syringe with a needle diameter of 0,4 µm was used to homogenize the samples via mixing 5-6 times, lysates were then transferred to an eppendorf tube.

Total RNA was extracted from lysed 2D models by using a RNeasy minikit (QIAGEN 74104). For the 3D-Biosilk, Biosilk debris were removed via centrifugation (12000g, 5 minutes), supernatant was then transferred into a new eppendorf tube and processed using the RNeasy microkit (QIAGEN 74034). To validate the quality and quantity of the RNA obtained from the cell cultures, Nanodrop measurements were performed.

Synthesis of cDNA was done using PrimeScript RT Kit (TaKaRa, RR037A) accordingly to the manufacturer’s instructions.

2.3.2 Quantitative RT-PCR

Reverse transcription quantitative real-time PCR (RT-qPCR) reactions were performed in a CFX96^TM Real-Time System, C1000^TM Thermal Cycler from BIO-RAD with a SYBR Green Universal Assay (BIO-RAD, #1725271). In each RT-qPCR well 18 μL of mix were added, the mix contained 10 µl of SYBR Green 2X, 0,6 µl of a forward and reverse primer stock at 10 M, and 7,4 µl of H2O. The primers had a final concentration of 0,3 µM/reaction. Before performing the RT-qPCR using the protocol below 2 µl containing 10 ng of cDNA were added to each well.

RT-qPCR protocol

1. 95°C for 30 seconds

2. 39 cycles of: 95°C for 10 seconds 60°C for 30 seconds 3. 65°C for 5 seconds

4. 95°C for 50 seconds

Normalized values were calculated by dividing the mean expression value by a factor equal to the geometric mean of the normalization genes (i.e. beta-2 microglobulin (B2M), transferrin receptor (TFRC) for MCF-7, SKBR-3, MDA-MB-231 and beta-2 microglobulin (B2M), glyceraldehyde-3-phosphate dehydrogenase (GAPDH) for the Wood cell line) and applying the ΔΔCt method [30].

(14)

12 2.3.3 Primer design

Identification of relevant genes was done through browsing existing breast cancer gene panels, followed by a literature search. The primer pairs were created using the website

‘Primer3’ [31] and the generated primers were tested in UCSC In-Silico PCR prior to ordering.

In table 1 a list of all genes for which primers were created as well as the primer sequences can be found. All primer pairs were initially tested in all three breast cancer cell lines using RT-qPCR.

Primer efficiency was tested through a series of cDNA dilutions in a RT-qPCR. The dilution series consisted of cDNA concentrations; 50 ng/reaction, 5 ng/reaction, 0,5 ng/reaction, 0,05 ng/reaction, and 0,005 ng/reaction.

Gene Name Forward Primer Reverse Primer

VCAN Versican GTCTTTACCGCTGTGACGTC AAACAAGCCTTCTGAGCAGC HIF1A Hypoxia inducible factor 1

subunit alpha

TGAGAGAAATGCTTACACACAGA CTTCCTCGGCTAGTTAGGGT

IGF1A Insulin-like growth factor 1 CAGCAGTCTTCCAACCCAAT ACAGCGCCAGGTAGAAGAGA IGF1B Insulin-like growth factor 1 CTCAGACAGGCATCGTGGAT GGTGCGCAATACATCTCCAG FGFR3 Fibroblast growth factor

receptor 3

TCCTCGGAGTCCTTGGGG CCGAAGACCAACTGCTCCT

CD24 CD24 molecule TCCAACTAATGCCACCACCA GACGTTTCTTGGCCTGAGTC JAM2 Junctional adhesion molecule 2 GCCAAAACCTGGAAGAGGAT CCACAGTTCCACTCAGAGCA ZO-1 Zonula occludens 1 CGTCCTTTTCCTGCTTGACC TCTGATTCTACAATGCGACGA MMP3 Matrix metallopeptidase 3 TGCTTTGTCCTTTGATGCTG AAGCTTCCTGAGGGATTTGC MMP7 Matrix metallopeptidase 7 TGCTCACTTCGATGAGGATG TGGGGATCTCCATTTCCATA MMP9 Matrix metallopeptidase 9 TAAGGACGACGTGAATGGCA CTCTGAGGGGTGGACAGTG MMP14 Matrix metallopeptidase 14 GTGACGGGAACTTTGACACC TTATTCCTCACCCGCCAGAA CLDN1 Claudin 1 CCCTATGACCCCAGTCAATG CCAGTGAAGAGAGCCTGACC CLDN3 Claudin 3 AGGCGTGCTGTTCCTTCTC CACCACGGGGTTGTAGAAGT CLDN4 Claudin 4 CTGTGGCCTCAGGACTCTCT ACCCTCCCAGGCTCATTAGT CLDN7 Claudin 7 GGGGATGATGAGCTGCAAAA CACAAACATGGCCAGGAAGC

DSC2 Desmocollin 2 TGATTTAGCCCAGCAGAACC CCTCCGTTTTTGATTCCTGA CD44 CD44 molecule (Indian blood

group)

CGGACACCATGGACAAGTTT TGGAATACACCTGCAAAGCG

(15)

13 ITGAV Integrin subunit alpha V TTGCCCTCAGTGAAGGAGAT AGCACTGAGCAACTCCACAA

Table 1: Name of breast cancer markers with the sequence for forward and reverse primer.

(16)

14

3 Results

3.1 Evaluating the quality of 3D-Biosilk breast cancer constructs

The first step was to investigate if the 3D-Biosilk constructs were reproducible and of good quality. To achieve this aim several parameters were evaluated, for instance, the scaffold structure was monitored by brightfield microscopy, the construct’s area was measured, presence of cells and attachment were confirmed by phalloidin staining, proliferation rate was measured with Alamar blue, and the expression of a biomarker affected by culturing condition was evaluated.

For the proliferation and biomarker testing only the MCF-7 cell line was used since, from results previously generated in the laboratory, MCF-7 cells have shown to give the most robust and reproducible changes when comparing 2D and 3D culturing conditions.

3.1.1 Brightfield images of the 3D-Biosilk structure

To create the 3D-Biosilk model, a solution comprising cells and FN-Biosilk was manipulated via introduction of air bubbles through rapid pipetting, allowing the formation of a foam. After stabilization of this structure, cells were integrated and could grow onto the 3D-Biosilk.

However, cell growth could be affected by their proximity to air bubbles, as it does not recapitulate a physiological condition. To reduce the potential negative effect of air bubbles, a pressure difference was applied twice, to remove the air bubbles present into the 3D-Biosilk models after the stabilization step. To ensure that the 3D-Biosilk had a structure free of air bubbles, pictures were taken at various time points as shown in figure 2. At day 0, the 3D- Biosilk already showed an initial collapse in the bubbles following the pressure difference application (fig. 2ii). After approximately 48 hours most of the air bubbles had left the constructs and the 3D-Biosilk had an ECM-like structure similar to the one visible at day 7.

(17)

15

Figure 2: i) To the top left: day 0 before pressure difference is applied, ii) top center: day 0 after the pressure difference has been applied, iii) to the top right: 3D-Biosilk structure after one day of cultivation, iv) bottom left:

two days of cultivation, v) bottom center: the structure after three days of cultivation, vi) bottom right: seven days of cultivation. All pictures were taken with 4X magnification, scale bar: 203,237 µm.

3.1.2 Area measurement

Area measurements was performed on 3D-Biosilk models to estimate the reproducibility among the created foams. The 3D-Biosilks were placed under a Nikon stereo microscope and pictures of 3D-Biosilk models, here with cell line SKBR-3, were obtained. To analyze the area, each 3D-Biosilk model were manually encircled and analyzed with ImageJ software. The results indicated that most of the 3D-Biosilk models had similar sizes, with a few bigger or smaller models. The models did not vary greatly in size throughout the cultivation period (fig. 3).

(18)

16

Figure 3: The picture in the top left corner were taken of SKBR-3 3D-Biosilk models at day 7 and the picture below visualize how the area was circled manually (green) and analyzed with ImageJ. The growth area of 3D- Biosilk models in the graph shows three time points when areas were measured during a 7-day cultivation period of SKBR-3 cells.

3.1.3 Staining with phalloidin and DAPI

Staining of the actin filaments and nuclei was performed in order to compare how the breast cancer cells grow in 2D and 3D-Biosilk models. MDA-MB-231 cells were fixed at one and seven days after seeding and then stained using phalloidin, a bicyclic peptide which stains the actin- filaments, and DAPI (fig. 4). The results indicated that cells proliferated in both 2D and 3D conditions, since a higher cell density can be seen when comparing day 7 and day 1.

Additionally, different patterns of proliferation could be observed, with the cells growing in a 2-dimensional fashion in 2D and in an ECM-like way in 3D-Biosilk.

Figure 4: Staining of actin filaments (green) and nucleus (blue) at day 1 and day 7 in both 2D and 3D-Biosilk models with cells from cell line MDA-MB-231. Pictures are taken with a 10X magnification, scale bar: 100 µm.

(19)

17 3.1.4 Alamar blue

To validate cell growth of MCF-7 in both 2D and 3D cultures Alamar blue, an assay which determines cells viability through measurement of the metabolic activity, was performed.

Alamar blue was chosen since the reagent is not toxic for the cells and can hence be used to analyze cell growth during a time course. The viability results obtained showed a similar growth in both models during a 7-day period, see figure 5 below.

Figure 5: Cell viability comparison between 2D (blue) and 3D (orange) models, with measurements for day 1, 4 and 7.

3.1.5 Measuring N-cadherin as biomarker for 3D-Biosilk quality

In previous experiments done in the laboratory, CDH2 had been identified as a gene upregulated in 3D, when comparing 2D and 3D cultured MCF-7 (fig. 6). For this reason, CDH2 was used as a benchmark to estimate the quality of the new sets of 3D-Biosilk. For sample preparation, MCF-7 were cultured in 2D or 3D-Biosilk and lysed after 24 hours and 7 days in culture. Total RNA was extracted, having three technical replicates for each condition.

Samples were then analyzed by RT-qPCR in order to measure their relative expression of CDH2 (fig. 6). The results showed that CDH2 levels increased over time for both 2D and 3D samples, with the upregulation in 3D being significantly higher than in 2D at day 7.

0 1000 2000 3000 4000 5000 6000 7000 8000

D1 D4 D7

A.U.

MCF7 Time course

2D 3D

(20)

18

Figure 6: To the left a time course of MCF-7 with an upregulation in CDH2 gene in 3D models compared to 2D.

Relative mRNA levels of CDH2 is displayed following qPCR analysis in MCF-7 cultured in 2D or 3D-Biosilk for one, four or seven days. To the right is a time course of MCF-7 with an upregulation in CDH2 gene in 3D models compared to 2D performed for this report. Values are represented as fold-change of the mean± SD (n=9). Fold- change was calculated comparing to the control MCF-7 2D at day 1. ***P < 0,001; ****P < 0,0001.

3.2 Investigating how culturing cells in 3D-Biosilk affects gene expression changes Once confirmed that the 3D-Biosilk model could be recreated, the next step was to investigate at transcriptional level whether the 3D culturing condition impacted biomarkers relevant for cancer biology. To answer this biological question, a set of primers to measure the expression level of relevant genes was designed and tested. Additionally, RNA samples derived from the three cell lines SKBR-3, MCF-7 and MDA-MB-231 and representative of two different time points (i.e. day one and day seven) as well as two culturing conditions (2D and 3D-Biosilk) were prepared and used for RT-qPCR analysis.

3.2.1 Selection of relevant targets to use as read-out for 3D-Biosilk effect on transcription Markers for angiogenesis, adhesion and migration, and epithelial-mesenchymal transition (EMT) were selected and tested to evaluate gene expression of the three cell lines. As can be seen in figure 7, several genes do not affect only one cellular function. For simplicity, only the main function affected by the gene product is exhibited here and 9 genes were identified to have an impact on both adhesion and migration as well as EMT.

(21)

19

Figure 7: A VENN-diagram of biomarkers and their associated cell function. Genes listed in the intersection of two circles are associated to both cellular functions.

An initial analysis with RT-qPCR was performed to investigate which primer pairs were working and if the cell lines were expressing the corresponding genes. For this evaluation, melting curves and quantification cycle (Cq) values were considered. A total of 19 primer pairs were designed and ordered, after an initial test 4 primer pairs were removed. VCAN and MMP9 were removed due to not clear peaks in the melting curves suggesting primer-dimers and/or problems in the amplification of their targets. Additionally, IGF1α was not expressed and MMP7 had a very low expression in all three cell lines, with Cq-values above 36 indicating a very low expression of target genes. In table 2 the Cq-values for each cell line and all primer pairs can be seen.

GENE NAME Cq-VALUE

MCF-7

Cq-VALUE SKBR-3

Cq-VALUE MDA-MB-231

VCAN Versican 32,2 36,5 35,0

HIF1α Hypoxia inducible factor 1 subunit alpha

23,3 24,3 22,4

IGF1α Insulin-like growth factor 1 - 18,4 -

(22)

20 IGF1β Insulin-like growth factor 1 - 32,1 34,3

FGFR3 Fibroblast growth factor receptor 3

30,5 33,1 34,4

CD24 CD24 molecule 20,7 20,3 31,1

JAM2 Junctional adhesion molecule 2 32,1 33,9 37,0

ZO-1 Zonula occludens 1 24,2 25,2 24,7

MMP3 Matrix metallopeptidase 3 35,6 31,6 36,0 MMP7 Matrix metallopeptidase 7 - 36,8 36,2 MMP9 Matrix metallopeptidase 9 30,5 30,8 37,2 MMP14 Matrix metallopeptidase 14 34,4 36,4 23,9

CLDN1 Claudin 1 27,7 27,8 31,5

CLDN3 Claudin 3 25,0 26,7 32,1

CLDN4 Claudin 4 24,2 24,4 26,2

CLDN7 Claudin 7 22,9 22,7 26,6

DSC2 Desmocollin 2 27,8 23,8 27,7

CD44 CD44 molecule (Indian blood group)

25,5 26,1 21,7

ITGAV Integrin subunit alpha V 24,0 25,2 26,1 Table 2: Data from initial testing of ordered primer pairs in all three cell lines.

The remaining 15 primer pairs were tested for efficiency to evaluate how efficiently the PCR product double in size during the logarithmic phase of a PCR reaction. Efficiencies between 95-110% were deemed as good, meaning that they were considered being comparable to each other [32]. Table 3 show an overview of the efficiencies obtained. IGF1β and FGFR3 were excluded from further analysis due to their efficiencies.

GENE NAME EFFICIENCY CELL LINE

VCAN Versican - -

HIF1α Hypoxia inducible factor 1 subunit alpha 98% MCF-7

IGF1α Insulin-like growth factor 1 - -

IGF1β Insulin-like growth factor 1 139% SKBR-3 FGFR3 Fibroblast growth factor receptor 3 -90% MCF-7

CD24 CD24 molecule 100% MCF-7

JAM2 Junctional adhesion molecule 2 109% MCF-7

ZO-1 Zonula occludens 1 95% MCF-7

(23)

21 MMP3 Matrix metallopeptidase 3 112% SKBR-3

MMP7 Matrix metallopeptidase 7 - -

MMP9 Matrix metallopeptidase 9 - -

MMP14 Matrix metallopeptidase 14 97% MDA-MB-231

CLDN1 Claudin 1 99% MCF-7

DSC2 Desmocollin 2 113% SKBR-3

CD44 CD44 molecule (Indian blood group) 95% MCF-7

ITGAV Integrin subunit alpha V 99% MCF-7

Table 3: Gene and gene name with corresponding Cq-value from initial testing in all three cell lines. The efficiency was tested in selected cell lines for which the Cq-value was good.

3.2.2 Identification of genes differentially expressed following culturing in 3D-Biosilk

The next step consisted in testing if the culturing condition was having any impact on gene expression. To generate the samples, cells were grown in 2D or 3D-Biosilk and lysed after 24 hours or 7 days in culture; for each condition three replicates were prepared. Using this experimental design, three independent experiments were performed using MCF-7 and MDA- MB-231, while two independent experiments were done with SKBR-3.

For all the cell lines, the genes which had a Cq lower than 33 were analyzed for relative expression. Genes with a Cq higher than 33 were excluded since they correlate with low abundance proteins, and a challenge to detect by immunofluorescence. In more details, for MCF-7, MDA-MB-231 and SKBR-3 11, 8 and 10 genes were tested by RT-qPCR, respectively.

The results were analyzed using the ΔΔCt method and were plotted with mean and standard deviation in GraphPad Prism. A two-way ANOVA having as variable time and culturing condition was done to estimate the impact of these two parameters and their interaction [33].

The data analysis revealed that only few of the analyzed genes had statistically significant differences between 2D and 3D-Biosilk models. Culturing condition was found to modulate gene expression in a similar fashion over time, with transcripts increasing (e.g. CD44, MMP14) or decreasing (e.g. ZO-1, CLDN3) consistently at day one and day seven. Only for two genes, being JAM2 and DSC2, growing cells in 3D-Bioilk had an effect only at the seventh day.

(24)

22 The regulated transcripts varied among cell lines with the exception of CD44, which was found upregulated in both SKBR-3 and MCF-7 kept in 3D (fig. 8 and 10). MCF-7 had also an increase in JAM2 and a decrease in CLDN3 and ZO-1 caused by culturing in 3D (fig. 8). In MDA-MB-231 the culturing format led to higher levels of MMP14, ITGAV and HIF1α in 3D samples compared to 2D samples (fig. 9). In SKBR-3, only the previously mentioned CD44 and DSC2 were upregulated as a result of culturing cells in 3D (fig. 10).

Figure 8: Genes whose expression is affected by culturing condition in the cell line MCF-7. Relative mRNA levels of the indicated genes are displayed following qPCR analysis in MCF-7 cultured in 2D or 3D-Biosilk for one or seven days. Values are represented as fold-change of the mean ± SD (n=6), with each data point also depicted. Fold- change was calculated comparing to the control MCF-7 2D at day 1. *P < 0,05; **P < 0,01; ***P < 0,001;

****P < 0,0001.

(25)

23

Figure 9: Genes whose expression is affected by culturing condition in the cell line MDA-MB-231. Relative mRNA levels of the indicated genes are displayed following qPCR analysis in MDA-MB-231 cultured in 2D or 3D-Biosilk for one or seven days. Values are represented as fold-change of the mean ± SD (n=6), with each data point also depicted. Fold-change was calculated comparing to the control MDA-MB-231 2D at day 1. *P < 0,05; **P < 0,01;

***P < 0,001; ****P < 0,0001.

Figure 10: Genes whose expression is affected by culturing condition in the cell line SKBR-3. Relative mRNA levels of the indicated genes are displayed following qPCR analysis in SKBR-3 cultured in 2D or 3D-Biosilk for one or seven days. Values are represented as fold-change of the mean ± SD (n=6), with each data point also

depicted. Fold-change was calculated comparing to the control SKBR-3 2D at day 1. *P < 0,05; **P < 0,01; ***P

< 0,001; ****P < 0,0001.

(26)

24 3.2.3 Validation by immunofluorescence of the identified decrease of ZO-1 expression To investigate if the results obtained from the RT-qPCR could be validated at the protein level, the gene with the most significant change due to culture conditions in the cell line MCF-7 (i.e.

ZO-1) was chosen for immunofluorescence staining in an attempt to visualize the downregulation of ZO-1 protein in 3D-Biosilk. MCF-7 cells were cultured either on glass coverslips or 3D-Biosilk for seven days, after which fixation and immunostaining with ERα, DAPI and ZO-1 were performed. In figure 10 are pictures from immunofluorescence staining with ERα, DAPI and ZO-1 from each channel respectively. Both ERα (orange) and DAPI (blue) stains the nuclei while ZO-1, a tight junction protein, stains the borders of the cell (green). In 2D all cells whose nuclei were stained by DAPI and ERα were also stained with ZO-1. The 3D- Biosilk model however revealed areas stained by both DAPI and ERα, but not by ZO-1.

Figure 11: Top row are pictures from a 2D model and bottom row are from 3D-Biosilk model. The ERα (orange), DAPI (blue) and ZO-1 (green). In the last picture all three channels are shown in one picture. Pictures were taken with 20X magnification, scale bar: 100 µm.

3.2.4 Application of the established 3D-Biosilk model to the novel Wood breast cancer The novel ‘Wood breast cancer’-cell line was generously donated to My Hedhammar’s group by Cellaria to perform some initial testing of gene expression differences when comparing 2D and 3D. A previously performed experiment to observe the growth of the novel cell line in 3D- Biosilk indicated that the cells grow well throughout the model, see figure 12. Phalloidin was used to stain the actin-filaments and DAPI stained the nuclei.

(27)

25

Figure 12: Wood breast cancer cells stained with phalloidin (green) and DAPI (blue) on 3D-Biosilk. Scale bar:

100 µm.

The initial test of gene expression with relevant biomarkers resulted in an overview of how much the genes were expressed (table 4). Biomarkers from previous experiments in the laboratory had generated several biomarkers which were of interest, hence a decision to test the most interesting markers were made.

GENE NAME Cq-VALUE

Wood cells MMP11 Matrix metallopeptidase 11 32,5

SNAI1 Snail Family Transcriptional Repressor 1 32,2

VIM Vimentin 26,1

EpCAM Epithelial Cell Adhesion Molecule 25,9

CDH2 N-cadherin 31,7

SOX2 Sex determining region Y-box 2 - OCT4 Octamer-binding transcription factor 4 30,5

NANOG Nanog homeobox 31,2

CDH1 Cadherin 1 24,1

CDH11 Cadherin 11 38,4

SNAI2 Snail family transcriptional repressor 2 24,7 BMP7 Bone morphogenetic protein 7 - HIF1A Hypoxia inducible factor 1 subunit alpha 23,7

CD24 CD24 molecule 23,6

JAM2 Junctional adhesion molecule 2 36,1

ZO-1 Zonula occludens 1 25,7

MMP14 Matrix metallopeptidase 14 25,1

CLDN1 Claudin 1 31,9

(28)

26

CD44 CD44 molecule (Indian blood group) 21,9 ITGAV Integrin subunit alpha V 26,7

Table 4: Values from initial testing of relevant biomarkers in the novel Wood breast cancer.

Results from RT-qPCR analyzed with the ΔΔCt method were visualized with GraphPad Prism showing the mean and standard deviation. Initial data indicates that some genes were upregulated when cultured in 3D-Biosilk (e.g. CHD1, CDH2, OCT4, VIM, HIF1α, MMP14, ZO-1) compared to 2D (fig. 13 and 14). The gene CD44 could be seen to become upregulated in 3D- Biosilk after seven days (fig. 13), which correlates with results obtained from cell line MCF-7 (fig. 8) and SKBR-3 (fig. 10).

(29)

27

Figure 13: Graphs from an initial test of gene expression in the novel ‘Wood breast cancer’-cell line to visualize relative mRNA levels of interesting genes obtained from qPCR analysis of cells cultured in 2D or 3D-Biosilk for one or seven days. Values are represented as fold-change of the mean ± SD (n=3), with each data point also depicted. Fold-change was calculated comparing to the control Wood cell 2D at day 1.

(30)

28

Figure 14: Graphs from an initial test of gene expression in the novel ‘Wood breast cancer’-cell line to visualize relative mRNA levels of interesting genes obtained from qPCR analysis of cells cultured in 2D or 3D-Biosilk for one or seven days. Values are represented as fold-change of the mean ± SD (n=3), with each data point also depicted. Fold-change was calculated comparing to the control Wood cell 2D at day 1.

4 Discussion

4.1 Reproducibility estimation of the 3D-Biosilk model

Ensuring the accurate formation of the 3D-Biosilk was the first key step, needing to determine if the constructs were a reliable source for samples preparation in the advanced stages of the project.

(31)

29 4.1.1 3D-Biosilk structure and area

The protocol to form 3D-Biosilk was quite straightforward to follow, however it required practice, especially in the foaming step. In the beginning of the project, being a beginner at pipetting, it was tricky to move the pipette tip in a circular motion while at the same time move in a vertical motion to create a width and height of the 3D model. Nevertheless, with practice it became easier. Another problem associated with the foaming step was to create the same size of the foams, several ended up being too small or falling apart but as mentioned above with practice it became easier and the foams became more reproducible. As air is introduced via pipetting, larger bubbles could occur beneath the foams and hence it was of importance when moving the foam from the 24-well plate to the 96-well plate with the metal spoon to try and place the foam with the air bubble facing upwards allowing the foams to

“bloom out” as the bubble burst and obtain a more open structure. If the air bubble were placed facing down the foam would instead obtain a smaller more folded structure.

Brightfield pictures of the 3D-Biosilk were taken with high magnification to look in details at the silk structure (fig. 2). To take these pictures, medium had to be gently removed in order to have the 3D-Biosilk (at least partly) adhering to the well bottom and thus on focus. The media must be carefully removed since silk easily adheres to plastic, e.g. pipette tips, and hard to remove without breaking or deforming their structure. While this set of pictures allowed us to monitor changes in the 3D-Biosilk, this approach had two main limitations 1) the lowest available magnification was 4X, still too high to be able to see the full foam and 2) light transmitted through the specimen made it difficult to focus on more macroscopic parts of the foam. To overcome these issues and be able to image the full construct a stereo microscope was used, as it allows low powers of magnification (e.g. 1X) and it reflects light from the surface of an object. It was then possible to visualize the whole 3D-Biosilk, which also kept floating in the medium, ensuring nutrients to cells. To take pictures with this microscope the cap of the 96-well plate needed to be removed, thereby exposing the models to the air in the room and possibly affecting the sterility of the models. The area measurements performed with ImageJ indicated creation of foams which had similar sizes and hence could be viewed as reproducible. The slight difference in size might be due to the orientation and the handling of 3D-Biosilk models. It is important to notice that this set of experiments were done only using SKBR-3 cell line, since from previous tests done in the lab it is known that there are no significant macroscopic changes in silk structure due to the use of different cells.

(32)

30 4.1.2 Cell viability and growth onto 3D-Biosilk

After having confirmed the 3D-Biosilk structure reproducibility and correct appearance, we then focused on measuring cells viability and visualizing how they grow onto silk. For the growth curve only MCF-7 were used, since they were known to be the most reproducible among the available breast cancer cell lines in terms of proliferation pattern. Viability was inferred by measuring the cells metabolic rate with the Alamar blue assay. The principle of an Alamar blue assay is that a reagent called resazurin becomes reduced in response to cellular metabolic events, the reduced form has a highly fluorescent pink color and its intensity corresponds proportionally to the amount of living, healthy cells which are metabolically active [34]. After 3 days most of the air bubbles trapped in the silk scaffold were gone and the cells could grow better without the presence of air bubbles (not common in their natural environment) and reach a high cell density at day 7, indicating that the cells are alive and growing in both models (fig. 5).

Staining the models with phalloidin was done to see the spread of cells throughout the model, and since phalloidin staining is easy and fast to perform with a clear result this method was chosen. Growth could be seen in both models when staining with phalloidin. After seven days the 2D model was completely covered with cells growing flat to the bottom of the 96-well plate. Within the 3D-Biosilk model on the other hand, the cells seem to have regained their physiological form and spread throughout the silk scaffold.

To visualize and photograph the phalloidin staining of both models, an ANDOR camera, Leica DMI6000B was used. Photographing the 2D models were fast and easily done due to the cells growing in a monolayer. However, obtaining a good picture of the 3D-Biosilk model was hard and time consuming. Imaging 3D-Biosilk have some limitations, the 3D-Biosilk needs to be placed on a microscopy glass slide in order to be imaged, this was done with a metal spoon and the addition of PBS using a pipette. Ideally the model should be placed flat on the glass slide without folding itself for better pictures and as mentioned previously, silk sticks to plastic and care was needed to avoid the model sticking to the pipette tip. If taking too long time trying to take a picture, the PBS evaporates and more needs to be added to prevent the foams from drying out since it can alter the structure of the foam. Since it is in 3D it was hard to focus the microscope on the cells within the silk scaffold. An alternative to facilitate obtaining pictures a confocal microscope or sections of the 3D model, making it more like 2D models.

Another challenge with silk is that it contains aromatic amino acids which have strong

(33)

31 excitation and emission properties in a broad spectrum of wavelengths, this is known as autofluorescence [29]. To reduce the autofluorescence and make it easier to see fluorescence signals from secondary antibodies the 3D-Biosilk models were treated with Sudan Black.

Optimization tests for Sudan Black were performed to find the optimal staining process to avoid debris can ruin the image quality but still obtain a reduction in autofluorescence. First a combination with 15 minutes staining before permeabilization and another 15 minutes before DAPI was compared to only staining once for 15 minutes before DAPI. Not a big difference so the protocol with only one Sudan Black staining was kept. Also, phalloidin conjugated to different fluorophores were tested but the best results were for channel 488 in the green spectra. This is in line with what was previously reported by Neo 2015 [29].

4.2 Analyzing how culturing cells in 3D-Biosilk affects the gene expression

Even though breast cancer biology involves multiple cell processes, such as epigenetic regulation, escaping apoptosis and avoiding immune system, etc. We decided to focus on genes representative of cell adhesion, migration and epithelial mesenchymal transition (EMT) as we hypothesized, supported by previous literature [35], that culturing in 3D should affect the migration and adhesion of cells. Moreover, these biological processes are intertwined and play a central role in cancer progression. During EMT the epithelial cells loosen their adhesion to each other and acquire invasive properties, becoming capable to metastasize and resistant to cancer therapies [36].

4.2.1 Transcriptional changes driven by culturing in 3D-Biosilk

To test the effect of 3D-Biosilk on gene expression of cells kept in culture for one or seven days, 19 biomarkers relevant for breast cancer were selected. The statistical analysis, done to measure the effect of culturing condition (i.e. 2D and 3D), time (i.e. day 1 and day seven), and their interaction, revealed that two to four biomarkers were significantly modulated by 3D culturing in each cell line. In the ERα positive MCF-7 cell line, four biomarkers significantly affected by culturing condition could be identified: JAM2, CLDN3, CD44 and ZO-1.

Both JAM2 and CLDN3 are constituents of tight junctions between cells. JAM2 has been found to have a crucial role for the cellular migration and aggressiveness of tumors [37]. We could see that in MCF-7 cells cultivated in 3D-Biosilk, JAM2 was upregulated compared to the 2D model, indicating a more aggressive phenotype of the cells. Claudin 3 (CLDN3) is a

(34)

32 transmembrane protein which if mutated can facilitate malignant transformation of tumors [38]. The expression level of CLDN3 was downregulated in 3D-Biosilk.

ZO-1 was found downregulated and will be commented in more depth in the next subchapter.

CD44 is a transmembrane protein which, together with CD24, is used to identify cancer stem cells since it activates several pathways leading to cell proliferation, adhesion, migration and invasion [39]. Since previous studies associated high CD44/CD24 ratio with increased cancer malignancy [40], we decided to measure both markers. Our result showed that CD24 was unaffected whereas CD44 was upregulated in both MCF-7 and SKBR-3 cultured in 3D-Biosilk, thus leading to an overall higher CD44/CD24 ratio in 3D cultures. These observations are of particular interest and should be followed up as they suggest the acquisition of an aggressive phenotype when cells are cultured in 3D. Another biomarker with a significant increase due to 3D culturing found for SKBR-3 was DSC2. DSC2 is a desmosomal protein acting as tumor suppressor, it is critical for desmosome formation and inhibition of cell motility [41]. Based on previous literature, DSC2 downregulation correlates with high tumorigenicity so our observed increase of DSC2 may suggest a less aggressive SKBR-3 phenotype when cells are kept in 3D- Biosilk. This speculation is difficult to conciliate with the opposite conclusion of a more aggressive tumor due to CD44 increase. To clarify whether SKBR-3 are more or less aggressive when cultured in 3D the experiment should be repeated and also accompanied by quantification of changes at protein level of the identified biomarkers.

Biomarkers tested in MDA-MB-231 identified three genes considered interesting due to their expression changes during a 7-day cultivation, for instance MMP14, HIF1α and ITGAV. ITGAV has been reported to be overexpressed in breast cancer tissue and is associated with metastasis and tumor progressiveness [42]. Matrix metalloproteinase 14 (MMP14) does also promote metastasis [43] and was seen to be upregulated in 3D-Biosilk compared to 2D samples, however it is worth mentioning that MMP14 increases during time also in the 2D model, still remaining overall lower than in 3D. To investigate if the increase in MMP14 in 3D- cultured cells is biologically relevant and robust, it would be useful to perform longer trials as well as measure the protein levels.

Oxygen concentration in tumoral tissues has been found to be of importance when it comes to if tumors metastasis or not. MDA-MB-231 is highly proliferative and has a high demand for oxygen to fuel metabolic processes. The further away the cells are from a blood vessel (or the media in a 3D model), the higher likelihood of hypoxic conditions, in which an accumulation of HIF1α occurs. Previous studies have identified that higher levels of HIF1α can be associated

(35)

33 with poor patient prognosis since it promotes EMT by regulating EMT-associated transcription factors or repressors [44, 45]. Our data, as shown in figure 9, revealed that HIF1α is upregulated in 3D-Biosilk models after both one and seven days in culture compared to 2D, indicating a higher level of hypoxia in 3D models. This could be explained by the fact that cells in the 2D model have more access to oxygen present in the medium. However, even in 2D there was an increase of HIF1α over time, which is probably due to the limited space in a well- plate leading to low oxygen accessibility.

For drug testing and modeling of biological events and mechanisms, cell lines are used and viewed as a crude model since they cannot capture all tumoral events and the heterogeneity of tumors. Despite the number of established breast cancer cell lines, very few are frequently used for modeling and drug testing (e.g. MCF-7, MDA-MB-231) due to easiness of cultivation, etc. Selective cultivations in 2D models might render the cell lines less representative of breast cancer cells at all [11].

Initial analysis of a novel ‘Wood breast cancer’-cell line indicated few markers which might be more stably expressed when cultivated in 3D-Biosilk model compared to cultivation in a 2D model. Biomarkers SNAI2 and MMP14 kept their level of expression in 3D-Biosilk almost constant, while a downregulation occurred in the 2D model between day one and day seven.

Since only one independent experiment was performed, no conclusions can be drawn with statistical certainty, and hence further testing is needed to obtain more data.

4.2.2 ZO-1 is a biomarker downregulated in the MCF-7 3D-Biosilk model

ZO-1 is a scaffold protein which anchors tight junction transmembrane proteins to the actin filaments of the cytoskeleton. It has previously been shown that breast cancers which are highly invasive and have a decreased differentiation, also have downregulated levels of ZO-1 [46, 47]. Our initial investigation of gene expression changes when comparing cells cultured in 2D and 3D, indicated that ZO-1 is down-regulated in 3D-Biosilk models compared to 2D models (fig. 8). To investigate if the observed changes were also detectable at protein level, an immunofluorescence staining was performed. In a first pilot experiment, where only ZO-1 and DAPI were stained, all 2D models indicated a clear green ZO-1 stain while 3D-Biosilk samples fixed at day 7 had areas positive only for DAPI.

To verify the findings and ensure that the initial observation was not due to antibody issues, e.g. not able to reach a certain area in the 3D sample, the ZO-1 staining was complemented by a staining of the positive control ERα. It is important to mention that, for this second

(36)

34 experiment, the choice of secondary antibody required an optimization step to achieve the condition with the lowest autofluorescence from silk and the brightest signal for the selected primary antibodies. During the optimization several combinations of secondary antibodies were tested, leading us to select goat anti-rabbit 546-labeled for ERα mixed with goat anti- mouse 488-labeled for ZO-1. The staining confirmed the initial results, showing only on 3D- Biosilk regions with cells stained for both ERα and DAPI but not ZO-1 (fig. 11). These results overall demonstrated a good correlation between the decrease at transcriptional level of ZO- 1 and its downregulation at protein level. Moreover, the data suggest the MCF-7 cells cultured in the 3D-Biosilk might be more invasive than cells from then same cell line cultivated in 2D models. In future, to avoid the difficulties to focus 3D specimens, it would be important to perform similar staining on sections taken from 3D-Biosilk or to take advantage of confocal microscopy.

Identification of changes in biomarkers relevant for breast cancer biology occurring in a novel 3D-Biosilk model

Identification of changes in

biomarkers relevant for breast

cancer biology occurring in a

novel 3D-Biosilk model

EMMY STÅHL

Identification of changes in

biomarkers relevant for breast

cancer biology occurring in a

novel 3D-Biosilk model

Emmy Ståhl

Supervisor: My Hedhammar

Co-supervisor: Caterina Collodet

Abstract

Sammanfattning

Table of Contents

1 Introduction

2 Materials & Methods

3 Results

4 Discussion